Nature Vol.

289 22 January 1981

211

22 January 1981

Towards what shining city, which hill?

The American presidency is an impossible job which has an
enlivening effect on those who d o it. New encumbents embark on
tasks widely held to be beyond the resources of one man with the
zeal and enthusiasm of an army. Both Presidents Kennedy and
Johnson, when new to office, offered the promise of a fresh start
on old problems, domestic and international. Given the persistent
importance of the United States in the affairs of most other
communities, the promise of a change of direction echoes around
the world. Some of the consequences are unsettling; changes o f
direction are disconcerting. Some are also enlivening, as for the
new incumbent himself. It is too soon to know what will be the
consequences of Mr Reagan's accession to the presidency, but a
single inauguration speech is not a sufficient guide to what will
follow in the months ahead. The safest assumption is that Mr
Reagan, like all his predecessors, will be good at some things and
less good, or even frankly bad, at others. There is no shame in
that, given that the job is everywhere acknowledged to be
impossible. After a spell in which the president seemed to be bad
- but not for want of trying to be good - at most things, it is no
wonder that Mr Reagan's coming has cheered even his political
opponents. For a time, at least, people will be prepared to give
President Reagan the benefit of the doubt. The vacillations and
exaggerations of the election campaign will be forgotten. T o that
rule, there will be only one exception. In his acceptance speech in
Detroit last summer, Mr Ronald Reagan promised that he would
lead the American people to a "shining city on a hill". That
phrase, redolent of Blake or Bunyan, is too arresting to forget.
President Reagan, no doubt less innocent than he seems, will
not be starved of advice, solicited and otherwise. Politics being
what it is, his presidency may well be made or broken in the next
few months by almost accidental issues - the inflation rate, o r the
problems of Central America. In these connections, n o new
president would choose to start from where President Reagan
must. One of the perils of the job is that he has no choice. Another
is that the issues, although recently made clear, are by no means
irrelevant to what may happen on the wider stage. If inflation
continues as it has been in the past few years, American
institutions will be weakened as surely as they have been
undermined elsewhere, then the foundations of the shining city
will be found not to exist. Already many of the new occupants of
offices in the White House must be regretting that there had to be
a campaign, and a promise of a tax-cut to go with it. If the
problems of democracy in Latin America are ignored or, worse,
supposed to be familiar problems susceptible to oldfashioned
solutions, the roof could fall in. Even President Reagan's friends
will be well aware of the seriousness of these dangers. Those who
heard his acceptance speech may be forgiven for looking further
ahead.
The Reagan position, so often defined in the past several
months that it has become unclear, is ambiguous on two crucial
issues - arms control and academics. In the parochial spirit in

which trade magazines such as Nature grind their own axes on

occasions such as the inauguration of a new president, special
pleading may be excusable. O n the face of things, for example,
there is no reason why the new Adminstration should give an
instant's thought to the special interest of the scientific
community in the management of the arms race. Scientists, the
argument might go, have (among other things) a special
competence in military research and development. But only
taxpayers have a right to say what should be done about
relationships between the superpowers. President Reagan has in
any case come a long way since the beginning of the election
campaign, when he was bent on seeming a 1950s hawk. Now (the
argument continues), he is for renegotiation of Salt I1 and other
accommodations of that kind. The argument misses a point that
will quickly become apparent, that the special but not exclusive
interest of the scientific community in arms control is not merely
academic. There is a sense in which J. Robert Oppenheimer was
right in saying that "the physicists have known guilt" - hawks
among them have persuaded doveish graduate students to work
on unwelcome projects since time immemorial. The point has
been reached at which both hawks and doves share with the
general public the belief that it would be best if the consequences
of the products of their joint endeavours could be contained
within bounds, and that more or less any bounds would be better
than none. This does not imply that all scientists share with the
Bulletin of the Atomic Scientists the view that international
tensions have worsened in the past few weeks - the Bulletin
advanced its notorious minute hand one minute last week (and
was mentioned on the British Broadcasting Corporation's
morning news programme as a consequence) but did not
convincingly explain why it had done so. Yet there is a great
majority in the scientific community that would wish something
sensible done about arms control, and which knows that such a
course is not unattainable. The political question for the new
president is whether he can see this as an opportunity.
Another issue for President Reagan, not too long to be left on
the back burner, is that of the research community itself. Mr
Carter (or, more probably, Dr Frank Press) has left the incumbent
President with an awkward choice. Should an expansionary
science budget be cut in the interests of financial prudence, or left
as it is so as not to give offence? Neither of these questions should
bother President Reagan in these early heady days. There is a
simpler question he must answer. Sincc his early days as Governor
of California, he has cut the figure of an adversary of the
universities. At one stage, he forced Mr Clark Kerr out of his post
as president of the University of California. Since then, he has
made clear his disdain for academic preoccupations - decision is
what matters. Such a position can be defended but is not
defensible. The President needs urgently to build a bridge to the
academics. T o fail to d o so will be to alienate some of those who
might build a city, or find a hill.

Responsibility for trust in research

The public complaint, fashionable in the 1970s, that scientists
could not be trusted when making statements about the hazards
of pollution or of genetic manipulation, is now mercifully
abating. Is the complaint now about to be replaced with the
suspicion that scientists cannot trust each other? This will besome

people's first reaction to the article by Harris et al. on page 228,

and our Washington Correspondent's commentary on it (page
227). The reaction is understandable, even forgivable. For what
has emerged is that three out of four laboratory cultures of cells
originally described in 1977 as derived from the spleens of patients

Nature Vol. 289 22 January 1981

212

with Hodgkin's disease have nothing to d o with human

malignancy but are, instead, almost certainly derived from owlmonkey kidney tissue. This improbable conclusion h8s been
established by means of as neat a piece of detective work as can be
found in the recent literature. The fact remains that the four cell
cultures have been the basis, in the past four years, for aspattering
of spurious data in the scientific literature. Some of these
unwarranted conclusions have been used by others in unrelated
work towards the understanding of Hodgkin's disease.
Elsewhere, laboratories have tried (usually without success) to
establish their own analogous cell cultures from patients with the
disease. But Dr John Long's apparent success with his supposedly
authentic cultures appears to have led to the award of at least one
substantial research grant (for $500,000 over three years),
afterwards rescinded. Especially because the principal
investigator had been accused of falsifying data in a different
context, many will now be tempted to suppose that four years of
work with Hodgkin's disease has been a kind of hoax.
T o jump to that conclusion would be wrong. Those who work
with cells in culture are continually aware of the risks of
contamination. Especially in laboratories specializing in this
work, one established cell culture may be contaminated with cells
from another by a variety of means - contaminated glassware,
people's hands or defects of laboratory plumbing. Almost by
definition, the most esoteric kinds of cells, requiring the most
elaborate mixture of nutrients to make them grow, are those
which are the most susceptible to contamination. HeLa cells, the
malignant cells derived from a single tumour half a century ago,
and widely used as models of malignant cells, are frequent
contaminants. Inevitably, when a cell culture that grows only with
difficulty is contaminated by cells that grow more vigorously, like
HeLa cells, it is a matter only of time, perhaps days, before the
original cells disappear, swamped by their more strongly growing
competitors. Naturally, people with cells at risk are constantly on
the look-out for contamination. This is obviously easiest when the
cells at risk can be easily recognized, perhaps because of some
distinctive biochemical characteristic. Again, however, and
almost by definition, novel types of cells are likely to be least easily
recognizable. In other words, it is entirely possible that people
may work for weeks or months with cells from a tissue culture
without knowing that the originals have been supplanted. It is less
easy to see how meaningful results can be wrung from such a
culture, but nothing is impossible. The fact that Harris etal. have
been able to say no more about the fourth of the cell lines than that
it is of human origin emphasizes the difficulties of identification.
There is therefore no reason to doubt Dr Long's assertion (see
page 227) that he believed the cell lines to be authentic; but his
response to the doubts raised within and from outside his
laboratory seems to have been too casual.
The significance of the paper by Harris et al., for all its interest,
should therefore not be exaggerated. Almost certainly, this is not
the first occasion on which research reports based on themistaken
identity of tissue culture cells have appeared in the scientific
literature. Most of the articles concerned have probably
disappeared from people's citation lists as their data have been
recognized as spurious or simply puzzling. It would be foolish to
ask that the scientific enterprise should be carried on in such a way
that such unwanted and unrecognized noise is entirely eliminated.
Would intending authors be required to submit with their
manuscripts samples of their tissue culture cells for validation by
the referees? Would those working with supposedly pure and
authentic strains of laboratory mice be expected to follow suit?
And how would the scientific journals cope with the problems
that would follow? The only reasonable conclusion is that the
ideal is probably unattainable, that the scientific literature is
based to a very large extent on trust in what its authors say, that
the system works surprisingly well and that its occasional failures
can be accommodated without too much difficulty. If it follows
that the scientific literature is not to be regarded as a collection of
Mosaic tablets, nobody will be the loser.
The problems created for laboratories by such developments as
those identified by Harris et al. are more immediate and in some

ways more difficult. The necessarily close working relationship

between people sharing the same facilities and even the same
objectives in research is one of the most powerful safeguards of
the authenticity of the scientific literature. However great may be
the temptation for individuals to make too much of preliminary
findings or to apply cosmetic treatments to otherwise untidy data,
the knowledge that their most candid critics are likely to be their
closest colleagues is constantly an influence towards sobriety. It is
true, of course, that individuals who have misled themselves can
usually succeed in misleading more junior colleagues as well, but
even this is an unsure calculation now that many graduate
students are zealous custodians of what is right and proper. The
more certain safeguard, however, is that people in laboratories
where all researchers are in the habit of talking freely with their
immediate colleagues about their work, not merely at formal
colloquia but over lunch, are imperceptibly persuaded to couch
their interpretations of their work in moderate language. (The socalled sociologists of science, always on the look-out for subjects
to study, could profitably pay some attention to the sociology of
the laboratory as a formative influence on the pattern of science.)
No doubt one of the reasons why the most successful laboratories
are frequently large laboratories derives from the influence of
these local invisible colleges.
Why, then, does this system of polite and imperceptible
persuasion by colleagues not always function as it should? In the
past few months, the grant-making system has frequently been
blamed for departures from the path of strict propriety. The
competition for grants, the argument goes, is now so fierce, and
the consequences of failing to secure a grant can be so serious for
the career of an individual, that people are tempted beyond
endurance. This danger is more apparent than real. If the
competition for grants is now more fierce, so is the scrutiny of the
peer-review committees on whose recommendations most grants
are awarded. A more likely explanation is that the grant-making
system has disturbed the balance there would otherwise be among
colleagues working within the same laboratory. A successful
grant-holder tends, when grants are hard to come by, to acquire a
status and a degree of autonomy that may not be entirely justified
by his colleagues' estimate of his work. And people who are seen
to enjoy the goodwill of the grant-making agencies are, if they so
choose, to some extent immune from the necessarily searching
scrutiny of their immediate colleagues. The Massachusetts
General Hospital, one of the best known components of the
Harvard medical complex, moved quickly after Dr Long's
resignation last year to promise an audit of his work, and the
paper by Harris et al. is one result. The hospital has probably for
the time being done everything it can. But, in the long run, there
and elsewhere, the objective should be to make sure that
laboratory communities remain as rigorous as they should be in
their internal assessment of all their members.
The responsibility lies fairly and squarely with heads of
laboratories and heads of university departments. It is an
unpalatable responsibility. But unless it is exercised diligently, the
result will be unpalatable for the scientific community as a whole.
One obvious danger is that of the emergence within the scientific
profession of groups of zealous and often over-zealous people
dedicated to rooting out what they consider to be dishonesty.
Already there are signs of such developments. Some of the
allegations of plagiarism in the past year or so appear to have had
their roots as much in self-righteousness as in a concern for
honesty in the literature. Some of the consequences of these
allegations have been unfair to those accused. Yet there is no
reason to suppose that the few cases of dishonesty that have come
to light are in any sense the tip of an iceberg. On the contrary, as
the day to day affairs of reputable scientific journals bear out,
there is a general and quite remarkable concern for truth. Wouldbe authors are forever going back to their benches to check small
points raised by referees, who in turn appear almost
masochistically prepared to recommend the publication of data
or arguments conflicting with their own positions. It would be
tragic if these civilized habits were to be corrupted by the activities
of self-appointed vigilantes.

O 198lNature Publishing Group

Nature Vol. 289 22 January 1981

213

More pressures on Rothschild system

Geology project
threatened by
more cuts
The Rothschild principle, whereby
British g o v e r n m e n t d e p a r t m e n t s
("customers") commission research from
research councils ("contractors"), already
corrupted in medical and agricultural
research, is now showing signs of strain
elsewhere. The latest development is that
two of the customers of the Natural
Environment Research Council, the
Department of the Environment and the
Department of Industry, are threatening to
cut the amount of geological research they
are prepared to commission.
At stake are the Geological Survey and
the mineral reconnaissance programme,
both run by the council's Institute of
Geological Sciences. The Department of
the Environment, which currently
contributes 1.5 million to the 4.5 million
annual budget of the Geological Survey, is
cutting its contribution to geological
generally from 3.5 million in 1980-81 to
2.6 million in 1981-82. Although the
department has not yet specified its
priorities, it is expected that the cuts will be
made in the Geological Survey and the
engineering geology and bulk minerals
resources programmes. The Department of
Industry, which is the sole supporter of the
mineral reconnaissance programme, is
threatening to cut that budget from about
1.2 million in 1980-81 to 0.8 million in
1981-82.

Both departments say that they no

longer want to fund research which is not
directly relevant to their work. Thus the
Department of the Environment will only
fund those parts of the Geological Survey
related to planning permission. Alternative
funds for the survey are unlikely to be
forthcoming and it will have to make do
with less. Dr G. M. Brown, director of the
Institute of Geological Sciences, finds his
planning hampered by the Department of
the Environment's delay in deciding its new
priorities.
The Department of Industry is hoping
t h a t t h e mineral r e c o n n a i s s a n c e
programme will find alternative funds
from industry. Since 1972, this unit has
been searching for metalliferous ores in
Britain, especially in the Scottish
Highlands, North Wales and Devon and
Cornwall. Although deposits of several
minerals - in particular copper and
tungsten - have been found, only one, a
barites deposit in Scotland, has led to an
application for mining rights by private
industry. Nevertheless, industry is
interested. Companies consider that the
0028-0836/81/04021302$01.00

results provide useful information for

assessing mining profitability in the light of
future market trends. But hitherto, there
have been no direct commissions from
industry.
The Natural Environment Research
Council has been trying to establish
whether it could act a contractor for private
industry. Meetings with the mining houses
have thrown up several problems such as
that of confidentiality (reports have
previously been publicly available) and the
assignment of mining rights. Industry itself
is keen that the programme should

continue, but considers that financial

support for it is properly the duty of central
government.
Time is now running out. The new
financial year begins in April, and twenty
of the mineral reconnaissance staff have
already moved to oil surveys of the North
Sea, paid for by the Department of Energy.
The Department of Industry has not finally
decided to cut its support, but is thought by
several observers to be in a mood to say that
if private industry does not produce some
cash, it will a sign that the programme has
little practical relevance. Judy Redfearn

Carter's last budget asks for more

Washington
Determined to enter the history books on
an optimistic note, the outgoing Carter
Administration has proposed to Congress
a budget for the next fiscal year that
contains a 4.3 per cent real growth in
support for basic research.
"This is the best budget for the past four
years for science and engineering
research", said Dr Frank Press, the
President's ScienceAdvisor and director of
the Office of Science and Technology,
commenting on the budget proposals last
week. He added that it confirmed
President Carter's commitment to the
support of science and technology "as an
investment in the future".
In addition to the intended growth above
the expected level of inflation, Dr Press
singled out several initiatives that were
being p r o p o s e d by t h e o u t g o i n g
Administration "to solve problems that we
have known about for a number of years".
One of these is the proposed inclusion in
the budget of the National Science
Foundation of a $75 million fund to
improve university research equipment and
laboratories. Their rapidly deteriorating
state was highlighted in a recent report
from the Association of American
universities.
Two other proposals involve traineeships and other awards to avoid potential
manpower shortages in fields such as
computer science and energy engineering,
and efforts to meet the present difficulties
of engineering schools where, according to
Dr Press, "faculty and equipment are not
on a par with what one would expect in
industry".
The big question, of course, is how much
of this increase will escape the rapidly
sharpening budget knife of the incoming
Administration of President Ronald
Reagan. For many social welfare
programmes, the writing is already on the
wall, but for science and technology the
signals are mixed.
On the one hand, Mr Reagan's budget
director, Mr David Stockman, has been
talking about the need to rewrite the Carter
budget "from top to bottom". Mr

Stockman has previously placed the

N a t i o n a l A e r o n a u t i c s a n d Space
Administration (NASA) among his "low
priority" agencies which might absorb
significant cuts. In the same vein, a group
known as the National Tax Limitation
Committee put out a report last week
suggesting, for example, reduced spending
on the Galileo mission to Jupiter -already
being delayed a year by NASA because of
delays with the launch vehicles - on its list
of "expenditure control opportunities".
The group is also raising questions about
the appropriateness of government
support for the space shuttle, and
suggesting a re-examination of the future
cost-benefit ratio.
At the same time, however, members of
the various transition teams which have
been established by t h e Reagan
Administration to look at research and
development programmes have been
making optimistic noises, insisting that the
new president is committed to the support
of
"science,
technology and
productivity".
Part of the increase in defence spending
supported by both the Carter and the
Reagan Administrations, for example, is
likely to have beneficial spin-offs for
research support, particularly in areas
where the large aerospace companies have
astake. "I have beenlookingin teacups for
the past two months, and each time I look I
see a different picture", said outgoing
NASA director Dr Robert Frosch last
week.
The details of the Carter budget
proposals contain several proposed new
scientific starts. The NASA budget, for
example, contains funding for the
development of the Venus Orbiting
Imaging Radar - announced by President
Carter shortly before the election - as well
as a new Geological Applications Program
(GAP) which will use remote sensing
satellites to study geological resources
which might contribute to the discovery of
new oil and gas deposits.
Reflecting the delays in the space shuttle
programme, the proposed commitment to
start work on a fifth shuttle orbiter, the
@I981 Macmillan JournalsLld

Nature Vol. 289 22 January 1981

214
additional costs of science projects such as
the space telescope, and the new projects
described above, the total requested
budget for NASA comes to $6,700 million.
This would be a 20 per cent increase over
the budget for the fiscal year 1981 which
began last October, and if it is allowed to
stand, would be the largest annual increase
in the agency's budget since the early 1960s.
The National Science Foundation has
also put in for a hefty 23.5 per cent
increase, from $1,096 million in the current
year to $1,353.5 million next year. Most of
this reflects the Carter Administration's
keenness to support both research and
training in engineering fields. The new
engineering directorate will receive a 20 per
cent increase in its research budgets.
At the National Institutes of Health
(NIH), the proposed increase for
biomedical research is less spectacular. On
the basis that whatever the president asks
for is traditionally increased by Congress,
Mr Carter is suggesting that the NIH
budget for basic research be raised by 9.4
per cent. Allowing for inflation, this would
result in a drop of 1.1 per cent between 1981
and 1982.
At the Department of Energy, increased
support for research into synthetic fuels
and nuclear power - particularly fusion
energy - has resulted in a requested
increase of 9.4 per cent in real terms for
basic research, second only to that of
NASA.
Many of these figures will remain only as
indications of the "good intentions" with
which the Carter Administration is leaving
office. Perhaps of more lasting significance
are the figures prepared by Dr Press to
demonstrate the main trends of federal
support for science during Mr Carter's four
years in the White House.

These reveal, for example, that overall

the biggest winner as far as support for
basic research is concerned has been the
Department of Defense. If the proposed
1982budget figures are taken into account,
the Pentagon's basic research efforts will
have grown by almost a quarter - 22.6 per
cent - between 1978 and 1982.
Next come N I H . Congressional
enthusiasm has raised the NIH research
budget by 13.3 per cent over the four years,
compared with a growth of 12.8 per cent at
the Department of Energy.
The National Science Foundation, even
if it is granted this year's large increase, will
still only have seen its basic research grow
by 9.2 per cent. And at NASA, reflecting
the pressures which the space shuttle has
imposed on the space science programmes,
the basic research budget actually fell, in
real terms, by 0.6 per cent over the same
four years.
Overall, the growth in basic research,
including the 1982 proposals, would come
to 10.8 per cent for the period of the Carter
Administration. In current dollars, the
budget would grow by 58.2 per cent. from
$3,704 million to $5,801 million.
Ironically, the research and development
budget shows an identical increase of 58.2
per cent from $26,388 million to $41,734
million.
"Anyone who says that we do not
engage in long-term planning is proved
wrong by these figures", quipped Dr Press
- expected soon to be elected president of
the National Academy of Sciences although he added that the agreement was
actually fortuitous and that "the figures
just happened to fall our this way". In
practice, he will not be required to explain
why it should be otherwise.
David Dickson

Promises for President Reagan to deny

Specific proposals included in the budget
are as follows.
.Major difficulties with the development
of a vehicle to launch the two Galileo
spacecraft on their journey to Jupiter from
the space shuttle have caused NASA to
propose delaying the Galileo project for
one year and switching to a new launch
vehicle, a converted Centaur rocket.
.The Carter Administration proposes that
the National Institutes of Health should
aim to stabilize support for both
competitive research grants and research
traineeships. Last year, the Administration
promised to provide enough money to keep
the number of new and renewing
competitive research grants constant at
about 5,000. Given general fiscal
constraints, however, this meant cutting
back severely o n the number of
traineeships, a move which brought strong
protests from various sectors of NIH, and
was subsequently overturned by Congress.
.Defense
Department support for
research on US campuses seems destined to

continue to grow faster than support from

any other federal agency. According to the
budget request for the fiscal year 1982,
military funding for research and
development at US universities and
colleges will be 21 per cent greater than in
1981, totalling $639 million.
The proposed figure, most of which will
be spent on unclassified basic research
projects, is part of a 16 per cent increase in
all military-sponsored basic research
expenditure.
.One major new start proposed by the
National Science Foundation (NSF) is the
detailed design and initial construction of a
25-metre millimetre-wavelength radiotelescope which is planned for installation
at Mauna Kea in Hawaii.
Plans for the new telescope have
generated widespread support in the
radioastronomy community. NSF has
asked Congress for funds as part of a 29 per
cent increase in the NSF budget for
astronomical sciences, rising from $58.5
million to $75.6 million.

Argentinian power

Soviets help
Argentina has bought five tonnes of
heavy water from the Soviet Union under
International Atomic Energy Agency
safeguards, the Argentinian Comision
Nacional de Energia Atomica announced
last week. It is intended for "topping up"
the Atucha-1 nuclear station, which needs
on the average an annual heavy-water
replacement of 1.5 tonnes.
The sale is part of the growth of SovietArgentinian trade since January 1980,
when Argentina refused to back President
Carter's embargo of grain sales to the
Soviet Union. The Soviet Union is now
Argentina's main market for agricultural
products. In July of last year, Argentina's
Secretary of Commerce, Alejandro
Estrado, signed an agreement to supply the
Soviet Union with 20 million tonnes of feed
grain and soya beans during the next five
years. There are also persistent rumours
that a major agreement to export meat to
the Soviet Union is now being negotiated.
In return, the Soviet side has shown
considerable interest in Argentina's
nuclear programme. Argentina has been
conspicuous among third world countries
since the early 1950s for its nuclear
programme aimed at ultimate autonomy in
both research and technology. The
commission has announced that the target
will be for practical purposes attained by
the end of 1981, when the Cordoba
uranium processing plant will begin
producing an estimated annual production
of 150 tonnes. Rafael Coppa, the director
of the plant, said last year that Argentina
will then have full control of the primary
uranium cycle, from prospecting for
.The Carter Administration is proposing
t o cut s u p p o r t f o r research a n d
development in the Environmental
Protection Agency (EPA) from $364
million to $345 million. At the same time,
the outgoing Administration wants to
earmark an extra $28 million to improve its
review of the environmental impact of
proposed major energy projects in the west
of the country -particularly in connection
with the synthetic fuels programme - and
to launch a government-wide research
programme on the effects of acid rain.
.The Carter Administration wants to give
a major boost to research into magnetic
fusion. The 1982 budget proposals include
an increase of 28 per cent in fusion
research, to a total of $520 million in
budget obligations. $32.8 million of this
would be spent on a new centre for
magnetic fusion energy, as proposed by the
Administration following a thorough
review of the magnetic fusion programme
last year. Research o n magnetic
confinement systems would increase from
$1 19 million to $151 million.
David Dickson

Nature Vol. 289 22 January 1981

215
-~

uranium to putting the fuel elements into

the reactor.
Nevertheless, the search for selfsufficiency does not rule out cooperative
projects. Indeed, on returning from the
World Energy Conference in Munich last
September, Admiral C a r l o s C a s t r o
Madero, the president of the Argentinian
commission, openly attacked the
"negative aspects" of restrictions of
nuclear technology transfer, adopted first
by the "London Club" of nuclear
suppliers and then (at the end of 1979) by a
wider circle of industrialized Western
states.
Soviet interest has grown in the past
year. Last April, during a visit to Buenos
Aires, the Soviet foreign trade viceminister Aleksandr Manzhelo suggested a
major nuclear cooperation between the
two countries, stating at the same time that
he thought that Soviet-Argentinian trade
could well double in the next few years. At
the end of July, Yurii Fokin, Secretary of
the Soviet Foreign Ministry, visited the
Vera Rich
Atucha-1 plant.

Large electron project

Swedish cloud
Sweden is having second thoughts about
participating in LEP, the 500-cm 500-GeV
electron-positron colliding machine
which, at a cost of 900million Swiss francs,
is planned to be the next major project of
the European centre for high energy
physics research, CERN. At a meeting of
the Swedish Natural Sciences Research
Council last month, 90 per cent of those
present expressed doubts about the
arrangements for funding the project.
Delegates from CERN's member states
are expected formally to approve the
building of LEP at the next meeting of
CERN council in June. The plan is to
finance the project out of CERN's annual
budget by reducing expenditure on other
programmes such as the intersecting
storage rings and the synchrocyclotron.
How quickly LEP can be built will depend
on how much of the budget - SwFr610
million this year - can be diverted to it
each year. What seems to be worrying
Sweden is that the CERN council, which
requires a two-thirds majority vote to
approve budgets, could demand that
Sweden pay more if the cost of LEP rises
above initial estimates.
One faction of the Swedish research
council says that LEP is simply too
expensive to be built now. Another would
agree to the project with some concessions
- either that Sweden .be made exempt
from budget increases approved by CERN
council, or that the CERN budget be
divided into LEP and other programmes,
giving Sweden the option of leaving LEP
while remaining a full member of CERN's
other activities. Under the present
arrangement, a decision not to participate
in LEP would effectively be a decision to

--

opt out of CERN altogether, An incidental

factor which seems to have added weight to
the arguments of LEP's Swedish opponents is the feeling that Swiss industry
has reaped unfair benefit from contracts
arising out of CERN's work.
Sweden's objections, which have come
rather late in the negotiations (most of
CERN's members have informally agreed
to the L E P proposal), could delay official
approval for the project. Although Sweden
is not alone in wanting some guarantee that
costs will not get out of hand, making it
exempt from cost increases approved by
CERN council is unlikely to be popular
with other members: neither is splitting
LEP from the rest of CERN's budget.
Many see incorporating L E P into the
annual budget as a way of controlling the
rate at which money is spent on it.
At present Sweden contributes slightly
less than 4.3 per cent of CERN's annual
budget. If it decided to leave LEP, the
other member states would have to decide
how to redistribute the costs amongst
themselves or whether to extend the time
taken to build the machine. The chairman
of the Natural Sciences Research Council,
Mr Mats Lemner, expects that the council
will shortly discuss the contribution with
the Minister of Education. Depending on
the outcome of this discussion, parliament
may have to make the final decision. If it
does, the Swedes would to be able to meet
CERN's June deadline for a decision on
Judy Redfearn
the contribution.

Artificial hormones

European register?
Brussels
The European Commission is making
heavy weather of its plan to ban the use of
certain artificial hormones in animal farming. A meeting of agricultural ministers
planned for last week was cancelled after
the death of Mr F.O. Gundelach, the
Danish agricultural commissioner. But the
signs were that the meeting would have
failed to reach an agreement.
The Commission decided last September
that something should be done about
hormones after t h e discovery that
diethylstilboestrol was still being used for
veal production in Italy. The hormone is
banned in the United States and also in
many European countries, but is so
effective at increasing weight-gain in calves
that, where its use is banned, black markets
such as that in Belgium spring up.
The agricultural ministers were to have
discussed two proposals elaborating on an
original proposal made last December. The
first calls for a register to keep track of all
hormones used as medicinal products,
whether for human or animal use, from
manufacture and storage to distribution
and final use. Veterinarians would be
required to control all administration of
hormones to animals. The proposed rules
say that banned hormones can be used only
for "therapeutic treatment" o f

pathological cases diagnosed by

veterinarians, and not for chronic use in
preventive treatments. T h e second
proposal is that there should be a
comprehensive sampling system for testing
animals and carcasses
The proposals as drafted involve the
"positive listing" of those hormones
considered safe for use in meat production.
This is opposed by Belgium and the United
Kingdom. There is no dispute over banning
compounds such as diethylstilboestrol, but
it is held that apositive list would inhibit the
development of new materials.
A system of "negative listing" would
overcome some of these problems, but this
would have to be continually revised as
alternative hormones came on the market.
Given the political need for action, the
Commission favours the more cautious
positive listing system. Whichever route is
followed, the cost to European farmers of
making the necessary adjustments will be
Jasper Becker
substantial.

Soviet chemical industry

Effective economy
The Soviet chemical industry is rapidly
acquiring prestige status in the Soviet
media and ranks, according to a Pravda
article last week, together with nuclear
energy, space research and electronics, as
one of the hallmarks of twentieth century
progress. The proximate source of this
accolade is not hard to identify: almost
every article cites, at some point, Mr
Brezhnev's dictum that "there can be no
effective economy today without amodern
large-scale chemical industry".
When Mr Brezhnev made this pronouncement at last October's plenum of
the Central Committee of the CPSU,
however, he was not so much commending
the industry as calling for a programme of
"resolute measures" for overcoming
major shortfalls in chemical production,
ranging from chemical fertilizers and plant
protection agents to synthetic fibres, dyes
and household detergents. The 33 per cent
production increase specified in the guidelines for the new Five Year Plan for the
chemical and petrochemical industries
(recently placed under separate ministries)
is, say the planners, essential if theshortfall
is to be eliminated.
The chemical industry does not shoulder
full responsibility for thegap. At theend of
December, a Pravda editorial shifted at
least part of the blame to other sectors.
Fertilizer plants, said Pravda, were held up
by insufficient supplies of natural gas and
"inaccurate" planning by the light
metallurgy sector. Plants with processes
requiring high temperatures and pressures
often cannot obtain corrosion-proof
equipment. In some cases new factories
have been built, without the necessary
equipment being forthcoming, while, on
other occasions, expensive installations
have been purchased before it has been

216
finally decided just what the plant is to
manufacture. The state supply board
"Gossnab", Pravda concludes, must
supervise the whole supply process more
closely to eliminate such discrepancies.
F a r m o r e serious, however, a r e
suggestions raised last week by a group of
scientists from Byelorussia that the
implementation o f radical new
technologies is being blocked by interdepartmental wrangles. In a major Pravda
article they describe two such cases, in
sectors which all planners consider to be of
the highest priority - energy and
agriculture. The first project, highly
thought of by the Institute of High
Temperatures of the Soviet Academy of
Sciences, would have combined an electrolysis unit with a nuclear power plant so that
hydrogen for fuel or industry could be
produced during "off-peak" times. This
combination of chemistry, electrochemistry and nuclear power engineering
ran into the stumbling block of what the
group describe as the "excessively rigid
specialization o f t h e b r a n c h e s o f
industry".
The second project, for the production
of protein biomass for animal feed using
hydrogen-consuming b a c t e r i a , was
worked out jointly by teams from the
Byelorussian and Moldavian academies as
long ago as 1976. T o be cost effective,
however, the byproducts of the synthesis
- oxygen, nitrogen and possibly ammonia
- would also have been exploited. All
these are prime requisites of the mineral
fertilizer industry, which itself produces
waste gas with a usable hydrogen content.
Yet although plans have been drawn up for
a pilot biomass plant to be run in
conjunction with the "Azot" fertilizer
plant in Grodno, neither the Ministry of
Chemical Fertilizer Production n o r
GIavmikrobioprom - the body in charge
of microbiological production - can see
Vera Rich
its way t o go ahead.

European research

Counting costs
Brussels
The knotty problem of putting a value
on the European Community's research
and development programmes is the
subject of new proposals which the
European Commission has just put
forward. The objective of this soulsearching is to define criteria and methods
f o r both evaluating a n d exploiting
research. The idea stems not from a fit of
self-doubt but from the Council of
"Research" Ministers in December 1979.
There are no hard and fast ideas in the
document, but areas of investigation are
instead defined. Four pilot projects have
b e e n a p p r o v e d , w h i c h will g i v e
independent experts free rein to put certain
programmes under scrutiny. T h e
commission proposes then to use the
experience of the experts to formulate a
0018-0836/81/040216.01S01.00

Nature Vol. 289 22 January 1981

policy for exploiting and evaluating
research results. This is to be ready by the
end of the year but, before then, the first
outlines of the policy will be presented at
the International Congress for Research
Evaluation in October.
One pilot project has already been
completed by six independent experts, who
looked into the Community's energy
conservation and solar energy activities.
Four other expert groups are to investigate
programmes of research and development
in fusion, radioactive waste management,
the "Reference Bureau Programme"
concerned with physical standards and
critical data, and geothermal energy, the
use of hydrogen as a fuel and work on the
systems analysis of energy use.
T h e tightening of t h e E u r o p e a n
Community's budgetary belt is one motive
for the new initiative, but another is the
fear that the results of research are being
ineffectively disseminated. The experts will
be concentrating o n the diffusion of
information and in particular how to make
the results available t o the layman,
requiring that more outlets should be
found and language barriers overcome.
Indeed, one member of the European
Parliament has asked in a written question
why the Commission does not publish the
results of research in English as well as in
the author's mother tongue. But the
commission says that unless the results of
research are first digested, readership is
always restricted to specialist circles.
The stickiest part of the process on which
the Community has embarked is the
evaluation of research whose impact is
likely to be long term or even unforeseen.
Simplifying the results may help. Thus the
report on solar energy and conservation
says that the success of the programme
cannot be measured in "tonnes of oil
saved" and that the use made of research
carried out may require further action and
will take time. Even so, the chairman of the
evaluation team, Ugo Forinelli, is
confident that European research in this
area is already competitive with that in the
United States even though the European
budget is much smaller than the American.
Forinelli's team has made a number of
recommendations about the management
of the Community's programme, including
the need for more explicit statements of
objectives when putting contracts out to
tender and some suggestions of areas where
research might be concentrated or even
discontinued.
The commission seems modestly aware
of the inherent limitations of plans to
e v a l u a t e research. T h e underlying
objective is to strengthen the link between
research and development and industrial
innovation, a n d t h e commission
acknowledges that its own research cannot
be an important source of commercial
inventions. But restrictions on Community
expenditure outside the agricultural field
are likely increasingly to stimulate the
Jasper Becker
justifying of research.
O 198lNature Publishing Group

Swedish research

Lucky science
Stockholm
Research is one of the very few areas to
be given more money in the budget bill for
the fiscal year 1981-82, presented this week
in Stockholm. Against the background of a
budget deficit amounting to about $15,000
million, cutbacks in social services and
mounting economic gloom, research
received a n extra $28 million.
Considering that this money is to be
divided between the projects of eight
ministries, individual increases will not be
very great, and in real terms may well be
eaten u p by inflation, which was 14per cent
in 1980. One staff member at the Natural
Sciences Research Council estimated that
the net results of the council's $2 million
increase will be an unchanged level of
activity. But this is better than a cutback,
the fate of nearly all other sectors.
The government sees research strengthening Sweden's industrial competitiveness
in the long term. The Minister of Industry,
Nils Aasling, wants to develop sophisticated chemical and mechanical products,
and is to make specific allocations for this
purpose in the spring. One sign of this
policy is the changing proportions of the
space budget being allocated between international and national programmes. In the
1980-81 fiscal year, the Swedish Space
Corporation's budget of about $44 million
was equally divided between international
and national activities, but in future the
accent will be on national programmes.
T h e corporation's budget has been
increased overall t o about $58 million - a
real increase in spite of inflation - about
half of this will be used to build the
country's first space satellite, Viking,
which is to be launched from the European
rocket Ariane in May 1984. It is hoped to
produce the Viking at half the cost of
comparable satellites from other European
countries, and the project is planned to give
Sweden a profitable national space
industry. Another project being planned is
a n experimental telecommunications
satellite.
The lion's share of the increase in
research expenditure - about $16 million
- is to be disbursed by the Education
Minister, Jan-Erik Wickstroem, who
wants to increase research capacity across
the board. He is proposing an increase of
about 12 per cent for mathematics and
natural science faculties, an average of
about 9 per cent for the four research
councils as well as 90 new P h D places to be
spread over all subjects, $2 million for
particularly expensive equipment, $2
million to subsidize lecturers who want to
take time off for research and $1million for
research libraries. Fifteen new chairs will
be set up, among them molecular genetics
(Karolinska Institute), medicine (Uppsala)
and astrophysics (Gothenburg).
Wendy Barnaby

Nature Vol. 289 22 January 1981

Graduate education

Too many laggards

The Science Research Council, the chief
source of public support for British
graduate students, is alarmed at the
lengthening time taken to complete P h D
courses. A preliminary survey of 25 higher
education institutions has shown that on
average only 60 per cent of those holding
SRC studentships complete their PhDs
within four years. According to Sir
Geoffrey Allen, chairman of the council, a
figure of 80 per cent would be respectable,
but 90 per cent would be the ideal.
The issue has come to the surface after an
investigation by a working party of the
Advisory Board for the Research Councils,
which has been looking into the broader
question of postgraduate education and
manpower needs. That in turn was
stimulated by the revelation of Sir Michael
Posner, chairman of the Social Science
Research Council, to the Public Accounts
Committee of the House of Commons last
summer that fewer than 30 per cent of his
council's graduate students complete their
PhDs within four years.
Given pause by that statistic, the Science
Research Council made a rapid survey of
five institutions where it supports students
and arrived at the figure of 60 per cent,
which a subsequent survey o f 25
institutions has upheld. The advisory
board's working party, which is also
concerned with the performance of
postgraduates supported by the Social
Science, Medical a n d A g r i c u l t u r a l
Research Councils, has commissioned a
more detailed study intended to throw
some light on why so many students take so
long to complete their theses, or even fail to
complete them at all.
Particular attention is likely to be paid to
the performance of the Science Research
Council's students, if only because there
are more of them than of the other councils
- on the average, 2,350 new science
studentships are awarded each year. Most
of these are in the gift of university
departments, to which studentships are
allocated on a quota basis. Studentships
are worth about 3,500 a year, and are
tenable for three years, the estimated time
for completing a research project.
The reasons for these delays are still
obscure. The Science Research Council
expects to find marked differences of
performance in different institutions and
subject areas. PhDs in pure science may
more often be completed than those in
applied science - applied scientists and
engineers are more likely to find jobs in
industry, where writing a thesis may seem
irrelevant and where there is little time for
writing up anyway. It is also suspected that
institutions and departments with a large
number of P h D students will have a better
track record than those with relatively few.
The issue also, however, raises questions
concerning the meaning and purpose of a

217
PhD, which the advisory board's working
party is looking into under its broader
remit. Should a P h D for example, be a
thorough and lengthy investigation of a
detailed scientific problem, or more simply
a means of training a student in the
techniques of research? The approach is
bound to have implications for the
completion time.
As yet, no reliable pecking order of
institutions has been established, but the
early surveys d o suggest t h a t t h e
Universities o f Birmingham, Cambridge,
East Angliaand Bristol, and King's College
in the University of London, have the best
completion records a n d that the
Universities of Newcastle upon Tyne,
Sussex and Bradford, together with
Imperial College, London, and most of the
polytechnics, have the worst.
The poor track record of Imperial
College, regarded as a highly prestigious
scientific and technological institution,
may seem surprising. Lord Flowers, rector
of the college, says that the explanation
may be that large numbers of its
postgraduates rapidly find employment in
industry, leaving them little time for
writing up.
The advisory board's working party has
yet to decide what should be done.
Sanctions against departments with poor
track records have been mentioned. Cutting quotas of studentships is an obvious
device. Sir Geoffrey Allen, however, hopes
to avoid such heavy-handed treatment.
Most academics, he says, are willing to
accept genuine criticism and put their
houses in order. The peer review system
should take care of that.
Judy Redfearn

British universities

More confusion
Confusion among British universities
about their financial prospects appears to
have been further deepened by the letter
from thechairman of the University Grants
Committee, Dr E. S. Parkes, circulated t o
vice-chancellors on 30 December. The
letter contained a warning that the
resources available for the 1981-82
academic year may be reduced by between
5% and 6 per cent compared with the
a m o u n t s advertised in t h e P u b l i c
Expenditure estimates a year ago. The
committee's latest estimate of the shortfall
next year is a n amalgam of a 3 % per cent
cut estimated to be the universities' share of
the 30 million cut for higher education
announced last November and the still
incalculable effect on university finances
of the partial disappearance of overseas
students, some of whom have been
frightened away by "economic" fees.
Some universities regard Dr Parkes's
warning as a signal for drastic belttightening. Last week, for example, Lord
Annan, Vice-Chancellor of the University
of London, told the university senate that

the total budget of 200 million might be

reduced by between 15 and 20 million in
1982. Other universities appear to be taking
a more phlegmatic line, believing that they
cannot know the worst until the individual
allocations of funds for 1981-82 are made
in April o r soon thereafter.
Between now and then, the committee
itself will have several difficult questions to
decide. One possibility raised in the letter
from Dr Parkes is that the committee may
keep back until later in the academic year a
proportion of the funds made available by
the Department of Education and Science,
using the reserve to make good deficiencies
that have by then appeared. There are
precedents for such a reserve, but no
decision has yet been made, nor have
criteria for deciding how to use the money
been defined.
The Parkes letter also promises explicit
guidance in the spring on the numbers of
home students at which British universities
should aim in the coming academic year,
and on their desired distribution among
different kinds of courses. Although such
"guidance" has accompanied previous
financial allocations to British universities,
some universities now apparently fear that
the committee intends t o b e more
"dirigiste" than in the past - while others
remark that Dr Parkes's letter is a good
deal less so than his speech to a closed
meeting of the Committee of ViceChancellors at the beginning of December.
It does however seem clear that the
University Grants Committee will take
steps to ensure that universities d o not earn
their way out of trouble by recruiting more
home students than at present f ~ the
r sake
of the extra fee income that would bring.
Planning for the year ahead has been
further confused by reports that Mr Mark
Carlisle, Secretary of State for Education
and Science, told a conference in the north
of England on 6 January that he saw no
reason why the cuts now proposed should
be matched by "a reduced provision".
Universities, on the other hand, say that
there will have to be a reduction of student
entry this October if they are to live within
their straitened budgets.
The future constitution of the University
Grants Committee itself also appears to be
in question. Taking its cue from a
recommendation of the House of Commons Select Committee on Education
towards the end of last year that the cornmittee should cultivate more independence, the department has raised the
possibility that the committee's staff (at
present seconded from the Civil Service)
should become direct employees. Some of
those concerned wryly reflect that this is
not quite the independence that the House
of Commons committee had in mind, but
they acknowledge that such a step would
reduce the size of the Civil Service and
transfer part of its cost away from central
government. The fine print in the proposal,
which is said to be "very small", is being
read carefully.

CORRESPONDENCE
Badgers guilty
SIR- In his letter published on 11 December
1980 (page 532), Stephen Harris, among
many other misleading and biased remarks,
stated that the rate of decline in the incidence
of tuberculosis (TB) in badgers in the South
West was "paralleled" by a decline in the
incidence of reactors in cattle herds not only in
the South West but also in the rest of England.
This statement was based on a graph which
was attached to a copy of the letter sent to
Lord Zuckerman and which is now reproduced
(Fig.1). It was deduced from this that there
might well be some common cause
unconnected with the gassing campaign that
was reducing the incidence of TB both in cattle
and badgers throughout the country.
This "parallelism", however, was only
secured by expressing all plotted values as percentages of the corresponding 1974 values. In
Fig.2 I have plotted the breakdowns for the
South West and for the rest of England on the
same scale, subdividing the South West into
(a) Gloucester, Avon and Wiltshire; (b)
Cornwall; and (c) Devon, Dorset and
Somerset. The radical differences between the
counties that have a high incidence of herd
breakdowns, (a) and (b), and the rest of
England (d) is immediately apparent.
Table 10 of the report (Badgers, Cattle and
Tuberculosis, Lord Zuckerman; HMSO,
London, 1980) shows that the sources of
infection are indeed different in the South
West and in the rest of England. In groups (a)

and (c) three-quarters of the infections are

attributed to badgers, but only 1 per cent to
imported Irish cattle, whereas in the rest of
Great Britain one half are attributed to
imported Irish cattle and none to badgers.
In Cornwall (b), as Harris observes, only 15
per cent were definitely attributed to badgers,
but as almost three-quarters of the causes of
Cornish breakdown were recorded as
"unknown" and the remainder were either
from purchased cattle or contiguous premises,
it may be presumed that infection from
badgers was much greater than this.

Fig.2 The percentages of herd breakdown are

shown by full lines (from Table 19 of the
report). The percentages of badgers found to be
infected are shown by the broken line (Table
15). The dotted line gives some supplementary
information from Table 18 (updated).

Cattle in SW

Fig.1 Annual changes in the incidence of TB in

badgers sampled at the Gloucester Veterinary
Investigation Centre (page 62 of the report) and
herds of cattle in South-West England and
herds of cattle in the rest of England (page 64).
The first year for which data were available in
the report from all three samples (1974) was
taken as 100 per cent; for subsequent years, the
level of TB was expressed as a percentage of the
incidence in 1974.

The fall in the incidence of herd breakdowns

following the introduction of the gassing
campaign in Gloucester/Avon agrees well with
the reduction in the incidence of TB in badgers
(broken line), bearing in mind that in addition
to the reduction in the percentage incidence in
badgers the numbers of badgers in close
contact with herds must have been
substantially reduced by the gassing.
The dotted line is derived from the latter
part of Table 18, updated to September 1980,
and indicates the rise in TB incidence in
badgers following the cessation of gassing.
(Updated figures for herd breakdowns in
Gloucester/Avon and Wiltshire are not yet
available).
The only support that Fig.2 gives to the
suggestion that there was a common cause,
other than appropriate action by ministry
officials, affecting the decline in TB operating
throughout the country is the substantial rise
between 1974 and 1975. This cannot be related
to infection from badgers. It may possibly
have been due to climatic factors or to some
improvement in the tuberculin test.
Figure 1 is a sad illustration of the way in
which distorted and partial graphical
presentation can be used for propaganda
purposes.
FRANK
YATES
Rothamsted Experimental Station,
Harpenden, UK

Rothschild retreat

SIR- It is not clear from your leading article in

the January 1/8 issue (page 2) whether you regret
the government's "retreat from Rothschild" or
just the manner in which it is happening.
One of the difficulties for the Medical
Research Council (MRC) at the time of the
Rothschild report was the uncertainty as to the
failures in the then current system which it was
intended to correct. The Ministry of Health
did not identify any omissions on the part of
the MRC in their contribution to improving
the health of the nation, nor did they
subsequently propose contracts in fields of
work outside the previous commitment of the
MRC. Rothschild, in a thoughtless comment,
referred to the very low expenditure on
research into ageing, one of the fundamental
aspects of all living things, but the
understanding of which is nowhere in sight.
He failed to ask the practical question as to
how much of the MRC budget was spent on
investigating diseases most common in old
age. The answer would have been a major
part.
There are, at present, three developments in
medical research on which great hopes of
practical gain are based: interferon, antibody
production by hybridomas and DNA
technology. How far the expectations will be
justified remains to be seen, but the heavy
investments of pharmaceutical companies
throughout the world in these fields is evidence
that the hopes are held widely. MRC
laboratories were responsible for the first two
discoveries and made a very big contribution
to the third. None came from planned research
but from "Dr Gowans' doctrine of investment
in good ideas and good people".
It has been argued strongly that much more
attention shouldbe given to public health and
~reventivemedicine and that the MRC's
expenditure has the wrong emphasis. But one
of the major advances in this field, the
correlation of the incidence of lung cancer
with cigarette smoking, came from work in an
MRC unit.
No doubt governmental administration of
civil science could be improved substantially,
but the next enquiry which you appear to be
advocating must surely start by identifying the
omissions and failures of the present system
whether in medical, agricultural or any other
field of research before recommending
changes to overcome them. Your generalized
denigration of a system which has achieved
obvious successes is not enough.
R.R. PORTER
Department of Biochemistry,
University of Oxford, Oxford, UK

Creating problems
SIR- Your readers should know of the court
action by creationists against the State of
California, now expected to begin on 2 March.
The suit, which is sponsored by the Creation
Research Society of San Diego, is in the names
of Segraves et al., students in California
schools, including children of Mrs Nell
Segraves of the society. The suit charges that
the State of California has violated the
Continued on page 335

Nature Vol. 289 22 January 1981

219

Deep-ocean hydrothermal
vent communities
from J.T. Enright, W.A. Newman, R.R. Hessler & J.A. McGowan

THEtheory of plate tectonics has led to a

completely unexpected bonanza of
research problems for marine biologists.
The theory predicted major heat input to
the ocean in areas where the sea floor is
spreading, and in the search for those
sources, geophysicists discovered not only
deep-ocean hydrothermal vents, where
warm water emerges from subterranean
sources, but also extraordinary concentrations of large and unusual animals
surrounding the vents's2. The bottom of
the deep sea typically has a sparse fauna
dominated by tiny worms and crustaceans,
with an even sparser distribution of macrofauna, including clams, crabs a n d
echinoderms. Near the hydrothermal
vents, however, are bizarre ensembles of
abundant large animals - communities
which resemble nothing ever before encountered by marine biologists. The most
conspicuous and unusual animals around
the warm-water vents (- 20C) are huge
gutless tube-worms (Vestimentifera), some
of which are over 3m long and up to 10 cm
in circumference; but there are also
remarkable densities of huge clams (25cm
length is typical), blind crabs, fish and
polychaetes (see figure). More than 100
species have been collected2, many of
which are elsewhere unknown.
How do the animals in these deep-sea
communities find enough to eat? Elsewhere, the deep benthic fauna relies on
particulate matter, ultimately derived from
photosynthesis, falling from above. The
food supplies necessary to sustain the vent
communities, however, must be many
times that ordinary fallout. In the first
report which described the vent fauna,
Lonsdalel proposed two possible sources
of nutrition: advection of food materials
from surrounding regions and bacterial
chemosynthesis. Shortly thereafter three
kinds of evidence were presented3 which
support the idea of intense local
chernosynthesis: (1) hydrogen sulphide, as
a potential basis for fixation of CO,, was
found in vent water; (2) some of the
bacteria isolated from a vent site were
found to be capable of chemosynthesis on a
thiosulphate medium; and (3) bacterial
W28-0836/8 1/040219035OI .00

concentrations of 108 to 109 cells per ml

were found in samples thought to be pure
vent water. This final observation seems
decisive: if such astonishing concentrations
of bacteria - difficult to achieve even
under optimal laboratory conditions
(D.M. Karl, personal communication) are indeed typical of steady-state vent
outflow, then food from within the vent
would be dwarfed by any contribution
from advection. Hence, the widely quoted
conclusion was reached, that bacterial
chemosynthesis provides the foundation
for hydrothermal-vent food chain^^-^ -an
exciting prospect because nowhere else on
Earth is a major community of diverse
organisms known which does not depend,
at least indirectly, on photosynthesis.
There are, however, certain difficulties
with this interpretation. Some of the large
sedentary organisms associated with vents
a r e found a t ordinary deep-sea
temperatures many metres from the
nearest hydrothermal sources6, a n d
therefore would not directly encounter any
bacteria from within the vent. It is conceivable, however, t h a t bacterial
aggregates, which have grown
subterraneanly, in the rising plume of vent
water might rain back out in the peripheral
areas t o nourish these filter-feeders.
Another difficulty is that similarly dense
populations of large animals have been
found2,' in the proximity of 'smokers' hot-water vents where the dominant,
conspicuous flow emerges at temperatures
up to 350C. No bacteria can be expected
to multiply or even survive such heat, and
none was found in the hot vent water7.
Unless the smokers are consistently
associated with smaller but more
hospitable warm-water vents2 - a
distribution that has not actually been
observed (F.N. Speiss, personal
communication) - the chemosynthesis
hypothesis could at best account for only a
fraction of the vent faunas.
The food resource question has been
further complicated by a recent and more
extensive investigation5 of t h e
microorganisms associated with one of the
warm-water vents in the Galapagos Rift

region. Bacterial concentrations were

measured in near-vent water which were
approximately three orders of magnitude
lower than those initially reported
(5 x I@-l@ cells per ml). These values
were confirmed by replicate estimates of
biomass based on ATP concentrations in
the water5 (100-250 pg of microbial carbon
per litre, compared with the previous
estimate3 of 100-1,000 mg of bacteria per
litre). Although the authors of that report
still endorsed the chemosynthesis
hypothesisS, the available data suggest that
Lonsdale's first alternative1 deserves more
attention than it has received. Could
advection conceivably supply the near-vent
fauna with enough food organic matter
ultimately derived from photosynthesis?
Because of its temperature, water from a
vent is very buoyant, densities as low as
0.62 gm cm3 having been reported for a
'black s m ~ k e r ' The
~ . rising plumes of vent
water are, however, rapidly diluted. Spiess
et aL7 report that a vent discharging at
23OC into water at 1.8OC produced a
temperature only 0.3OC warmer than
ambient at a height of 10 m above the vent.
These values indicate that within 10 m of
the bottom, the advected diluting water
was about 70-fold the discharge volume
[(AT,-AT,)/AT,] . Stability considerations
dictate that such advective flow into the
vent area would originate near the watersubstratum interface. where sus~ended
organic matter accum;lates and epibenthic
animals are concentrated. Typical values
for suspended particulate matter in deepsea waters within a few metres of the
b o t t ~ m are
~ . of
~ ~the order of 50-500 mg
m-3, of which several per cent is organic
matter. Taking a conservative estimate of 5
mg of particulate organic carbon per cubic
metre and the calculated dilution factor of
70, we find that for every cubic metre of
vent discharge, 350 mg of particulate
organic carbon would be advected into the
vent area. If discharge volume were to be,
say, 1m3 s l , advection could provide more
than 30 kg per day of potential food. In
addition, there is a good chance that small
live animals in the advected water might be
killed or stunned by thermal and/or
01981 Macmillan Journals Ltd

220

Nature Vol. 289

22 January 1981

Examples of hydrothermal-vent fauna from Galhpagos Rift area. Left, vestimentiferan worms, about 3 cm in diameter.
Middle, dense group of encrusting serpulid polychaetes, about 5 cm long, some with tentacular crown extended. Right,
vesicomyid (white) and mytilid (grayish) bivalves adjacent to an aggregation of anemones. The tip of the probe is 1 cm
in diameter.
chemical shock, and subsequently rain
down, thereby contributing to the vent
food supply. Hence, it appears that an
input of non-photosynthetic food might
not, at least in principle, be required by the
vent animals.
Such i n p u t calculations, however,
ignore what may be the most important
aspect of advective processes, that the
near-vent area should accumulate and concentrate most advected organic matter.
Entrained food particles which are not consumed in their initial passage through the
vent area would be wafted upwards, but
ought thereafter (given a density greater
than ambient seawater) to rain out predominantly on the near-vent substratum.
Once on the bottom, this food material
would not be lost to the vent ecosystem
because of resuspension. Sediment-trap
d a t a H indicate t h a t t h e particulates
suspended in near-bottom water of the
deep sea must, on the average, be resuspended every few days. Because of such resuspension, the initially advected material
would be subject to repetitive re-entrainment, thereby enriching the advected
waters. Hence, filter-feeding animals near
a vent should experience not only increased
flow rate of waters bearing organic
nutrients, as suggested by Lonsdale', but
also experience unusually high concenI . Lonsdale. P. Deep-sea Res. 24, 857 (1977).
2. Ballard, R.D. & Grassle, J.F. Naln. Geogr. 156.689(1979).
3. Corliss, J . n . el a/. Science203, 1073 (1979).
4. Galapagos Biology Expedition Participants Oceanus 22
110.2. 2 (1979).
5. Karl, D.M., Wirsen, C . O . & Jannash, H.W. Science207,
1345 (1980).
6 . Killingley, J.S., Berger. W.H.. MacDonald. K.C. &
Newman. W.A. Nature 287. 218 (1980).
7. Spiess. F . N . el 01. Science 207, 1421 (1980).
8. MacDonald. K.C., Becker. K., Spiess. F . N . & Ballard, R.
Eurlh planet. Sci. Letl. (in the press).
9. E~ttreim. S.L & Ewing. M. in Studies m Physrcal
Oceanography Vo1.2 (ed. Gordon, A.L.) 123 (Gordon &
Breach, London. 1972).
10. Jacobs. M.B., Thorndike. E.M. & Ewing, M . Mar. Geol.
14. 117 (1973).
I I . Gardner, W.D. tliesis,Massachusetts Inst.Technol.(IY77).
12. Rau. G.H. & Hedgec. 1.1. Science203. 648 (1979).
13. Rau. G.H. Nature(in the press).
14. Williams. P.M.. Smith. K.L., Druffel, E.M. & Linick,
T.W. Nafure (in preparation).
15. Williams. P.M., Stenhouse, M . C . . Druffel. E.M. & Koide,
M . Nature 276. 698 (1978).

trations of suspended particulates from

non-vent sources, food which is ultimately
derived from sea-surface photosynthesis.
The nutrition of filter-feeding animals is
even more dependent o n food concent r a t i o n s t h a n o n t h e flow r a t e o f
surrounding waters. At any new vent site,
even if chemosynthesis were t o b e
negligible, this sort of accumulation should
progress until catabolic degradation, by
both micro- and macroorganisms,
balances the net rate of advective input.
Both Lonsdale's suggestion that food
materials w o u l d b e a d v e c t e d f r o m
extensive areas surrounding deep-water
hydrothermal vents, and the enhancement
of the local density of these food materials
due to gravitational fallout and recirculation, appear t o be reasonable, if not
logically inescapable, expectations. The as
yet unresolved issue is the quantitative one
- what proportion of the food supply for
macroorganisms is provided by such
advective input and what proportion by
vent-related chemosynthesis? T h e
suggestion that advective input alone might
provide an adequate nutritional basis for
near-vent food chains implies that animals
relying on advected biomass ought to d o
well around any intense persistent deepwater source of heat - an idea which may
be worthy of direct experimental test.
By similar reasoning, the near-vent area
should also represent a zone of
accumulation for any organic materials
which might be produced by chemosynthesis, and therefore it may prove to be
extremely difficult to distinguish between a
photosynthetic and a predominantlv chemosvnthetic basis for the
near-vent food chains by measuring the
local density of organic matter and
associated bacteria, or by examining the
spatial gradients in those densities. Even
samples thought to be pure vent water3 are
not easy to interpret; the possibility of
dilution or contamination by materials of
non-vent origin will remain troublesome,
because of the intense mixing around the
vent5.
O 198lNature Publishing Group

S o m e of these uncertainties a n d
paradoxes may s o o n be resolved by
carbon-isotope data. The stable isotope
ratio, I3C t o 12C, depends on the nature of
an organism's carbon sources, and is also a
sensitive indicator of alternative metabolic
pathways involved in photosynthesis and
of heterotrophic fractionation. In the
tissues of some of the vent fauna, the
' 3 C P C ratio differs significantly from
values typical f o r o t h e r deep-sea
animal^'^-'^, suggesting that some sort of
unusual carbon sources o r metabolic
processes - perhaps chemosynthesis may be important in near-vent areas. The
finding that the I3C ratios differ between
vent bivalves and vestimentiferan worms in
the direction of this anomaly (H. Craig,
p e r s o n a l c o m m u n i c a t i o n ) is s o f a r
unexplained, the one group having smaller
and the other larger values than expected.
Also of great interest are data on the ratios
of I4C to 12C in the tissues of near-vent
animals, which ought to reflect their food
resources, a n d which indicate14 a n
apparent age of about 2,500 yr. Intra-vent
chemosynthesis would incorporate the
dissolved inorganic carbon of vent water,
80% o r more of which apparently comes
from magmatic sources (H. Craig,
personal communication). If one assumes
that magmatic CO, is very old and contains
negligible amounts of I4C, then in-vent
chemosynthesis should lead to I4C ages
(>15,000 yr) very different from those
obtained for the tissues of vent macrofauna (- 2,500 yr). So far, however, no
actual measurements of I4C in vent water
are available, and other interpretations
about its probable I4C age have been
proposedi4. The apparent age of 2,500 yr
of the circumvent fauna is clearly much
o l d e r t h a n c o n c u r r e n t sea-surface
photosynthetic production (expected age
-O), and much younger than the available
data for organic carbon incorporated into
the top few millimetres of deep-ocean
The authors are professors at the Scripps
Institution of Oceanography, La Jolla,
California.

Nature Vol. 289 22 January 1981

221

A new stingray from South Africa

from Alwyne Wheeler

ICHTHYOLOGISTS
are accustomed to the
regular description of previously unrecognized species of fishes, which if not
a daily event a t least happens s o
frequently as not to cause great comment.
Previously undescribed genera are likewise not infrequently published, but
higher categories are increasingly less
common. T h e discovery of a new
stingray, which is so different from all
known rays as to require both a new
f a m i l y a n d a new s u b o r d e r t o
accommodate its distinctive characters, is
therefore a remarkable event.
A recent paper by P.C. Heemstra and
M.M. Smith (Ichthyological Bulletin o j
the J. L. B. Smith Institute of Ichthyology
43, 1; 1980) describes this most striking
ray as Hexatrygon bickelli and discusses
its differences from other batoid fishes.
Surprisingly, this remarkable fish was not
the result of some organized deep-sea
fishing programme, but was found lying
on the beach at Port Elizabeth. It was
fresh but had suffered some loss of skin
by sand abrasion on the beach, and the
margins of its fins appeared desiccated in
places. The way it was discovered leaves a
tantalising question as to its normal
habitat, but Heemstra and Smith suggest
that it may live in moderately deep water
of 400-1,000m. This suggestion is
supported by its general appearance
(small eyes, thin black dorsal skin, flacid
snout) and the chemistry of its liver-oil.
T h e classification of Hexatrygon
presents some problems, t h e most
obvious of which is that it has six pairs of
gill openings, whereas all other batoid
fishes (sawfishes, rays, skates, stingrays
and electric rays) have only five paired gill
openings. Another striking feature is that
the spiracular openings behind the eyes
are closed dorsally by a cartilagesupported flap to form a n oblique slit. It
seems that Hexatrygon may be able to
shut its spiracles d o w n externally,
something that other batoids cannot do.
Internal features a r e as strikingly
different from its relatives, and the shape
and structure of the snout are also
unique. The snout is elongate, of uniform
thickness (16-18mm deep in the fish of
103cm total length) a n d flaccid.
Internally it is supported by lightly
calcified rostra1 cartilages and filled with
sediment (apparent agem8,OOO yr)ls . For
t h e advective c o n t r i b u t i o n t o vent
nutrition, however, t h e a p p r o p r i a t e
comparison would be with the I4Ccontent
of suspended particulate organic matter
ordinarily present in near-bottom waters,
which ought to be somewhere between
these extremes, but again, appropriate
measurements are not yet available.

Ventral view of Hexatrygon bickelli

c h i m a e r o i d s Rhinochimaera a n d
Harriota, and there can be little doubt
that it is an adaptation for finding food in
the soft-bottom ooze of deep water.
One final point that Heemstra and
Smith note is the remarkably small brain
of Hexatrygon (it is about 3% of the
c r a n i a l volume); in shallow-water
stingrays it is about 80%. As they point
o u t t h e c o e l a c a n t h , Latimeria
chalumnae, has a similarly small brain (as
does the basking shark, Cetorhinus
mauimus, according to Scott J. Fish Biol.
16, 665; 1980). This may also be an
adaptation to deep-water life.
If nothing else the discovery of
Hexatrygon off South Africa, as with the
first coelacanth, reminds one of the tag ex Africa semper aliquid novi.

an acellular jelly, while the underside is

richly supplied with well developed
ampullae of Lorenzini. These ampullae
are known to be electroreceptors and
assist in finding buried food organisms in
shallow-water rays. The combination of
characters persuade Heemstra and Smith
to
erect
a
new
suborder
(Hexatrygonoidei) within t h e order
Myliobatiformes to contain this fish. The
order contains the other stingrays, eagle
rays and mantas, most of which, like
Hexafrygon, have one or more serrated
dagger-like spines on the back of the tail.
The interest of this fish is not just that it
represents a novel group within a well
known order, but that all theother myliobatiformes are shallow-water bottomliving species o r surface-living fish.
Hexatrygon seems to be the first deepwater stingray known. Its elongate, jellyfilled snout richly supplied with electroreceptors is similar to the deep-water

Alwyne Wheeler is in the Department of

Zoology, British Museum (Natural History),
London.

In a d d i t i o n t o t h e u n c e r t a i n t i e s
associated with the two initial hypotheses
for vent nutrition considered here, a third
hypothesis has now entered the scene.
Recent evidence (G.N. Somero and H.
Felbeck, p e r s o n a l c o m m u n i c a t i o n )
indicates that significant bacterial chemosynthesis may be taking place outside the
w a r m - w a t e r vents, b y s y m b i o t i c

microorganisms within the tissues of some

of the macro-fauna. Hence, the basic
problem of how the hydrothermal-vent
fauna gets enough to eat remains an open
and challenging question.
W e are most grateful t o the many
colleagues who have shared their ideas with
us, as well as those who havegiven us access
U
to their unpublished data.

222

Nature Vol. 289 22 January 1981

Waldmann also presented evidence to suggest that the DNA sequences necessary for
IgA production are present in AT patients
and there was some speculation about the
possibility that the cutting and splicing of
compared with normal cells (Taylor, DNA required to produce the specific imBirmingham). This, together with the munoglobulin genes may be impaired. The
observation that caffeine can potentiate rec A gene of E. coli also controls the
DNA damage to visible chromosome cutting and splicing involved in genetic
damage a t least 24 hours after Go recombination. The relationship betwen
irradiation of A T cells (Natarajan, the immune defect and the occurrence of
Leiden), suggests that there are long-lived lymphoid cancers was considered. Unlike
unrepaired lesions. The exact nature of the the other immunodeficiency syndromes
DNA repair defect is still not clear and it is AT patients show a range of epithelial cell
possible that more than one pathway may tumours in addition to the preponderance
be affected. Paterson detected a reduction, of l y m p h o i d c a n c e r s ( S p e c t o r ,
in some AT cell lines,of y ray-induced Minneapolis). The pattern is also different
repair replication,and also a deficiency in from that seen in irradiated populations,
their capacity to excise and repair specific y where there is a preponderance of
ray-induced base damage, although there leukaemias rather than lymphomas. No
has been difficulty in reproducing some of conclusion was reached, but it seems likely
these results elsewhere. Inoue reported a that the cancers in AT are not related in a
deficiency in cell-free extracts of all AT cell simple way to either the radiosensitivity or
lines studied of an activity which modifies the disordered immune system, but rather
y ray-damaged 3' sites to permit the action to some combination of the two.
of DNA polymerase. Some, but not all,
The neurological defects in AT patients,
laboratories could confirm these results.
particularly progressive cerebellar ataxia,
The other abnormal radiation response could not at present be readily linked with
which was agreed by several groups is that the biochemical or immune defects. There
AT cells show a much less marked decrease was,
h o w e v e r , evidence
that
in DNA synthesis than do normal cells differentiation may be imperfect in many
following y-irradiation (Edwards, tissues and organs especiafiy gonads, liver.,
Birmingham; Lehmann, Sussex; Jaspers). the immune system and the central nervous
Similarly, Scott (Manchester) finds that system and this may in some way be related
AT cells do not show the normal post- to the repair defect. The uniformly raised
irradiation mitotic delay. It is not clear how AFP levels in these patients may represent
these defects relate to the repair defect.
some immaturity of the liver. The deficient
Unlike the other cancer-prone DNA thymus appears embryonic rather than
repair deficiency syndrome, xeroderma atrophic. Kidson (Brisbane) suggested that
pigmentosum, where there is a hypersen- failure of the nervous system to
sitivity of the cultured cells for mutagenesis differentiate properly might in timelead to
by UV light, there is no evidence for any progressive loss of neurones.
spontaneous o r radiation-induced
There was a great deal of interest in the
hypersensitivity for the induction of gene possibility of detecting individuals heteromutations in AT (Cox; Simons, Leiden). zygous for the AT gene. This stems at least
Furthermore, Arlett (Sussex) was unable to in part from the observation that heterodetect any radiation-induced gene zygotes had an increased risk for largely the
mutation in AT cells. One of the most inter- same spectrum of tumours as the
esting features of AT patients is the high homozygotes (Swift, North Carolina). If
level of spontaneous chromosome current estimates of the AT gene frequency
aberrations in circulating lymphocytes and are correct, heterozygotes would comprise
particularly the tendency of cytogenetically one per cent or so of the general population
abnormal clones t o proliferate in the a n d t h i s would h a v e p r o f o u n d
absence of any haematological abnor- implications. It could be, for example, that
mality. The formation of these clones AT gene carriers may account for a
following breakage at specific points on substantial proportion of early cancers of
chromosome 14 was emphasized and it was the lymphoid system. Suggested methods of
suggested (Hecht and McCaw, Tempe, detecting heterozygotes were low-dose
Arizona) that aberrations involving the chromosomal radiosensitivity (Kidson)
breakpoint at or near 14q32 may have and cell survival under hypoxic conditions
particular neoplastic potential.
(Paterson). The differences between
There is definite evidence for a profound obligate heterozygotes and normals were
immunological defect in AT patients. All small, but may be significant in some cases.
have a defect of cell-mediated immunity However, overlap between these two
and some have a deficiency of IgA and IgE. groups means that those tests would not be
Waldmann (Bethesda) showed that the IgA s u i t a b l e f o r t h e r e c o g n i t i o n of
deficiency is due to lack of synthesis rather heterozygotes in the general population.
than abnormal metabolic rates. There is
diminished helper T-cell function, but even Bryn Bridges is Director of the MRC Cell
Mutation Unit anda Professor at the University
stimuli which do not require a helper T of Sussex. David Harnden is Professor of
function fail to provoke an IgA response, C a n c e r S t u d i e s a t t h e U n i v e r s i t y o f
suggesting a defect of B-cell maturation. Birmingham.

Untangling ataxia-telangiectasia
from Bryn A. Bridges and David G. Harnden

THE rare autosomal recessive disorder,

ataxia-telangiectasia (AT), has attracted
attention because it is associated with an
unusual sensitivity to ionizing radiations
and an increased susceptibility to cancer,
particularly of the lymphoid tissues.
Patients also have a variety of other
problems, including a progressive
neurological disorder, a substantial
immunological deficiency, spontaneous
chromosome instability, a defect of DNA
repair, premature ageing, raised serum
alpha fetoprotein(AFP) levels and other
less regular features. How are all these
diverse manifestations of the genetic defect
interrelated? In an attempt to answer this
question and to resolve inconsistencies in
reported results, workers from a variety of
different fields came together at a recent
conference* at the University of Sussex.
Many laboratories have now confirmed
the cellular radiosensitivitv of AT cells to
ionizing radiation and there was agreement
that AT cells are also unusually sensitive to
the chromosomally radiomimetic antitumour agent, bleomycin. Not all agreed
that AT cells are sensitive to a range of
other chemical mutagens, suggesting that
increased sensitivity to any of these agents
cannot be very marked although some of
the disagreements may be d u e t o
differences between A T lines. Three
different groups (Paterson, Chalk River;
Inoue, Mishima; Jaspers, Rotterdam) all
reported complementation between
different AT lines for some radiosensitive
function and agreed that AT3BI could be
distinguished from AT4BI and ATSBI.
Attempts to elucidate the molecular basis
of the radiosensitivity failed to detect
differences between AT cells and normals
in the repair of single-strand or doublestrand breaks in the DNA using presently
available techniques. Two different types
of abnormal molecular response were
described, a failure to inhibit DNA
replication after y- or X-irradiation, and a
defect in one or more repair pathways. A
possibly analagous situation exists in
Escherichia coli where mutations at the rec
A gene block both inhibition of DNA
replication after UV exposure and several
pathways of DNA repair. The existence of
a repair defect was demonstrated in
different ways by several different groups.
Cox (Harwell) showed a failure to repair
potentially lethal damage in AT cells
following X-irradiation. Indirect evidence
of a repair defect comes from cytogenetic
studies following Go irradiation of AT
lymphocytes which reveal a large increase
in fragments and chromatid aberrations as
' I he u v r l \ h o p unc held from h o ~ e m b e 6th
r to 7th, 1980, and
u a s organlied jointly by the MRC Cell M u t a t ~ o nUnlt and the
Department of Cancer Studies, University of Birmingham.
Sponsorrhip was provided by the MRC, the Cancer Research
Campaign and the International Cancer Research Workshops
programme (ICREW) of theUICC.

Nature Vol. 289 22 January 1981

The emphasis of recent research on AT
has been on the sensitivity of these cells to
various agents and their effect on DNA
synthesis and repair and progress in these
areas has been substantial. While these are
undoubtedly important, more effort could

223
in future be directed towards relating these
areas to the defect of the immune system
and the neurological disorder. Such a
relationship, though surprising, must exist
since the evidence that AT is due to a single
0
autosomal recessive gene is good.

The spruce invasion

from Peter D. Moore

ONEof the greatest weaknesses of palaeoecology is its necessary reliance upon

circumstantial rather than experimental
evidence. Yet it is this weakness which also
opens the way to interminable speculation
on the basis of the temporal correlation of
events in the past. A very good example is
the debate concerning the expansion of
spruce (Picea abies) in Europe since the last
glaciation.
Pollen analysis shows that, following the
final glacial retreat, Picea immigrated into
the eastern Alps very rapidly, accompanied
by deciduous tree genera such as alder
(Ainus), hazel (Corylus), oak (Quercus)
and elm (Ulmus). It was preceded by birch
(Betula),pine (Pinus)and larch (Larix).In
Austria, for example, this deciduous forest
type expanded into the south-east around
10,200 BP (Fritz CarinthiaI 1 83,277; 1973)
and by 9,500 BP it had expanded westwards
in t h e Tyrol (Bortenschlager Ber.
naturw. -med. Ver. Znnsbruck 63, 105;
1976). Although Picea was a member of the
early post-glacial forests of Austria, even in
the upper valley of the Inn, it is interesting

to note that its immigration into Switzerland was considerably later; and it is here
that scope for speculation begins.
Markgraf (Nature 288, 249; 1970)
collated the Swiss data concerning the
movement of Picea into Switzerland and
she found that its expansion was not very
regular. Only in the south-east had it
arrived by about 6 . 0 0 0 ~but
~ . between that
date and about 4,500 BP Picea pollen
increased very markedly in a series of sites
scattered all over the country, extending
right to the western borders, north of
Zurich. There remained some sites,
however, even in the south of the country,
in which the expansion was delayed until
about 3 , 0 0 0 or
~ ~even later. The expansion
of spruce was often accompanied by a
decrease in other tree species, including fir
(Abies), pine and the deciduous forest
genera. Also there was a simultaneous
increase in herbaceous pollen, largely
grasses, which suggested that the forests
had been disturbed by human activity and
that this provided spruce with a n
opportunity t o establish itself where

New vistas in soil research

from Hamish Anderson

FRANCIS
BACON'S assertion, that "there is
nothing makes a man suspect much, more
than to know little", gives an adequate
summary of the present state of soil
organic chemistry. Soils produce a vast
array of organic and inorganic entities many formed only in the soil environment
- which resist synthetic mimicry, long
accepted as the ultimate basis of
structural proof. There remain few
separation techniques or analytical
methods which have not been used in humic
substance research, and the structural
chemistry of soil extractives has
uncertainly advanced, mainly by
degradative techniques (see Schnitzer and
Khan Soil Organic Matter Elsevier,
1978).
Proton magnetic resonance spectroscopy is an obvious tool but initial results
were beset with line-broadening problems
arising from chemical shift anisotropy,
even in the rare cases where samples were
sufficiently soluble for examination. The
advance in NMR capability has now led
to the threshold of the soil organic
chemist's dream, the possibility of
examining humic substances in situ
0028-0836/81/040223-02WI .00

(Barron et al. this issue of Nature, page

275). The Antipodean researchers have
shown that the use of cross-polarization
I3C NMR techniques has allowed the
generation of useful spectra from whole
soil samples. Conventional I3Cspectra of
amorphous solids usually give no useful
structural clues, mainly due to large
dipolar interactions. The I3C crosspolarization technique leads to increased
sensitivity by the elimination of
heteronuclear dipolar broadening, this
being achieved by transferring
polarization from IH to I3C (Mehring
High Resolution NMR Spectroscopy in
Solids Springer-Verlag, 1976). The
transfer from the more abundant and
more polarized IH spins leads to a
possible quadrupling of the I3Cspins and,
more importantly, the I3C spin-lattice
relaxation rate now occurs at that of the
proton population. Since this is usually
an order of magnitude faster than the
conventional I3C rate, the d a t a
accumulation frequency can be greatly
increased. Even allowing for this
'speeding-up' process, it is sobering to
realize that the number of free induction

previously it had been excluded by

competitive pressures. The combination of
these pieces of circumstantial evidence led
Markgraf to reject the earlier theories of a
climatic change which provided the right
conditions for Picea invasion into
Switzerland and to propose that the tree
followed in the wake of Neolithic agricultural progress.
One still needs to explain the delay in
e x p a n s i o n between A u s t r i a a n d
Switzerland. Here Markgraf suggested that
the early invasion of Abies, Lark and
Pinus cembra (the only central European
member of the Haploxylon subgenus of
Pinus and therefore distinguishable on the
basis of its pollen) may have formed a
dense canopy forest into which Rcea could
not immigrate. All three of these species
were present in Austria, but either arrived
later or were in lower frequency. It is
claimed that such a forest may have a
certain degree of inertia which renders it
resistant to invasion by new species and
also resilient in the face of climatic
fluctuations (Smith Proc. R. Soc. B161,331;
1965).
Spruce has also moved westwards in
Fennoscandia in the latter half of the postglacial, but here alternative explanations
have been proposed. Moe (Bot, Notiser
123, 61; 1970) collected the dates for the
rise in Picea pollen in this area and showed
that its expansion began in eastern Finland
by 5,000 BP, reached the Baltic coast of
Finland by 3,500 BP and subsequently
spread through Sweden and into Norway.
It has an unbroken distribution, reaching
the west coast of Norway only in the central
decays collected by Barron et al. during
whole soil spectrometry ranged from
2 x 10" to 2 x loS.One major defect of the
present system is that the chemical shift
anisotropies of aromatic and carboxylic
carbon lead to signal broadening and loss
of resolution. However Barron et al. are
confident that the combination of the
present high-field cross-polarization
technique with 'magic-angle' sample
spinning will resolve this problem, as has
been demonstrated with isolated humic
substances (Hatcher et al. Org. Geochem.
2 , 87; 1980).
Humic substances research has long
been dogged by the uncertainty of
structural alterations occurring during
extraction, and the proposed techniques
offer answers to this problem. Indeed,
the fact that modern NMR spectroscopy
offers an ever-widening scope for
examining a variety of nuclei instils hope
that whole soils and their physical
separates can be studied, resolving the
interactions between humic substances
and silicates, along with the involvement
of nutrients such as nitrogen, sulphur and
phosphorus.
Hamish Anderson is in the Department of Soil
Organic Chemistry, Mucaulay Institute for
Soil Research, Cruigiebuckler, Aberdeen.
Cj 1981 MacmillanJournals Ltd

224

Nature Vol. 289 22 January 1981

-

region. There are some similarities to the

Swiss data in that the expansion is nonsynchronous, but in this case there is a
more distinct pattern of progressive
westward movement.
Tallantire (Norw. J. Bot. 19, 1; 1972)
assessed the possible courses of the late
post-glacial movements of spruce in
F e n n o s c a n d i a a n d h e regarded t h e
expansion as a step-wise development with
phases of expansion, and plots them in this
form on his maps. He quotes in full the
dates upon which he bases this idea but, if
the dates are considered in conjunction with
their errors, it cannot be fully supported; it
is possible to interpret the spread of spruce
as a gradual process. Tallantire considers
the expansion a consequence of climatic
changes and bases this partly upon his
assumption of phased movements, but also
upon the inconclusive nature of the
palynological evidence for coincident
human activity. Further analysis of the
climatic requirements of spruce ( J .
Biogeogr. 4, 219; 1977) led Tallantire to
regard the critical parameters as ample
early-summer water supply and stable, low
winter temperatures. In the absence of
either of these factors, the young spruce
(2-3 years) is selected against. H e
concludes "the conformable arrangement
o f t h e d a t e s o f s p r u c e s p r e a d in
southwestern Finland and its relationship
to present winter climatic data is surely too
marked to be due to a chance distribution
of the investigated sites."
Since we are dealing with correlations,
however, it is worth pointing out that
correlation between pollen evidence for
human activity and the spread of the spruce
is becoming increasingly convincing as
more detailed d a t a accumulate,
particularly from Finland. In the Kuusamo
district of the north-east, Picea expanded
at the time when tree pollen was falling and
i n c r e a s i n g G r a m i n e a e , Calluna,
Vaccinium and Empetrum indicate ground
flora changes (Hicks Trans. Inst. Br.
Geogr. NSl, 362; 1976 and Commentat.
Biol. 8 0 , l ; 1975). Although Hicks does not
place such an interpretation upon the data,
it is possible to support the idea of human
agency in Picea spread.
Similarly in southern Finland, Tolonen
(Ann. Bot. Fenn. 15, 4; 1978 and 17, 15;
1980) produces pollen stratigraphic curves
in which the indications of human activity
at the times of Picea expansion are strong.
Many weed species, in addition to cereals,
show short-duration peaks at precisely the
time of the Picea rise at Lake Lammijarvi.
At Lake Ahvenainen, Picea pollen rises in a
series of episodes at each of which there are
also cultural indicators. It is also
noteworthy that considerable increases of
charcoal are found in the latter site
following the horizon at which Picea
begins its spread.
Tallantire has claimed that the ecological
Peter D. Moore is in the Department of Plant
Sciences, Universityof London, King's College.
00280836/81/040224-02$0I .00

conditions in Fennoscandia were

essentially different from those in central
and southern Europe in that Abies, Fagus
and Carpinus were not found in the prePicea woodlands, whch consisted largely
of Betula, Pinus and Sorbus. These forests
would therefore have offered little competitive resistance to Picea, it may be
supposed. The correlation between forest
disturbance and spruce invasion certainly
exists, however, and it may be that nutrient
release from litter caused by disturbance
and especially burning was influential in
allowing spruce to invade. In the current

management of Fennoscandian spruce

forest, for example, Viro (in Fire and
Ecosystems (eds T.T. Kozlowski & C.E.
Ahlgren), Academic Press, New York, p.7;
1974) recommends the use of fire for rejuvenating old stands.
We are thus faced with the task of
balancing alternative correlations, bearing
in mind that a strong probability exists that
both processes were operative. In the
absence of any possibility of experimental
confirmation of either hypothesis, the
ground is well prepared for lengthy debate
0
and speculation.

Inclusion compounds
from J.E. D. Davies
C O M P O U N D Swhich s h o w u n u s u a l
properties, such as a preference for
methanol rather than for ethanol, for the
para rather than for the ortho o r meta
isomers of C6H4CI2, for the axial rather
than for the equatorial conformer of
C,Hl,C1, for the trans rather than for the
gauche rotamer of I-chlorooctane, and
even a preference of C6D4(CD,), rather
than for C,H,(CH,),, were the subject of a
recent symposium* in Poland.
These interesting compounds are the
inclusion compounds, which are twocomponent solid-state systems where the
'guest molecule' is accommodated in voids
formed in the crystal structure of the 'host
lattice'. These host-guest complexes (for
which the term 'hogiplexes' was suggested
[J.E.D. Davies, L a n c a s t e r ] ) c a n b e
classified according to the type of void
available for the guest molecule. The guest
species is accommodated in interlayer
spaces in the intercalates; in the clathrate
compounds the guest species is completely
enclosed in a cavity; and the term inclusion compound is usually used where
the guest species is accommodated in any
other type of void, such as long channels
(see figure).
After a historical introduction by W.
Kemula (Warsaw) and a general survey of
thedifferent types of host-guest complexes
by J . E . D . Davies ( L a n c a s t e r ) , t h e
symposium heard about current work
using a variety of host lattices.
The greatest number of papers was
devoted to the Werner-type host lattices,
M(X-py), (NCS),, where M = Mn(II),
Co(II), Ni(II), Fe(I1) and X = 3-Me, 4-Me,
4-Et, whose selectivity towards isomeric
disubstituted benzenes has long been
known (Schaeffer et al. J. Am. Chem.
S o c . 7 9 , 5870; 1957) a n d r e c e n t
selectivity studies were reported by S.
Brzozowski (Warsaw), J.E.D. Davies
(Lancaster) a n d A. Lewartowska
(Warsaw). This selectivity can be put to
good use, and D. Sybilska (Warsaw)
'First International Symposium o n 'ClathrateCompounds and
Molecular Inclusion Phenomena' organized by the Institute o f
Physical Chemistry o f the Polish Academy of Sciencesand held
at Jachranka, Warsaw, 22-26 September 1980. The author
acknowledges receipt of a British Council Travel Grant.

showed how it is possible to separate

isomers chromatographically on columns
consisting of the Werner complexes. The
chromatographic use of a wider range of
inclusion compounds was reviewed by E.
Smolkova-Keulemansova (Prague).
Undoubtedly the most unexpected
selectivity shown by these host lattices is
towards isotopic molecules. N.O. Smith
(New York) described his careful work in
this area and was able to show that when
Ni(4-Mepy),(NCS), is equilibrated with a
mixture of the protonated and deuterated
guest, there is a definite selectivity towards
the deuterated species with benzene, xylene
and naphthalene.
T h e related compounds, Ni(NCS),(a-arylalkylamine),, also show remarkable
selectivities and P. de Radzitzky (Brussels)
described the work of the Labofina group
on these compounds and illustrated a plant
using these compounds to separate the
dimethylnaphthalenes on an industrial
scale.
Crystallography has been an important
technique in the study of inclusion compounds, particularly since it was t h e
pioneering crystallographic studies of
H.M. Powell which first elucidated the
nature of these compounds, and G.D.

Cross-sectional view of urea-n-paraffin

complex. (Fig. 11 from Smith Acta crystallogr.
5, 224; 1952. Reproduced by permission of
Acta crystallographica.)
Q 1981 Macmillan Journals Ltd

Nature Vol. 289 22 January 1981

225

The source of meteorites in time and space

from Robert Hutchison
VARIOUSfalse alarms have occurred in
the past over alleged discoveries of fossil
meteorites. For example, a squarish piece
of iron weighing 7858 from Tertiary
lignite mined at Wolfsegg, Austria, was
originally reported as meteoritic (C.r.
hebd. Seanc. Acad. Sci., Paris 103, 702;
1886; N a t u r e 35, 36; 1887). More
recently, in 1930, a fist-sized piece of
iron-nickel metal was said to have been
recovered from a borehole at a depth o f
1,525 feet in Eocene sediments; this
'Zapata County', Texas, iron has since
been lost (Lovering N a t u r e 183,
1664; 1959). Until the discovery of the
remains of a chondrite in Ordovician
limestone in Sweden by Thorslund and
Wickman (see this issue of Nature, p.
285), the only probable fossil meteorite
material known to the writer was the
occurrence of chondrule-like structures
in Mesozoic sediment from the Urals
(Yudin Meteoritics 6, 99; 1971). The
O r d o v i c i a n c h o n d r i t e h a s been
reasonably ( b u t n o t conclusively)
identified as a member of the high-iron or
H-group of the ordinary chondrites. Its
fall to Earth some 460 Myr ago has
considerable significance.
There are compelling arguments that
most meteorites come from asteroids.
But although it now seems probable that
there are sufficient numbers of the Earthcrossing Apollo asteroids and the related

Mars-crossing Amor asteroids to account

for the observed flux of meteorites, a
steady state can only be maintained if
these groups are fed from some source
beyond Mars (Wetherill Geochim.
cosmochim. Acta 40, 1297; 1976). The
small Apollo objects havea mean lifetime
of only a few tens of millions of years
before they are either captured by a
planet or ejected from the Solar System
by Jupiter. About 30% of observed
meteorite falls are H-group chondrites.
Most of these meteorites have cosmic-ray
exposure ages close to 4 Myr (see,
f o r example, Ziihringer Geochim.
cosmochim. Acta 32,209; 1968), which is
taken to mean that they were released as
metre-sized objects from some larger
body some 4 Myr ago. Release was
presumably the result of a collision
involving the parent body. No stony
meteorite has an exposure age greater
than 100 Myr. Thorslund and Wickman's
discovery indicates that there was
H-group chondrite material a m o n g
Apollo asteroids some 460 Myr ago, and
the possibility then arises that the
meteorite flux has not changed
significantly in composition with time.
If this is so there appears to be a
problem. From spectral reflectance data
Gaffey and McCord (Proc. 8th lunar Sci.
Conf. 1, 113; 1977) found that belt
asteroids rarely have surfaces with the

mineralogy of ordinary chondrites; such

surfaces are common among Apollo and
Amor objects. Wetherill concludes that
to maintain a steady-state population of
these bodies, belt a s t e r o i d s a n d ,
dominantly, comets must have orbits
which can evolve into Earth- or Marscrossing orbits. However, belt asteroids
appear to have surfaces which are almost
exclusively not ordinary chondritic, and
Anders (Icarus24,363; 1975) has argued,
on the basis o f the ancient bombardment
undergone by meteorites of different
types by particles emitted by solar flares,
that most stony meteorites cannot have
come from the outer Solar System. The
problem of the maintenance of the
meteoritic flux was recently discussed by
Napier and Clube (Nature 282, 455;
1979), who proposed that a proportion of
meteorites come from outside the Solar
System, but the meteorites themselves
have no major chemical or isotopic
anomalies consistent with different
regimes of nucleosynthesis; so we are left
with a Solar System origin. The question
of where in the Solar System that might be
has, if anything, been aggravated by the
new discovery.

Andreetti (Parma) dealt with crystallographic studies of a whole range of

inclusion compounds. I t was therefore
surprising t o learn of t h e lack of
crystallographic studies o n W e r n e r
inclusion compounds and J. Lipkowski
(Warsaw) reported his recent studies which
aimed at trying to understand the factors
governing host lattice selectivity. A
knowledge of the crystal structure is also
necessary to understand the energytransfer processes between host and guest
using UV and y-ray radiation reported by
the group from the Institute of Nuclear
Chemistry, Rome. Another new term
suggested at the meeting was 'excitrate' (A.
Guarino, Rome) to denote a clathrate with
either the host or guest, or both, in an
excited state.
The Hofmann inclusion compound,
Ni(NH,),Ni(CN),. 2 C 6 H 6 , has been
known since 1897, but its properties have
only been recently studied extensively. The
most prominent worker in this field, T.
lwamoto (Tokyo), described his work on
modifying these host lattices by replacing
the a m m o n i a by bidentate bridging
ligands, and by replacing the square planar
Ni(CN),2- group by tetrahedral M(CN),2groups. These modified host lattices show
interesting properties such as the ability of

CdenHg(CN), t o stabilize the cjrclohexadienyl free radical, produced by the

y-ray irradiation of the benzene guest
molecule. This free radical is stabilized up
to the thermal decomposition temperature
of the host lattice, 423K. Vibrational
spectroscopic studies of these systems were
reported by J.E.D. Davies (Lancaster) and
stability studies were reported by A.
Sopkova (~oyice).
It is true t o say that the majority of
inclusion compounds known have been
discovered by accident, b u t D.D.
MacNicol (Glasgow) gave a fascinating
talk outlining his strategy for the synthesis
of a variety of compouds which should be
expected to act as host lattices. Equally
fascinating was the talk by J.L. Atwood
(Alabama) describing his work o n 'liquid
clathrate' systems. These are non-stoichiometric compounds which form in the liquid
phase on the interaction of aromatic molecules and M[AI,R6X] moieties. The mole
ratio of aromatic t o anion can vary
enormously, the present record being
[N(C,H,,),][A12Et61].39.6
o-xylene. A few
of these liquid-phase systems have yielded
crystalline p r o d u c t s , f o r e x a m p l e
Cs[A12Me6N3] .2p-xylene, in which the
p-xylene molecules sandwich the caesium
ion.

A recent interesting application of

inclusion compounds is the realization that
a chiral host lattice might show enantiomer
selectivity, and G. Tsoucaris (ChatenayMalabry) gave examples of the enantiomeric selectivity of the tri-o-thymotide host
lattice. Another recent development is the
application of solid-state NMR spectroscopy to the study of inclusion compounds.
J.A.Ripmeester and D.W. Davidson
(Ottawa) illustrated the use of 129XeNMR
in the study of the Structure I xenon
clathrate deuterohydrate, where it was
possible t o observe well s e p a r a t e d
resonances from the atoms in the two types
of cavity.
In conclusion, this was a well timed
symposium, as it is apparent that the field
of inclusion chemistry has advanced
considerably in the 33 years since H.M.
Powell first characterized this type of
compound, and it is equally apparent that
further advances remain to be made. Some
of the papers presented at this meeting will
be published in a special issue of the
Journal of Molecular Structure, and it is
hoped to hold another symposium in
C
Parmainabout two years time.

Robert Hutchison is at the British Museum

(Natural History), where he is responsiblefor
the curation of the national collection of
meteorites.

J.E.D. Davies is in the Department of

Chemistry, University of Lancaster, UK.

Nature Vol. 289 22 January 1981

226

Holographic bubble chambers

from David J. Miller

THEnew particles have tantalisingly short

lifetimes. Even the D+ , the longest-lived
charmed particle, lasts for only about 1012
seconds - time to travel 300 pm at the
velocity of light, or a few millimetres with
relativistic time dilation. But the Do,the
A,,,,,
the F meson, the T lepton and the
'beauty' mesons are predicted to travel
only some tens of microns before decay.
Nuclear emulsions have come back into
popularity in the past five years because
they gave the only chance of studying such
short tracks, and now a new generation of
tiny bubble chambers is developing which
can resolve bubbles of less than 25 pm in
diameter. Each technique has shortcomings - the emulsion requires many
scanner-years of work t o find the
interesting events; a bubble chamber with
conventional cameras has such a poor
depth of field when the resolution drops
below 25 pm that it becomes impossible to
focus all of the beam particles into the
sensitive volume.
At the Fermilab workshop* Le Coq
(CERN) presented impressive pictures and
videotapes that had been reconstructed
from holograms made with a 6.5 cm
diameter by 3.5 cm deep heavy liquid
bubble chamber ('BIBC', the Berne
Infinitessimal Bubble Chamber). The
holographic technique gave clearly
resolved images of bubbles only 8 pm in
diameter over the whole depth of the
chamber. This work has inspired a number
of other developments. Pless (MIT)
presented scaling rules for the change of
resolution when Le Coq's 'straightthrough' holographic layout is modified to
medium-sized bubble chambers with m 1
m diameter. His collaboration is now
presenting a proposal to Fermilab for a
study of the T neutrino in such a chamber,
using a neutrino beam from a beam dump.
A proposal by a European collaboration
for a small holographic hydrogen bubble
chamber was discussed by Fisher
(Rutherford). This would be a continuation of the work being done with
conventional optics in proton and pion
beams at CERN. He warned that, although
the desired resolution can almost certainly
be achieved over a useful volume, and
although there are techniques available for
detecting, measuring and identifying the
secondary particles outside the chamber,
one serious problem remains for the
exploitation of holographic bubble
chambers - a very similar problem to that
faced by nuclear emulsions. The new particles are produced very rarely - in less
*A workshop on holographic bubble and streamer chamber
techniques was organized by L. Voyvodic at Fermilab, Batavia.
Illinois on II-12 November 1980.

than one event in a thousand. To make an

emulsion, or a hologram, really useful, an
electronic trigger must be devised which
will identify the subsample of events which
might contain the short-lived particle. The
tracks from these events must then be
extrapolated back into the emulsion or the
chamber so that the scanner knows where
to look. With a bubble chamber it should
be possible, in principle, to pulse a laser
and make the hologram only when a good
event has been detected, but Fisher warned
that no technique yet exists which can
separate the rare particles cleanly from the
background within the few microseconds
that it takes for bubbles to grow to the
desired size.
Some people think that an alternative
approach might bring good results. Instead
of looking for a 'needle in a haystack',
using small chambers with fine resolution
in proton, pion or photon beams, where
the backgrounds are enormous, it is perhaps more sensible to work in a neutrino
beam where charmed particles occur in 10
per cent of events. To have a reasonable
number of neutrino interactions a large
bubble chamber must be used, such as
BEBC at CERN or the 15 foot chamber at
Fermilab. At present these chambers grow
their bubbles to around 600 pm in diameter
in order to give a clear conventional image
from more than 10 m3 of sensitive volume.
A few D+ tracks have been resolved - the
very furthest tail of the length distribution
- but many more could be seen if 50 pm
resolution were possible. Very little work
has been done on the possibilities of using
holography in the large bubble chambers
but no one has demonstrated that it cannot
be done. A number of methods were

suggested: bright field 'Scotchlite'

illumination, though work at Imperial
College has shown that speckle may be a
problem; dark field with Scotchlite, but
can sufficiently powerful lasers be found;
retrodirecting mirrors, if a suitable
reference beam can be set up. Peterson
(Hawaii) drew attention to the relevance of
these ideas to proposals his collaboration is
making for a new chamber at Fermilab. Its
primary purpose is to allow secondary particles from neutrino interactions to leave
the bubble chamber and be identified, but
such a new device could also be designed to
incorporate whatever is found to be the
most favourable holographic layout.
We were reminded by Lederman, the
Fermilab director, that finer resolution
means smaller equipment which costs less
to build. Fine electronic counters would
enable experiments of all kinds to be
scaled-down proportionally t o the
resolution, giving more than proportional
reductions in magnet, material and
building costs. Such counters were outside
the scope of the workshop, but more than
one group is convinced that the solution to
the 'needle in a haystack' problem will
depend on a n emulsion o r small
holographic bubble chamber being
followed by electronic counters with a
resolution bf 20 pm which will both
improve the trigger and allow very precise
predictions to be passed on to the scanners.
The workshop ended with a theorist's
checklist of the effects that could be
studied at high resolution. Bjorken
(Fermilab) stressed that once the new
particles can be collected in reasonable
numbers they should show as complex and
instructive a phenomenology as did the
0
strange particles.

D. J. Miller is a lecturer in the Department of

Physics and Astronomy at University College
London, at present on sabbatical leave in
Fermilab.

100 years ago

A "Natural" Experiment in Polarised Light

Mr E. T. Sachs sends us some interesting notes

from Batavia: - "Within the past month I
have been so lucky as to make what I hope is a
very interesting if not remarkable discovery.
On the Island of Biliton, 200miles from here, I
found a fresh water fish which produces its
young living from its mouth. I am quite
prepared for the cry of incredulity that will be
raised; but I conducted my observations with
living fish and closed doors, and what I assert
is undeniable: the eggs are hatched in the lower
portion of the head of the fish, and are
projected out at the mouth and from nowhere
else. I have secured several specimens, which I
shall send to Dr. Giinther, who will of course
at once set the matter at rest."

Break off a plate of ice and hold it between the

sky and apool of water. Its reflected image will
show the beautiful colours due to polarised
light. The incident rays should come from a
part of the sky about 90" from the sun, and
reflection should take place at the polarising
angle for water, and the plate will probably
require adjusting to bring out the maximum
effect. Water, vaporous, solid, and liquid,
thus furnishes us with polariser, crystal, and
analyser. I do not remember to have read any
account of this very simple experiment, for
which Nature provides all the materials.
Chas. T. Whitmell
9, Beech Grove, Harrogate, January 10
From Nature 33,20 January, 268,275, 1881
@I981 Macmillan Journals Ltd

227

Nature Vol. 289 22 Januarv 1981

Contaminated cell lines

from David Dickson, Washington
YET another corruption of the scientific
literature has been uncovered with the
discovery that several cell lines widely
quoted in the scientific literature a s
originating from patients with Hodgkin's
disease have now been shown not to be
human cells at all, but to come from a strain
of owl monkey (see the accompanying
article, pages 228-230).
The cells were grown in culture by Dr
John C. Long, who resigned a year ago as
assistant pathologist at the Massachusetts
General Hospital and associate professor
of pathology at Harvard Medical School,
after admitting to having faked data in
a research paper published in the Journal
of the National Cancer Institute in 1979.
On resigning Dr Long admitted having
misled colleagues about carrying out
experiments to produce revised data on the
molecular weights of immune complexes.
The alterations were included in the paper
after it had been rejected by the referees of
another journal.
Dr Long said on the telephone last week
that during his time at the Massachusetts
General Hospital, he had no reason to
suspect t h a t his cell lines were
contaminated and welcomed the work of
Harris etal. However, when he resigned, he
did not reveal that six months earlier he had
received data from the National Cancer
Institute (NCI) casting doubt o n the
identity of the cells.
T h e first successful l o n g - t e r m
cultivation of cells from patients with
Hodgkin's disease was claimed by Dr Long
and colleagues from the hospital and the
medical school in Proceedings of the
National Academy of Sciences (PNAS) in
1973. Other groups have encountered
considerable difficulty in establishing
permanent lines from diseased spleen cells.
The procedures used to grow the original
cells through over 200 passages were
described in a 1977 paper in Journal of
Experimental Medicine as well as PNAS.
(Science Citation Index lists 30 citations of
this paper in 1979 and 1980, which its compilers say means the paper is "highly
cited".) In total four papers have appeared
in PNAS - three naming Dr Long as the
first author. Experiments using these cells
have a l s o been described in o t h e r
publications (see page 228).
All the papers referring directly to the
now-challenged cell cultures will have to be
reassessed to see if any data of scientific
value can be salvaged, since there is now
evidence that the cell lines were contaminated from an early stage.
Dr Long was awarded a three-year grant
for more than $200,00Oin 1976 to continue
his study of the Hodgkin's disease tissue
cultures. This was renewed in 1979 -based
on an application said to include the data

which he later admitted forging - and he

received $%,000 in that year, as the first part
of a three-year grant of almost $500,000.
This grant was terminated by the NCI at
the request of the hospital, after Dr Long
admitted falsifying data in the 1979 paper
in Journal of the National Cancer
Institute. According to the hospital, Dr
Long claimed at the time that "competition for federal research grants" was a
major factor in causing him to lie about
experiments he had not carried out.
However, he did not admit t o any other
impropriety and the hospital put out a
statement saying that the data he had
admitted falsifying was "obscure, and of
no known significance. . . no known harm
has resulted even in the field of science".
The hospital also said that further
suspicions about his research were being
followed up. There was already evidence o f
contamination of the cell lines - early in
1979 one of Dr Long's research associates
had grown suspicious about the true nature
of the cells, and sent samples to Dr Stephen
O'Brien at the NCI Laboratory of Viral
Carcinogenesis for genotypic evaluation.
The NCI scientists told Dr Long that
three of the cell lines carried the enzymatic
signature of the same individual. And they
also pointed out an unusual characteristic
they could not explain - a peculiarly fast
mobility in a key enzyme system.
Dr Long did not follow up this comment
with the NCI. And he told his research
workers that the NCI check had shown that
the cells were genuine - indeed that in one
case the signature of the cultured cells
corresponded directly with that of the red
blood cells of the patient from whom the
cells were said to have been taken (another
claim subsequently shown to be false).
Nothing further happened a t the time.
But after Dr Long's resignation, Dr Robert
T. McCluskey, head of the hospital's
pathology department, considered further
investigations necessary. Karyotypes and
cell samples were sent to the cell culture
laboratory at the University of California's
School of Public Health, where chromosomal analysis revealed that the cells were
not human. Several other laboratories were
subsequently sent samples to complete the
identification. And the karyotype was
recognized as that of the owl monkey by Dr
Bharati Hukku of the Child Research
Center of Michigan, Detroit.
The cells were then confirmed by the
New England Regional Primate Center as
coming from a Northern Columbian
brown foot owl monkey. Indirect evidence
that this was the likely source of contamination was the fact that this cell line
had been used for virus research in theearly
1970s in the laboratory in which Dr Long
was then working.

The fourth suspected cell line, started

later than the other three, has turned out to
be human. But there is no evidence of a link
with Hodgkin's disease tumours.
T h e full i m p a c t of D r L o n g ' s
publications remains unclear. His results
have been quoted as supporting a
particular view of the pathogenesis of
Hodgkin's disease which may now be
questioned, in particular suggesting a
macrophage origin of the peculiarly large
Reed-Sternberg cells found in Hodgkin's
patients.
Other research workers have been
careful not to put too much weight on Dr
Long's results, remaining sceptical in the
light of their own failure to culture the
same cells. And Dr Henry Kaplan of
Stanford University feels that the main
impact may have been to reduce the
incentive for others to tackle the problem
of growing the cells in vitro, since Dr Long
had already claimed success.
Investigators for the Department o f
Health and Human Services in Boston
must now decide whether there is any
evidence that should be passed to the
Department of Justice to charge Dr Long
wiih fraud against the federal government.
Under the US Code, consciously providing
false information when applying for
federal assistance is a criminal offence.
The journals which published Dr Long's
papers now have to decide what action to
take. A representative of the National
Academy of Sciences has said that the
situation was "unique" for PNAS, which
has never previously had to disown any
paper it has published.
Papers submitted to PNAS are not
formally reviewed if a member of the
academy is a co-author. In Dr Long's case,
each of the four papers that he published
was written with academy member Dr Paul
Zamecnik, who recently retired as
professor of ontological medicine and
director of the J.C. Warren Laboratories
at Haward Medical School, where Dr
Long was a research student.
The National Institutes of Health are
already discussing how to tighten up on
evaluating the scientific content of research
funded by federal grants. An audit of Dr
Long's g r a n t revealed n o fiscal
irregularities in the way that the money had
been used, but did not go into the
legitimacy of the resezrch.
I r o n i c a l l y , t h e recently c r e a t e d
President's Commission for the Study of
Ethical Problems in Medicine and
Biomedical and Behavioral Research is
now considering another Boston case, in
which a research worker at the Boston
University Medical School was shown to
have faked data in a cancer survey.
Research administrators admit that
cases of fraud - usually relatively minor
- are not unusual, although few reach the
public eye. "Whatever you see is probably
only a small fraction of what actually goes
on," according to Dr Donald Kennedy.
president of Stanford University.

228

Nature Vol. 289 2 2 January 1981

Contamination of Hodgkin's disease cell cultures

Nancy L. Harris*, David L. Gang*, Steven C. Quay*, Sibrand Poppema*,
Paul C. Zamecnik', Walter A. Nelson-Reed & Stephen J. O'Brien'
* Department of

Pathology, Harvard Medical School, Imrnunopathology Unit, Massachusetts General Hospital, Boston, Massachusetts 021 14
t Worcester Foundation for Experimental Biology, Shrewsbury, Massachusetts 01545
$ Cell Culture Laboratory, University of California School of Public Health, Naval Biosciences Laboratory, Oakland, California 94625
8 Laboratory of Viral Carcinogenesis, National Cancer Institute, Bethesda, Maryland 20014

Several laboratories have recently reported the establishment and characterization of long-term cell lines thought to be
related to the neoplastic cell of Hodgkin ' sdisease. Here, Harris et al. discuss evidence that some of these lines are, in fact, not
related to Hodgkin's disease but are non-human contaminants.
IN an effort to determine the origin of the malignant cell of
Hodgkin's disease, many investigators have attempted to establish permanent cell lines from tissues involved with the
disease. Early attempts to culture tissues involved with Hodgkin's disease yielded either mixed cultures which could not be
maintained in uitro, or Epstein-Barr virus-infected lymphoblastoid cell lines (see ref. 1 for a review). Recently, several laboratories have reported the establishment and characterization of
long-term cell lines thought to be related to the neoplastic cell of
Hodgkin's
Our studies indicate that several of these
cell lines are not related to Hodgkin's disease, but are nonhuman contaminants.
Long et al. reported the establishment of four permanent

monolayer cell lines, designated FQ, RB, SpR, and RY, from
tumour nodules in spleens of patients with Hodgkin's disease7.
These four cell lines were considered to be neoplastic by virtue
of their capacity for indefinite replication in uitro, aneuploidy
and the production of invasive tumours on subcutaneous
injection into nude mices. Although the cells had n6 surface
markers demonstrable by standard immunological techniques,
several features, namely the production of lysozyme and
esterases, the ability to phagocytose iron particles9 and the
ultrastructural morphology of the cells in c ~ l t u r e ' ~
were
, felt to
be consistent with a macrophage origin of the cultured cells. The
establishment of these permanent cell lines, which seemed to be
neoplastic and had features of macrophages, supported the

Fig. 1 Three marker chromosomes, conventional staining (insert). Trypsin-Giemsa

banded metaphase chromosomes of an FQ
(topfour rows) and an OMK-210 cell (bottom
four). The FQ culture exhibited polysomy for
many chromosomes including, in some
instances, duplicate marker chromosomes.
OMK-210 cells (kindly provided by Dr M.
Daniel) were near-diploid. Identical marker
chromosomes in both cells are indicated by
arrows. Most noticeable are isochromosomes
for 13q. A small, altered chromosome which
resembles the Y chromosome was noted
among the markers in the FQ cells [designated
Y(?)]. As the karyotype of OMK-210 is
female, and the presence of multiple identical
markers in the two cell lines indicates a common origin, it is unlikely that the small FQ
chromosome is a true Y chromosome.The FQ
karyotype was kindly constructed by Dr
N. S. F. Ma; its pattern was followed for the
OMK-210 cell.

0028-0836/8l/04022843$01.00

@ 1981 Macmillan Journals Ltd

229

Nature Vol. 289 22 January 1981

concept that the malignant-appearing cells observed by

pathologists in Hodgkin's disease tumours are neoplastic
macr~pha~es'.~.'".
Further studies of these cell lines appeared to offer a new
insight into the pathogenesis of Hodgkin's disease. It was reported that two of these cell lines (FQ and RB) specifically bound
immune complexes from the sera of patients with Hodgkin's
d i s e a ~ e " . ' ~As
. the cells lacked receptors for immunoglobulins
and complement and did not bind immune complexes from sera
of patients with other disorders, it was concluded that the
antibody portion of the immune complex was reacting with an
antigen on the cultured cells. This observation not only provided
the first direct evidence that the cells cultured in vitro were
related to the human disorder, but also suggested that some
patients with Hodgkin's disease were reacting immunologically
to a tumour-associated antigen.
A recurring problem associated with the use of long-term,
established cell lines in the study of human tumours is the risk of
contamination and overgrowth of cultures by unrelated cell
lines". Previous publications have shown that the techniques
of trypsin-Giemsa-banded karyotypingI4 and isoenzyme
(allozyme)
can detect both intra- and interspecies
contamination of cell lines. In addition to these methods of
genetic characterization of cell lines, the technique of subcutaneous transplantation into nude (athymic) mice can also be
used to investigate the origin of a cell line. When noncultured
tumour cells or established cell lines are injected into nude mice,
the tumours which form may be more differentiated than either
the original tumour o r the in vitro cultured cells, thus permitting
determination of the histogenesis of an apparently undifferentiated cell".
Using these three techniques, we have found that three of the
four cell lines (FQ, R B and SpR) reported by Long et a/.' are
identical to each other and are of owl monkey rather than human
origin. Furthermore, they have morphological features of epithelial cells rather than macrophages. The fourth cell line (RY) is
of human origin, but its relationship to Hodgkin's disease has
not been established.
Cultures of FQ, RB, SpR and R Y cells maintained by D r J.
Long, (Immunopathology Laboratory, Massachusetts General
Hospital) were studied, as were the earliest available cultures
from the laboratory of D r P. Zamecnik (spanning passages
14-500 of FQ, 300-380 of RB, 21-220 of SPR, and 5-130 of
RY). In FQ, R B and SpR cells the chromosome numbers ranged
from 7 5 to 100, with a modal number between 8 3 and 100
chromosomes. Conventional chromosome preparations of FQ,
R B and SpR metaphases revealed three to four copies of a
medium-size submetacentric chromosome with a pronounced
near-terminal secondary constriction; a chromosome peculiar to
normal owl monkey cells (Fig. 1, insert). Trypsin-Giemsabanded karyotopes of FQ, RB, and SpR exhibited non-human
chromosomes, singly or in multiples, and were recognized as
polyploid derivatives of cells of the owl monkey, Aotus trivirgatus (2n = 54)'"19. Cells at both early and late passages in
culture contained many normal chromosomes of the owl
monkey karyotype, as well as highly altered, marker
chromosomes. Most notable among the latter were one o r two
isochromosomes involving the long arm of chromosome number
13 (Fig. 1). A search for an owl monkey cell line that was well
adapted to in vitro culture led to the study of OMK-210 cells, an
established cell line derived from owl monkey kidneyz0.
Comparison of FQ, R B and SpR with OMK-210 cells showed
that identical marker chromosomes, as well as normal
chromosomes, existed in all lines.
Despite the presence of identical marker chromosomes,
further chromosomal alterations had clearly taken place following the separation of FQ, R B and SpR from OMK-210 cells,
leading to differences in their individual karyotypes. For example, although OMK-210 contained near-diploid numbers of
chromosomes, the FQ, R B and SpR cells were all near-tetraploid. Such chromosomal alterations are common occurrences
in cultured cell lines2'.

PEP8

ESD

SOD

GPI

MPI

I+)

PEPC
UI

----

(-)

----

GOT

NP

LDH

MDH

I+)

I
-

111

11111111

Fig. 2 Composite electropherograms of 10 gene-enzyme systems diagnostic for interspecies cell identification (refs 15, 16). The columns
represent: I. FQ patient's red blood cells; 2, FQ cell line; 3,OMK-210 cells:
4. RB cells; 5, RY cells; 6 , HeLa cells. The 10 diagnostic enzymes were:
PEPB, peptidase-B (EC 3.4.1 1.); GPI, glucose phosphate isomerase (EC
5.3.1.9); SOD, superoxide dismutase (EC 1.15.1.1); ESD, esterase-D (EC
3.1.1.1); MPI, mannose phosphate isomerase (EC 5.3.1.8); PEPC, peptidase-C (EC 3.4.1 1); LDH, lactate dehydrogenase (EC 1.1.1.27);NP, purine
nucleoside phosphorylase (EC 2.4.2.1); GOT, glutamate oxaloacetate
transaminase (EC 2.6.1.1);and MDH, malate dehydrogenase (EC 1.1.1.37).
* Indicates that no activity was observed in that extract for the enzyme. All
the systems except ESD vary between man (HeLa) and owl monkey (OMK210). In man ESD is polymorphic for two alleles ESD-I and ESD-2. The
RY line is heterozygousfor ESD- 1-2. The FQ red cell extract shows that the
FQ patient is heterozygous for ESD- 1-2 while the FQcell has an ESD with a
mobility of ESD-1, which is identical to that of OMK-210. The OMK-210
isozymes were each identical to those observed in extracts of OMT-1, an owl
monkey lymphoid cell line derived by Drs Louis Qualtiere and Gary Pearson
(data not shown).

The identity of the FQ, R B and SpR cell lines with OMK-210
cells was further established by determining the electrophoretic
mobilities of a panel of 1 0 isoenzymes selected to detect interspecies contamination (Fig. 2). The FQ, R B and SpR cell lines,
at the passages examined, were identical to each other and to
OMK-210 cells at each locus, and distinct from the human cell
line, HeLa, and the red blood cells of the patient from whose
spleen the FQ cell line purportedly arose.
Together, the chromosome and isoenzyme analyses indicated
that the three supposed Hodgkin's disease cell lines had been
cross-contaminated by OMK-210 cells. OMK-210 cells had, in
fact, been carried in the laboratory of Drs Long and Zamecnik at
the time when the FQ, R B and SpR cell lines were initiated. In
the original description of the cell lines, it was reported that the
Hodgkin's disease monolayer cultures underwent a morphological alteration in vitro after 3-10 months' propagation in
culture, with the appearance of polygonal cells that replicate in a
mosaic patterns. This appearance of cells with different
morphology from that first noted in the primary cultures or early
passages is a characteristic occurrence in the course of
contamination of cell lines14.
Subcutaneous tumours formed by the FQ, R B and SpR cells
in nude mice contained acinar structures with microvilli and
intercellular junctions (Fig. 3 a ) . These features indicated that
the cultured cells were epithelial, and not of monocyte-macrophage or lymphoid originz2. The ultrastructural features of the
nude mouse tumours contrasted with those of the cells when
examined in monolayer culture'", and indicated that additional
differentiation had occurred in the nude mouse. Surface microvilli were present on the monolayer cells, but acinus formation
and intercellular junctions were not seen in vitro"'. Although a
previous report o n the transplantation of the FQ, R B and SpR

O 198lNature Publishing Group

Nature Vol. 289 22 January 1981

230

Fig. 3 a, Electron micrograph of tumour formed 3 weeks after subcutaneous injection of 5 x lo6 FQ cells into a 6-week-old nude mouse
BR, Charles River), fixed in Karnovsky-I1 and stained
(CRL:nu/nu (CD-1)
by the OCUB method9. ~ 2 , 7 5 0 .Insert: intercellular junctions. ~ 2 7 , 5 0 0 .
b, Paraffin-embedded section of subcutaneous tumour formed in nude
mouse by FQ cells (from J. C. Long, as described in ref. 7) showing gland and
tubule formation (haematoxylin-edsin, x 185).

cell lines into nude mice stated that the morphology was that of
'reticulum cell sarcoma", review of the original slides from that
study revealed that tubular and acinar structures indicative of
epithelial differentiation were present in the tumours formed by
these three cell lines (Fig. 36). This observation indicates that
contamination of these cell lines had already occurred by then.
The origin of the fourth cell line, RY, could not be established
with certainty. The cells had approximately 60 chromosomes.
Conventional chromosome preparations did not contain the
characteristic owl monkey chromosome with the near-terminal
secondary constriction which was seen in the other three cell
lines. Trypsin-Giemsa banding revealed an abnormal human
karyotype, which was distinct from HeLa or any other
established cell line known to us. The isoenzyme genetic signature likewise indicated that the cell line was human (Fig. 2);
analysis of a different panel of isoenzymes, selected for their
Received 10 November; accepted 17 December 1980.
1. Kaplan, H. S. Hodgkin's Disease (Harvard University Press. Cambridge. 1980).
2. Boecker, W. R., Hossfeld, D. K., Gallmeier, W. M. & Schmidt, C. G. Nature 258,235-236
(1975).
3. Roberts, A. N., Smith, K. L., Dowell, B. L. & Hubbard, A. K. Cancer Res. 38,3033-3043
(1978).
4. Schaadt. M., Fonatsch, C., Kirchner, H. & Diehl, V. Blur 38,185-190 (1979).
5. Long, J. C., Aisenberg. A. C., Zamecnik. M. V. & Zamecnik, P. C. Proc. nam. Acad. Sci.
U.S.A. 70,1540-1544 (1973).
6. Long. J. C.. Aisenberg, A. C. & Zamecnik, P. C. Proc. narn. Acad. Sci. U.S.A. 71,
2285-2289 (1974); 71,2605-2609 (1974); 1. narn. Cancer Insr. 58,223-227 (1977).
7. Long, J. C., Zamecnik, P. C., Aisenberg, A. C. & Atkins, L. I. exp. Med. 145, 1484-1500

-- . . .

(1077\
\
,

8. Zamecnik, P. C. & Long, J. C. Proc. nam. Acad. Sci. U.S.A. 74,754-758 (1977).
9. Long. J. C. Clin. Haemar. 8.531-565 (1979).
10. Gang, D. L. Long, J. C., Zamecnik, P. C., Chi, S.-Y. &Dvorak, A.M. Cancer 44,543-557
(1979).
11. Long. J. C. eral. New Engl. 3. Med. 297,295-299 (1977).
12. Long, J. C., Dvorak. A. M., Quay. S. C., Stamatos, C. & Chi, S.-Y. I. narn. CancerInsr. 62,
787-797 (1979).

polymorphism within human

confirmed that
RY differed from HeLa or any other known cell line (not
shown). Tumours formed in the nude mouse by the RY cells, in
contrast to the other cell lines, did not have epithelial features;
the histogenesis of this cell line could not be determined from
the morphology of the nude mouse tumours. As previously
described7, the RY cell line did not have surface receptors
characteristic of macrophages. We were unable to confirm the
initial report of lysozyme production by this cell line (S. Quay
and S. Poppema, unpublished data). A review of the donor
patient's hospital record revealed that the spleen from which the
RY cell line originated contained no grossly or microscopically
identifiable tumour. As this cell line fails to meet a basiccriterion
for a tumour cell line-namely, that it be cultured from a
specimen known to contain tumour cellsz3-its relationship to
the malignant cell of Hodgkin's disease remains in doubt.
Although a few other laboratories have reported the establishment of permanent cell lines from patients with
Hodgkin's disease, no conclusions regardin the nature of the
malignant cell have been drawn from themk4. Although there
are indications from other sources that the malignant cell of
Hodgkin's disease is of macrophage origin, the evidence is not
conclusive (for a review, see ref. 24). As the only permanently
established Hodgkin's disease cell lines with macrophage
characteristics, the cell lines reported by Long et al. were
thought to provide substantial evidence for the macrophage
origin of the Reed-Sternberg cell. Our findings, therefore,
remove some of the support for the theory that the malignant
cell of Hodgkin's disease is a macrophage. Furthermore, results
obtained with these cell lines cast doubt on the specificity of the
properties which were considered to indicate that these cultured
cells were macrophages. Recently, the RY cell line was found to
produce endogenous pyrogen, an observation which it was
thought might explain the tendency of some patients with
Hodgkin's disease to develop feverz5. The significance of this
observation must be re-evaluated in the light of our findings.
Our results also indicate that the establishment of permanent
cell lines from tissues involved with Hodgkin's disease may be
more difficult than had been previously ~ u ~ ~ e s t eSeveral
d~-~.
major laboratories have experienced repeated failure in their
attempts to establish permanent cell lines from tissues involved
with Hodgkin's d i s e a ~ e ~ ~ The
. ~ " ~inability
~.
of these other
workers to establish permanent lines in vitro, despite numerous
attempts, may mean that appropriate methods for culture of
these cells have not yet been found. On the other hand, it is
possible that their very resistance to in vitro propagation
has some biological significance which has not yet been
recognized.
We specially thank Dr Bharati Hukku of the Child Research
Center of Michigan, Detroit, who first noted the similarity
between certain chromosomes in FQ and Aotus triuirgatus. Dr
Ward D. Peterson Jr provided additional immunological and
serological proof of contamination. Messrs R. R. Flandermeyer
and D. W. Daniels and Dr L. Atkins assisted with karyology as
did Dr N. S. F. Ma, whose classification of Aotus karyotypes was
followed. We also thank Drs Robert T. McCluskey, Sigmund A.
Weitzman, C. Alan Brown and Clive L. Hall for their support.
13. Stulberg, C. S. in Conraminarion in Tissue Culture (ed. Fogh. J.) 1-27 (Academic. New
York. 1973).
14. Nelson-Rees, W. & Flandermeyer, R. P. Science 191, 96-97 (1976); 195, 1343-1344
(1977).
15. O'Brien. S. J., Kleiner, G.. Olson. R. &Shannon, J. E. Science 195, 1345-1348 (1977).
16. O'Brien, S. J..'Shannon. 3. E. &Gail. M, H. I n Vitro 16. 119-135 (1980).
17. Sharkey, F. E. eral. in The NudeMouse in ExpcrimentalandClinical Research (eds Fogh, I.
& Giovanella, B. C.) 187-214 (Academic, New York, 1979).
18. Miller, C. K. er ol. Cyrogener. Cell Genet. 19,215-226 (1977).
19. Ma. N. S. F.. 1ones.T. C... Miller.. A. C. Mornan.
- . L. M. & Adams. E. A. Lab. AnimalSci.26.
1022-1036 (19;s).
20. Melendez, L. V. er al. Lob. Animal Care 19,372-377 (1969).
21. Nelson-Rees. W. A. in Oriuin and Natural Hisrorv of Cell Lines (ed. Bariaoui. C.1 25-79
(Liss. New York, 1978).
22. Mackay. B. & Osborne, B. M. in Pathology Annual 359-405 (Raven. New York, 1978).
23. Fogh. J., Goodenou. M., Loveless. 1. & Fogh. H, in I n VirmMerhods in Cell-Mediaredond
Tumour Immunity (eds Bloom, B. R. & David, J. R.) 677 (Academic, New York, 1976).
24. Kaplan. H. S. Cancer 45,2439-2474 (1980).
25. Bodel, P. Ralph, P., Wenc, K. &Long, J. C. I. clin. Inuesr. 65, 514-518 (1980).
26. Kaplan, H. S. & Gartner, S. In:. I. Cancer 19,511-525 (1977).
27. Sundstrom, C. & Nilsson, K. Acra parh. microbial, scand. 86, 173-184 (1978).

O 198lNature Publishing Group

Nature Vol. 289 22 January 1981

231

Multiple asperity model for

earthquake prediction
*Cooperative Institute for ~ e s e m c hin Environmental Sciences, University of Colorado, Boulder, Colorado 80309
t Tennessee Earthquake Information Center, Memphis State University, Memphis, Tennessee 38152
$ Hawaiian Volcano Observatory, US Geological Survey, Hawaii 96718

Large earthquakes often occur as multiple ruptures reflecting strong variations of stress level along faults. Dense instrument
networks with which the volcano Kilauea is monitored provided detailed data on changes of seismic velocity, strain
accumulation and earthquake occurrence rate before the 1975 Hawaii 7.2-mag earthquake. During the -4 yr of
preparation time the mainshock source volume had separated into crustal volumes of high stress levels embedded in a larger
low-stress volume, showing respectively high- and low-stress precursory anomalies.
OBSERVATIONS
of seismic velocity anomalies by Russian
seismologist^'.^ and the development of the dilatancy-diffusion
model of earthquake precursor^^.^, have encouraged research
into earthquake prediction. The simplicity of the microcrackpore fluid mechanism5, its foundation in repeatable observations from laboratory rock mechanics and the absence of
contradictory evidence led many seismologists to anticipate in
vain that definite prediction of earthquakes would soon be
possible. For many medium-size earthquakes (5 < M < 6),
mostly strike-slip events along the San Andreas fault, the
expected precursors could not be
These negative
findings bring into question the universal applicability of
dilatancy to the Earth's crust.
A new model of the preparatory process leading to earthquake ruptures is needed. However, field data for such a model
cannot be gathered rapidly and in abundance because we d o not
know where and when large earthquakes will occur. Without an
earthquake prediction capability we may have to wait several
decades for large shocks to occur in a location where we can
measure in detail the phenomena associated with their preparation process.
The Hawaii earthquake of 1975 occurred within a dense
network of geophysical observations. Geodetic and seismological measurements had been carried on systematically for many
years to monitor the volcano Kilauea. These data indicate that
earthquake precursory changes d o occur, but that they show
contrasting behaviour in different parts of the earthquake source
volume. We interpret this as meaning that the stress level
throughout the source volume varies strongly-in small subvolumes dilatancy-related precursory changes may occur, while
strain-softening phenomena (due to stress unloading by creep)
occur in other parts of the source volume.

intrusion. While the stress loading process was different we

believe that the rupture process was the same as that in other
tectonic earthquakes.
Seismic waves emanating from this rupture were recorded by
instruments on Hawaii which are designed to produce on-scale
records for the strong ground motions of large, local earthquakes. The seismic energy was radiated in a sequence of
high-amplitude pulses separated by comparatively long periods
of low-amplitude radiation (Fig. 2). Such records are often
observed for large earthquakes and are thought to represent
multiple rupture
As the rupture propagates along a
heterogeneous faulting surface, or through a heterogeneous
crust, it encounters high- and low-stress volumes. The rupture
may speed up and slow down, or stop temporarily, with little
radiation emitted from low-stress portions of the fault and large
pulses emitted from zones of high stress. The latter may form
around asperities on the fault surface, a term which we will use
loosely as meaning a strong segment of the fault along which
stresses will accumulate.
We conclude that the Hawaii earthquake source volume
contained high- and low-stress sub-volumes, based on the strong

r I n n e r Anomalous Areas

The Hawaii (Kalapana) earthquake

The Hawaii earthquake of 29 November 1975 of 7.2 mag.
occurred in the south-east coast of the island of Hawaii on
Kilauea's south flank. The rupture occurred along a nearhorizontal plane in the Earth's crust a t a depth of 8 1 km. Its
length was 40-50 km and its width 10-20 km (Fig. 1).Along the
south coast subsidence of up to 3.5 m occurred and tsunami
heights reached 14.6 m (ref. 6). Damage from the earthquake
and related catastrophic events was estimated at $4.1 million,
and two people were killed.
The source of stress in Kilauea's south flank is the intrusion of
magma into the volcano and its rifts".". The last earthquake of
similar size to the 1975 rupture seems to have occurred in 1868.
Since then repeated episodic intrusions of magma along the rifts
(dotted in Fig. 1) have provided the 'pushing hand' compressing
the south flank. The Hawaii earthquake was a rupture of the
crust under stress, not a volcanic quake caused directly by an

(1028-0836/81/04023144$01

00

Fig. 1 Map of the south flank of Kilauea volcano showing major

geological features, the aftershock zone of the Kalapana earthquake (dashed diagonal hatching), the outer precursor volume
(solid hatching), the areas designated as 'major asperities', and
the rupture area inferred from surface wave and tsunami data.
A, Seismograph stations where constant seismic velocities were
observed; A, the station showing the velocity anomaly; .,geodetic
line benchmarks; 0 , the site of the 1972 Mauna Ulu eruption. 0,
Immediate foreshocks; the 5.9-mag foreshock is the larger circle
and the 7.2-mag main shock is the largest circle. Some seismic and
geodetic stations are omitted for clarity.
% 1981 Macrnlllan Journals Ltd

Nature Vol. 289 22 January 1981

232

20

40

60

Time (s)

Fig. 2 Strong-motion seismogram of the Hawaii main shock of 29

November 1975, recorded at Hilo. Note that the main rupture
seems to be made up of several events emanating energy pulses
roughly comparable in amplitude and duration with that of the
fore~hock~~.

motion records (Fig. 2). It is therefore not surprising that in

different regions of the source area we found contrasting precursory changes which we interpret as reflecting high- and
low-stress regimes respectively.

Precursors to the Hawaii earthquake

The first documented precursor to the Hawaii earthquake was a
delay of teleseismic P-wave travel times by 0.2 s lasting from
3.8 yr to several months before the main shock15 (curve b of Fig.
6). According to Student's t-test this anomaly is significant at the
99.9% confidence level. Only the station nearest to the epicentre (A Fig. 1) exhibited delayed arrivals; six other stations (A,
Fig. 1) within 3 km of the rupture area showed no significant
change. Nearly horizontal rays from local earthquake sources
also showed constant travel times at all stations, indicating that
the anomalous volume did not extend to the surface16.
These observations indicate that the Hawaii earthquake was
preceded by a velocity decrease in a volume of -10-km
horizontal extent centred on the hypocentre, while the rest of
the source volume remained nondilatant.
The rate of earthquake occurrence may contain information
on the state of stress of a crustal volume. If the seismicity rate
changes significantly as a function of time, the stress or the
strength of the rock in the volume must have changed. Since
1968 all shocks with magnitude 2.0 or more are located routinely. During the 8 yr before the main shock -1,600 such
earthquakes occurred within the main rupture volume. This
abundance enabled us to examine separately the rate in subvolumes of roughly 10-km dimensions. Two types of seismicity
behaviour were found. In some volumes the seismicity rate
remained approximately constant. An example of this consistency is the seismicity rate of the foreshock volume (eastern
major asperity in Fig. 1) as shown in curve 2 in Fig. 3. In other
sub-volumes the seismicity rate dropped dramatically around
the year 1971 1 (see curve 1 in Fig. 3).
The combined seismicity of the volumes with approximately
constant rate (inner anomalous areas of Fig. 1) contrasts sharply
with that in the outer anomalous region (Fig. 4) where the rate
dropped in early 1972 by -50% from 120 events yr-' to 55
events yr-l.
The normal rate in the outer anomalous region is defined by
data for the period 1968-7 1. A longer observation period can be
opened if we raise the minimum magnitude in the data set. The
seismicity rate in the outer anomalous region for magnitudes 2.4
or more is shown in Fig. 3 (curve 3) for the period 1962-75.
Again we see that a highly significant decrease by about 50%
occurred. The time of the change in this data set seems to be

late 1970. We conclude that a 50% seismicity rate reduction

occurred in the outer anomalous volume in 1971+ 1yr and
lasted up to the time of the main shock.
Summarizing the geodetic data from 1914 to 1971, Swanson
et al." showed that the south flank was displaced several metres
to the south-east with respect to observation points to the north
and west of Kilauea's rifts. They also noted that the south flank
accumulated compressive strain in the NW-SE direction in steps
which coincided with volcanic eruptions and intrusions (Fig. 5).
They concluded that shallow (depth <5 km), forceful magma
intrusions under Kilauea and its rifts provided the 'pushing
hand' which strained and dislocated the south flank away from
Kilauea and the rest of the island. Based on these geodetic data,
Swanson et al." wrote that a large earthquake might be expected on the south flank.
All geodetic observation lines on the south flank showed
compression up to 1970. In late 1971 and late 1974 respectively
some lines reversed their behaviour, showing extension, that is
they indicated strain was released". The data for the past eight
years are shown in more detail in Fig. 6a. Open circles indicate
the average strain registered by lines 1 and 3 (defined in Fig. 1);
solid squares show the strain measured on line 2. The strain data
interpretation after 1972 is debatable because of the absence of
measurements until late 1974. During this period a major phase
of the east rift Mauna Ulu eruption commenced (large arrow in
Fig. 6; O in Fig. 1). If the associated intrusive activity of this
extrusive event compressed the south flank as observed for
previous eruptions", it could account for the approximate 30bar increase of line 2 observed in late 1974. Because no data are
available for 1972-74, we show in Fig. 6 the uncertainty of the
strain (stress) behaviour of this period with both a solid (preferred) and dotted line, depicting two alternative strain histories.
After the late-1974 survey, only line 2 in the western inner
region was resurveyed before the 1975 main shock. It revealed a
massive, pre-main shock strain relaxation, corresponding to a
stress drop of -30 bar. The earthquakes which occurred during
this time could not account for this stress release because they
were far too small. We conclude that the strain was released
aseismically by some kind of creep process, perhaps along the
future fault plane, and at a rate >lo-'* s-'. Significantly, line 2,
which accumulated stress up to the end of 1974, was located on a
volume of constant seismicity rate, supporting the idea that high
seismicity indicates high stress accumulation.

Time (yrj
Fig. 3 Cumulative number of earthquakes as a function of time
for three subregions of Kilauea's south flank. The slope of the
curves equals the seismicity rate. The lower two curves (scale to the
right) starting in 1968 are based on earthquakes with magnitudes
>1.9. Curve 1 represents the south flank volume east of longitude
154.9" E, curve 2 the foreshock volume. For a better definition of
the normal seismicity rate, the magnitude >2.3 seismicity from
1962-75 is shown as curve 3 (scale to the left) for the entire, outer
anomalous region as defined in Fig. 1.

O 198lNature Publishing Group

Nature Vol. 289 22 January 1981

I

233
I

Ms = 7.2 ~ a
earthquake

Average normal

ally. Because the microearthquake activity in this region could

not have accumulated the 10-bar stress drop throughout the
outer anomalous volume, we hypothesize that a seismic fault
creep along the nearly horizontal fault plane at 9-km depth
caused the strain relaxation and the relative seismic quiescence.
The stress decrease also explains the absence of travel time
anomalies in the outer volume. The two inner anomalous
volumes both have a constant and high seismicity rate. As this
contrasts with the observations in the strain-softening outer
anomalous volume, we suggest that the high seismicity rate
indicates high stress in the inner anomalous zones. This interpretation is supported by different observations in the eastern
and western volumes. In the eastern volume, near the point of
initial rupture, the travel time anomaly indicates the presence of

4.
I

Time (yr)
Fig. 4 Cumulative number of earthquakes (M2 2.0) in the south

flank of Kilauea volcano, Hawaii, as a function of time. The solid

curve, representing seismicity in the outer precursor volume (see
Fig. 1). shows a sharp (-50%) decrease in 1972. The dashed curve
indicates roughly constant seismicity rate in the inner precursor
volumes.
The strain release along line 1 (1971) and line 2 (1974) is
interpreted to indicate that the respective southeastern end
points of these lines were then located above the outer
anomalous volume. Therefore the outer anomalous volume was
apparently expanding at the expense of the inner anomalous
volume.

The asperity model

The model for the preparatory process for the Kalapana earthquake must be complex, because different precursory patterns
existed in subregions of the future rupture area. The characteristics of the outer anomalous zone are: decreased seismicity
rate, strain relaxation and the absence of a travel time anomaly.
One o r both of the eastern (E) o r western (W) inner anomalous
zones, exhibited: constant seismicity rate until shortly before the
main shock when the rate increased (E and W), a significant
P-wave velocity decrease (E),the largest surface displacements
during the main shock6 (W) and an increase of compressive
stress at least up to the fall of 1974 (W).
The decrease of seismicity in the outer volume seems to be
directly related to the stress decrease observed there geodetic-

I
I

MI = 7.2 Kalapana earthquake

I

68

70

72

74

Time (yr)
Fig. 6 Schematic summary of crustal parameters during the 8 yr

before the Kalapana earthquake. Arrows at the top indicate times

and relative size of volcanic intrusive episodes. The top curve
shows geodetically measured horizontal strain increase in the
western major asperity volume (m) and strain relaxation in the
The dotted lines represent a possible
outer precursor volume (0).
alternate strain history between 1972 and 1975 when no
measurements were made. The second curve shows the seismicwave travel-time anomaly observed in the eastern asperity and the
absence of such an anomaly in the outer precursor volume. The
third curve shows the seismicity rates in the inner and outer
precursor volumes (from Fig. 3). The final graph illustrates that the
(normalized) average distance of earthquakes from the location of
the large foreshock decreased beginning in 1972, which is an
alternate way of quantitatively measuring the contrast between
seismic activity in the inner and outer anomalous areas. Note that
the anomaly onsets of the independent parameters occur at
approximately the same time 1971 1 yr. This strongly supports
the idea that all of these events are related to the main-shock
preparatory process, which may have been initiated by the 1972
large eruption/intrusion event.

Time (yr)
Fig. 5 Strain accumulation on Kilauea's south flank from 1914
to 1975 (upper curve). 0,Averages from the geodetic line 1 and 2

(Fig. 1);0,strain along line 2. The cumulative eruption volume is

shown for comparison.The strain direction is approximately southeast and the loading cycle is assumed to have started in 1868 after
the last large earthquake on the south flank.

dilatancy, which requires high, deviatoric stresses. In the western volume, the stress increase of 30 bar between 1971 and 1974
shows that elevated stresses were present at that time. Thus, our
observations indicate that two areas of high stress accumulation
exist surrounded by a larger area where aseismic slip is occurring
o n the future rupture plane. By analogy with rock mechanics
studies, we term these small volumes of high-stress concentrations 'major asperities'.
The geodetic line 1 within the eastern asperity showed extension (relaxation) starting in 1970, and the line within the western
asperity extended in early 1975 (Figs 5 and 6). This may mean

O 198lNature Publishing Group

Nature Vol. 289 22 January 1981

234
that these two lines were contained entirely within their asperities until 1970 and 1975 respectively. After these times the
creeping portions of the fault plane extended under the respective southern geodetic stations and the associated line began
to lengthen. The central and northern parts of the asperities
may have been locked until the time of the main shock.
Unfortunately there is no geodetic data during the crucial period
1971-74, and the exact time of strain reversal is unknown.
This model of highly stressed asperities embedded in a strainsoftening region of lower stress was derived from precursory
geophysical data accumulated from the rupture zone of the
Kalapana earthquake. It could, however, be considered a corollary of the known multiple-event nature of the main rupture
(Fig. 2). Because the main rupture indicates that high-stress
volumes were mixed with low-stress ones, the state of stress
before the main shock must also have been one of contrasting
high- and low-stressed volumes.
Based on laboratory observations, models similar to ours
including strain-softenhgls and hard. inclusions within strainsoftening s u r r ~ u n d i n g s have
' ~ ~ ~been
~ proposed.

Conclusions
If our model is an accurate representation of the preparatory
process to major earthquakes, it may prove difficult to detect
high-stress-related earthquake precursors because they may be
developed only in small portions of the precursor volume. If our
observations can be scaled linearly to smaller and larger earthquakes in other tectonic environments, we expect that for a
5-mag earthquake the major asperities would have radii of
-1.5 km which may explain why no clear velocity anomalies
have been detected for such events. For great earthquakes of
Received 9 June; accepted 26 September 1980.
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.

Savarensky. E. F. Tecronopbysics 6, 17-27 (1968).

Semenov, A. N. Bull. Acad. Sci. USSR, Phys. Solid Earrh 3,245-248 (1969).
Whitcomb, J. H., Garmany, J. D. & Anderson, D. L. Science 180,632435 (1973).
Scholz, C. H., Sykes, L. R. & Aggarwal, Y. P. Science 181,803-810 (1973).
Nur. A. Bull. seism. Soc. Am. 62, 1217-1222 (1972).
Tilling, R. I.eral. U.S. Geol. Sum. Circ. No. 740, 33, (1976).
Kanamori, H. & Fuis, G. Bull. seism. Soc. Am. 66,2017-2037 (1976).
Bolt, B. A. Bull. seism. Soc. Am.67, 27-32 (1977).
Lindh, A. G., Lockner, D. A. & Lee, W. H. K. Bull. seism. Soc. Am. 68,721-734 (1978).
Ando, M. I. geophys. Res. 84.76167626 (1979).
Swanson, D. A,. Duffield. W. A. & Fiske. R. S. US. Geol. Sum. Profess. Pap. 963 (1976).

several hundred kilometre source length, the asperities might

have radii of -40 km, which means that high-stress precursors
might be detected with dense observation networks. The relationships between the number of asperities and earthquake size
is not known. Our data do not exclude the possibility that large
earthquakes contain numerous asperities with radii smaller
than 5 km.
The precursor time for the Kalapana earthquake agrees
approximately with the precursor time versus magnitude relations proposed e ~ s e w h e r e ~ . ~We
, ~ ' . have no evidence for
diffusion of pore fluids, so the precursor time will more likely be
a function of the ratio of locked-fault area to creeping-fault
surface.
We conclude that three independent precursors (Fig. 6) have
been observed for the Hawaii earthquake, which constrain the
anomaly onset to approximately 1971+ lyr. They are of two
types: high-stress precursors (decrease in P-wave velocity and
foreshocks) and by contrast low-stress precursors, here
observed as a 50O/0 decrease in seismicity and strain relaxation.
The high-stress phenomena are restricted to small volumes embedded in a larger volume characterized by lower stresses. The
main-shock nucleation point was located in one of the major
asperities and the area of largest surface deformations in the
other. Our precursor model of hard asperities surrounded by
creeping-fault area is analogous to the multiple rupture model
for earthquakes. We suggest that future earthquakes may be
predicted if patterns similar to those described here are
observed.
We thank Robert Koyanagi and the staff of the Hawaiian
Volcano Observatory for collecting and organizing data. This
work was supported by NSF grant EAR 780-3633.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.

Wyss, M. & Brune, I. N. Bull. seism. Soc. Am. 57, 1017-1023 (1967).
Trifunac, M. D. & Brune, J. N. Bull. seism. Soc. Am. 60,137-160 (1970).
Kanamori, H. & Stewart, G. S. I. geophys. Res. 83,3427-3434 (1978).
Johnston. A. C. Science 199, 882-885 (1978).
Johnston. A. C. thesis, Univ. Colorado (1979).
Lipman, P. W., Lockwood, J. P., Okamura, R. T.. Swanson, D. A. & Yamashita, K. M. US.
Geol. Sum. Profess. Pap. (in the press).
Mjachkin. V. I.,Brace. W. F.. Sobolev, G. A. & Dieterich, J. H. Pure appl. Geophys. 113,
169-181 (1975).
Brady, B. T. Pure appl. Geophys. 112,701-725 (1974).
Stewart. W. D. Tecronophysics 5 2 , 6 1 3 4 2 6 (1979).
Rikitake, T. Devl. SolidEanh Geophys. 9, 357 (1976).
Rojahn, C. & Morill, 8 . 1 . Bull. seism. Soc. Am. 67,493-515 (1977).
Wyss, M., Klein, F. W. & Johnston, A. C. I. geophys. Res. (in the press).

Upper mantle seismic anisotropy and

lithospheric decoupling
J. H. Leven, I. Jackson & A. E. Ringwood
Research School of Earth Sciences, Australian National University, PO Box 4, Canberra 2600, Australia

The lithosphere underlying stable continental shield regions is thought to extend to a depth of -200 km. Compressional
wave velocity models derived from a recent approximately NS refraction profile in northern Australia include a pronounced
high-velocity zone at a depth of around 200-250 km. Conventional petrological interpretations in terms of mineralogical or
compositional changes cannot readily explain this feature. Calculated velocities for the garnet pyrolite model (using a
derived geotherm), fit the velocities above and below this high-velocity zone, but cannot account for the high-velocity
(-8.6 km s-') zone itself. This feature is explained in terms of velocity anisotropy, which might reflect mechanical
decoupling of the continental lithosphere from the underlying mantle.
IT has been suggested that the lithosphere underlying continental shield regions extends to depths of the order of 200 km, and
that the lithosphere (tectosphere)'-3 beneath stable continental
shields is depleted in low melting point components, is cooler
than the normal mantle material at this depth, and hence would
translate with the continental plates by virtue of its higher
viscosity. Body and surface wave studies of the upper mantle
structure beneath shield regions have not indicated any

significant low-velocity zone above a depth of 200 km. A

detailed study of the shear wave velocity structure beneath the
Australian shield4 has not found evidence for any region of low
shear wave clocity above a depth of 150 km.
~ e h m a n n "first
~ postulated the existence of a seismic discontinuity in the upper mantle at a depth of around 200 km. This
feature has subsequently been observed in many upper mantle
refraction surveys, commonly at depths shallower than 200 km.
Q

1981 Macmillan Journals Ltd

Nature Vol. 289 22 January 1981

235

Distance (km)

Fig. 1 Vertical component record section of event 256 from the

Randa Sea Arc, with the travel time-distance curves of the model 8
(ref. 9) (dotted line) and CAPRI" (solid line) velocity models
superimposed. The branch labelled 200 km corresponds to rays
refracted below the 200 km discontinuity. This branch displays a
rapid decrease in amplitude beyond an epicentral distance of
1,400 km, corresponding to the onset of a negative velocity
gradient beneath the 200 km discontinuity (see Fig. 2).

~ n d e r s o n ' named this upper mantle feature the Lehman discontinuity. HalesR suggested that the relative motion of the
Australian continental lithosphere over the underlying mantle
material occurs "below a discontinuity in the P-wave velocity at
a depth of about 200 km". Continents can move with comparable velocities to oceanic plates, so where is this zone of relative
motion of the continental lithosphere over the underlying
mantle and what is the relation, if any, of the '200 km' discontinuity to this zone?

Seismic observations
Hales et al.' analysed data from a long range upper mantle
refraction profile conducted in Central Australia. Their upper
mantle velocity model (model 8 ) has a velocity discontinuity at a
depth of 200 km, with an increase in the compressional-wave
velocity of 0.38 km s-I. The depth of this feature has been
constrained by the recorded arrivals of an event o n the 14
October 1977 (event no. 256). The record (Fig. 1, see also Fig.
14 of ref. 9 ) of this event, which has been located in the depth
range of 176 5 km (on the basis of 164 readings), indicates that
the 200 km discontinuity is significantly below the 176 km
depth. The method of Gutenbergg." constrains the velocity at
the source depth of this event to be ~ 8 . 2 km
8 s-I. Beneath the
200 km discontinuity, the velocity is well constrained to be
8.62 km s-', indicating that the velocity increase across the
200 km discontinuity is >0.3 km s-'.
The 200 km prograde refraction branch, corresponding to
rays which bottom below the 200 km discontinuity, shows an
abrupt dccrease in amplitude at an epicentral distance of around
1,700 km (depending on the source depth). Figure 12 of ref. 9
shows a fit of the travel time-distance curve derived from their
model 5, to the record section for event no. 256. Model 5 did not
incorporate any discontinuous increase in velocity around a
depth of 200 km, so that this figure indicates the ease with which
the 200 km discontinuity can remain undetected, especially
when only shallow events are studied. Hales et al.' modelled the
200 km structure as a velocity knee with a layer having a
negative velocity gradient below. Leven", using synthetic
seismograms to model amplitudes, showed that the amplitude
decay of the 200 km prograde branch can be explained by a low
velocity zone which starts more sharply, around 30 km beneath
the 200 km discontinuity. The exact nature of the velocity
profile within the low velocity zone cannot be determined from
present refraction data. However, synthetic seismogram
modelling indicates that the negative velocity gradient com-

mencing at a depth of 230 km must be sharper than

that of model 8 (ref. 9), to explain this rapid decrease
in amplitude near 1,700 km distance.
A sharp velocity increase at a depth of around
200 km, followed beneath by a velocity decrease, is a
peculiar seismological feature. Similar high-velocity
zones in the velocity models of other upper mantle
studies can be found-but
usually at shallower
depths. For example, Lewis and Meyer12 have a
comparable structure at a depth of around 130 km,
while in the model of Wiggins and Helmberger13, this
structure is at a d e ~ t hof around 150 km. The abrupt
decrease in amplitude of the 200 km branch has algo
been observed in other upper mantle refraction
profiles, although the low velocity zone which is
implied by this decrease has not been incorporated in
the models resulting from these studies. Examples of
this abrupt decay in the amplitude of the prograde
200 km branch can be clearly seen in Fig. 8 of ref. 1 4 and in Figs
11 and 12 of ref. 15. However, no convincing evidence of this
type of velocity structure has been observed in the data of any of
the Early Rise experiment profiles, which traverse North
America with an excellent coverage of azimuth16.
The existence of a negative velocity gradient a t around
200 km depth is supported by the S to P conversion evident in
records from teleseismic events. Sacks et a/.'' studied these
converted phases recorded at the NORSAR array, and have
interpreted precursors to the S onset as being S to P conversions
from a negative velocity gradient at a depth of 250* 15 km.
Long range refraction surveys for oceanic paths have also
indicated a similar velocity structure, but at a considerably
shallower depth. Both Asada and ShimamuraI8 and Hales et
al." have an analogous feature in the velocity models derived
from oceanic refraction surveys. These velocity models both
have a discontinuous increase to a compressional velocity of
around 8.6 km s-I, immediately above a low-velocity zone, the
depth of this high-velocity zone in both cases being around
75 km. Leeds and other^^".^' have derived a model for the
variation of the lithospheric thickness over the Pacific basin from
surface wave studies. Their model indicates a thickening of the
oceanic lithosphere from around 7 6 km in the younger regions,
to 104 km in the older regions of the oceanic basin.

Conventional interpretations
Traditionally, seismic discontinuities in the Earth's mantle have
been explained in terms of isochemical changes in mineralogy,
changes in chemical composition, o r a combination of both. The
possible relevance of such phenomena to the 200 krn discontinuity is now reviewed.
(1) Exsolution of M:+Al2Si3Ol2(M = Mg, Fe, Ca) garnet from
aluminous pyroxenes. The pressure-sensitivity of the A1203
content of pyroxenes coexisting with garnet provides the basis of
pyroxene g e ~ b a r o m e t r ~ ~Along
~ . ~ ! the 1,200 "C isotherm, for
example, the A1,03 content of orthopyroxene coexisting with
garnet in both synthetic and natural systems decreases, rather
rapidly from 5-7% near 2.5 GPa to 1.5-2.5% near 4 GPa, and
then more gradually to -0.7% at 7 GPa and 0.4%, in Ca-poor
clinopyroxene at 1 4 GPa [ref. 231. Green and R i r ~ ~ w o (see
od~~
also ref. 25) have shown that the orthopyroxene AI2O3isopleths
in the garnet pyrolite stability field are subparallel to the Clark
and R i n g ~ o o dPrecambrian
~~
shield geotherm-suggesting
approximate constancy (at the 1.0-1.5Y0 level) of orthopyroxene alumina content over the 2-5 GPa pressure range.
The low alumina content of such orthopyroxene and the evidence for its gradual reduction with increasing pressure beyond
5 GPa exclude any possibility that exsolution of pyrope-rich
M:+Al2Si,Ol2 (M = Mg, Fe, Ca) garnet from aluminous orthopyroxene is responsible for the discontinuous increase in
compressional wave velocity near 200 km depth.
(2) Formation of complex M:' "'(At2, ~ ~ + S i ) Sgarnet
i ~ 0 ~ ~
solid solutions. Ringwood's2' study of the crystallization of a
glass of 90% MgSi03-10/~A1203composition revealed a rapid

O 198lNature Publishing Group

236

Nature Vol. 289 22 January 1981

increase in both the garnet lattice parameter and the proportion

of garnet near 10 GPa. More recently Akaogi and AkimotoZ8
have confirmed the high pressure solubility of MSiO, pyroxene
in the garnet structure, but suggest that such solution occurs
gradually over the pressure range 4-14 GPa, with a pronounced
increase in the proportion of garnet (and hence the density and
P-wave velocity) in the 14-19 GPa pressure interval. The fact
that the garnet lherzolite nodules of deepest origin (-200 km
depth, 7 GPa pressure) brought to the Earth's surface by the
~ e s o t h o ' ~Kao2'
,
and other kimberlite pipes contain 3.02-3.07
silicon atoms per 12 oxygen atoms is marginal evidence for the
initiation of M:'(AI2)Si3Ol2- M:'(M"S~) Si3012solid solution
near this depth. However, neither laboratory experiments nor
the garnet-bearing nodules indicate any abrupt increase in the
M:'(M2'Si)Si3OI2 content of garnet near 200 km depth in the
Earth's mantle. This, coupled with the absence of other
mineralogical changes in garnet lherzolite in the relevant pressure range, suggests that phase transformations are not responsible for the rapid increase in P-wave velocity near 200 km depth
in the sub-continental mantle.
(3) Compositional change: harzburgite to garnet Iherzolite.
The long and complex geochemical history of subcontinental
uppermost mantle, the geochemistry of ultramafic nodules from
kimberlite pipes, and the absence of significant seismic lowvelocity zones beneath Precambrian shields suggest that the
sub-continental uppermost mantle is likely to be strongly (but
variably) stripped of its low melting-point fraction. Under these
circumstances, a transition would be expected, perhaps near
200 km depth, from barren harzburgite above, to fertile garnet
lherzolite be low'^". The densities and P-wave velocities for
these assemblages have been estimated to be 3.31 g cm-, and
8.32 km s-' for harzburgite and 3.38 g c W 3 and 8.38 km s-' for
pyroliteZ5.~ o r d a n 'has
~ shown that removal (by 20% partial
melting) of the basaltic fraction of pyrolite, decreases the density
of the residium by -0.05 g ~ m - but
~ , leaves the P-wave velocity
almost unchanged. The small change in P-wave velocity
accompanying such differentiation is due to the near-cancellation between the velocity decrease associated with garnet
removal and the velocity increase associated with the more
magnesian composition of the residual olivine and orthopyr~xene~~.
Clearly, although the subcontinental upper mantle must be
chemically zoned as a result of differentiation, it is virtually
impossible that a P-wave velocity discontinuity of the order of
0.3 km s-' near 200 km depth, can be explained in these terms.
(4) Compositional change: garnet lherzolite (or harzburgite)
to eclogite. ~ n d e r s o n ~has
. ~ ' suggested that the Lehmann discontinuity might represent a change in chemical composition
from garnet lherzolite (or harzburgite) above the discontinuity
to eclogite below. Such eclogite, whether produced by subduction of oceanic basaltic crust (as in Anderson's model), or by
the necessarily relatively slow crystallization of basaltic magmas
within the mantle, would be expected to possess an average
mid-ocean ridge basaltic Mg number (100 Mg/(Mg+Fe2+))of
about 60 (see refs 31,32). In the former case, the basaltic major
element chemistry might be perturbed during subduction by
partial melting in the quartz eclogite stability field, resulting
principally in significantly increased (MgO+FeO)/CaO in the
residium (see ref. 33). Typical mantle-derived eclogites are
quartz-free, essentially biminerallic aggregates of pyroxene and
garnet (in approximately equal proportions) with Mg numbers
around 60.
Mean compressional-wave velocities for such mantle-derived
e ~ l o ~ i t e s(samples
~ ~ - ~ =11,14, and 15 of ref. 36 but not 8 which is
very magnesian (Mg number 71) and contains 74.0% clinopyroxene, 16.6% garnet, and 7.9% zoisite) span the range
8.27-8.50 km s-' at 1.0 GPa with measured velocity anisotropies of 0.8-3.4%. Shear-wave velocities are more variable
(4.49-4.86 km s-' at 1.0 GPa) with anisotropies of 1.5-4.9%.
The NovC Dvory (Czechoslovakia) eclogite studied by
Christensen to 3.0 GPa is representative of the highest P-wave
velocity eclogites with p = 3.56 g cW3, and extrapolated

compressional and shear wave velocities of 8.33 and 4.59 at

atmospheric pressure. Our preferred parameters for garnet
pyrolite are p = 3.40 cm-,, Vp = 8.31 km s-', and V, =
4.82 km s-' (these differ slightly from those of Jordanz9because
of our use of the pyrope elastic moduli measured by Leitner
et a/.,").
Our analysis thus strongly reinforces other observation^^^'^^^^
concerning the indistinguishability of garnet lherzolite and
eclogite P-wave velocities. We conclude that a change in chernical composition from garnet lherzolite to eclogite might
produce a significant decrease in shear wave velocity but that no
discernable change in compressional wave velocity would be
expected. Under these circumstances, it seems unlikely that the
Lehmann discontinuity reflects a chemical change of this type.
(5) Base of pronounced low-velocity zone. Many seismic
models for upper mantle velocities beneath tectonically active
continental regions (such as the western US) include a substantial and rapid increase in seismic wave velocities in the depth
range 150-200 km (refs 38, 39). In most such models, the
discontinuity represents the base of a pronounced low-velocity
zone, so that a large part of the velocity contrast across the
discontinuity is accounted for by a return to 'normal' upper
mantle velocities (for example, at substantially subsolidus
temperatures) below the discontinuity. The absence of
significant low-velocity zones, shallower than 200 km depth,
beneath Precambrian shield regions precludes such an interpretation for the Lehmann discontinuity.

Seismic-wave velocities for garnet pyrolite

The above arguments demonstrate the difficulty of accounting
for the observed 0.3 km s-I increase in Vp at -200 km in terms
of conventional explanations. To understand the nature of the
seismic feature near 200 km depth, we now compare the CAPRI
velocity-depth model1' with calculated P-wave velocities for an
assumed garnet pyrolite upper mantle.
For this calculation, we assume that the compressional-wave
velocity (Vp of 8.31 f0.05 km s-' at STP) varies linearly with
10 GPa) and temperature (to
1,400 O C ) .
pressure (to
Because volumetric strains will be no greater than 8% at
10 GPa, finite strain equations of state should not be necessary.
The general tendency for both pressure and temperature
derivatives of the elastic-wave velocity to decrease with increasing pressure and temperature suggests that linear extrapolation,
using derivatives measured near STP, may result in slight
overestimation of velocities at the upper end of the extrapolation. Any relaxation of ultrasonically determined shear moduli
for upper mantle seismic-wave propagation would similarly
result in an overestimation of the seismic velocities.
The pressure and temperature derivatives, used in the calculation:

represent a weighted average (57% olivine, 17% orthopyroxene, 12% clinopyroxene, 14% garnet) of the mineral
pressure and temperature derivatives. Data for Stillwater bronzitite (aVp/aP = 0.137 km s-' GPa-' (refs 35, 40), are used in
preference to the much higher value aVp/aP=
0.207 km s-' GPa-' derived from a study of single-crystal
bronzite4'. The latter value is grossly inconsistent with the
expectation4' that (11Vp)(aVp/aP)is approximately equal to

This relation yields an expected aVp/aP of 0.138 km s-' GPa-',

in agreement with the measured value for Stillwater bronzitite.
Furthermore, even if the unusually large single-crystal pressure
derivatives (dK/dP = 9.6) were, in fact, appropriate at STP, it
seems improbable that such a high derivative would be maintained over a substantial pressure range. Use of Frisillo and

O 198lNature Publishing Group

237

Nature Vol. 289 22 January 1981

Velocity (km s-'1

oceanic lithosphere, in terms of the calculated elastic properties

of petrologically appropriate isotropic aggregates.
The match between the calculated velocities and the CAPRI
model, other than in the 190-225 km depth range, precludes
neither a more barren harzburgitic chemistry above 200 km, nor
a gradual increase in the M:'(M~'S~)S~,O,~component of the
garnet with increasing depth below 200-250 km.

Seismic velocity anisotropy?

We are clearly concerned, not with a seismic discontinuity in the
usual sense, but rather with a zone of the order of 50 km thick,
characterized by higher velocities than both overlying and
underlying mantle.
W e have demonstrated the difficulty involved in explaining
the 0.3 kms-' velocity increase near 200 km, in terms of
mineralogical o r chemical changes. The negative velocity
eradient below 230 km is also not readilv- exdained
in con.
ventional terms. Leven" inferred that this velocity gradient
must be more negative than -0.002 km-' s-' km-' to produce
synthetic seismograms consistent with the observed rapid
decrease in first arrival amplitudes near A = 1,700 km. Using
P-wave velocity pressure and temperature derivatives appropriate for garnet pyrolite, we find that an unusually large (but not
impossible) temperature gradient, of at least 9 OC km-', would
be required if the negative velocity gradient below 230 km
depthy, were to be explained in terms of the combined effects of
pressure and temperature.
O n the other hand, this sub-continental 'high-velocity zone'
bears a strong qualitative resemblance to shallower high-velocity lids capping low-velocity zones in many velocity models for
the sub-oceanic upper mantle. These considerations, and the
variability of its observation suggest that this unusual seismic
feature might be explained in terms of seismic velocity anisotropy associated with deformation-induced preferred orientation of olivine (and possibly orthopyroxene) crystals in a shear
zone, at 200-250 km depth, which is responsible for decoupling
of the continental lithosphere from the underlying mantle. This
hypothesis is supported by the following observations.
(1) Extensive field and laboratory studies of deformed peridotites from the Lanzo massif in Northern Italy (believed to
represent subcontinental mantle), have established a firm relationship between the observed 7 % velocity anisotropy, the
fabric and the flow coordinate^'^^^'. The velocity anisotropy is
associated with preferred orientation of the [loo] axes of olivine
crystals at a low angle of inclination (6) = lo0, Fig. 3) to the flow
direction during high temperature (- 1,200 "C) deformation at
an inferred strain rate of 10-l4 s-'.
(2) Experiments conducted over a wide range of temperature
and strain
have demonstrated the development of
strong preferred orientation of olivine ([O1O]//ul and
[100]//u3) as a consequence of laboratory deformation of
dunite. At high temperatures and moderate strain rates
(moderate flow stress), the preferred orientation seems to be
derived primarily from syntectonic recrystallization driven by
strain accumulation, due principally to dislocation glide on the
system (010)[100]. The coincidence of principal axes of stress
and strain in most laboratory deformation experiment^^^.^^
results in an ambiguity concerning predicted fabric orientation
in natural triaxial deformations which may resolve the apparent
conflict between the orientations of the natural and laboratory
fabrics mentioned above.
(3) Studies of the azimuthal variation of cornpressional-wave
velocities in the oceanic upper mantles5.56have established the
velocity anisotropy of the sub-oceanic mantle. The fast direction
in each case is sub-parallel to the spreading direction, that is
normal to the mid-ocean ridge. The azimuthal range in P-wave
velocities (-0.3 km s-') and the association of fast velocities
with the direction of plate movement over underlying mantle are
both appropriate for explanation of the high P-wave velocities at
200-250 km depth along a NS profile beneath the Australian
plate (Fig. 3).

Fig. 2 A comparison of the model 8 and CAPRI compressionalwave velocity4epth models with calculated velocities ( D )
for a
garnet pyrolite upper mantle based on the geotherm presented in
the text.

Barsch data for orthopyroxene9 would increase the garnet

pyrolite aVp/dP to 0.128 km s-' GPa-' and hence raise the
velocity at 200 km depth by -0.14 km s-'. Further experimental work on single-crystal and polycrystalline orthopyroxene is
needed to resolve this uncertainty. Olivine and garnet pressure
and temperature derivatives were taken from refs 43-45. A
clinopyroxene (aVp/aP) of 0.119 km s-' GPa-' was obtained by
comparing christen sen'^^^ data for Nov6 Dvory eclogite with
data for garnet. Otherwise, it was assumed
the Bonczar et
that the orthopyroxene derivatives also apply to clinopyroxene.
In the initial calculations we used the geotherm of Harte4". The
resulting compressional-wave velocity profile was broadly
consistent with the model 8 of Hales er aL9 and the CAPRI
model in the 75-190 km and 225-320 km depth ranges but
failed to explain the high velocities in the 190-225 km interval.
Variation of the geotherm within reasonable limits29allowed a
more satisfactory fit (Fig. 2) to the CAPRI velocities in the
75-190 km and 225-320 km depth ranges. The required geotherm
depth (km):
temperature PC):

100; 150; 200; 250; 300

770; 1,050; 1,180; 1,320; 1,420

is between, and qualitatively similar to, Jordan's2" S1 and S2

conductive geotherms, and would intersect adiabats for the
oceanic upper mantle at depths between 300 and 400 km.
The comparison presented in Fig. 2 implies that the garnet
pyrolite composition and mineralogy, combined with a reasonable geotherm, explains compressional-wave velocities for all
but a narrow interval (190-225 km) of the 75-300 km depth
range. Similar difficulties in explaining high velocities just below
200 km are evident in the studies of Graham4', and of Ahrens4'
who noted that none of his calculated models "predict the
8.65 km s-' P-velocity maximum. . . at 200 km depth in profile
HWNE" (of Wi ins and HelmbergerI3). There have been
similar commentsf! on the difficulty of explaining compressionalwave velocities 2 8 . 5 kms-I in some seismic models of the

O 198lNature Publishing Group

Nature Vol. 289 22 January 1981

238
(4) The seismic implications of the development of such an
olivine preferred orientation are more fully explored in Fig. 3.
For small inclinations of the fast [loo] axis to the flow direction
(N), propagation directions within 45' of the flow direction
display compressional (or quasi-compressional) mode velocities
significantly higher than 8.42 km s-' which is the Voigt-ReussHill (VRH) average. Note that the considerable shear wave
birefringence-the difference in quasi-shear mode velocities is
as much as 0.7 kms-' for some propagation directions.
Thorough characterizations7 of an essentially monominerallic
sample of Twin Sisters dunite with a strong preferred orientation
revealed a high degree of seismic anisotropy (7.98 s V , s
9.24 km s-', 4.57 s V , S 5.01 km s-') and shear wave birefringence (maximum 8Vs= 0.38 km s-'). In upper mantle garnet
peridotite, the magnitude of these effects will be reduced
because of the reduced proportion of olivine. Nevertheless, the
results of Fig. 3 provide a firm semiquantitative basis for testing
the preferred orientation hypothesis.
The model provides a qualitative explanation of the high
compressional-wave velocities for approximately NS propagation at depths of 200-250 km in the Australian upper mantle.
The results presented in Fig. 3 (see especially 8 = g5 = 22.5')

Flow directlon

Mode velocities and displacement

vectors in olivine o r function8 of
prqxgotion direction 4
0 ="/a
0:n/4

suggest that significant shear wave birefringence might also be

observed (on both horizontal and vertical components) at
appropriate azimuths.
(5) Both the chemistries and textures of deep-seated
ultramafic modules contained in kimberlite pipes are consistent
with the view that decoupling of the continental lithosphere and
the underlying mantle is concentrated at depths near 200 km
(ref. 58). The observation that ultramafic nodules of deepest
origin (-200 km), brought to the surface by South African
kimberlites varying in age by more than 1,000 Myr (refs 22,59)
display more primitive chemistry than their shallower counterparts, suggests that the former are derived from a zone of the
mantle which is continuously replenished from a large mantle
reservoir. Furthermore, while the highly sheared mosaic textures of the deepest-seated xenoliths are probably not
representative of steady-state mantle deformation processes60,
the contrasting fabrics of these and the relatively undeformed
coarse-grained xenoliths4' of shallower origin, are suggestive, at
least, of very different modes of xenolith incorporation and
magma transport.
(6) The combination of more primitive chemistry below
200 km depth and the general characteristics of upper mantle
geotherms suggests that the zone between 200 and 300 km
depth may be characterized by higher homologous temperatures
(TIT,,,) for olivine than for the shallower and the deeper mantle.
Under these circumstances, this region is a plausible zone for
strain concentration, although the relative roles of dislocation
creep-syntectonic recrystallization (which may be subject to
strain hardening) and grain boundary sliding6', remain to be
determined. Anderson3' similarly concludes that the viscosity of
subcontinental mantle should attain a minimum value near
200-250 km depth.

Conclusions
We have demonstrated the difficulty involved in explaining high
compressional-wave velocities for approximately NS propagation at 200-250 km depth in the Australian upper mantle in
terms of various suggested variations of mantle mineralogy and
chemistry. Having rejected conventional interpretations, we
suggest that the observed high P-wave velocities may derive
from preferred orientation of olivine crystals in a deformation
zone responsible for the mechanical decoupling of the lithosphere from the underlying mantle.
Refraction profiles of the Australian upper mantle9.62"4,
while in general agreement on the need for a substantial increase
in compressional-wave velocity near 200 km depth for profiles
within 45' of NS, provide limited information for other azimuths. The only indication of a possible 200 km discontinuity on
an approximately EW profile is provided by a single first
arrivalb5 at A 2,100 km (Meekatharra, Western Australia)
from an explosion at Mt Fitton (South Australia). This datum,
while possibly compatible with NS profiles, does not provide a
strong independent constraint on the velocity below 200 km
depth for EW propagation azimuths. Clearly, the anisotropy
hypothesis needs further testing. Carefully controlled experiments over a wide range of azimuths (such as those conducted on
the suboceanic mantles6) will be required to test for anisotropy
and possible shear wave birefringence (see ref. 66) of the
continental upper mantle.
However, observation of subcontinental upper mantle seismic
anisotropy may be subject to considerable variability. Spatial
variation in the abundances of key volatile species (such as
H 2 0 ) , and of mantle geotherms and strain rates may result in
variability of the dominant upper mantle deformation mechanism, with some plausible flow mechanisms (such as grain
boundary sliding) having little tendency to produce a strong
preferred orientation. Even where an appropriate mechanism
(such as dislocation creep-syntectonic recrystallization) dominates, the development of an observable seismic anisotropy
may depend on a long, consistent and current history of relative
motion of the lithospheric plate with respect to underlying
mantle.

Fig. 3 Seismic anisotropy and shear-wave birefringence of

olivine. The crystal orientation is the expected preferred orientation of olivine (see text) in the Australian upper mantle: Row
plane horizontal, plate movement over underlying mantle due
North. The olivine c-axis (001) lies EW while the inclination 8 of
the a-axis (001) to the flow direction is varied between 0 and 4 4
(see text). Mode velocities and displacement vectors for horizontal
propagation vectors k are calculated as functions of azimuth 4 and
inclination 8 from the single crystal moduli43. The particle displacement vectors for the three independent propagation modes
for each (4, 8) pair are plotted (e) on a stereographic projection and labelled with the appropriate velocities (in km s-').
Note that off-axis propagation directions involve modes for which
particle displacement is neither parallel nor orthogonal to the
propagation vector. Such coupled modes are labelled QP, OSH,
QSVb7.The results may be applied in the SW quadrant by virtue of
the mirror symmetry of olivine about the plane normal to 001.

O 198lNature Publishing Group

Nature Vol. 289 22 January 1981

-

--

Such anisotropy beneath both continental and oceanic

lithosphere, would facilitate seismic mapping of the vertical
extent of the lithosphere, and provide considerable insight into
the rheological behaviour of the upper mantle.
We thank P. N. Chopra, D. A. Gust, A. L. Hales, and K. J.
Muirhead for useful discussion.

30. Anderson. D . L. Geophys. Res. Letr. 6,433436 (1979).

31. Cann, 1. R. Phil. rrans. R. SOC.A268.495-505 (1971).
32. Melson. W. G., Vallier, T. L.. Wright, T. L.. Byerly. G. & Nelen. J. i n 7he Geophysics
of rhe Pacific Basin and its Margin (eds Sutton. G. H . er 01.1 ( A G U Monogr. No. 19.
1976).
33. Stern,C. R. and Wyllie. P. J. Am. Miner. 63,641-663 (1978).
34. Kanamori, H . & Mizutani. H. Bull. Earthquake Res. Insrn. 43, 173-194 (1965).
35. Christensen, N. I.I. geophys. Res. 79.407412 (1974).
36. Manghnani. M . H., Ramananantoandro. R. & Clark. S. P. I. geophys. Rec. 79,5427-5446

Received 26 June: accepted IINovember 1980.

37. Leitner, B. J., We~dner,D. I.

& Liebermann, R. C. Phys. Earrh planer. Inrer. 22, 111-121
(1980).
38. Archambeau,C. B., Flinn, E. A. & Lambert, D. G. I. geophys. Res. 74,5825-5864 (1969).
39. Helmberger, D. V. & Wiggins, R. A. I. geophys. Res. 76,3229-3245 (1971).
40. Wang, C. I. geophys. Res. 79.771-772 (1974).
41. Frisillo. A. L. & Barsch. G. R. I. geophys. Re%77, 6360-6384 (1972).
42. Birch, F. i n The Earrh'sCrusr and UpperManrIr (ed. Hart, P. 1.) 13 (American Geophysical
IJnion. Washington DC, 1969).
43. Kumazawa, M . & Anderson, 0. L. b geophys. Res. 74,5961-5972 (1969).
44. Graham, E. K . & Barsch, G. R. I geophys. Res. 74,5949-5960 (1969).
45. Bonczar, L. I..
Graham, E. K . & Wang. H. I. geophys, Res. 82,2529-2534 (1977).
46. Harte. B. Phil. Trans. R. Soc. A288.487-500 (1978).
47. Graham. E. K. Geophys. I. R. asrr. Soc. LO, 285-302 (1970).
48. Ahrens, T. J. Phys. Earrh planer. Inar. 7. 167-186 (1973).
49. Green, D . H . & Liebermann, R. C. Tecronophysics 32.61-92 (1976)
Nicolas, A. & Stevenson, P. R. I geophys. Res. 79, 1175-1 182 (1974).
50. Peselnick, I*.,
5 I.Nicolas. A. & Poirier, J. P. Crysmlline Plasticity and Solrd-StareR o w i n Meramorphrc Rocks
(Wiley-lntersc~ence. London. 1976).
52. Ave'Lallemant, H . G. &Carter, N. L . BUN, geol. Soc. Am. 81, 2203-2230 (1970).
53. Nicolas, A.. Boudier, F. & Boullier, A. M . Am. I. Sci. 273, R53-876 (1973).
55. Hess, H. Narure 203,629631 (1964).
56. R a w R. W., Shor, G. G., Francis, T. I.
F. & Morris, G. B.I. geophys. Res.74.3095-3106
(1969).
57. BabuSka, V. I. geophyr. Res. 77,69554965 (1972).
58. Boyd, F. R. Geochim. cosmochim. Acra 37, 2533-2546 (1973).
59. Danchin, R. V. i n The Mantle Sample: Inclusions i n Kimbrrlrres and Orher Volcankr (eds
Boyd. F. R . & Meyer. H. 0. 4.)(American Geophysical Unlon. Washington DC, 1979).
60. Goetze. C. Geology 3. 172-173 (1975).
61. Chopra. P. N. & Paterson. M . S. Tecronophysrcs (in the press).
62. Muirhead. K. J.. Cleary. J. R. & Finlayson. D . M . Geophys. I. R. asrr. Soc. 48, 509-519
(1977).
63. Denham, D.. Simpson. D . W.. Gregson. P. J. & Sutton, D . J.. Geophgs I. R. asrr. Soc. 28,
225-235 (1972).
64. Simpson. D . W. thesis. Australian National Univ. (1973).
65. Finlayson, D . M., Cull, J. P. & Drummond. B . J. I. geol. Soc. Ausr. 21,447458 (1974).
66. Hirn. A . Geophys. I. R. asrr. Soc. 49.49-58 (1977).
67. Crampin, S. Geophys. I. R. asrr. Soc. 49.9-27 (1977).

. .- ,

11976)
*
.
\

Ringwood. A. E. I. geophys. Rep 67,857-867 (1962).

J0rdan.T. H . Rev. Geophys. Space Phys. 13, 1-12 (1975).
Jordan, T . H . Narure 274,544-548 (1978)
Hales. A. L.. Muirhead, K. J. & Rynn, J. M . W. Geophys. I. R. asrr. Soc. (submitted).
Lehmann. I. Ann. Geophys. IS, 93-118 (1959).
Lehmann, I.Bull. seism. Soc. Am. 52, 519-526 (1962).
Anderson, D . L. I. geophys. Res. 84,7555-7560 (1979).
Hales, A. L.Drrecror's Report, ResearchSchoolof Earrh SciencesAnnual Reporr (Australian
National University. Canberra, 1977).
9. Hales, A. L., Muirhead, K. J. & Rynn. J. M. W. Tectonophysics 63. 309-348 (1980).
10. Gutenberg, B. Bull. seism. Soc. Am. 3, 223-232 (1953).
II.Leven, J H . thesis. Australian National University (1980).
12. Lewis, B.T. R. & Meyer. R. P. Bull. srism. Soc. Am. 58,565-596 (1968).
13. Wiggins, R.A. & Helmberger, D . V. 3. geophyr. RPS.78, 1870-1880 (1973).
14. Masse, R. P., Landisman, M . & Jenkins, 1, B. Geophys. I. R. asrr. Soc. 30, 19-36 (1972).
IS. Johnson. L . R. I. geophys. Res. '12,63094325 (1967).
16. Warren, D . H . el al. Project Early Rise, Travel Times and Amplirudes (U.S. Geological
Survey Open File Report, Menlo Park. 1968).
17. Sacks, I.S., Snoke, 1. A . & Husebye, E. S. Tecronophysics 56. 101-110 (1979).
18. Asada, T. & Shimamura, H . i n The Geophysics ofrhe Pacific Ocean Basin and rrs Margin
(eds Sutton. G. H. eral.) 19 (American Geophysical U n ~ o n Washington
,
DC. 1976).
19. Hales, A. L., Helsley, C. E. & Nation, J. 8 . L geophys. Res. 75,7362-7381 (1970).
20. Leeds. A. R., Knopoff, L . & Kausel. E. G. Scrence 186,141-143 (1974).
21. Leeds. A. R. Phys. Earth planer. Inrer. 11,61-64 (1975).
22. MacGregor, I.D . i n The Manrle Sample: Inclusions i n Kimberlires and Orher Volcanics (eds
Boyd, F. R. & Meyer, H . 0. A , ) (American Geophysical Union. Washington DC. 1979).
23. Akaogi, M. & Akimoto. S. Phys. Earrh planer. Inter. 19, 31-51 (1979).
24. Green. D. H . & Rlngwood, A. E. Earrh planer. Scr. Lett 3. 151-160 (1967).
25. Ringwood, A. E. Composirion and Perrology of the Earth's Manrle (McGraw-Hill. New
York, 1975).
26. Clark. S. P. & Ringwood, A. E. Rev. Geophys. 2.35-88 (1964).
27. Ringwood, A. E. Eanh planer. Sci. Lerr. 2,255-263 (1967).
28. Akaogr. M . & Akimoto. S. Phys. Earrh planer. Inrer. 15.90-106 (1977).
29. Jordan. T. H . i n The Manrle Sample: Inclusions i n Kimberlires and Orher Volcanics (eds
Boyd. F. R. & Meyer. H . 0.
A,) (American Geophysical Union. Washington DC, 1979).
1.
2.
3.
4.
5.
6.
7.
8.

Regulation of transcription in expressed and

unexpressed mating type cassettes of yeast
Amar J. S. Klar, Jeffrey N. Strathern, James R. Broach* & James B. Hicks
Cold Spring Harbor Laboratory, Cold Spring Harbor, New York 11724

* Department of Microbiology, SUNY at Stony Brook, Stony Brook, New York 11790
The genes that control the a, a and a / a cell types in Saccharomyces are carried on transposable elements known as a and a
cassettes which reside at three different chromosomal loci. Examination of the transcripts by R-looping and filter
hybridization indicates that each cassette is capable of producing two divergent transcripts. Cassettes at the MAT locus are
transcribed constitutively. Transcription of cassettes at H M L and H M R is prevented by trans-acting negative regulators.
THE genes controlling mating type in Saccharomyces yeasts
reside on movable elements, referred to as cassettes, which
become activated on transposition from silent storage sites to
the mating type locus (MAT)'-5. The particular mating type
gene activated in a cell determines the mating type of that cell.
Cells expressing only an a mating type gene are phenotypically
u cells: that is, they can mate with a cells but not with other a
cells. Conversely, cells expressing only an a mating type gene
are phenotypically a cells: they can mate with a but not with a
cells. Cells that express both a and u mating type genesdisplay a
third phenotype (designated a l a ) : they can mate with neither a
nor a cells and, if diploid, can undergo meiosis and sporulation.
All cells contain three distinct genetic loci on yeast chromosome
111 at which mating type genes may reside. At two of these loci,
HML and HMR, the genes are normally unexpressed. Only the
gene present at the M A T locus (either a or a ) is expressed in
wild-tv~ecells.
We describe here observations on the transcription of mating
type cassettes located at both the MAT locus and the HML and
HMR storage sites in a variety of strains. The results of this

U028-0X36/X1/0402394)6$01.00

study have allowed us to refine current hypotheses which deal

with two specific questions of mating type regulation. First, what
is the mechanism by which MATa and MATn products
determine cell type? Second, what is the mechanism which
allows expression of a and n DNA at M A T but keeps them
unexpressed at HML and HMR?
Genetic analyses suggest that MATa and MATa do not code
for all the cell-type specific functions, but determine cell type by
regulating the expression of cell-type specific genes elsewhere in
the genome6. Several of these structural genes involved in the
sexual cycle have been identified as mutations that result in a - or
a-specific defects in mating or as mutations defective in meiosis
and sporulation of a l a cells. Two complementation groups in
MA Ta and one complementation group in MATa have been
identified and their roles in the regulation of a-specific, aspecific and alu-specific genes
MATa 1 is a positive
regulator of a-specific genes (matal mutants are sterile).
MATa2 is a negative regulator of a-specific genes (matu2
mutants express both a-specific and a-specific genes and hence
are sterile). MATa2 is also a positive regulator of a/a-specific
1981 Macmtllan Journals L t d

240

Nature Vol. 289 22 January 1981

HMLcx

MATa

HMRa

MAT 25

Fig. 1 Organization of the mating type cassettes on yeast chromosome 111. Open boxes represent sequences common to all three loci, MAT, HML and HMR.
Hatched regions are present at HML and MAT only. Sequences specific to a or a cassettes are denoted by dotted and zig-zag lines, respectively. The solid circle
denotes the centromere. Detailed restriction maps are presented elsewhere4. Only restriction fragments used as probes in subsequent figures are labelled. The BglII
site in the u cassettes labelled with an asterisk is presented only in strains derived from Saccharomyces carlsbergensis.

genes (mata21MATa diploids are sporulation defective).

MATa has no role in the mating ability of a cells (matamutants mate efficiently with a cells), but MATa is required,
along with MATa2, for the a l a phenotype (mata-/MATa
diploids behave like a cells)lO.The observations reported here
contribute further to our understanding of how MATa and
MATa regulate cell type.
There is substantial genetic and structural information available regarding the mechanism by which the mating type
cassettes at the storage sites are prevented from being expressed. The two storage sites for the a and a cassettes, HML
and HMR, are located at opposite ends of the same chromosome
in which the M A T locus resides". Each storage site contains
one copy of a mating type cassette, which can be either a or a3-'.
Recent structural analysis of cloned DNA molecules from M A T
and the storage loci, HML and HMR, has shown that all three
loci share two regions of sequence homology flanking a 600- or
750-base pair sequence unique to a or a , r e s p e c t i ~ e l ~The
~'~.
physical structure of HML, HMR and M A T and their orientation on the chromosome are shown in Fig. 1. Finally, genetic
analysis indicates that HML and HMR are under apparent
negative control by the combined action of several distinct genes
variously known as MAR, SIR and CMT12-14.These genes are
genetically unlinked to each other or to HML, HMR or MAT:
The MAR 1 locus has been mapped to chromosome IV (ref. 12).
MAR2 has been mapped to chromosome XI1 and is allelic to
CMT (unpublished results). A mutation in either of these genes
is sufficient to cause phenotypic expression of HML and HMR.
These genes, therefore, operationally act as co-repressors of
HML and HMR. In strains with the most common arrangement
of storage genes, that is, HMLa and HMRa, the m a r l - 1
mutation yields a non-mating, ala-like phenotype, presumably
because both a and a functions are being expressed12.
Thus, the mating type cassettes encode products which
regulate expression of unlinked genes involved in the mating
and sporulation pathways and are themselves regulated when
located in the storage sites. To examine the nature of these
controls further, we have examined the transcription patterns of
the MATa and MATa genes in a, a and a l a cells. We have
mapped these transcripts and assayed their expression in mutant
straints defective in normal control functions. We have also
examined transcription from the a and a elements residing at
HML and HMR in normal strains and in strains carrying marmutations which result in the expression of these loci.

Identification of transcripts
The MATa and MATa alleles each produce two mature RNA
transcripts. These transcripts were detected by electrophoretic
separatibn of poly(~)'-sefected RNA on agarose gels followed
bv transfer to diazotized filter oaverlS and hvbridization with
3i~-labelleda and a DNA cloneh bacterial &asmids. Figure 2
displays a series of RNA gel blots from normal haploid strains in
which only the M A T locus is expressed and marl- non-mating
strains in which all three loci containing mating type cassettes

are expressed (discussed below). Tracks a and c of Fig. 2 contain

RNA from normal MA Ta and MA Ta haploid strains probed
with an EcoRI-Hind111 fragment of HMLa DNA, and tracks e
and g contain the same RNAs probed with a EcoRI-Hind111
fragment of MATa DNA. The sizes of the labelled transcripts
were determined relative to pBR322 restriction fragment
markers run on the same gel. A comparison of these four tracks
yields the following conclusions: (1) there are two transcripts
from MATa, one of 650 bases (ax) and one of approximately
450 bases (ay); (2) only the 650 base MATa transcript is
homologous with a DNA from HMLa (Fig. 2, track a ) ; (3)
MATa produces a broad band (tracks c, g) which consists of two
transcripts of similar size, -800 bases ( a 1) and 750 bases (a2).
An additional band high in gel tracks e-h represents a transcript
not associated with mating type but which overlaps the cloned
M A T fragment used as a hybridization probe. The MATa
transcripts, which are not separated on the gel shown in Fig. 2,
can be distinguished both by R-loop analysis in the electron
microscope and by hybridization mapping, as described below.
More detailed mapping of these transcripts is also presented
below.
In addition to the normal haploid strains we have also
analysed the patterns of transcripts in marl- haploids. As
described above, genetic characterization of strains harbouring
the marl- allele indicates that such strains express the storage
loci, HML and HMR, in addition to the M A T locus12.Hybridizations to RNA from marl- strains of genotype
HMLaMATaMATa
(Fig. 2, tracks d, h) and
HMLa MATa HMRa (tracks b, f ) are shown. In the tracks
probed with MATa DNA both characteristic a transcripts
appear as well as at least one a transcript. This pattern of bands
is identical with that obtained using RNA from normal a l a
diploid (see Fig. 4, track j). Therefore, we assume here that the
state of mating type expression in non-mating mar- cells is
equivalent to that in a a l a diploid. The relative intensities of the
bands with tracks of RNA from marl- strains correspond
roughly to the gene dosage of a and a cassettes in each strain,
suggesting that all three loci are being expressed at comparable
levels. Comparison of marl- and MAR 1' strains indicates that
the HML and HMR loci are apparently under transcriptional
control through the MAR functions. Similar results have been
obtained for mutants carrying the mar2- allele which defines a
MAR function mapping on chromosome XI1 (data not shown).
The tracks probed with HMLa yield a similar result with the
exception that the small a transcript is missing, indicating that
this 450-base transcript comes from the sequences unique to the
a cassette.

Mapping the transcripts

The positions of both a transcripts ( a 1and a 2) and the longer of
the two a transcripts (ax) were determined by electron microscopy of R-loops formed by hybridization of poly(A)' RNA to a
EcoRI-Hind111 fragment containing MATa or MATa DNA
cloned in the bacterial plasmid pBR322. A histogram of

O 198lNature Publishing Group

Nature Vol. 289 22 January 1981

241

R-looped molecules is shown in Fig. 3. One a transcript was

mapped to the X region of the M A T locus and the other extends
from approximately the middle of the Ya region rightwards
through Z. Both a transcripts would therefore be expected to
show homology with the MATa gene as well as MATa because
the X and Z regions are common to both (see Fig. 1). One of the
MATa transcripts (ax) was also found to map in the X region of
the MAT locus. It seems to share the same left-hand end point
as the homologous a transcript ( a l ) but is somewhat shorter
than a 1, as expected from the gel blotting experiments. Only
two molecules containing the a transcript (a y) were found in our
R-loop preparations. However, blotting results in Fig. 2 and
additional studies described below clearly indicate that a y is not
homologous to the MATa gene and must therefore lie in the Ya
region of the MATa gene.
Additional mapping data were obtained by probing RNA
blots with purified restriction fragments from cloned mating
type genes as shown in Fig. 4. Tracks a-f in Fig. 4 represent a
single pair of blotted tracks hybridized sequentially to three
different probes. Tracks a, c and e contain RNA from m a r l non-mating ala-like strain, tracks b, d and f contain RNA from

-w
a

probe: HMLa

probe: MATa

Fi& 2 Polyadenylated RNA samples (20 pg per track) were obtained from
strains 2180-1A ( a MAR', tracks a , e ) , 2180-1B ( a MAR', tracks c, g).
K55 ( n marl -,tracks, b, f ) and K54 (a m a r l - , tracks d, h ) as previously
described2', were fractionated on a 1.5% agarose/methylmercury hydroxide gel" and transferred to diaiobenzyloxymethyl (DBM) paper20.The
immobilized RNA was hybridized with DNA, "P-labelled by nick-translation. spanningeither HMLa (plasmid 26.3, in Fig. 1; tracks a-d)or MATa
(plasmid MAT1, Fig. 1; tracks e-h) as described previously2'. The washed
filter was autoradiographed with Kodak XR-1 film and Dupont Lightning
Plus intensifier screens for 24 h. The relevant genotypes of the strains from
which the RNA wasobtained are indicated above each track and indicate the
mating type cassette ( a or a ) present at HML (bottom designation), M A T
(middle designation) and HMR (top designation). Brackets around the
designation indicate that the cassette at that particular locus is unexpressed.
as established by genetic criteria. The absence of brackets indicates that the
cassette is expressed at that particular locus. Bandscorresponding to a 1, a 2 ,
a x and ay transcripts, identified asdescribed in the text, are indicated to the
right of the autoradiogram. Molecular weights were estimated using restriction fragments of bacterial plasmid pBR322 as standards (not shown).

EcoRI

MATa

Hind 111

-Eco R I

MATa

Hind 111

Flg. 3 Histogram of R-looped plasmid molecules containing the MATa

gene (pBR-MATI) hybridized with RNA from a MATa strain ( a ) and the
MATa gene (pBR-MAT25) hybridized with RNA from a MATa strain (b).
Both cloned M A T genes were inserted in pBR322 as EcoRI-Hind111
fragments and the hybrid plasmids were digested with EcoRI. Dotted lines
indicate individual molecules containing both R-loops, measured molecules
represent approximately 50% of those observed to contain R-loops.
Bacteriophage @X-double stranded DNA was included as a size standard.
R-looping was performed according to the method of Kaback el aL2' and
measurements were made on a Numonics electronic planimeter.

a normal a haploid. Tracks a and b are probed with a fragment

which covers Ya, Z and overlaps the X region of MATa. The
probe reveals a single band in the a / a tracks, at -750 bases, but
shows a doublet at 800/750 bases in the a track. We interpret
this result to mean that two transcripts are made from the MATa
locus in strains exhibiting the a phenotype but only one in a / a
strains. This interpretation was confirmed by additional hybridization studies described below. Knowing the positions of the a
transcripts, we could confirm that only one of two a transcripts is
present in a / a cells. This was done by testing the original filter in
Fig. 4a,h with probes containing only X-region homology
(103.4) and Z-region homology (103.5). Using the high molecular weight band associated with the right-hand end of the
M A T probe as an internal standard it is clear (Fig. 4, tracks c-f)
that the Z-region transcript ( a l ) is not present in the a / a strain
but the X-region transcript ( a 2 ) is present in both a and a / a
cells. The presence of two transcripts from MATa is consistent
with the observation of two complementation groups MATa 1
and MATa2 as described above. Further, the absence of the a 1
transcript in a / a cells correlates with the fact that a functional
MATa 1 allele is not required for the a / a phenotype. On the
other hand, MATa2 is a required function in a / a diploids. We
therefore identify the message which is present only in a haploid
strains ( a l ) as the transcript for the M A T a l gene and the
message present in both a and a / a strains ( a 2 ) as the transcript
of MATa2. This interpretation is consistent with genetic
recombination results suggesting that MATa2 is proximal to
M A T a l (refs 7, 9) and has been confirmed by in vim
mutagenesis by Nasmyth et al. (see accompanying article").
Figure 4 also yields information concerning the location of the
a transcripts. Both a x and a y transcripts are clearly present in a
HMLa MATa HMRa marl- strain when probed with whole
MA Ta DNA (Fig. 2, track h ; see also the a l a strain in Fig. 5 ) .
However, when probed with fragment 103.8, only the smaller of
the a transcripts, a y, is revealed in the a l a track, indicating that
the 650-base transcript (ax) lies mainly to the left of the HhaI
restriction site in the region X. Furthermore, as the small a
transcript does not hybridize significantly with HMLa DNA
(Fig. 2), it must lie mainly within the Ya region. This interpretation is confirmed by additional hybridization of a / a and a
strains with the X-region probe 103.7 (Fig. 4g,h) and an a strain
with the Y-region probe 103.8 (Fig. 4i). In additional experiments (data not shown) the W region probe 103.6 was used
against a and a RNA. No hybridization was detected with this

0 1981Nature Publishing Group

242

Nature Vol. 289 22 January 1981

probe to any of the RNA species. Because either mating type

cassette can be expressed at HMR", even though the W region
is not represented there4'5, it was not required that those
sequences be transcribed.

Direction of transcription
The direction of transcription of the mating type genes was
determined by hybridization of RNA gel blots to separated
strands of A bacteriophage carrying cloned HMLa or HMRa
sequences followed by subsequent hybridization to wild-type A
DNA labelled with 32Pby nick translation. The orientation of
the cloned genes in the A vector was determined by heteroduplex analysis in the electron microscope. The result of these
experiments is shown in Fig. 5. The tracks on the left represent
parallel electrophoresis of poly(A)' RNA from an a haploid
strain S245c hybridized with the Rand Lstrand, respectively, of
the hybrid A clone MAT25 (MATa). The a 1band hybridizes to
the R strand of the A clone, that is, it is transcribed from Y a
towards Z, whereas a 2 hybridizes to the L strand and is thus
transcribed leftward as the locus is drawn in Figs 1 and 5. The
pair of tracks on the right contain poly(A)' RNA from a a / a
diploid strain probed with separated strands from A clone
HMR15c (HMRa). As described previously, this RNA preparation would be expected to contain the transcripts designated
ax, a y and a 2 , all of which should be detected by the HMR 15c
probe. These tracks show that a x as well as a 2 hybridizes to the
L strand and is transcribed leftward in the X region of the
cassette. The a y transcript hybridizes to the opposite strand and
is thus transcribed to the right in region Ya. Thus, both a and a
cassettes are transcribed divergently from within the middle of

ax -

-v

ay-

103.8

d
103.4

103.5

HMR- R

MAT 25

Fig. 5 Orientation of the a and a transcripts was determined using

separated strands of hybrid bacteriophage A molecules22 carrying cloned
mating type cassettes as hybridization probes against RNA gel blots of the a
haploid strain DC12 and the a / a diploidstrainDC13 followed by 'sandwich'
hybridization with 32P-labelled A DNA. Isolation of A clones was described
previously4. MAT25 carries a EcoRI fragment containing the MATO
cassette. HMRl5 contains an HMRa cassette. Strand separation and
hybridization have been described p r e v i ~ u s l y ~The
~ . orientation of the
cassette in each clone, and hence the designation of R and L strands, was
determined relative to the short and long arms of the A cloning vector
( A ~ ~ W EbySheteroduplex
~~)
analysis in the electron microscope. Unlabelled
bands represent non-mating type transcripts originating from flanking
sequences carried on the cloned molecules.

the cassette. At least for a 1and a y the 5' end of the transcript is
inside the unique Y a or Ya sequence and therefore inside the
transposed portion of the cassette. In addition to the direction of
transcription, a comparison of tracks a and c in Fig. 5 provides
additional evidence that the a 1transcript is found only in cells of
the a mating type and not in a / a diploids.

probe: MAT- R

g h
103.7

MATa

Control of a 1 transcription in a / a cells

Flg. 4 Mapping the a and a transcripts by hybridization of RNA gel blots

with specific probes. tracks a-f represent the same filter probed sequentially
with the probes indicated. Between hybridizations the previous probe was
removed by treatment with 99% formamide and 10 mM HEPES buffer, pH
7.5, at 65 "C for 1 h. Relevant genotypes of the strains used are indicated
above each track as described in Fig. 2 legend. Parentheses denote silent
HMLa and HMRa cassettes.

The transcript mapping results described above showed that one

of the two a transcripts (a1) is not present in a / a (non-mating)
cells. This is consistent with genetic results indicating that
MA Ta 1is not required for expression of the a / a cell type but is
required in a cells to turn on a-specific mating functions. We
have therefore investigated whether MATa 1is turned off by the
products of the MATa allele (ax and/or a y) or by the combined
action of MATa and MATa2. Poly(A)' RNA was isolated from
mata2/MATa diploids and tested for expression of the a1
transcript by hybridization with the whole MATa probe
(homologous to a 1, a 2 , a x and ay) and a purified probe
containing only the MATa yz region specific to a l . The

O 198lNature Publishing Group

243

Nature Vol. 289 22 January 1981

autoradiographs in Fig. 6 show sequential hybridizations of
these two probes to the same filter containing two
mata 2 - / M A Ta diploids, a normal M A Ta/ MA Ta diploid and
a M A T a haploid strain. The a 1 and a 2 bands are not resolved,
but the positive signal at the a l l a 2 position in the mutant
diploids when probed with M A T a Y Z indicates that the a 1
transcript is present. As expected, no signal is seen at that
position in the M A T a I M A T a diploid. Thus, it is clear that both
M A T a 2 and the product(s) of the M A T a gene are required for
turn-off of MA T a 1. The possible mechanisms of M A T a 2 and
MATa action are described further below.

o / a specific

The results presented here, combined with genetic results

described elsewhere, can be used to generate a circuit diagram
for the control of cell type in Saccharornyces. Figure 7 shows the
inferred regulatory pathways in a strains, a strains and a / a
strains. In the a strain shown in Fig. 7 a both M A T a l and
M A T a 2 have active regulatory roles. Because M A T a 1 is
necessary for efficient mating but not for sporulation it is proposed to be a positive regulator of a-specific functions. M A T a 2 ,
on the other hand, is proposed to act as a negative regulator of
a-specific functions. A mara2- mutant is presumed to be defective in mating as both a and a mating functions are turned on.
In the mating type a strain shown in Fig. 76, no control
functions are assigned to the products of the MATa allele. This
is because no recessive mutations at the MATa locus have been
shown to cause defects in a mating ability. W e presume, therefore, that the a-specific genes are constitutive in the absence of
M A T a 2 , and that even though M A T a is transcribed in both a
and a l a strains, its products only exert an effect in a / a strains.
Regulation in a M A T a I M A T a diploid strain is shown in Fig.
7c. In addition to the turn on of ala-specific functions by
M A T a 2 and MATa, some control mechanism shutting off aand a-specific functions is required. Several genetic arguments
suggest that M A T a 2 is responsible for shutting off a-specific
functions in a / a cells just as it does in a cells9. W e show here
that a-specific functions are turned off in a / a diploids by the
combined action of M A T a 2 and M A T a repressing the tran-

o specific

Control of cell type

W
.

d specific

a/d specific

-/

o specific
d specific

Fig. 7 Control of cell type in Saccharomyces. ( a ) o cells; ( b ) a cells; (c) a l a

cells. Straight arrows indicate positive control. T-bars indicate negative
control. Wavy arrows are the transcripts described in the text. Other symbols
as in Fig. 1. The requirement of both a x and a y in control of cell type has not
been determined. Genetic analyses of control of cell type are described
elsewhereh-'".

scription of the positive regulator of a genes, M A T a I . This

analysis does not determine whether it is a x o r a y (or both) that
are required for the a / @ cell type. I n virro mutagenesis of cloned
MATa-containing plasmids by Nasmyth et al. (see accompanying articleI6) indicates that only a y is necessary for this interaction. No function has been established for ax. W e therefore
propose that an interaction occurs between the M A T a 2 and the
M A T a product(s) which changes their regulatory specificities.
This interaction creates a new regulatory molecule which serves
to turn off M A T a 1 and also to turn on the ala-specific genes
whose functions are required for meiosis and sporulation. Of
course, it is also possible that M A T a 2 and the products of
M A T a d o not interact with each other but must act at independent sites on the D N A so as to regulate a 1 and the a / a specific
genes.

ax'
'JY-

Control of the silent genes

probe :

M ATa

M ATuY Z

Fig. 6 Sequential hybridization of R N A from a series of strains hlotted on

diazotized paper2uto probes containing a complete MATa cassette ( M A T I ,
Fig. I ) and the right half of a MATa cassette (MATaYZ). Two independently isolated mafa2 - / a diploids (DCIOO, DC101). a normal a / @ diploid
(DC13) and an a haploid (DC12) were used. The MATa YZ probe was
made hy cutting a cloned n cassette from strain CB11 at the BgnI site in the
Y region4 yielding a probe specific for the maru 1 transcript.

In addition to the control of cell type itself the transcription

mapping presented here provides insight into the mechanism by
which the storage cassettes, H M L and HMR, are kept silent.
There is much genetic evidence indicating that H M L and H M R
are under negative control and the results presented here show
that control is apparently at the level of transcription. A t least
four distinct genetic loci, known as M A R o r SIR, have been
identified which are required to prevent expression of HML and
HMR. In both m a r l and rnar2- mutants mature transcripts are
produced at H M L and H M R and these transcripts seem to be
identical to their mature counterparts from MAT. These results
are particularly surprising considering the S e n d of at least one a
transcript ( a 1) and one a transcript (a y) map within the region
of the cassette which must b e transposed during mating type
switching.

0 1981Nature Publishing Group

Nature Vol. 289 22 January 1981

How, then, can that region of the cassette be silent at one
chromosomal locus and active at another? O n one hand, it is
possible that the 5' end of the observed transcripts d o not reflect
the actual transcription initiation site. The transcripts could start
near the ends of the cassettes and proceed in opposite directions,
crossing each other to reach the regions maintained in the
mature transcript. In this model the transcription initiation sites
could be in or near regions not shared by MAT and the storage
loci and hence subject to different control. O n the other hand,
transcriptional initiation may occur near the 5' ends of the
mature transcripts. In this case, differential regulation of the
cassettes could be effected by differential accessibility of the
promoter site depending on whether the cassette is located at
MAT or at HML o r HMR. The accessibility of the promoter
would be established by the interaction of sites outside the
regions of homology with the MAR and SIR products and could
reflect differential chromatin structure18, phasing of
n u c l e o s ~ m e s 'or
~ differential methylation. The structure of
certain intrachromosomal rearrangements4 suggests that the
sites of action of the negative regulators lie to the left of the W-X
regions at HML and HMR. Deletion of these sequences results
in the expression of the normally silent HML and HMR
cassettes.
Regardless of the particular mechanistic model, it is clear that
mating type expression in yeast is an example of a genetic
positional effect; that is, the expression of a D N A sequence
depends on its location in the genome. The availability of
mutants which allow expression of the silent cassettes and the
existence of cloned D N A from MAT, HML and HMR should
permit a detailed analysis of the molecular basis of this positional effect.

W e thank David Kaback for assistance with the R-looping

analysis, Ira Herskowitz for comments on the manuscript, Jean
McIndoo, Carolyn McGill, Victoria Guarascio and Sajida Ismail
for technical assistance, Louisa Dalessandro for preparation of
the manuscript and Mike Ockler for the figures. This research
was supported by NIH research grant GM25678-03 to A.K.,
GM25634-03 to J.H. and J.S and GM24226 to J.B.

Received 8 September; accepted 14 November 1980

1. Hicks, J. B., Strathern, J. & Herskowitz. I. in DNA Insertion Elements, Plasmids and
Epiromes (eds Bukhari, A,, Shapiro, J. & Adhya. S.) 475-462 (Cold Spring Harbor. New
York. 1977).
2. Harashima, S., Nogi, Y. & Oshima, Y. Generics 77,639650 (1974).
3. Hicks, J., Strathern, J. N. & Klar, A. J. S. Nature 282, 478-482 (1979).
4. Strathern, I., Spatola, E., McGill, C. & Hlcks. J. B. Proc. nam. Acad. Sci. U S A . 71,
2839-2843 (1980).
5. Nasmyth, K. A. & Tatchell, K. Cell 19, 753-764 (1980).
6. McKay, V. & Manney, T. R. Genetics 76,273-288 (1974).
7. Strathern, J. N. thesis, Univ. Oregon (1977).
8. Hicks, J. B. thesis, Univ. Oregon (1975).
9. Strathern,J., Hicks, J. & Herskowitz, I. J. molec. Biol. (in the press).
10. Kassir, Y. & Simchen, G. Genetics 82, 187-206 (1976).
11. Harashima, S. & Oshima, Y. Genetics 84,437451 (1976).
12. Klar, A. J. S., Fogel, S. & Macleod, K. Genetics 93, 37-50 (1979).
13. Rine, J., Strathern, 1.N., Hicks, J. B. & Herskowitz, I. Generics 93, 877-901 (1979).
14. Haber, J. E. &George, J. P. Generics 93, 13-32 (1979).
15. Alwine.1. C., Kemp, D. J. & Stark,G. Proc. natn.Acad. Sci. U.S.A.74.5350-5354(1977).
16. Nasmyth, K. A., Tatchell, K., Hall, B. D., Astell, C. & Smith, M. Nature 289, 244-250

(1981).
Klar, A. J. S., McIndoo, 1..Hicks. J. 8 . & Strathern, J. N. Genetics (in the press).
Wu, C., Bingham. P. M., Livak, K. J., Holmgren, R. & Elgin. S. Cell 16,797-806 (1979).
Gottesfeld, J. & Bloomer, L. S. Cell 21, 751-760 (1980).
Broach, J. R., Atkins, J. F., McGill, G. & Chow, L. Cell 16, 827-839 (1979).
Kaback, D. B., Angerer, L. M. & Davidson, N. Nucleic Acids Res. 6. 2499-2517 (1979).
Szybalski, W., Kubinski, H.,
Hradecna, Z. & Summers, W. C. Meth. Enzyrn. 21,383-413
(1971).
23. Leder, P., Tiemeier, D. & Enquist, L. Science 196, 175-177.

17.
18.
19.
20.
21.
22.

A position effect in the control of transcription

at yeast mating type loci
Kim A. ~ a s r n ~ t hKelly
* , ~atchell*,
Benjamin D. all', Caroline Astell'
& Michael Smitht
* Department of Genetics, SK-50, University of Washington, Seattle, Washington 98195
t Department of Biochemistry, Faculty of Medicine, University of British Columbia, 2075 Westbrook Place, Vancouver,
British Columbia, Canada V6T 1W5
The two mating type loci MATa and MATa each produce two mRNAs that are transcribed in opposite and diverging
directions from central promoters. Silent copies of MATa ( H M R a ) and MATa ( H M L a ) contain identical D N A sequences
throughout the transcribed region, yet are not transcribed. It is concluded that sequences to the left of HMRa (and probably
H M L a ) must somehow affect transcription initiated at the centre of each locus 700 to 1,400 base pairs away. A possible
mechanism for this position effect is discussed.
THE mating type of a yeast cell, a o r a , is determined by a single
locus on chromosome 111called MAT (ref. 1) which may contain
one of two types of D N A
In heterothallic cells,
MATa and MATa, are stable mendelian 'alleles' which when
hemizygous or homozygous (as in a o r a haploid cells o r a / a o r
a / a diploid cells) confer an ability to mate with cells of the
opposite mating type and when heterozygous (as in a / a diploid
cells) turn off mating functions and allow the cell to sporulate'.
wver, in cells containing the homothallism gene (HO), MATa
and MATa are interconvertible, switching mating type at
almost every cell division until conjugation leads to the formation of a stable a / a diploid whereupon the MAT alleles are

table^'^.
In addition to the HO gene, an a to a switch requires another
locus, HMLa, located on the left arm of chromosome 111.
Similarly, an a to a interconversion requires HMRa, located at
the far end of the right arm of chromosome II16.'. Geneticn and
physica12.3analyses of the MAT loci have revealed that interconversion occurs by a replacement of information present at
MAT with a copy of information of the opposite mating type

present at one of the HM loci. In other words, HMLa is a silent

o r unexpressed copy of MATa, and HMRa is a silent copy of
MATa. The physical structures and chromosomal location of
MATa o r M A Ta, HMLa and HMRa are depicted in Fig. 1. At
the centre of each locus lies a sequence specific to the mating
type, either a 650-base pair a specific ( Y a )o r 750-base pair a
specific ( Y a ) sequence. In most strains HML contains Y a and
HMR contains Y a . The sequences flanking Y are found at both
silent and active loci-a 700-base pair region to the left of Y
(called X) and a 230-base pair sequence to its right (called Z l ) .
Two further blocks of D N A are found at both MAT and
HML-700 base pairs adjoining the left end of X (called W) and
9 0 base pairs adjoining the right end of Z 1 (called 22). W and 2 2
are found at MAT and HML but not at HMR, irrespective of
whether a o r a informatvn is present.
MAT seems to determine cell type by the control of unlinked
a o r a specific genes rather than by direct expression of all
structural information required for mating9. MATa performs
this function through the action of at least two genes, a 1 and a 2
(ref. 9). The MATa 1 gene product is necessary for the expres@ 19R1 Macrnillan Journals Ltd

245

Nature Vol. 289 22 January 1981

Table 1 Relative level of MAT transcripts in different strains
RNA
Genotype

a1

HMLa MATa
HMLa MATa
HMLa MATa/ MATa
HMLa MATajmaia2
HMLa mura l / M A T a
HMLa MATa SIR 1
HMLaMATaSIR3or4
23C
HMLaMATaSIR3or4
34C
HMLa MATaI MA Ta
SAD 1 / S A D 1
HMLa MATala-lethal

++

a2

a1

++
++

++
++

++
-

++

++

HMRa
HMRa
HMRa
HMRa
HMRa
HMRa
HMRa

ND
ND

++
++

HMRa

ND

++

HMRa

HMRa

++

++

++

++

++

a2

++
+
++
++
i

++

++

(HA)
The relative levels of each MAT transcript in different strains are shown. + +
Signifies the highest level seen for a given transcript, + signifies a significant or 3to 10-fold reduction,
signifies a 10- to 100-fold reduction and - signifies our
inability to detect it. No correspondence is intended for comparisons between
transcripts; for these, see the text. Most of the data summarized here can be found
by examining Fig. 3. Sometimes is written when no such transcript is visible in
Fig. 3. This is due to inadequate reproduction of the latter. The Hawthorne
deletion removes all DNA on chromosome I11 between MAT and HMR, due toan
intrachromosomal recombination between these two loci3. The SAD1 mutation is
also due to a recombination between MAT and HMR, but is an interchromosomal
recombination that results in the duplication of all DNA on chromosome I11
between MAT and HMR (J. Hicks, personal communication). ND, not determined.

sion of genes unlinked to M A T that specify an a mating type.

MATa2, on the other hand, is necessary for repressing a specific
genes which are otherwise expressed constitutively and would
therefore antagonize the expression of an a mating type. This
model9 for the determination of mating type implies that MATa
plays no part in the specification of an a mating type as the genes
necessary are constitutively expressed in the absence of the a 2
gene. A gene at MATa (a 1) is, however, necessary for specifying, in conjunction with the a 2 gene, the non-mating a / a
diploid cell type and therefore the ability to sporulate".
The mating type genes, although also present at the silent loci
HMR and HML, are normally exclusively expressed at MAT.
Recessive mutations at a number of unlinked loci(SZR1,2, 3
and 4)".12 allow the silent copies to be expressed, suggesting
that HML and HMR are under negative control of the SIR gene
products.
Our aim here is to define the process whereby ty genes at
HMRa and HMLa are activated on transposition to MAT. We
have therefore focused on the location and control of transcripts
made from mating type loci. Our results reveal that MATa and
MATa each produce two transcripts (a 1 and a 2 , a 1 and a 2 )
neither of which is made at their corresponding silent copies
HMRa and HMLa, even though all transcript sequences and
their immediate surroundings are present there. More precisely,
the sequences flanking the left-hand end of each silent locus is

apparently responsible, in conjunction with SIR gene products,

for preventing the initiation of transcription within the Y regions
at the centre of each locus. This situation is reminiscent of
position effects in Drosophila I' and similar phenomena recently
discovered at the haemoelobin locus in humans14.
In addition to defining &e control of transcription at the silent
( H M ) loci, our results have revealed that M A T controls its own
transcription; MATa 1 and MATa2 together repress the transcription of M A T a l in a / a diploids, thereby ensuring their
non-mating phenotype.

MA T transcripts
We have mapped RNA transcripts in the region of MAT by
two different methods, R - l o ~ p i n and
~ ' ~ S, nuclease protectionI6. We have used the former to estimate the number and
approximate position of transcripts and the latter to discover
their direction of transcription and the position of their termini
at a nucleotide level.
The details of our R-looping are described in Fig. 2 and its
legend. Figure 2A shows typical examples of R-loops formed
between linearized MATa or MATa plasmid DNA and
poly(A)'RNA from a or a cells. Figure 2B shows our
conclusions, which can be summarized as follows. First, a cells
contain two distinct RNA species homologous to MATa in
roughly equal abundance. One, which corresponds to the a 1
gene", is about 450 base pairs and is transcribed from the Ya
region with its (centromere proximal) left-hand terminus about
100 base pairs to the right of the X-Ya boundary. The other
transcript, which we call a 2 , is about 600 base pairs long and is
found within the X region with its right-hand (centromere distal)
end near the X-Ya boundary. Neither a transcript seems to
have an intervening sequence sufficiently large to be detected by
electron microscopy. Second, a cells also contain two distinct
RNA species homologous to MA Ta. One, which corresponds to
the a 1 gene'7, is about 700 base pairs long and is transcribed
from a region extending from the middle of the Ya to the end of
2 2 . The other, which corresponds to the a 2 gene (ref. 17), is
about 720 base pairs long and is transcribed from the X region
with its right-hand end just within the Yaregion. Neither of the
a transcripts seem to have intervening sequences.

S1 mapping
Figure 3 shows the strategy used to map the exact 5' and 3'
termini of all four M A T transcripts. We have used the extent of
protection from S1 nuclease, by RNA of single-stranded endlabelled restriction fraements. as a measure of the distance from
A
The details of this
a labelled restriction site to ~ ~ ' R Nterminus.
analysis are described in Fig. 3 and its legend (Table 2 shows the
genotype of the strains used in Fig. 3). In the text, therefore, we
will confine ourselves to its conclusions.

HML

MAT

HMR

Fig. 1 An overview of yeast

mating
type
loci
on
chromosome I11 and the control
of cell type by MATa and
MATa. The boxed area refers
to the activity of MATa and
MATa in their determination
of a, u or u / a cell types. The
rest of the figure sets these
activities in the perspective of
other 'silent' copies of MAT on
the left (HMLa) and right
(HMRa) arms of chromosome
111. A description of the
comparative
structures of
MAT, HML and HMR is given
in the text. Positive controls are
represented
by connecting
arrows and negatlve controls by
arrows interrupted by a bar.

constitutive g mating

HAPLOID

I'

repression of &

O 1981Nature Publishing Group

"

"

'

repression of

HMR

246

Nature Vol. 289 22 January 1981

Table 2 Genotype o f strains used

No.

Strain

S1
S2
S3
S4
S5
S6
S7
S8
S9
S10
S11

XP8-4A
XP8-1OB
XPll
XR29-10C
2049-1-4a
2050-5-la
265-7-4
H293-4-2
K205
K210
2760

S12

X999-1

S13
S14
S15

K79
17-16
G200-15a

Genotype

Source

isogeneic derivatives
MATa ho his6 leu1 met1 trp5-1 gal2 can 1
M A T a ho his6 leu1 met1 trp5-1 gal2 can1 o f XT 1172-S245C
XP8-4A x XP8-1OB
M A Ta ho sir1 - 1 ade6 arg4 leu2 trp 1 R M E
MATa ho ste8-59a (sir3)trpl ade2 leu2
M A T a ho ste9-62c (sir4) trpl ade2 leu2
M A Ta ho hom 3 ilu 1
M A Ta ho trp 1
H293-4-2x9200-15a homozygosed for trpl by X rays
K79 x 17-16 homozygosed for ade2 by X rays
M A T a ho gall2 G A L 1 suc ma1 ural ade7
matad
ho gall2 gall suc ma 1 U R A 1 A D E 7
M A T a S A D 1 ho ade2 his4-12 leu2-27 L Y S 2 T R P l ura3 gal2
M A T a S A D l ho A D E 2 his4-12 leu2-27 Iys2 trpl ura3 gal2
M A Ta ho leu2 trp 1
matal ( a * )hoade2 ura3 trpl can1 cyh2
mata2-1 ho sirl-1 ade6 met1 his6 R M E

(1) The a 1 and a 2 genes are transcribed in opposite and

diverging directions. The a 1 transcript is initiated about 90 base
pairs to the right of the X-Ya (see Figs 3, 4) boundary and
extends rightward 510-530 base pairs through the Ya region
(see Fig. 3, fragment 6). The a 2 transcript, on the other hand, is
initiated about 50 base pairs to the left of the X-Ya boundary
(Figs 3 , 4 ) and extends leftwards through the X region for 600
base pairs (Fig. 3, fragment 1).
(2) The a 1 and a 2 genes are also transcribed in opposite and
diverging directions. Both are initiated within the Y a regionthe a 2 gene at a point about 90 base pairs (Figs 3,4) and the a 1
gene at a point about 300 base pairs from the X-Ya boundary
(Figs 3, 4). The a 2 and a 2 transcripts have exactly the same
termination point (Fig. 3, fragment 1).
(3) All genes, to a greater or lesser extent, have heterogeneous 5' and 3' termini. The 5' end of the a 2 transcript is a
good example of this heterogeneity. Essentially the same spectrum of termini is found, barring a consistent 5-base pair
difference, whether SI or the single-strand specific exonuclease
ExoVII (ref. 16) is used to degrade unprotected fragment 2
DNA (Fig. 3). This suggests that the pattern seen is a true
reflection of RNA termini and not an S, degradation artefact.
We do not know, however, whether this heterogeneity of RNA
termini is due to genuine heterogeneous initiation (or termination) or to processing or degradation.
(4) We have mapped all transcript sequences between their
termini by the S1 protection method. This has been achieved
either by mapping both 5' and 3' termini from a common site
within the gene (for example, MATa2, Fig. 3) or by also
analysing the protection by RNA of end-labelled DNA fragments internal to the gene (such as MATa 1, Fig. 3). As a result,
we have shown that none of the M A T transcripts has an
intervening sequence between the termini defined by our
analysis. This, however, does not exclude splicing of terminal
sequences. To exclude the latter in the case of the 5' end of the
a 2 gene, we have directly compared the results of S1and ExoVII
digestion and have found no discrepancy except for a consistent
5-7-base pair difference in the length of the protected fragment.
ExoVII apparently cannot degrade the last 5 base pairs of a
single-strand tail adjacent to duplex DNA (Fig. 3, fragment 2).

Control of MAT transcripts

We have studied two aspects of the control of M A T transcripts:
(1) the regulation of transcription at the 'silent' copies of MAT,
HML and HMR and (2) the autoregulation of M A T transcription by M A T itself.
MATa cells contain neither a 1 nor a 2 transcripts. Likewise, MATa cells contain neither a 1nor a 2 RNAs. This implies
that the 'silence' of HMLa and HMRa is due to a failure in
transcription. Recessive mutations at any one of four loci (SIR
1, 2, 3 or 4) result in the expression of HMLa and HMRa".12.
Figure 3 (fragment 9) and Table 1 show that, at least in the case
of the a 2 gene, this is due to a failure to repress HMLa
transcription. The a 2 transcript found in MA Ta SIR 1, SIR 3 or

P. Kushner
P. Kushner
P. Kushner
G . Sprague
L. H. Hartwell
L. H. Hartwell
L. H. Hartwell
L. H . Hartwell
This work
This work
D. C. Hawthorne
A . Hopper
This work
A . Hopper
G. Sprague

SIR4 strains has identical 5' termini to those found in a / a

diploids (the most comparable genotype). In the case of SIR3
and SIR4, we have analysed temperature-sensitive alleles of
these loci and have shown that a 2 transcription is also temperature sensitive.
SIR gene products, therefore, repress transcription at HMLa.
It is not clear, however, how this repression is mediated. MATa
and HMLa have identical DNA sequence^'^^'^ from the left end
of W to the right end of Z (a total of 2,500 base pairs), with the
apparent promoter region in the middle of this homology. The
specific repression of HMLa by SIR gene products must
presumably be effected by flanking sequences outside this
homology, beyond the 3' ends of both a1 and a 2 transcripts.
The same is presumably true for HMRa.
To characterize further the mode of SIR gene product action,
we have analysed the levels of a 1 and a 2 transcription from two
variants of HMRa which have different flanking sequences; the
which replaces
first is that caused by the Hawthorne
the left-hand flanking sequences with those from M A T but
leaves the right-hand flanking sequences intact, and the second
is the converse rearrangement, called SAD1 (refs 21,22), which
replaces the right-hand flanking sequences with those from
M A T but leaves the left-hand flanking sequences intact
(J. Hicks, personal communication). Table 1 shows that the
former produces the same amount of a 1 and a 2 mRNA as
MATa whereas the latter produces much less of each transcript
(<I%). This implies that sequences flanking the left end of
HMRa are necessary for the SIR-mediated repression, whereas
those flanking the right end are not. Further explanation of these
variant HMRa loci is provided in Table 1 legend.

Regulation of MAT transcripts

by MATal and MATa2
Figure 3 and Table 1show that the spectrum of M A T transcripts
in MATaIMATa diploids is not the sum of its MATa and
MATa parents. Whereas the a 1 transcript is present in
comparable amounts in a and a / a cells, the a 2 transcript is
reduced at least fivefold and that of a 1 is entirely absent in a / a
diploids. The a 2 transcript starts at three discrete positions in
haploids (Fig. 3 and ref. 18), one ( 4 % ) 8 base pairs upstream
from the AUG, a second (>90%) 12 base pairs upstream and a
third (<So%)28 base pairs upstream from the AUG. Only the
first and second of these transcripts are subject to repression
(about 10-fold reduction) in a / a diploids. The repression of a 1
and a 2 transcription in a / a diploids requires the action of both
MATa 1 and MATa2 genes (Fig. 3, Table 1).

DNA sequence of putative control

regions in MATa and MA Ta!
We have determined the sequence of MATa DNA between the
starts of a 1 and a 2 , of MATa DNA between the starts of a 1
and a 2 , and the corresponding region of HML (see Fig. 4). The
following points are noteworthy. First, the region between a 1

O 1981Nature Publishing Group

Nature Vol. 289 22 January 1981

247

and a2 in Y a and the corresponding region in Y a have neither

extensive sequence homology nor analogous sequence patterns;
second, the sequence of the region between a 1 and a 2 in MATa
is identical to that in HMLa; third, the regions between the
structural genes of both MATa and MATa contain symmetrical structures (boxed sequences of Fig. 4).

Control of silent copy transcription:

a position effect
MATa produces two transcripts that are initiated within 250
base pairs of each other within the unique a region ( Y a )and are
transcribed in diverging directions. We have shown by D N A
sequence analysis'8.1Ythat for a 1 and a 2 the entire coding
region of the mature transcripts and the sequences between
them is virtually identical in MATa and HMLu. However,
MATa and HMLa differ drastically in their expression of
identical MAT genes. HMLa will produce neither a 1 nor a 2
mature transcript except when mutations at unlinked SIR loci
are present. The sequence of MATa and HMLa does diverge

beyond the 3' ends of the transcripts, 800 base pairs to the left of
the a 2 transcript termination point and at the termination point
of the (Y I transcript. Somehow, then, sequences downstream of
the 3' ends of the a 1 and a 2 genes are responsible for the
differential repression of transcription at M A Ta and HMLa by
SIR gene products.
A similar situation exists for MATa and HMRa, although
here we have not demonstrated identity at the D N A sequence
level. Comparison of the levels of MATa transcripts in the
Hawthorne deletion2" and the SAD1 m u t a t i ~ n *(Table
~ ~ ~ ~1 )
suggests that sequences flanking the left (centromere proximal)
but not the right of HMRa are primarily responsible for mediating the repression by SIR gene products (Fig. 1).
It is possible that the observed transcripts made from MATa
and MATa are initiated not, as they seem, within Y a and Y a ,
but from the sequences flanking MAT. By this hypothesis,
M A Ta 1 would be processed from a long precursor initiated in
flanking regions to the left of MAT, and MATa2 would be
processed from a precursor initiated in flanking regions to its
right. One might then interpret the regions of pronounced dyad
symmetry in ~a and Y a as being for R N A secondary structures
involved in processing. The most serious empirical objection to
this hypothesis is our observation that HMRa with its left-hand
flanking sequences replaced by those from M A T (the
Hawthorne deletion) produces MAT levels of both a 1 and a2
transcript, whereas with its right-hand flanking sequences
replaced by those from M A T (SADI),it barely produces either
transcript. In addition to this, the R N A processing hypothesis
does not have a convincing explanation for the phenotype of SIR
mutants. For instance, one would have to postulate that the
promoters flanking M A T also existed in the nonhomologous
sequences flanking HMLa and HMRa but that they were
specifically repressed by SIR gene products.
If we reject the processing hypothesis, we are forced to
conclude that sequences flanking the left end of HMR can

1
C

okb

MAT a

ZiZZ

Yo

H,.~OIII
1

~ k b

"

2kb

MAT a

'

'

~ k b

H ~ ~ ~
.

'

4kb

Fig. 2 R-loops formed between linearized MATa or a plasmids and

poly(A)+ RNA from a and a cells. ( a ) MATa DNA-a RNA: shows a 1; ( b )
MATa DNA-a RNA: shows a 2 ; (c) MATa DNA-a RNA: shows a l (note
the tail): ( d ) MATaDNA-a RNA: shows a 1 and 0 2 : ( e ) as (d): (f)
MATa DNA-a RNA; shows a 2 (note the tail). All R-loops were formed
between pBR322 containing a 4.3-kilobase MATa or MATa HindIII
fragment (inserted into its Hind111 site with the centromere distal side
towards the BamHi siteof pBR322) and poly(A)' RNA made from a ( S l )or
a 6 2 ) cells. The procedure used was essentially that described by Kaback er
a/.''. This method involves the use of cross-linked DNA which occasionally
results in R-loops being interrupted by a cross-link; see, for instance, ( d ) .
Plasmid DNA was linearized by restriction with endonuclease BamHI,
extracted once with phenol/chloroform, ethanol precipitated, cross-linked
with trioxsalen (2-3 cross-links per molecule) and hybridized at 52C for
15-18 h at 5 pgml-I with 1 mg ml-' poly(A)' RNA in a total volume of
25 p l sealed in a glass capillary. After hybridization, 4 p l of deionized
glyoxal (Baker) was added to the pre-cooled (to OC) hybridization mixture.
which was then incubated at 12-14'C for 2 h. Spreading was perfomed
directly without prior purification of RNA loops from the unhybridized
RNA, by diluting an aliquot of the glyoxalated mixture with 9 volsof OCTE
buffer (10mM Tris-HCI, 1 mM EDTA, p H 8.0) and adding 1-2 p1 of the
diluted R-loops to 100 p l of spreading solution (50% formamide to be
spread on a 20% formamide hypophase). Spreads, their preparation for
electron microscopy, thcir viewing and photography were performed as
previously described2. A quantitative analysis of the R-loops seen is shown
in the lower half of the figure. We have seen R-loops within most regions of
the M A T HindIII fragment, but have only heen able to characterize
properly transcripts from its central M A T region, the others being too rare
and having insufficiently precise end points to be characterised by this
technique (see later S, mapping). MATa and MATn both produce two
discrete transcripts that are apparent in the histograms beneath the R-loop
pictures. The apparent difference between MATa and MATa in the occurrence of flanking transcripts is probably due to the M A T specific transcripts
from the former being at a lower level (of detection, possibly) than those
from the latter. Underneath the histogram is a map of M A T with the
position of M A T transcripts placed on it. These positions were estimated by
averaging the end points of the molccules observed. The MATatranscripts
both overlap homologous (X or Z) and unique ( Y uj regions. Thus, one might
predict the occurrence of different types of R-loop if a RNA was hybridized
instead to MATa DNA. One should find in the case of a 2 a sliehtlv
- ,shorter
R-loop with a short RNA tail and in the case of a 1 a very much shorter
L
R-loop
~
with a long RNA tail. We have consistently found the latter structure
in the predicted position but have only found individual examples of the
former, (see ( c ) and ( f ) ,respectively).

O 1981Nature Publishing Group

248

Nature Vol. 289 22 January 1981

Ya

Xbal

fragment l

fragment 2

Xbal

fragment 4

,Hlnfl

7
t
t

fragment 6

Hln f l

fragment 9

Taql

Hinf l

frogment !O

fragment 12

Fig. 3 S, transcription mapping. The upper part of the figure shows the strategy by which we have mapped the exact 5'and 3' termini of all four M A T transcripts. By
hybridizing total poly(A) RNA to end-labelled separated strands of restriction fragments, then measuring the lengths of the 32P-DNA strands remaining after
nuclease digestion, we have measured the distance from each RNA terminus to a physically mapped restriction site. The single-stranded, end-labelled fragments used
as probes are given numbers (1-12). The XbaI, HinfI and TaqI fragments were derived from wild-type MATa or MATa DNA (Hind111 fragments) cloned in
pBR322. Fragment 2 was derived from an in oitro manufactured mutant with aXho1 site within Y a (mata23, ref. 17). 5'* at the end of each fragment indicates an end
labelled by polynucleotide kinase and 3" indicates an end labelled with Klenow DNA polymerase. Solid lines indicate stretches of DNA protected from S, nuclease by
RNA whereas dashed lines indicate sensitivity. Apart from the four major transcripts presented here, we have evidence for at least three more transcripts within the
M A T region (W-Z of fig. 2). One of these, made at least in a cells (see R-looping), covers part of the W region. Another (called MAT3), made from both MATa and
MATa in a / a cells, starts some point before the TaqI site of Z 1 and ends at the same points as the M A T a 1transcript. S, protection experiments comparing a / a cells
with a lethal (Hawthorne deletion)la ~ e l l s ~ ~ s hthat
o whalf of these M A T 3 transcripts cease to have any homology to MATa from apoint 8 8 basepairsfrom the TaqI
rite. This implies that the 21-22 boundary occurs at this point. A third minor transcript, made at least in a cells, starts somewhere to the right of the TaqI site to the
right of the Z1 Taql site, and ends somewhere near or close to the left of the latter. We do not know the biological significance of these minor transcripts. Only those
restriction sites used for S, mapping are shown. The lower half of the figure presents the autoradiograms of end-labelled fragments. The markers are either @XI74or
pBR322 cut with HinfI and endlabelled with Klenow DNA polymerase. The numbers in parentheses after the M A T genotypes refer to the strain number of Table 2.
Unless otherwise stated, all digestions were performed with S, nuclease and cells were grown at 30C in YEPD (1). Fragment 1(or 7): The 3' end of MATa2 and
MATa2. Protection of fragment 1 or 7 by RNA at 1 mgml-I. (1) @X174, (2) MATa (Sl), (3) MATa (S2), (4) MATaIMATa (S3). Fragment 2: The 5' end of
MATa2. Protection of fragment 2 by RNA at 1 mg ml-' (lanes 2,3)or 2.5 mg ml-' (lanes 5-10). (l)@X174; (2) MATa (S1)-S, nuclease; (3) MATa (S1)-ExoVII;
(4) @X174; (5) MATa 6 1 ) ; (6) MATa 6 2 ) ; (7) MATaIMATa 6 3 ) ; (8) mata1IMATa (S10); (9) MATalmataZ 6 9 ) ; (10) MATaIMATa S A D l I S A D l (S12);
(1 1) no RNA. The intense band(s) at 400 nucleotides in lanes 6-10 corresponds to a 2 RNA, which is also homologous to the probe up to the X-Y boundary. Fragment
4: The 5'end of M A T a l . Protection by RNA at 1 rng ml-I. (1) MATa (Sl); (2) MATa (S2); (3) @X174.Fragment 6: The 3'end of M A T a l . Protection by RNA at
2.5 mgml-'. (1) @X174; (2) MATa 6 1 ) ; (3) MATa (S7); (4) MATa 6 2 ) ; (5) MATaIMATa 6 3 ) ; (6) m a t a l l M A T a (S10); (7) MATalmata2 (S9); (8)
MATaIMATa S A D 1ISAD1 (S12); (9) no RNA. Fragment 9: The 5'end of MATa2. Protection by RNA at 1 mg m1-I. (1) @X174; (2) MATa (Sl) grown at 2 3 T ;
(3) MATa (Sl) 34C; (4) MATaSIR3 (S5) 2 3 T ; (5) MATaSIR3 (S5)34"C; (6) MATaSIR4 (S6) 23C; (7)MATaSIR4 (S6) 34"C;(8) M A T a S I R l - 1 (S4)23'C; (9)
MATa (S2) 23C; (10) MATa (S2) 34C; (11) MATaIMATa (S3) 30C. Fragment 10: The 5'end at MATa 1. Protection by RNA at 1 mg ml-'. ( I ) MATa (S2); (2)
MATaIMATa 6 3 ) ; (3) pBR322. Fragment 12: The 3' end of MATa 1. All lanes except lane 2 show protection of fragment 12 by RNA at 2.5 mg ml-'. Lane 2 shows
protection of the opposite strand to fragment 12. (1)@X174; (2) MATa (S2); (3) M A T a (S2); (4) MATa (S2) 0.5 mg ml-' RNA; (5) MATa (Sl); (6) MATaIMATa
6 3 ) ; (7) m a t a l l M A T a (S10); ( 8 )MATalmata2 (S9); (9) MATalMATa S A D l I S A D l (S12); (10) MATa sir1 - 1 (S4); (11) MATa sir3 (S5) grown at 3 4 T . Lanes 5
and 6 show bands characteristic of a 1 termini even though no transcripts with the 5' sequences of a 1 have been detected in these strains (see fragment 10). These
termini probably belong to a transcript (MAT3?) that starts in Z and terminates at the same points as a 1 (see below). Klenow DNA polymerase and T4
polynucleotide kinase were used to label the 3' and 5' ends, respectively, of restriction fragments which were then strand separated by electrophoresis. Kinase
reactions and strand separations were done as described by Maxam and Gilbert2* and the Klenow polymerase I reaction and all further procedures as described by
K.A.N. et al.". Poly(A)+ RNA and single-strand end-labelled DNA fragments were annealed at 65C followed by S, (or ExoVII) digestion. After digestion the
samples were ethanol precipitated, redissolved in NaOH-urea loading buffer and run on 6.0% or 8.0% acrylamide sequencing gels2'. Gels were exposed to X-ray film
plus intensifying screen at -70C for 1-14 days.

0 1981Nature Publishing Group

Nature Vol. 289 22 January 1981

249

somehow affect the initiation of transcription within Y a , a

'~
situation reminiscent of position effects in D r ~ s o ~ h i l aand
similar phenomena at the haemoglobin locus in humans14.

Possible mechanisms for the position effect

Hypothetical mechanisms for the position effect can be divided
into two distinct types: one in which the DNA sequence is solely
responsible for the effect and another in which chromosome
structure is also involved.
The first type of mechanism could, for example, involve the
action of an enzyme like the bacterial type I restriction
e n d o n u ~ l e a s e swhich
~ ~ , can recognize a particular sequence on a
DNA molecule and then modify or cut the same molecule at a
point distant from the recognition site. It is possible that the
recognition site for such an enzyme (a SIR gene product?) exists
within the left-hand flanking region of HMLa and HMRa and
that DNA modification (or some other interaction)
subsequently occurs within Y a or Y a .
There are, of course, many variants of the second type of
mechanism in which chromosome structure has an important
role. These range from the very general, that HML and HMR sit
within a heterochromatic region, to the very specific, that the
precise composition or phasing of nucleosomes is responsible.
One can, in fact, distinguish two general pathways whereby the
left-hand flanking sequences of HML and HMR could influence
the structure of Y a and Y a . Information from the flanking
sequences could be transmitted to Y a or Y a either directly by
some form of physical bridge or indirectly along the axis of the
DNA molecule separating them. Higher order chromosome
folding might be involved in the case of the former, whereas the
precise phasing of nucleosomes might be involved in the latter.
We propose that of all the possible explanations for the
position effect, the most readily tested is the notion that the
sequences flanking the left end of HML and HMR are responsible, possibly in conjunction with SIR gene products, for
initiating, from left to right, a precise phasing of nucleosomes
such that promoters of transcription at Y a or Y a are either
hidden or sites for repressors of transcription (SIR gene
products) are exposed at these two loci but not at MAT. MAT,
having different flanking sequences, might either be phased
differently or not at all (as might be the case for a repressor
model). The phasing of nucleosomes could affect the availability or activity of control sequences at Y a or Y a at two levels.
At a gross level, it could be responsible for bringing sequences
into juxtaposition or not; for instance, by their being wrapped
around the same rather than a neighbouring core particle. At a
finer level, it could also, or instead, be responsible for the exact
orientation of sequences with respect to their coiling around a
core particle; for instance, whether they face in or out.
Possibly relevant to .the phasing hypothesis is the recent
observation by Hicks (personal communication) that restriction
fragments containing HMLa and HMRa both contain replication origins whereas that containing MAT does not. If the
inheritance of nucleosomes is conservative, one must postulate
some means by which the naked daughter chromatid is rapidly
covered with in-phase nucleosomes. This could be more readily
achieved if the replication origins are also orgins of phasing, as
found in the case of SV402'. The proximity of replication origins
at the silent copies may be sufficient to effect the correct phasing.
On the other hand, one or more SIR gene products may be
involved in initiating as well as maintaining phasing.
Whatever the mechanism of the position effect on the expression of MAT genes, it is worth pointing out possible parallels
with various phenomena associated with mating type interconversion. For instance, how does a homothallic yeast cell distinguish MAT from HML as the target of an interconversion
when they are so extensively homologous? It is tempting to
speculate that the same property that distinguishes their transcription may also be used to distinguish their roles as donor and
recipient in an interconversion of mating type. This notion raises
the question of the role of recombinant mating type loci (such as
the MATIHMR recombinant caused by the Hawthorne

deletion) in an interconversion; that is, whether they serve as

donors or recipients. Pursuing the analogy still further, a
conservative inheritance of nucleosomes at MAT or its silent
copies may be responsible for the ability of mother cells, but not
daughter cells, to undergo mating type interconversion4.'.

Regulation of MAT transcription and

the-determination of cell type in yeast
Genetic analyses have identified three genes involved in the
specification of cell type in yeast: a 1 , a 2 and a 1 (refs 9,lO).The
present investigation has shown that MATa and MATa each
produce two major transcripts starting in or near their Y regions,
one going leftwards and another rightwards. A series of
experiments in which plasmids containing MATa or MATa
were mutated in vitro and subsequently tested for in vivo
function have shown that the left a transcript corresponds to the
a 2 gene, the right a transcript to the a 1 gene, and the right a
transcript to the a1 gene17.This leaves the left ( a 2 ) transcript
unassigned to a known complementation group. The same series
of experiments implied that this transcript was, however, not
essential for any of the known functions performed by plasmids
carrying a wild-type MATa fragment; that is, deletions within
the a2 region had no effect on the plasmid's ability to suppress a
mating functions or to induce sporulation (in a / a cells). We
therefore believe the a 1 , a 2 and a 1 genes to be both necessary
and sufficient for specifying all three yeast cell types: the a
mater, the a mater and the non-mating but sporulating a / a
cell1.
How, then, do these three genes determine cell type?
Strathern et a1.' have suggested that the a 1 gene is necessary for
promoting the expression of unlinked a specific mating genes
and the a 2 gene for repressing the expression of unlinked and
constitutive a specific mating genes. With both of these active, as
in a haploids, a specific mating genes will be expressed but a
specific mating genes not. In the absence of either, as in a
haploids, only the constitutive a specific mating genes will be
expressed. The presence, albeit lower, of a 2 and complete
absence of a 1 transcripts in a / a diploids therefore accounts for
the non-mating phenotype of these cells (Fig. 1 ) .
The repression of MATa transcription that results in the
complete absence of a 1 and lowering of a 2 transcripts in a / a
cells requires both the a 1 and a 2 genes. The sequence between
the promoters of the a 1 and a 2 genes contains a symmetrical
region that may be the target of the hypothetical repressor of a 1
and a 2 . No similar structure is seen near the start of the MATa 1
transcript, which we have shown to be refractory to the repression by a1 and a 2 ,
The a 1 and a 2 genes are necessary not only for suppressing
mating functions in a / a diploids but also for inducing their
ability to sporulate. This function cannot be just a consequence
of turning off a 1 and turning down a 2 because it can be
expressed in circumstances in which the repression of a 1 and a 2
is absent2'. It is, therefore, an independent property of the
a l / a 2 state (Fig. 1 ) . Our analysis of transcription in
a / a S A D l I S A D l strains suggests that the induction of sporulation by a l / a 2 requires less a1 gene product than does the
repression of a 1 and a 2 transcription. The existence of recessive mutations (for example, CSP, ref. 26) that allow sporulation
of a / a or a / a cells suggests that sporulation is normally under
negative control. It is tempting to speculate that the a l l a 2
repression of a 1 transcription and induction of sporulation are
mediated by a common mechanism; that is, the transcription of
CSP is repressed by a l l a 2 in a manner similar, though not
identical, to that of a 1.
As the amino acid sequence of all MAT gene products has
recently been deduced from their DNA
we may
speculate on the physical mechanisms by which they control cell
type. We know that some of the consequences of their action are
regulation of transcription. However, we do not yet know
whether they act so themselves or through intermediates. With
this in mind, it is interesting that the MAT gene products are all
basic proteins that may be capable of binding to DNA.

O 1981Nature Publishing Group

Nature Vol. 289 22 January 1981

250
MATa
-

S t a r t a 1 RNA

~ T ~ C G C ~ - G K ~ B T G T A T A A A A C T T C C A A A A A G G A A A A G T A A A A C A A TnTCTCCTTRT
A C A ~ - - - - - - ~TC

Y) D

A E fieTGFe~Gg1,
AAhTC
,,rA8'
A GAGCGGAI
G \AACATA
AG T C T T T A G T T C T T C C G-G

GA k $ . C C C C T T T I G A C A T A T T T T G A A G G T T T T I C C T T T T C A T T T T G T T A T G T G A A

CC

CTATAAACAT ATACCGCCTTTTGTAT

AAGGTCTTTAATGGGTATTTTATTCATTTTTTCTTGCTTATCTTCCTTTTTTTCTTGCCCACTTCTAAGCTGATTTCAATCTCTCCTTTATATATPTTTTTAAGTTCCAACATTTTATGT
TTCCAGAAATTACCCATAAAATAAGTAAAAAAGAACGAATAGAAGGAAAAAAAGAACGGGTGAAGATTCGACTAAAGTTAGAGAGGAAATATATATAAAAATTCAAGGTTGTAAAATACA
L D K I P I K N(M)
-start
a2 RNA

T T C A A A A C A T T A TGATGTCTGG T TGT TGGG T CAAT T TTGC CCCA T G T G AAAGT C A T C A T T G A A A C A A C T T A A A C T C T T A A C T A C T T C T T T T A A C C T T C A C T T T T

AAGTTTTGTAAT~~GACC~ABCJACCC~A~~A~GGGT~ACA~C~TTTCA~~ACTTTGTTGAATTTGAGAATTGATGAAGAAAATTGGAAGTCAAAA
S t a r t a1 RNA

TATGAAATGTA~CAACCATATATA~~~-AACTTA~GA~GA~ATT~A~AATATGTTTA~TT~GAAG~~TG~
ATACTTTACATAGTTGGTATATATTATTGAATTATCTGCTGTAAGTGTTATACAAATGAAGCTTCGGACGAAAGTTTTAATTCTTGTTTCGTAGGTTTAGTATGTCTTTGTGTCGCCAAA

Fig. 4 DNA sequences around the starts of MAT transcripts. The DNA sequences around and between the starts of MATa and MATa transcripts are shown. The
sequences were determined by the methodof Maxam and Gilbertz8. The sequencingstrategy will be described elsewhere along with a more complete sequence of the
regionI9. In MATa, the X-Ya boundary is marked by a vertical bracket. The equivalent boundary for MATO and HMLa ( X - Y a ) is 48 base pairs beyond the left end
of the sequence shown. Putative coding sequences are signified using the one-letter amino acid code. Those of the a2 transcript are not marked because there is no
evidence that this transcript encodes a functional protein. Arrows mark the start points of RNAs as determined by S, mapping. Each transcript has two or more start
points. The boxed sequences are regions that have a dyad symmetry around an axis marked with an asterisk. At each end of the a symmetrical region is a 7-base pair
inverted repeat sequence that would extend the region of perfect symmetry except for 7 extra base pairs inserted between it and the left-hand end of the symmetrical
region. The dashed boxes in the MATa sequence are regions of symmetry with respect to purine and pyrimidine content.

We thank Hal Weintrub for helpful discussions on nucleosome

phasing, Don Hawthorne and George Sprague for supplying
crucial strains, and Jane Garnett for help in compiling the paper.
This investigation was supported by a grant to B.D.H. from the
National Institute of Medical Sciences, a Jane Coffin Childs
postdoctoral fellowship awarded to K.A.N. and a Damon
Runyon-Walter Winchell Cancer fund postdoctoral fellowship
to K.T.
Received 5 August; accepted 17 October 1980.
1. Mortimer, R. K. & Hawthorne, D. C. in The Yeasts Vol. 1 (eds Rose, A.H. & Harrison, J. S.)
3 8 5 4 6 0 (Academic. New York 1969).
2. Nasmyth, K. A. & Tatchell, K. Cell 19,753-764 (1980).
3. Strathern, J. N., Spatola, E., McGill. C. & Hicks, J. B. Proc. narn. Acad. Sci. U.S.A. 77,
2839-2843 (1980).
4. Hicks, J. B. & Herskowitz. I. Generics 83,245-258 (1976).
5. Strathern, I. N. & Henkowitz, I. CeN 17,371-381 (1979).
6. Naumov. G. I. & Tolstorukov. I. I. Genetika 9.82-91 (1973).
7. Harashima, S., Nogi, Y. & Oshima, Y. Generics 77. 6 3 9 4 5 0 (1974).

8. Hicks, J. B., Strathern. J. N. & Herskowitz, I. in D N A Inserrion Elemenrs Plasmids and

Episomes (eds Bukhari, A. I., Shapiro, J. A. & Adhya, S. L.) 4 5 7 4 6 2 (Cold Spring
Harbor Laboratory, New York, 1977).
9. Strathern, J. N.. Hicks, J. & Herskowitz. I. I. molec. Biol. (in the press).
10. Kassir, Y. & Simchen, G . Generics 82, 187-202 (1976).
11. Rine. I., Strathern. I., Hicks, 1. & Herskowitz, I. Generics 93, 877-901 (1979).
12. Klar, A. J. S., Fogel, S. & Macleod, K. Generics 93.37-50 (1979).
13. Lewis, E. B. A d v . Gener. 3.73-113 (1950).
14. Van der Ploeg, L. H. T. et a / . Narure 283,637-642 (1980).
15. Thomas, M., White, R. L. & Davis, R. W. Proc, narn. Acad. Sci. U.S.A. 73, 2294-2298
(1976).
16. Berk, A. 1.& Sharp, P. A. Proc. narn. Acad. Sci. U.S.A. 75, 1274-1278 (1978).
17. Tatchell, K., Nasmyth, K. A.. Hall. B. D..Astell. C. &Smith, M. Cell. (submitted).
18. Nasmyth, K. A., Tatchell, K., Hall, B. D., Astell, C. & Smith, M. Cold Spring Harb. Symp.
quanr. Biol. 45 (in the press).
19. Astell, C. el a / . Cell. (submitted).
20. Hawthorne, D. C. Generics 48,1727-1729 (1963).
21. Hopper, A. K. & Mackay, V. Molec. gen. Gener. (in the press).
22. Kassir, Y. & Herskowitz, I. Molec. gen. Genet. (in the press).
23. Roberts, R. J. C R C C r i r . Rev. Biochem. 4,123-164 (1976).
24. Leffak, I. M., Grainger, R. & Weintruh, H. Cell 12, 837-845 (1977).
25. Saragosti, S., Moyne, G. & Yaniv. M. Cell 20.65-73 (1980).
26. Hopper. A. K. & Hall, B. D. Generics 80.41-59 (1975).
27. Kahack, D. B., Angerer, L. M. & Davidson, N. Nucleic Acids Res. 6, 2499-2517 (1979).
28. Maxam, A. & Gilbert W. Merh. Enzym. 65,499-560 (1980).

Homology and non-homology at the

yeast mating type locus
George F. Sprague Jr, Jasper Rine" & Ira Herskowitz
Institute of Molecular Biology and Department of Biology, University of Oregon, Eugene, Oregon 97403

Four mutations of the a mating type locus of Saccharomyces cerevisiae have been analysed to determine their relationship
+
are produced by mata2-/MATa but not by matal -/MATa diploids.
to the a mating type locus. ~ a t a recombinants
MATa and MATa thus contain regions of homology (coding for at least part of MATa2) and regions of non-homology
(codingfor at least part of MATa1 )-the genetic determinant for cell type is larger than the non-homologous sequence seen
by DNA-DNA heteroduplexes and genetic analysis. The segment transposed in mating type interconversion includes both
types of sequence.
THE mating type locus determines which of three distinct and
complex cell types (a, a and a / & )is expressed by the yeast,

* Present address: Department of Biochemistry, Stanford University

School of Medicine, Stanford, California 94305.
0028-0836/81/040250--03$01.00

Saccharomyces cereoisiae. It has been proposed that alleles of

this locus, MATa and MATa, code for different regulators
controlling the expression of unlinked genes necessary for mating and sp~rulation'-~.
Several lines of evidence suggest that
MATa and MATa are 'complex' alleles, that is, are genetic
@ 1981 Macmillan Journals Ltd

Nature Vol. 289 22 January 1981

alternatives coding for several different functions. First, there
are at least two complementation groups in M A T a , a 1 and a 2 ;
only one complementation group is known for M A T a . M A T a 1
and M A T a 2 are required for mating in haploids; M A T a 2 and
M A T a are required for sporulation in diploids2-'. Second,
heteroduplex analysis of cloned M A T a and M A T a D N A (as
well as cloned cryptic mating type information from the H M L a
and H M R a loci) reveals a region of non-homology (region Y)
between a and a information of -750 base pairs (Ya for a and
Y a for a ) . Heteroduplexes between H M L a and M A T a and
between H M R a and M A T u show that the non-homology
region is flanked by regions of apparent homology of at least
700 base pairs on one side (the X region) and at least 200 base
pairs on the other side (the Z
The physical picture of
the mating type locus derived from heteroduplex analysis does
not provide information on the coding regions within the mating
type locus. Thus, they could be entirely within the nonhomology (Y), entirely within the regions of apparent homology
(X and Z ) or distributed throughout X, Y and Z regions. We
have made detailed studies of four mutations of the a mating
type locus, two in each of the known complementation groups,
to determine their genetic relationship to the a mating type
locus. T o investigate whether mata mutations are in regions of
homology or non-homology with respect to M A T a , we have
constructed a series of m a t a - / M A T a diploids and have
analysed the meiotic products for M A T a ' recombinants. If a
mutation is in a region of M A T a that is non-homologous with
M A T a , genetic rescue of the wild-type allele from M A T a
should not be possible, but genetic rescue should be possible if
the mutation is in a region that is homologous with M A T a . The
mutations analysed here (isolated by MacKay and ManneyR)all
differ: mata 1 - 5 causes a temperature-sensitive mating phenotype (G. F. S. and J. R., unpublished), m a t a l - 2 is an amber
mutation9, and mata2-1 and m a t a 2 - 4 have different effects on
sporulation (G. Sprague and J. Strathern, unpublished). We
report here results obtained with three types of diploid:
mata 1-/MATa, rnata2-/MA Ta, and m a ta 1-/mata2-, which
show that the wild-type alleles of m a t a 2 mutations but not of
mata 1 mutations are rescued from MA Ta.

Production of MA T a + segregants from

mata-/MA Ta diploids
Segregants that mate as a cells are found at the same low
frequency (-5 x 1 0 5 , among the meiotic products of two
mata I - I M A T a diploids carrying different mata 1 mutationsm a t a l - 5 and mata 1-2 (strains D G l and DG2, respectively,
Table 1). Subsequent crosses show that the a mating phenotype
is due to the presence of a wild-type a mating type locus in these
recombinants (see Table 1 legend). Because these MATa'
segregants may have arisen by recombination between the
m a t a l allele and cryptic a information at the H M L a locus
rather than by recombination between m a t a l and M A T a ,
we have constructed a mata 1-5/MA T a diploid lacking
cryptic a information. Such a strain (DG3, m a t a l / M A T a
H M L a I H M L a H M R a I H M R a ) does not yield MATa' segregants. This result indicates that the wild-type allele of the
mata 1 - 5 mutation cannot be rescued from M A T a .
In contrast to these results, M A T a ' segregants arise at high
frequency (-2 x lo-') from mata2-/MA T a diploids carrying
either of two different m a t a 2 mutations, matn2-4 and mata
2-1. This result has been documented most thoroughly for the
m a t a 2 - 4 mutation. In particular, the production of M A T a '
segregants from m a t a 2 - 4 / M A T a diploids is not dependent on
the presence of the HMLu allele (Table 1 : compare strain DG5,
which is homozygous for H M L a and H M R a , with strain DG4,
which is H M L a HMRa). Two observations indicate further that
the M A T a ' segregants are the products of recombination
between m a t a 2 - 4 and MATa. First, the frequency of M A T a '
segregants is at least 100-fold higher than the reversion
frequency of the mata2-4 mutation (HG2, Table 1). More-

-.

251

recombinants

recombinants

Fig. 1 Interpretation of genetic rescue experiments between

mara mutations and MATa. X, sites of mafa1 or MATa mutations. a, Failure to obtain MATa' recombinants in crosses
between mata 1 - 5 and MATa. Filled and open rectangles indicate
non-homologous segments of MATa and MATa, respectively. 6,

Production of MATa' recombinants in crosses between MATaand MATa. Upper diagram: symbols as in a. Lower diagram: a
region of homology is located between sequences that are nonhomologous in MA Ta and MA Ta.

over, half of the M A T a ' segregants are recombinant at the

closely linked C R Y 1 locus, which indicates that M A T a '
has been generated by a recombination event rather than by
reversion.
It is possible that the observed frequency of MATa' segregants from mata2-4/MA Ta diploids is an overestimation of the
true frequency, because D G 4 and D G 5 are defective in the
M A T a 2 function and therefore sporulate poorly. Wild-type
M A T a recombinants or revertants that arise in D G 4 o r D G 5
may be enriched among the spore progeny due to efficient
MATa'-promoted sporulation. T o test this possibility, we
determined the frequency of MATa' segregants from a mata24 I M A T a diploid (DG6) homozygous for a mutation, s i r l - 1,
which permits in situ expression of the cryptic mating type
information". This diploid sporulates well because wild-type
M A T a 2 function is provided by a information at HMLu.
M A T a ' segregants arise from D G 6 at the same frequency as
from D G 4 a n d D G 5 (Table 1). A m a t a 2 - l / M A T a s i r l - l l s i r l 1 diploid likewise yields M A T a ' recombinants at high
frequency (DG7, Table 1). Consequently, the frequency of
M A T a ' segregants from diploids D G 4 , 5 , 6 and 7 seems to be a
true reflection of recombination between M A T a and the
MAT& mutations and is not biased by a selection for revertants
o r recombinants.
T o evaluate the significance of the recombination frequencies
reported above, we have determined the frequency of recombination between two mata mutations. The frequency of
recombination in mata 1-/mata2- diploids (-10 ', D G 8 and
DG9, Table 1) is similar to that obtained from MATa-/MATa
diploids. This value for recombination across homologous
mating type loci places an upper limit on the frequency of

O 198lNature Publishing Group

Nature Vol. 289 22 January 1981

Table 1 Frequency of a mating type clones obtained

Strain
DG 1

Genotype

Frequency of a-maters
Clone
Frequency

matal-5

Regions of homology and non-homology

DG2

DG3

HMLa mata 1-5

HMRa
HMLa MATa HMRa

DG4

DG5

DG6

DG7

HMLa mata2-4
HMRa
HMLa MATa HMRa

MATa sirl-1

MATa sirl-1

DG8

DG9

HG1

HG2

MA Ta 2' recombinants expected from mata 2- 1/MA Ta or

mata2-4/MATa diploids (for example, D G 4 , 5 , 6 , 7 above), in
which MATa2' recombinants are formed by a recombination
event occurring between the mata mutation and the end of the
MATa non-homology region.

HMLa mata2-4 HMRa

Strains and markers: Diploid strains were constructed by prototroph

selection. Genotypes were confirmed by analysis of the segregants.
HG1 and HG2 were the mata parents of DG1 and DG5. All diploids
except DG1 were heterozygous at the CAN1 locus, which segregates
independently frwn MAT (ref. 8). DG1, 2, 3 and 5 were crylICRY1,
with cry 1 coupled to MATa in DG2 and CRY1 coupled to MATa in
the others. DG3 and DG5 were shown to be homozygous for HMLa and
HMRa; other strains are assumed to have the standard HMLa HMRa
genotype. Isolation of meiotic products: Haploid segregants were
isolated from two clones from each diploid (DG1, 2, 3, 4, 5, 6, 7) by
random spore analysis: asci from sporulated cultures were treated with
Glusulase (Endo), sonicated and plated onto medium containing cryptopleurine (for DG1) or canavanine (for the other diploids), which
selects for haploids because resistance alleles cry 1 and can1 are
recessives.15. Haploid segregants were isolated from sporulated cultures
of DG8 and DG9 by ascus dissection. Assay of MATa segregants:
Spore clones with a mating type were identified by replica plating
colonies onto a lawn of a cells and assaying mating by prototroph
formation. Frequency was the number of a colonies per total number of
spore clones tested. Numbers in parentheses indicate the number of a
colonies observed. Subsequent crosses showed that all a colonies tested
were MATa. Tests were performed on all a segregants from DG1,
DG2, DG8 and DG9; on 3 each from DG4A, DG4B, DGSB, DG6A
and DG6B, and on 19 from DG5A. Sporulation: MATa2IMATa
strains DG1, DG2 and DG3 sporulated efficiently (60% for DG3).
mara2/MATa Strains DG4 and DG5 sporulated poorly (4% for DG5).
mata2/MATa strains DG6 and DG7 sporulated efficiently due to
homozygosity for sirl-1 (47% sporulation for DG6). matallmata2
strains DG8 and DG9 sporulated (albeit inefficiently) due to homozygosity for rmel, which bypasses the requirement of MATa2 and
MATa 1 for sporulation4.

Our genetic results indicate that MATa and MATa contain

regions of homology as well as regions of non-homology: two
sites in MATa 1 are non-homologous with MATa, and two sites
in MATa2 are homologous with MATa (Fig. 1). Studies are in
progress to determine whether MATa' recombinants can be
obtained from mata-/MATa diploids and m a t a recombinants
from MATa-/MATa diploids. Although the extent and position of the homology between MATa2 and MATa cannot be
rigorously determined from our data, the similar frequencies of
MATa+ recombinants from mata2/MATa and mata2/mata 1
diploids lead us to favour the view that the homologous
sequence is outside the non-homologous sequence (Fig. lb, top)
rather than being within it (Fig. lb, bottom). These results
therefore suggest that the genetic information determining cell
type is larger than the non-homologous sequence observed in
heteroduplexes between MATa and MATa (refs 6, 7). This
view is reinforced by the observation that the a mating type
locus codes for two transcripts, one primarily within the nonhomologous region, and the other primarily outside this
region11v12.
Our observations on homology and non-homology
between MATa and MATa lead us to propose that the latter
RNA codes for MA Tcu 2 and the former for MA Ta 1.
Yeast cells can change the information present at M A T (an a
or a 'cassette') by replacing it with a cassette from HML or
HMR (reviewed in ref. 3). This change reflects a transposition of
at least the Y region information from HML or HMR to M A T
(refs 7, 13). Is the Y region the only information that is transposed? Our findings indicate that homologous sequences are
also transposed. This conclusion follows from the observations
that MATa mutations can be switched to MATa2' by mating
type intercon~ersion'~
and, as shown here, that MATa mutations are outside the non-homology region. Hence, the cassette
that is transposed during mating type interconversion must
include not only Y information but also flanking sequence
information (from X or Z).
We thank Kim Nasmyth and Jim Hicks for communicating
their results before publication, Karen Sprague for comments on
the manuscript, Julie Dunn for preparation of the manuscript,
and David Hagen for preparation of the figure. This work has
been supported by Research Career Development Award AI00163 and USPHS research grant AI-13462 to I. H., by PHs
Molecular Biology Training Grant GM-00715 (P. von Hippel,
Director), and by American Cancer Society postdoctoral
fellowship PF-1282 to G. S.
Received 11 August; accepted 4 November 1980.

1. MacKay, V. L. & Manney, T. R. Generics 76,273-288 (1974).

2. Strathern, J. N., Hicks, J. B. & Herskowitz, I. I. rnolec. Biol. (in the press).
3. Herskowitz, I. eral. in The Molecular Genericsof Deuelopmenr (eds Loomis. W. & Leighton,
T.) 79-1 18 (Academic. New York, 1980).
4. Kassir, Y. & Simchen, G. Generics 82, 187-202 (1976).
5. Klar, A. J. S., Fogel, S. & Radm, D. N. Generics 92,759-776 (1979).
6. Nasmyth, K. A. & Tatchell, K. Cell 19,753-764 (1980).
7. Strathern, J. N., Spatola. E., McGill. C. & Hicks. J . B. Proc. narn. Acad. Sci. U.S.A. 77,
2839-2843 (1980).
8. MacKay, V. L. & Manney, T. R. Genetics 76,255-271 (1974).
9. Herskowitz. I., Rine, J., Sprague, G. Jr & Jensen. R. in Mobilization and Reassembly of
Generic Informarion Vol. 17 (eds Scott, W. A,, Werner, R., Schultz, 1. &Joseph, D. R.)
133-151 (Academic. New York, 1980).
10. Rine, J., Strathern. 1.N., Hicks, J. B. & Herskowitz. I. Generics 93,877-901 (1979).
11. Klar. A. 1.S., Strathern, 1. N., Broach, 1. R. &Hicks, 1. B. Narure 289, 239-244 (1980).
12. Nasmyth, K. A., Tatchell, K., Hall, B, D., Astell, C. & Smith, A. Nature 289, 244-250

(1980).

13. Hlcks. J. B. & Herskowitz. I. Generics 85, 373-393 (1977).

14. Strathern, J. N., Blair, L.C. & Herskowitz. I. Proc. narn. Acad. Sci. U.S.A. 76,3425-3429
(1979).
15. Skogerson, L., McLaughlin, C . S. & Wakatama. E. J. Bacr. 116,818-822 (1973).

Publishing Group

253

Nature Vol. 289 22 January 1981

Nucleotide sequences at host-proviral

junctions for mouse mammary tumour virus
John E. Majors & Harold E. Varmus
Department of Microbiology and Immunology, University of California, San Francisco, California 94143

Proviruses cloned from rat cells infected with mouse mammary tumour virus, a B-type retrovirus regulated by glucocorticoid
hormones, show the structural features of transposable elements: short inverted repeats conclude long direct repeats at the
ends of viral D N A , and short sequences of cellular D N A are duplicated during integration and flank each provirus. The
integrative mechanism joins a precise site in viral D N A to non-homologous sites in host D N A .
sequences, DNA from each clone was digested with EcoRI,
INTEGRATION
of a DNA copy of viral RNA into the genome of
electrophoresed through agarose, blotted on to nitrocellulose
the host cell is a central event in the life cycle of retroviruses (see
paper by the technique of SouthernI4 and hybridized with
Fig. 1; for a review see ref. 1). The viral genome is composed of
"P-MMTV cDNA. Because there is a single recognition site for
two identical subunits of single-stranded RNA, each 5-10 kiloEcoRI within the MMTV(C3H) genome, each normal provirus
bases long. On infection, a linear duplex form of viral DNA,
should yield two hybridizing fragments, the sizes of which are
slightly longer than a subunit of viral RNA and bearing long
determined by the location of recognition sites in neighbouring
terminally repeated units ( L T R ~ )is~synthesized
~,
in the cytocell DNA (Fig. 1b shows a restriction map of MMTV DNA).
plasm. The LTRs are composed of sequences unique to the 3'
and 5' ends of viral RNA (U3and Us) and a single copy of a short
sequence (R) repeated at the ends of viral RNA, in a manner
represented by U3RUS(Fig. 1). Some of the linear unintegrated
DNA is carried to the nucleus where a portion at least is
converted to closed circular species5, mostly monomers which
LTR
contain either one or two copies of the LTR ~ e ~ u e n c e ~ . ~ . ~ - ' .
Integration of viral DNA into the host genome can occur at a
wide variety of sites in cellular DNA, possibly at random; the
structure of integrated (proviral) DNA is invariably very similar,
if not identical, to that of unintegrated linear DNA'.", although
the immediate precursor of the provirus has not been determined. The primary product of transcription is likely to be
equivalent to a subunit of genomic RNA, which can be processed to subgenomic, spliced mRNAs for expression of
internal cistrons'.
Mouse mammary tumour virus (MMTV) is a type B retrovirus
with at least three genes (gag, pol and env) on genomic RNA
subunits of eight to nine kilobases (see Fig. 1). Although its
generally poor infectivity in tissue culture has impeded biochemical study, MMTV seems to replicate like other retroviruses and has the experimental advantages of causing mammary carcinomas in mice and showing dramatic regulation of
viral RNA synthesis by glucocorticoid hormones in many types
of infected cells (for reviews see refs 11,12). The mechanisms of
carcinogenesis and steroidal regulation by this agent are poorly
understood, and an analysis of integration sites of MMTV DNA
seemed likely to clarify these aspects and also the general
Fig. 1 a, Steps in the replication of retroviruses. The RNA
mechanism of integration of retroviral DNA.
subunit of the MMTV genome (1) is shown with features relevant
We used molecular cloning and DNA sequencing to analyse
to the present study: the site for priming of the negative ( - ) DNA
the structure and integration sites of three proviruses of MMTV
strand by ~RNA'.'' bound near the 5' t e r m i n ~ s ~(shown
" ~ ~ on the
DNA introduced into the genome of rat cells (XC cells) by
left); the site probably used for priming (+)strand ~ ~ n t h e s i sthe
~.~,
infection in culture. Our analysis demonstrates the remarkable
short direct repeat (R,probably -13 bases for MMTV) at both
precision of the integrative mechanism and reveals additional
ends of viral RNA; and the regions unique to the 3' and 5' ends of
striking similarities between the structure of proviruses and that
viral RNA located between the priming sites and R (U3,-1,300
of several transposable elements: short inverted repeats (six
bases; US, -130 bases). At an early step in DNA synthesis these
two regions are fused via the R sequence to form the LTR U3RUS
base pairs) within the LTRs (1.4 kilobases) terminate each
which is in turn duplicated during the completion of linear duplex
provirus, and duplications of five or six base pairs of host cell
D N A ~ ~ 'The
" linear DNA (2) is a precursor to the two circular
DNA are generated during the integration process.
forms
which
contain either two copies (3) or one copy (4) of the
To optimize the possibility of correlating structural and
LTR nit^.^'^". One of these unintegrated forms (2, 3 or 4) is
functional properties of proviruses, we isolated a set of clonal rat
presumably the precursor to the integrated (proviral) f ~ r m ' . ' ~ . ~ ~
(XC) cell lines which contain single MMTV proviruses. The lines
( 5 ) . which serves as the template for synthesis of viral RNA. b, A
were chosen from a larger set of clonal lines of XC cells which
map of restriction sites in Mh4TV-C3H DNA which are relevant to
had been infected by about one infectious unit per cell with the
the present
AS described in the text we have been unable
to clone the region marked 'poison'.
C3H strain of MMTV13. To analyse these clones for MMTV
Q 1981 Macrn~llanJournals Ltd

254

Nature Vol. 289 22 January 1981

a

kiloba

Fig. 2 a, Identification of MMTV proviruses in clones of infected rat cells. Rat (XC) cells46were infected with MMTV-C3H, seeded in

multiple-well plates at a concentration < 1 cell per well, and propagated in Dulbecco's minimal essential medium plus 10% calf serum (without
added hormones). Purified DNA prepared from each clonal line (-10 kg) was digested with EcoRI, electrophoresed in 0.8% agarose,
transferred to nitrocellulose sheetsI4 and hybridized with 32~-labelledcDNA randomly transcribed from MMTV-C3H R N A ~ . ' ~The
. figure
shows an autoradiogram of an analysis of DNAs from four clones, including 8, 15 and 34, described in the text. b, Strategies for cloning the
virus-cell junctions of the proviruses from lines 8 and 34. DNA (mg aliquots) from each line were cleaved with EcoRI and passed through 0.8%
agarose in a fractionating gel electrophoresis apparatus4'. The eluted fractions were analysed for fragments of interest as described for a. To
isolate the junctions from line 8, DNA from the fractions containing the 12-kilobase EcoRI fragment was ligated with purified Charon 4A arms
' . assembled phages were plated onto the strain DP50SupF (ref. 18) and screened
and then packaged in vitro as described by Blattner et ~ 1 . ~The
for proviral inserts by the procedure of Benton and ~ a v iusing
s ~ MMTV
~
32~-labelled
cDNA,,,. A positively hybridizing plaque was isolated for
further analysis. The insert from this phage was transferred into pBR322 using vector DNA that had been cleaved with EcoRI and treated with
calf intestinal alkaline phosphatase. DNA from one such clone was isolated and used as a substrate for restriction cleavage analysis. For
sequencing, the indicated SacI fragments were purified by preparative electrophoresis through 0.8% Sea-Plaque (Marine Colloids)agarose gels,
then further restricted for the construction of fine structure maps. The two junctions from the provirus in line 34 were isolated separately. DNA
from fractions containing the right end 4.4-kilobase EcoRI fragment was ligated with purified arms from the vector AgtWES. AB, then
packaged, plated and assayed as described above. After isolation, the purified insert was then transferred into pB322 and used in restriction
enzyme analysis. DNA from fractions containing the six-kilobase EcoRI fragment from the left end of the provirus was further digested with
SacI, then ligated to purified arms from the vector Charon 10 (ref. 20). Packaging, plating and screening were as described above. After
preliminary characterization and preparative electrophoresis through 0.8% Sea-Plaque agarose, the inserted SacI fragment was cleaved with the
restriction enzyme Sau3A, and the cleavage products ligated to BamHI-cleaved pBR322 that had been treated with calf intestinal alkaline
phosphatase. After transformation of the bacterial strain HBlOl (ref. 50), ampicillin-resistantrecombidants were screened with a probe specific
for the MMTV LTR, using the procedure of Hanahan and ~eselson", for those containing the junction.

Using this criterion, of the 4 5 clones analysed, 1 2 had two o r

more proviruses, 1 4 had a single provirus and 19 had escaped
infection. Figure 2a shows the results of such an analysis of four
of the clones carrying single proviruses. We chose three of these,
designated 8, 15 and 34, for further analysis. O n e (clone 8)
carried a truncated provirus (see Fig. 2b) expressed under the
control of glucocorticoid hormones (our unpublished data and
those of D . Robertson); the second (clone 34) contained a
full-sized provirus from which little, if any, stable R N A was
transcribed in the presence or absence of hormone (Fig. 2b,
unpublished data); the third, not fully analysed here, contained a
full-sized provirus under steroid control (unpublished data).

Cloning unintegrated and proviral

MMTV DNA
Our choice of strategies for cloning MMTV proviral D N A was
influenced by a difficulty we encountered in attempting to isolate
unintegrated forms of MMTV DNAs by molecular cloning in
Escherichia coli (similar difficulties have been encountered in

attempts to clone integrated MMTV D N A (ref. 15 and D.

Ucker, personal communication)). Circular MMTV molecules
were isolated from dexamethasone-stimulated rat XC cells that
were chronically infected with the C 3 H strain of MMTV.
(Heterologous cells chronically infected by MMTV generally
contain 1-10 copies of unintegrated DNA, apparently
synthesized by reverse transcriptase, when grown in the
presence of g l ~ ~ o ~ o r t i ~ o i d s After
~ ~ ' ~ partial
~ " . ) purification by
Hirt fractionation and CsC1-propidium diiodide equilibrium
centrifugation6'16, the viral D N A was cleaved with EcoRI and
cloned into the vector AgtWES.AB (ref. 18). Although the
virus-specific material that we ligated to the vector arms seemed
to be primarily intact EcoRI-cleaved circular molecules, each of
the positively hybridizing AgtWES . MMTV recombinants had
suffered deletions o r rearrangements of MMTV sequences. The
deletion end points were variable, but each lesion covered a
common set of MMTV sequences that lie within a region of the
genome (which is designated 'poison' in Fig. 1) which probably
codes for gag proteins.
T o investigate this phenomenon further, we cleaved the same
D N A preparation with PstI, an enzyme with several recognition

O 198lNature Publishing Group

255

Nature Vol. 289 22 January 1981

sites within the MMTV genome, and cloned the fragments in the
PstI site of the plasmid vector pBR322. W e isolated recombinants containing all the expected MMTV PstI fragments
except the 1-kilobase fragment containing the 'poison'
sequence. We concluded from these experiments that there is a
sequence within the 1-kilobase,Pst I fragment which propagates
poorly, if at all, in the strains of E. coli used. We cannot explain
this result and have not yet isolated those sequences. In analysing integration sites, we were forced to use strategies which
allowed us to circumvent this cloning difficulty.
Figure 26 shows partial restriction maps of the two proviruses
and describes the cloning strategies. The provirus from clone 34
seems normal, that is, the restriction fragments generated from
internal regions of the provirus were identical to those of
unintegrated linear DNA. The provirus in clone 8 is abnormal in
that cleavage with EcoRI released only a single 12-kilobase
fragment which hybridized with MMTV cDNA,,,. Further
analysis with PstI, BamHI, Hind111 and SacI showed that the

(I)

Strong stop

3'.

. .CCGGCTGACGCCGTCG

5'

Rimer bindine site

Fig. 4 Schematic diagram of an MMTV provirus. The inverted

repeats at the ends of the LTRs are shown as inverted solid
triangles. The direct repeat of the cell sequence at the junction site
is indicated by C within the open arrow. Viral DNA is shown as a
double line, flanking cellular DNA as a single line. Other symbols
are as described for Fig. 1.

clone 8 provirus had incurred a large deletion of viral sequences,

probably covering the gag-pol region of the MMTV genome;
this deletion was fortuitous in that it eliminated both the poison
sequence and the internal EcoRI site, and we were able to clone
the entire steroid-responsive provirus with its flanking cell
sequences in the EcoRI site of the A phage vector Charon 4A
(ref. 19).
T h e strategy for cloning the ends of the proviruses in clones 34
and 15 was dictated by the need to circumvent the poison
sequence. The EcoRI fragments containing the right ends were
cloned in AgtWESSAB. The SacI site in the U3 region of the
LTR at the left end separates the virus-cell junction from the
poison sequence; therefore the left-hand virus-cell junction
from line 34 was cloned as a SacI fragment in the lambda phage
vector Charon 1 0 (ref. 20) (Fig. 2b).

Determining the boundaries of the LTR

.. .. ..

T A G T T C A C T TA
ArcnnGTcnf~~QE

Fig. 3 a, Junctions at the right ends. The end of US was deter-

mined from the sequence of 'strong stop' ( - ) strand DNA (RU5)

synthesized in oirro (1) and from the sequence of the junction
between the left-hand LTR and the region encoding the binding
~ ~The
' latter was derived from a PstI subclone of
site for ~ R N A (2).
the relevant region from line 8 provirus. These sequences were
then compared with the sequences of the junctions between viral
and host DNA at the right ends of proviruses from lines 8, 15 and
34 (3). The sequence of strong stop DNA was determined by two
methods: a modification of the di-deoxy chain termination
methods2 was used to analyse products synthesized by purified
avian myeloblastosis virus (AMV)DNA polymerase on a template
of MMTV-C3H 70s RNA; the chemical sequencing technique of
Maxam and Gilberts3 was used to analyse strong stop DNA
synthesized by detergent-disrupted virions of MMTV-GR (provided by Dr Nurul Sarkar). For each junction sequence, both viral.
cell and Us:primer binding site, DNA was labelled at a HpaIl site
within the U5 sequence, 30 base pairs from its right end. b,
Junctions at the left ends. The end of the Us sequence was
estimated from the position of the indicated PsrI site and from a
comparison of sequences from the Schmidt-Ruppin A strain of
' and from MMTV line 8 proviral DNA at the
ASV D N A ~ ~ "(4)
leftward boundary of the right-hand LTR (5) (see text). The
11-base homology is boxed in and the arrow indicates the boundary of the ASV U3 region. These sequences were then compared
with those from the viral-host junctions at the left ends of the two
proviruses (6).The ( + ) primer:U3 junction was derived from DNA
labelled at the BglIl site within the line 8 proviral sequences. The
cell-virus junctions were derived from DNA labelled at sites within
the host flanking sequences, a HaeIII site for line 8 and a Sau3A
site for line 34 (see Fig. 26). Presumed viral US and U3 sequences
are enclosed within solid lines and the repeated cell sequences are
indicated by large letters. Sequenceswere obtained by the chemical
cleavage method of Maxam and Gilberts3, using DNA fragments
that had been labelled at 3' ends either with AMV reverse transcriptase or E. coli DNA polymerase I.

T o determine the fate of viral sequences during integration we

had to know accurately the structure of viral D N A before
integration. It was particularly necessary to define the boundaries of the long terminal repeat, the right and left ends of
which are also the right and left ends of the linear form of
unintegrated DNA (Fig. 1).
The right end should be the site of initiation of viral reverse
transcription2', a sequence determined by direct DNA
sequencing of the initial or 'strong stop' ( - ) strand D N A made
in vitro (unpublished). Figure 3 a shows the sequence of the
region in the provirus from clone 8 covering the boundary
between the left LTR and viral sequences outside the LTR, that
is, the boundary between US (the strong stop sequence) and the
tRNA primer binding site. The strong stop transcript begins
GCTGC, which is immediately preceded by 18 bases identical to
the 3' end of tRNA:-Y', the primer for MMTV negative strand
synthesis22.
Present replication models argue that the left end of the LTR
should correspond to the initiation point for ( + ) strand strong
stop synthesis4. We have no direct evidence with which to define
this point; therefore we have resorted to indirect arguments.
There is a PstI site within the LTR, near the left end of this
previously described repeat unit of -1.4 kilobases (see Fig. 1
and refs 6, 10, 23). Figure 3b shows the sequence extending
leftwards from that PstI site, towards the viral BglII site, which
is on the 5' side of the rightward LTR. This sequence should
contain the left boundary of the LTR, and within it are two
features which seem useful for defining the boundary: ( I ) in the
(+) strand there is a polypurine tract of 20 bases ending -10
bases from the PsrI site; (2) a sequence of 11 base pairs,
including the last nine of the polypurine tract, has precise
homology with a sequence from the analogous region of avian
sarcoma virus (ASV) D N A 24.2s.Swanstrom ~ t a 1 . 'have
~ shown,
by sequencing the junction between the LTRs of a cloned ASV
circular molecule containing two copies of thc LTR, that thc left
end of the ASV LTR is probably at the position marked with an
arrow in Fig. 36. W e suggest that the left end of the MMTV LTR
is at the homologous site and that the polypurine tract is
involved with plus strand strong stop priming. Comparison of
the proposed sequences at the left and right ends of the LTR unit
show them to be related by an inverted repeat of six bases
(AATGCCGC . . . . GCGGCAGC), starting two base pairs
from the ends. This supports our choice of boundaries because
short inverted repeats of various lengths have been found at the

O 1981Nature Publishing Group

256

Nature Vol. 289 22 January 1981

for MMTV proviruses in clones 8 and 34. There seem to be

ends of the LTRs of murine leukaemia and sarcoma v i r ~ s e s ~ ~ - ~ ' sites
,
no common features between these sites or between either site
spleen necrosis virus (SNV)" and ASV (ref. 24).
and the ends of the proviruses. In addition, there is no obvious
Sequences at the host-viral junctions
relationship between these sequences and those flanking the
right end of the provirus in clone 15 (Fig. 3 a ) or those flanking
Sequencing of virus-cell junctions was facilitated by transferring
the right ends of three proviruses present in a mouse mammary
the EcoRI fragments from Charon 4A.8, AgtWES.34R and
tumour ceIll5.This limited survey is thus consistent with the idea
AgtWES.15R into the plasmid pBR322. To analyse the cellthat integration occurs randomly within the host genome;
virus junction in Charon 10.34L, the Sac1 insert was cleaved
however, our small sample size makes the detection of any
with Sau3A and the fragment containing the junction cloned
subtle
specificities unlikely. For instance, fine structure analysis
into the BamHI site of pBR322. Restriction maps derived from
of the target sites for a large number of insertions of the
certain of these plasmid clones are shown in Fig. 2. Sequences
transposable elements Tn3 and Tn9 shows a correlation
from the right-end junctions were obtained by labelling at a
between the probability of integration and both base composiHpaII site in the Us region and sequencing across the junction
tion and homology with end sequences3'.
into host DNA. Figure 3 compares those sequences with that
from the boundary between the right end of the leftward LTR
and the tRNA primer binding site. Two proviruses are missing at
Models for integration of retroviral DNA
least one base of Us sequence and the right end of the third
This analysis of MMTV DNA and similar analyses of other
provirus (from line 15) is missing at least two bases from the Us
retroviral
argue that any model for retroviral
sequence. (All three probably lack the first two bases of Us at
integration should account for the following observations: (1)A
this end.) Also shown are the viral-host junction sequence
precisely defined region of viral DNA, corresponding within a
derived from the 34L Sau3A subclone and that of the left
few base pairs to the ends of linear DNA, seems to be invariably
junction of the provirus from line 8 derived from the EcoRI
linked to host DNA. This implies either that a mechanism exists
fragment cloned into pBR322. A comparison with the expected
to recognize a specific site in viral circular DNA or that linear
left end of the LTR shows that each provirus is lacking two base
DNA integrates directly. (2) A limited number of base pairs,
pairs from the proposed terminus of the LTR.
probably two, are lost from both the right and left ends of viral
Comparison of the cellular sequences at the right and left
DNA during integration. Note that these bases are 'lost' only
junctions of the provirus from both lines 8 and 34 shows that
with respect to the probable structure of the unintegrated linear
there are direct repeats in the cellular sequence at the site of
DNA. The proximal precursor to the provirus could be a circular
integration-these comprise six base pairs for both proviruses.
form with a different complement of sequences in the relevant
In each case, the assignment of a GC base pair to cellular or viral
region. Furthermore, the lost base pairs do not affect the genetic
sequences is ambiguous. However we argue that symmetry and
consistency make a six-base pair repeat of cell DNA, with two
base pairs missing from both ends of viral DNA, the logical
provisional assignment.
The two base pairs removed are those which are mismatched
in the inverted repeats at the ends of the LTR, and their removal
leaves a perfect inverted repeat of 6 base pairs in the viral
sequences at the ends of the provirus. Thus an MMTV provirus
has the structure shown in Fig. 4: unique coding sequences,
flanked by large terminal direct repeats, which in turn are
flanked by short inverted repeats, with a direct repeat in the cell
sequence at the integration site. These features are strikingly
analogous to those of several transposable elements, parti. ~ ' ,of D r o ~ o ~ h i l a and
~'-~
Tyl
~ of
cularly Tn9 of E. ~ o l i ~ ~copia
yeast35-37.

The repeated cell sequence flanking the

provirus is duplicated due to integration
It has been shown for several transposable elements that the
direct repeat in the host sequence which flanks the element is the
result of a duplication generated during integrative recombination3'. To verify that a similar process is involved in retrovirus integration and to exclude the formal possibility of
integration into a duplicated region of host DNA, we cloned the
EcoRI fragment into which the provirus from line 8 was
integrated. A restriction map of that fragment is indistinguishable from that of the flanking sequences in the fragment
containing the line 8 provirus (Fig. 5). The sequence of the
integration site shows that the six-base pair repeat which flanks
the provirus in clone 8 was present only once in the integration
site before insertion; thus we conclude that the duplication was
generated during the integration process.

Specificity of integration sites

Analyses of proviruses of several retroviruses indicate that there
must be many integration sites available for each virus in each
host Ce119.10.23.39(see Fig. 2). However, mapping procedures are
inadequate to determine whether a repeated host sequence is
common to the integration sites used by any of these viruses.
Figure 6 shows a comparison of the sequences for the integration

Fig. 5 The target site for the provirus in line 8. A hybridization

probe made against the entire 12-kilobase EcoRI fragment from
the line 8 provirus cloned in pBR322 was used to isolate fractions
containing the target site fragment from a preparation of fractionated EcoRI-cleaved D N A from line 34. D N A from those
fractions was ligated to purified AgtWes . AB arms and a recombinant phage carrying the fragment which contained the target site
was isolated as described in Fig. 2 legend. The purified recombinant D N A was used for restriction enzyme cleavage analysis. For
sequence analysis the EcoRI insert was first transferred into
pBR322; the PstI fragment containing the target site was then
sub-cloned into the PsrI site of pBR322. The sequence was
obtained by Maxam-Gilbert chemical cleavage analysis of the
Sau3A-PstI fragment which had been labelled at the 3' terminus of
the Sau3A site. The relevant part of a 1O0/0 acrylamide sequencing
gel and the inferred sequence are shown. The band present in
all four lanes is the result of a partial Sau3A cleavage 5' to the
sequence C A T C .

0 1981Nature Publishing Group

Nature Vol. 289 22 January 1981

34

257

. . T T C A C G G G C T G A A A G T C T G G A T EtEMGC AT TATACGATTAGCAT TA TAXXCGAT ,, ,. ,,

, , ,AAGTGCCCCACTTTCAGACCTA

Fig. 6 A comparison of two targets for MMTV integration into

rat XC cell DNA. The sequence of the integration site used for the
provirus in line 8 was determined directly as described in Fig. 5
legend. The sequence of the site used in line 34 was deduced from
the sequences flanking the ends of the provirus (Fig. 3).

capacity of the provirus because they originate from regions

which are repeated in linear D N A but present uniquely in viral
RNA (see Fig. 1). (3) Host sequences are duplicated at the siteof
integration. (4) Viral D N A terminates with short inverted
repeats. (5) A large number of regions in host D N A , including
sites with sequences unrelated to each other o r to viral DNA,
can accommodate proviral DNA.
Although proviral elements from all retroviruses analysed
have shown these general features, there are potentially
significant differences in their fine structures: (1) The terminal
base pairs which are lost from MMTV D N A after integration are
not included within terminal inverted repeat sequences, whereas
those from other retroviral DNAs are part of an inverted
repeat27-29.(2) The size of the inverted repeats in proviral D N A
varies from 20 base pairs (with four mismatches) in Moloney
murine leukaemia virus (Mo-MuLV) D N A ~ 'to three base pairs
in SNV DNA2'. (3) The length of duplication in cellular
sequences generated by integration seems to be specific for each
retrovirus; thus, Moloney strains of MuLV and murine sarcoma
virus (MSV) generate a four-base pair repeat2'.", SNV
generates a five-base pair repeatz9 and MMTV probably
generates a six-base pair repeat.
The duplication of cellular sequences strongly suggests that an
early step in the integrative process will prove to be the introduction of a staggered scission in host chromosomal D N A , as
proposed for transposable element^^^.^'. The lack of homology
between the two MMTV integration sites described here
indicates that the enzyme responsible for this nucleolytic activity
may have little sequence specificity. No such host o r virus-coded
activity has yet been described, although an endonucleolytic
activity, encoded in the viral pol gene, has been found in virions
of avian myeloblastosis virus42.
Unfortunately, our structural studies shed little light on the
identity of the viral integrative precursor. Both the linear and
the two circular forms of unintegrated DNA can be accommodated as precursors within credible models for integrationz8,
each of which is closely related to that proposed by Shapiro4' for
the transposition of bacterial transposons. Recent work on the
mechanism of restriction of MuLV replication by alleles at the
Fv-1 locus can be interpreted as supporting a circular preRecewed 28 August: accepted 13 November 1980
1. Bishop, I. M. A . Reo. Riochem. 47, 35-88 (1978).
2. Shank. P. R. r r a l . Crll IS, 1383-1396 (1978).
3. Hsu,T W..Sahran. J L .Mark.G. E..Guntaka,R. V. &Taylor. J. M. J. Virol 28,810-818
(197x1
4. Gilboa. E Mitra. S. W.. Goff. S. & Haltimore, D Cell 18.93-100 (1979).
5. Shank, P. R. & Varmus, 11. 1:. I . Virol. 25, 104- 114 (1978).
6. Shank. P. R., ('ohen. I. C Varmus. H. E.. Yamamoto, K. R. & Rmgold. G.M Pror nnm.
A r a d . Srr U . S . A 75, 21 12-2116 (19781.
7 Yoshtmura. F. K. & Weinberg. R. A. Cell 16, 323-332 (1979).
8 Kung. H. I.. Shank, P. R., Bishop, I. M. & Varmus. H. E. V~rulogy103,425-433 (19811).
9. Hughes, S. H. PI a/. Crll IS, 1397-3410 11978).
10. R~ngold,(;. M..Shank.P. R.. Varmus. H. E.. Riny. I. & Yamamoto. K . R. Pror naam. Acod.
Sri C I S A 76.665- 669 (1979).
11 Moore, D. H. el 01. Adv. C u n r ~ Re?.
r
29, 347-418 (1979).
12 Varmus. H. E., R~naold.(i & Yamamoto. K. R In Glucvcomcord Hormonr Arrion ( e d ~
Baxter. J. 11. & Rousseau. G. G.) 253-278 (Sprmger, Herlln. 1979).
13. Fine, D. L., Plowman. I. K.. Kelley, S. P., Arthur. L. 0 .& Hillman. E. A. J. natn. Cancer
Insr. 52. 1881-1886 (1974).
14. Southern. E. M. J. molrr. H i d . 98, 503-517 (19751.
15. Hager, G. I.. & Donehower. L A in Animal VirusGenetics: I C N - 1 l C I . A Symp. molrr rell.
H i d . (eds Fields. B.. Jaenisch. R. & Fox, C. F.) (Academic. New York, in the press).
16 Ringold. G. M., Yamamutu, K. R., Shank, P. R. & Varmus. H. E. Cell 10, 19-26 (1976).
17. Ringold, G . M., Shank, P. R. & Yamamoto, K. R. J. Virul. 26.93-101 (1978).
18. Lcdcr. P . . 7'1emeier. D. & Enquist. L. Scknrr 1%. 175-177 (1977).
19. Rlattner. F R cr nl Srienre 196. 161- 169 (1977).

c u r ~ o r ~ In
" ~some
~ . cases, Fv-1 alleles inhibit both the accumulation of circular forms of MuLV D N A and the integration of
MuLV D N A into mouse DNA. However, it is possible that
these two observations both reflect the defective nature of linear
D N A ; thus the identity of the proximal precursor to the provirus
remains an open issue.

Functional implications of proviral structure

The significance of the structural similarities between proviruses
and transposable elements (Fig. 4) is uncertain in the absence of
mechanistic details about retrovirus integration o r about transposition of bacterial o r eukaryotic transposons. Duplications at
the sites of integration argue that the generation of staggered
single-stranded scissions in host D N A is an early event in both
processes. Terminal inverted repeats should allow an enzyme
with a single recognition specificity to operate on both ends of an
integrating molecule; however, this feature may not be essential
for transposition because the T y l elements of yeast lack
terminal inverted repeats36.".
There is no information concerning the possibility that proviruses can be relocated at new sites in host genomes by means
other than transcription into RNA, subsequent reverse transcription and reintegration. Experiments to test for direct
transposition are difficult to design and unlikely to yield unambiguous results. However, there is evidence to suggest that
retroviruses can show other forms of behaviour observed for
transposable elements, including deletion formation, insertional
mutagenesis and promotion of transcription into flanking cellular DNA45. Each of these may have significant genetic
consequences for the host, apart from the expression of viral
genes.
The structural studies reported here d o not directly further
our understanding of hormonal regulation or carcinogenesis in
MMTV-infected cells. However, the provirus from clone 8 is
under glucocorticoid control and has been cloned intact with
large regions of flanking cellular D N A ; it should serve as a
useful reagent for examining the role of viral and context
(host) sequences in the mediation of the hormone response.
The sequence of the MMTV LTR (studies to be presented
elsewhere) includes an A+T-rich region characteristic of
putative eukaryotic promoters. This region, which reads
T A T A A A A G A , is positioned 24-32 bases from the probable
initiation site for R N A synthesis, the site corresponding to the 5'
terminus of genomic RNA. Thus we favour the notion that
LTRs, like repeated units in transposons, may provide signals
for gene e x p r e ~ s i o n ~ ~ - ' ~ .
We thank Bertha Cook for secretarial assistance. This work
was supported by USPHS grants C A 12705, C A 19289, Training Grant IT32 Ca 09043 and American Cancer Society VC-70.
J.E.M. was supported by a fellowship from the Jane Coffin
Childs Memorial Fund for Medical Research.
20,
21.
22.
21.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34
35.
36.
37.
38.
39

deWet. J. R. ur ul. J. Vrrol. 33,401-410 (IYXO).

Taylor, J . M. Brochrm. hiophys. Arra 473.57-72 (1977).
Peters. Ci & Glover. C C 3 Virol. 35. 31 40 (1980).
Cohen. I. C Shank. P. R.. Morris. V. L., Cardiff, R. & Varmus. H. E (211 16. 333 -346
(1979).
Swanstrom, R., 1Iel.nrhe. W l h h o p , I M & Varnlus. H E. Pruc, nurn. Acod. Srr. 1I.S.A.
(In the press).
Czernilofsky, A P.. Delorbe, W.. Swanstrom. R.. Varmus, H I:. & Hichop. I M N d r . i c
Acids Res 8. 2967-2984 (1980).
Sutcliffe. J . G., Shlnnick, T M.. Verma, l M. R l.erner. R A P r w nurn. Ar-ud. Scr. U.S.A.
77, 3.102-33Oh (1980).
Ilhar, R.. McClements. W. L Enquisl. L. W. & Vandc Woudc. (i. I-. Pror nnm A r a d Srt
U S A . 77,3937 3941 (1980).
Shoemaker. C. el ul. Pruc. narn. A r a d . Sci. f1.S.A. 77. 3932-3936 (1980).
Shimotohno. K., Mtzutdnl. S. & Iemin. 11. M Narrm 285. 550-554 119801.
MacHathe. I.. A. & Jackowski. J B In D N A In,r~riun E l r r n m r ~ .P1usmrd.i and Epi~orntr
(eds Bukhari. A . I.. Shap~ro.I. A. & Adhya, S. L.) 219-228 ((:old Sprlng Harbor
Laboratory. New York, 1977).
Juhnsrud, I.., Calos. M. & Miller, J. W. Cell IS, 1209-1219 (1978).
Strohel, t.,Ilunsmu~r,P. & Ruhin. G . M Cell 17,429 439 (1979).
Dunsmuir, P.. Brorein. W I. Jr, Simon. M. A , & Rubin, G . M. Crll 21, 575-579 (198Ol.
Levis. R., Dunsmuir. P. & Rubin, G . M. Cell 21, 581-588 (1980)
Cameron, J. R.. Loh. E. Y. & D a m . R . W. ( P l l 16,7111-751 (1979)
Farahaugh. P. I. & Fink. G.R Narure 286. 352-356 (1980).
Gafner, J. & Phillipsen. P. N a r ~ r e286.414 418 (19801.
Calos. M P & Miller. J. H. Cell 20, 579-595 (1980).
Steffen. D. & Wrinbcrg. R. A. Cell 15, 1 0 0 3 1 0 1 0 (1978).

0 1981Nature Publishing Group

Nature Vol. 289 22 January 1981

2 58
40. Grindley, N. D. F. & Sherratt, D. I. Cold Spring Harb. Symp. quanr. Biol. 43, 1257-1268
(1978).
41. Shapiro. J. A. Proc. naln. Acad. Sci. U.S.A. 76, 1933-1937 (1979).
42. Golomb, M. & Grandgenett, D. P. I. biol. Chem. 254, 1606-1613 (1979).
43. Yang, W. K. eral. Proc. narn Acad. Sci U.S.A. 77,2994-2998 (1980).
44. Jolicouer, P. & Rassart. E. I. Virol. 33, 183-395 (1980).
45. Majors. J. E. er al. Cold Spring Harb. Symp. quanr. Biol. 45 (in the press).
46. Svoboda. J. Narure 186. 980-981 (1960)

Leder, P. eral. Cold Spring Harb. Symp. quanr. Biol. 42, 915-920 (1978).
Blattner, F. R. era/. Science 202, 1279-1288 (1978).
Benton, W. D. & Davis, R. W. Science 196,180-182 (1977).
Boyer, H. W. & Roulland-Dussoix, D. I. molec. Biol. 41,459-472 (1969).
Hanahan, D. & Meselson, M. Gene 10,63-68 (1980).
Sanger, F., Nicklen, S. & Coulson, A. R. Proc. nam. Acad. Sci. U.S.A. 74, 5463-5467
(1977).
53. Maxam, A. &Gilbert, W. Proc. narn. Acad. Sci. U.S.A. 74, 560-564 (1977).

47.
48.
49.
50.
51.
52.

Nucleotide sequence and formation of the

transforming gene of a mouse sarcoma virus
Charles Van Beveren, Janice A. Galleshaw, Vivian Jonas, Anton J. M. Berns*,
Russell F. ~oolittle',Daniel J. Donoghue & Inder M. Verma
Tumor Virology Laboratory, The Salk Institute, PO Box 85800, San Diego, California 92138
+Departmentof Chemistry, University of California, San Diego, La Jolla, California 92093

The complete nucleotide sequence of the transforming gene of a mouse sarcoma virus has been determined. It codes for a
protein of 374 amino acids. The nucleotide sequence of the junctions between a murine leukaemia virus and cellular
sequences leading to the formation of the viral transforming gene have also been elucidated. The viral transforming
sequence and its cellular homologue share an uninterrupted stretch of 1,159 nucleotides, with few base substitutions. The
predicted amino acid sequence of the mouse sarcoma virus transforming gene was found to share considerable homology
with the proposed amino acid sequence of the avian sarcoma virus oncogene (src) product.
MANY of the retroviruses capable of cellular transformation
have acquired a normal cellular sequence as an integral part
of their genome1. The acquired cellular sequences can be expressed either as an independent viral gene or as a product
fused with viral structural proteins2-'0. With the exception of
Rous sarcoma viruses (RSV), acquisition of cellular sequences
occurs at the expense of viral genomic sequences required for
Not surprisingly, most retroviruses capable of
inducing cellular transformation are unable to propagate
without helper viruses.
We were interested in the biogenesis of a transforminggene at
the level of nucleotide sequences. We chose to study Moloney
mouse sarcoma virus (Mo-MSV) which is able to transform
fibroblasts in vitro and induce neoplasia in vivo, but is unable to
repli~ate'~-'~.
It was obtained from sarcomas induced in
BALB/c mice after passage of high doses of Moloney murine
leukaemia virus (Mo-MLV)22.The genome of a cloned isolate
of Mo-MSV (clone 124, ref. 23) has been characterized
extensively. Briefly, the Mo-MSV genomic RNA is about
5.3 kilobases long and shares 60-7O0/0 sequence homology with
Mo-MLVZ4.Figure 1 shows a schematic heteroduplex formed
between a genome-length Mo-MLV cDNA transcript and MoMSV 124 genomic RNA. The Mo-MSV genome has a sequence
substitution of about 1.2 kilobases and two deletions of approximately 1.2 and 1.6 kilobases. About 2.0 kilobases at the 5' end
and 0.6-0.7 kilobases at the 3' end are shared between the two
We have molecularly cloned the unintegrated circular forms
of Mo-MSV clone 124 and Mo-MLV DNA in bacteriophage A
(refs 25, 26). by using molecularly cloned Mo-MSVWc-specific
DNA fragments as probe, a 14.0-kilobase EcoRI fragment from
uninfected mouse cells was identified and subsequently molecularly cloned in bacteriophage A (refs 27, 28). The cloned
cellular DNA fragment contains an uninterrupted stretch of
about 1.1 kilobases homologous to Mo-MSV-specific
sequences. The Mo-MSV-specific sequences will be referred
to here as v-mosMO[viral (v), mouse sarcoma virus (mos),

* Present address: Laboratory of Biochemistry, University of Nijmegen,

Nijmegen, The Netherlands.
0028-0836/81/04025845$01.00

Moloney (Mo)] and their cellular homologue is referred to as

c-mosMO[cellular (c)]. We were interested in determining the
nucleotide sequence in the region of the junctions of c-mosMO
and Mo-MLV sequences to learn more about the process leading to the formation of the v-mosMOgene. We present here the
following results: (1) the complete nucleotide sequence of the
v-mosMOgene of Mo-MSV clone 124; (2) the predicted amino
acid sequence of the v-mosMOgene product; (3) nucleotide
sequences of the 5' and 3' extremities of c-mosM0and Mo-MLV
leading to the formation of the v-mosMOgene; (4) nucleotide
sequence comparison of the portions of v-mosMOand c-mosMO;
(5) comparison of the predicted amino acid sequence of the
v-mosMOgene product with the proposed avian sarcoma virus
oncogene (src) product.

Construction of plasmids
The following plasmids were constructed for the purpose of
determining the nucleotide sequence of the appropriate regions
(Fig. 2). The plasmid, p.MSV-1, contains the entire Mo-MSV
DNA, whereas the plasmid, p.MSV-21, contains a majority of
the v-mosMOgene sequences. The cellular DNA fragment
containing the c-mosMOsequence was cloned in bacteriophage
Charon 4A and portions of it were subsequently subcloned in
the plasmid pBR322. The plasmid, p. CS-1, contains the 5' half
of the c-mosMO
and cellular sequences, whereas p .CS-2 contains
the entire coding region of c-mosMOand adjacent cellular
sequences. The plasmid, p.MLV-1, contains the entire MoMLV genome. The approximate position of the junction of
Mo-MLV and v-mosMOsequences was obtained from the
heteroduplex map (Fig. 1) and physical maps generated by
restriction e n d o n u ~ l e a s e s ~ ~ , ~ ~ .

Nucleotide sequences of the v-mosMOgene

The Maxam-Gilbert method of DNA sequencing was used. In
general, 5'+ 3' sequences were determined by labelling with
polynucleotide kinase and [y-32P]ATP, and the 3'+5'
sequences were determined by labelling the 3' tQd with [a"P]CTP and terminal transferase. The complete nucleotide
Q 1981 Macmillan Journals Ltd

Nature Vol. 289 22 January 1981

5' k

909

Pp.1

259
env

13'

Fig. 1 Schematic heteroduplex formed between genome-length

Mo-MLV DNA and Mo-MSV 124 genomic RNA". kb, Kilobases.

sequence of a DNA fragment encompassing the v-mosMOgene

of Mo-MSV 124 is shown in Fig. 3. The nucleotide sequence
from just before the first BglI site through the HindIII site
(1,356 base pairs) was obtained from p-MSV-21 (Fig. 2). The
nucleotide sequence of v-mosMOpast the HindIII site was
determined from p .MSV-1.
There is only one major region in which a reading frame in
either direction is open for translation. It starts with an ATG
about 10 nucleotides downstream from the single XbaI site (Fig.
2), and can encode a protein containing 374 amino acids. The
open reading frame terminates with an opal codon (TGA)within
the acquired cellular sequences. Figure 3 also shows the predic-

ted amino acid sequence of the v-mosMOgene product. Based on

the predicted sequence, the largest possible v-mosMOgene
product has a molecular weight of 40,992, which is in good
agreement with the estimated size of a 37,000 molecular weight
(37 K) protein and a related family of proteins synthesized in an
in uitro translation of Mo-MSV clone 124 genomic RNA3'. The
synthesis of the 37K protein and a related family of polypeptides
is specifically inhibited when the Mo-MSV 124 genomic RNA is
preannealed to v-mosM0-specificDNA fragments (J. Papkoff,
T. Hunter and I. M. V., unpublished results). The synthesis of
other viral proteins, like the p60 gag gene product, is not
inhibited by the v-mosM0-specificfragments. Note, however,
that the 37K and related polypeptides synthesized in uitro have
not been conclusively shown to be in viuo v-mosMOgene
products, as no antisera against these polypeptides are yet
available. Based on its amino acid sequence, the v-mosMOgene
product has a net positive charge, in agreement with the mobility
of the in vitro 37K motein on isoelectric focusine
" eels
(J. Papkoff and T. ~ u n t d rpersonal
,
communication).

Nucleotide sequence of the c-mosMOand

Mo-MLV junctions
To determine the nucleotide sequences at the 5' and 3' junctions
of the c-mosMOand Mo-MLV leading to the formation of the
v-mosMOgene, we used DNA fragments from the plasmids
p.CS-1, peCS-2 and p-MLV-1 (Fig. 2). Figure 4 shows the

Flg. 2 Construction of plasmids. a, p.MSV-1 and p.MSV-21. The unintegrated circular forms of Mo-MSV DNA from freshly infected
NIH/3T3 cells were isolated and cleaved at the unique HindIII site3*. The DNA was molecularly cloned in the unique HindIII site of
bacteriophage Charon 21A as described elsewhere2'. The insert DNA containing the entire Mo-MSV DNA with two long terminal repeats was
subcloned in the Hind111 site of plasmid pBR322. About 1.0 bg of p. MSV-1 was cleaved with restriction endonucleases HindIII and SalI and
gene was eluted from the gel and joined to
fractionated on 0.7% agarose gels. The 2.2-kilobase HindIII-SalI fragment containing the v-mosMO
HindIII- and SalI- cleaved pBR322 by T4 DNA ligase as described elsewherez5. The recombinant plasmids were used to transform Escherichia
coli C600SF and colonies were screened both for drug resistance ( ~ m ~ ~ ~and
e hybridization
t ' )
to MO-MSVcDNA probes. The bracket denotes
the v-mosMOsequences, b, p. CS-1 and p . CS-2. The 14.0-kilobase EcoRI fragment from uninfected NIH/3T3 cells containing the c-mosMO
sequences was cloned in bacteriophage Charon 4A. The recombinant clone, A MTX-1,was analysed by restriction endonucleasez8. The clone,
A Mrx-1, was cleaved with KpnI which generated an expected fragment of abaut 1.2 kilobases, containing the 5' half of c-mosMO
and adjacent
cellular se uences. It was eluted from an agarose gel and a stretch of oligo(dC) residues was added using terminal transferase, as described
previously8, and cloned in B e RlI-cleaved [dG]-tailed pBR322. To make the plasmid p . CS-2. the recombinant clone, A MTx-1. was cleaved
and adjacent cellular sequences was isolated from agarose gels and subcloned in
with BglI and a 5.5-kilobase fragment containing the 3' c-mosMO
sequences. c, p.MLV-1. The details of molecular
the plasmid pBR322 by the [dC]-[dG]-tailing methodd3. The solid box indicates the c-mosMO
cloning of unintegrated circular forms of Mo-MLV are described elsewherez6. Briefly, the Mo-MLV DNA was cleaved at an unique site with
HindIII and cloned in the unique HindIII site of bacteriophage Charon 21A. The recombinant clone containing the Mo-MLV DNA was cleaved
with HindIII and the insert Mo-MLV DNA further subcloned in the HindIII site of pBR322. The orientation of the insert in the plasmid was
determined by restriction endonuclease analysis. The wavy lines show vector DNA.

O 198lNature Publishing Group

260

Nature Vol. 289 22 January 1981

GGCCCCCATGGCCTCACCCCATATGAGATCTTATGTGGGGCACCCCCGCCCCTTGTMCTTCCCTGACCCTGACATGACAAGAGTTACTAACAGCCCCT

I
11
(thr)
21
met ala his ser thr ro cys ser gln thr s e r leu ala val pro asn his phe ser leu val ser hi5 val
201

AC

ATG GCG CAT TCA A

A TGC TCC CAA ACT TCC CTG GCT GTT CCT Y T CAT TTC TCC CTA GTG TCT CAT GTG
L

41
thr val pro ser glu gly val met pro ser pro leu ser leu cys arg tyr leu pro arg glu leu s e r pro ser

275

ACT GTC CCA TCT GAG GGT GTA ATG CCT TCG CCT CTA AGC CTG TGT CGC TAC CTC CCT CGT GAG CTG TCG CCA TCG

(val) 51
61
71
val asp ser arg ser cys ser i l e pro leu val ala pro arg lys ala gly lys leu phe leu gly thr thr pro
350

GTA GAC TCG CGG TCC TGC AGC ATT CCT TTG GTG GCC CCG AGG AAG GCA GGG AAG CTC TTC CTG GGG ACC ACT CCT

c-l

81
91
pro arg ala pro gly leu pro arg arg leu ala t r p phe ser i l e asp t r p glu gln val cys leu met his arg

425

CCT CGG GCT CCC GGA CTG CCA CGC CGG CTG GCC TGG TTC TCC ATA GAC TGG GAA CAG GTA TGT CTG ATG CAT AGG

soo

101
(tyr)
111
121
leu gly ser gly gly phe gly ser val tyr lys ala thr tyr his gly val pro val ala i l e lys gln val asn
CTG GGC TCT GGA GGG m GGC TCG GTG T A ~AAA GCC ACT TAC CAC GGT GTT CCT GTG GCC ATC AAG CAA GTA AAC

575

(1~s)
131
141
(ard
lys cys thr glu asp leu arg ala ser gln arg ser phe t r p ala glu leu asn i l e ala gly leu arg h i s asp
AAG TGC ACC f A G GAC CTA CGT GCA TCC CAG CGG AGT TTC TGG GCT GAA CTG AAC ATT GCA PGA CTA CGC CAC GAC

151
161
171
asn i l e val arg val val ala ala ser thr arg thr pro glu asp ser asn ser leu gly thr i l e i l e met glu
850

AAC ATA GTT CGG GTT GTG GCT GCC AGC ACG CGC ACG CCC GAA GAC TCC AAC AGC CTA GGT ACC ATA ATC ATG GAG

725

181
191
phe gly gly asn val thr leu his gln val i l e tyr asp ala thr arg ser pro glu pro leu ser cys arg lys
m GGG GGC AAC GTG ACT CTA CAC CAA GTC ATC TAC GAT GCC ACC CGC TCA CCG GAG CCT CTC AGC TGC AGA AM

201
211
221
gln leu ser leu gly lys cys leu lys tyr ser leu asp val val asn gly leu leu phe leu his ser gln ser
800

CAA CTA AGT TTG GGG AAG TGC CTC AAG TAT TCC CTA GAT GTT GTT AAC GGC CTG CTT TTT CTC CAC TCA C M AGC

231
241
i l e leu his leu asp leu lys pro ala asn i l e leu i l e ser glu gln asp val cys lys i l e ser asp phe gly
875

ATT TTG CAC TTG GAC CTG AAG CCA GCG AAC ATT TTG ATT AGT GAG CAG GAC GTT TGT AAG ATC AGT GAC TTC GGC

850

TGC TCC CAG M G CTG CAG GAT CTG CGG GGC CGG CAG GCG TCC CCT CCC CAC ATA GGG GGC ACG TAC ACG CAC CAA

1025

GCT CCG GAG ATC CTA AAA GGA GAG ATT GCC ACG CCC AAA GCT GAC ATC TAC TCT TTT GGA ATC ACC CTG TGG CAG

251
261
271
cys ser gln lys leu gln asp leu arg gly arg gln ala ser pro pro his i l e gly gly thr tyr thr his gln

281
291
ala pro glu i l e leu lys gly glu i l e ala thr pro lys ala asp i l e tyr ser phe gly i l e thr leu t r p gln

(thr) (arg)
311
(val)
321
aw)
net thr thr arg glu val pro tyr ser gly glu pro gln tyr val gln tyr ala val val ala tyr asn leu (arg
1100

ATG ACT ACC A? GAG GTG CCT TAC TCC GGC GAA CCT CAG TAC GTG CAG TAT GCA GTG G T t GCC TAC M T CTG CG:
t
G
T
C

1175

CCC TCA CTG GCA GGA GCG GTG TTC ACC GCC TCC CTG ACT GGA AAG GCA CTG CAG M C ATC ATC CAG AGC TGC TGG

331
(thr)341
pro ser leu ala gly ala val phe thr ala ser leu thr gly lys ala leu gln asn i l e i l e gln ser cys t r p

1250

351
(ala)
(91~)
361
371(ala)
glu ala arg gly leu gln arg pro ser ala glu leu leu gln arg asp leu lys ala phe arg gly thr leu gly
GAG GCC CGC GGC CTG CAG AGG CCG AGT GCA GAA CTG CTC CAA AGG GAC CTC AAG GCT TTC CGA GGG qCA CTA GGC
A
G
G

OP
1325

1421

TGA

CTCCATCGAGCCAGTGTAGAGATAAGCTTTTGTTTCTGTTTATTT~~~\TGW;ACCCCTTATTGTACTCCTAATGATTTTCCTCTTCGGACCC

--I

L-t

TGCATTCTTMTCGATTAGTCCAATTTGTTAAAGACAGGATATCAGTGGTCCAGGCTCTAGCTTTGACTCAACAATATCACCAGCTGAAGCCTATAGAGT

O 198lNature Publishing Group

Fig. 3 Nucleotide and

predicted amino acid
sequence of v-mosMO
gene.
The
complete
nucleotide sequence of
Mo-MSV DNA fragment
containing the v-mosMO
gene is shown. The
numbers in italics on the
left-hand margin refer to
the nucleotide numbers
(1-1,547). The numbers
in the figure (1-374) refer
to the predicted amino
acid sequence of the vmosMOputative gene product, starting with Met as
number 1 and terminating
with an opal codon, TGA.
The thick solid arrows
under the boxed nucleotides show the recombination 'unctions between
c-mosLO and Mo-MLV
sequences; the horizontal
arrowsindicate theregions
of the c-rnosMO whose
nucleotide sequence was
determined; the solid triangles and the nucleotides
underthemdenotethebase
alterations in the c-mosMO
sequences; the amino
acids in parentheses above
the predicted amino acid
sequences of v-mosMO
gene product are the predicted changes in the cmosMO product;
the
closed
bar
between
nucleotides 184 and 187
shows the position of the
192-nucleotide Mo-MLV
sequence deleted in MoMSV. The locations of
some endo. R cleavage
sites are: AcyI: 986;
A d : 385, 426; BalI:
557; BglI: 8,448; BglII:
26; ClaI: 1,432; HhaI:
208,681; HindIII: 1,351;
HinfI: 353, 692, 739,
1,083, 1,326, 1,486;
HpaI: 844; HpaII: 435,
448, 776, 979, 1,028,
1,125; KpnI: 711; PstI:
368, 794, 966, 1,227,
1,266; PvuII: 789, 1,504;
SsrII: 1,258; XbaI: 192.
A complete list of endo. R
restriction sites for this
sequence is available on
request.

Nature Vol. 289 22 January 1981

Fig. 4 The 5' and 3' recombination
junctions between Mo-MLV, MoMSV and A .MTx-1.The 5' and 3'
houndaries of the recombination
junctions between Mo-MLV and cmos Mo leading to the formation of
the v-mosMOgene sequence are
shown. The open boxes denote
homology whereas the hatched areas
indicate non-homology. The A T G
(below the arrow) indicates the
position of the predicted first amino
acid (Met) of the v-mosMOgene
product (Fig. 3).

261

p . ~ ~ ~ - l

p .MSV- 1

X.MTx-1

~'--~A~G~~AA/T~~A,GCCATGCTCCCA---------------TGTTTATTTTTT~~G~GJA~--~'

sequence of c-mosM0and p.MLV-1 at the 5' and 3' ends of the

v-mosMOgene sequence. The 5' junction of c-mosMOwith MoMLV occurs 12 nucleotides downstream from the initiation
codon of the putative v-mosMO
gene product. At the 3' end of the
junction there are 19 nucleotides to the right of the Hind111 site
at the end of a stretch of six T residues. The N-terminus of the
v-mosMOgene product extends five amino acids into the MoMLV genome past the c-mosM0junction. There is no direct
sequence homology between Mo-MLV and c-mosMojunction
sequences. However, the sequence TTCA is present immediately to the left of the 5' junction in c-mosMO,and two nucleotides to the left of the 5' junction in Mo-MLV (Fig. 4). This kind
of sequence homology may suggest that v-mosMOoriginated by
homologous recombination between c-mosM0 and Mo-MLV
sequences. The exact mechanism of recombination, however,
remains obscure.

Comparison of portions of c-mosMO

and
v-mosMOsequences
It was of interest to determine the sequence divergence between
c-mosMOand v-mosMOsequences. A cloned isolate of Mo-MLV
(Mo-MLV clone 1) has been ~ b t a i n e d ~We
~ .have
~ ~ . propagated
Mo-MSV clone 124 and Mo-MLV (clone 1) in our laboratory
since 1973. Figure 3 compares the nucleotide sequence of
portions of v-mosMOand c-mosMOsequences. In a stretch of
about 500 nucleotides at the 5' end there seem to be only five

base changes. Similarly, in a stretch of 300 nucleotides at the 3'

end there are only eight single base substitutions. The nucleotide
sequence homology between the v-mosMO and c-mosM"
sequences is in complete agreement with the results obtained
from nuclease SI analysis of hybrids formed between cloned
A .MT.x-l DNA (containing c-mosMOsequences, Fig. 2) and
Mo-MSV genomic RNA~'.The initiation codon ATG (at position 203 in Fig. 3) of the longest possible v-mosMOgene product
is not present in the c-masM0sequences (Fig. 4). The sequences
in c-mosMO
alone are not able to transform cells; however, when
portions of Mo-MLV containing the long terminal repeat (LTR)
are attached to it, cellular transformation can be observed
(W. M. McClements and G. F. van de Woude, personal communication). Thus, it seems that the proteins encoded by
v-mosMOand c-mosMOsequences are very similar.
The nucleotide sequences adjacent to the 3' end of the
v-mosMOgene in Mo-MLV and Mo-MSV are identical.
However, a t the 5' end, 29 nucleotides from the 5' junction of
c-mosMOand Mo-MLV, the Mo-MSV seems to have undergone
a previously undetected deletion of 192 nucleotides (shown as
closed bar in Fig. 3). Analysis of the sequences on the 5' and 3'
ends of the deletion reveals a common sequence, GACC. This
small deletion was not observed in the heteroduplexes formed
between Mo-MLV cDNA and Mo-MSV 124 RNA".". As
there are only 29 base pairs between this deletion and the major
substitution loop, it was perhaps scored as a part of the major
substitution loop, P, (ref. 11).

00

[I]

S L C R Y L P R E L S P S V D S R S C S I P L V A P R K A G K L F L G T
K A ~ Q [ v A Y y ~ K H A ~ G L c H R L T N v c P

Q S C W E A R G L Q R P S A E L L Q R D L K A F R G T L G

374

D P G G A ~ H F S S Y L Q ~ Q ~ P A C V L E V530
A E
--

O 198lNature Publishing Group

Fig. 5 Comparison between the

proposed amino acid sequences for
the v-mosM" and ASV src gene
products. The amino acid sequences
(Fig. 3) and ASV src
of the v-mosMO
(ref. 36) gene products were aligned
for optimum identity as described
e~sewhere~'.~! The portions of the
sequences presented here include
the entire v-mosM"sequence, and
ASV src sequence from amino acid
188 to its carhoxy terminus (residue
530). Arcas of identity are depicted
by boxes. The letter code for the
amino acids is: A, alanine; C,
cysteine; D, aspartic acid; E, glutamic acid; F, phenylalanine; G ,
glycine; H, histidine; 1, isoleucine;
K, lysine; I., leucine; M, methionine;
N, asparagine; P, proline; Q, glutamine; R, arginine; S, serine; T,
threonine; V, valine; W, tryptophan;
Y, tyrosine.

262

Nature Vol. 289 22 January 1981

The sequence in Fig. 3 was examined for features which have

been implicated in RNA synthesis in other systems. Neither the
v-mosM0 nor the c-mosMo sequences seem to contain the
canonical sequence (TATAAA, Hogness-Goldberg box) often
implicated in promotion of transcription. Examination of the
A+T-rich region of the sequence shown in Fig. 3 reveals a
sequence, TTGTAAA, beginning at nucleotide 53, which might
be used as a DNA-dependent RNA polymerase I1 binding site.
Further sequence analysis of Mo-MLV, 5' from nucleotide 1,
reveals another potential promoter sequence (data not shown).
The importance of the TATAAA sequence is uncertain,
however, as it has recently been reported that removal of
TATAAA sequences does not eliminate the transcription of
histone genes in oivo3". TO the 3' side of the v-mosM0coding
region, 210 nucleotides from the TGA stop codon, there is the
sequence AATAAA which has been associated with the addition of poly(A) tails onto eukaryotic ~ R N A S ~ ~ .

Comparison with other proteins

The amino acid sequence of the v-mosM" gene product, as
derived from the DNA sequence, was subjected to a computer
search for sequence resemblance among 1,300 known amino
acid sequences, including those for the T-antigens of polyoma
and SV40. Although several sequences were found to have some
marginal resemblances to portions of the v-mosMogene product,
only the recently reported oncogene product (deduced from
DNA sequence studies) of avian sarcoma virus (ASV)36had a
genuinely homologous sequence.
A very significant homology was uncovered when overlapping
segments of the two proposed amino acid sequences were
compared (Fig. 5). Residues 1-374 of the v-mosMogene product
were placed into optimum alignment with residues 188-530
(C-terminus) of the ASV src gene product using a program
which has an algorithm similar to that described by Needleman
and W u n s ~ h ~ ~and
. ~ ' ,which uses an empirically derived gap
penalty. After an alignment score had been obtained, the two
sequences were randomly jumbled, 36 comparisons of jumbled
sequences executed and the alignment scores again computed.
A mean value of the alignment scores was, finally, determined
and the standard deviation calculated. The alignment score
obtained for the two authentic sequences was about 7 s.d. above
the mean of the jumbled comparisons, a highly significant
result38.
The optimum alignment of the two sequences revealed that
they are about 23% identical in the region of the 371-residue
overlap shown in Fig. 5. The strongest homology occurs over the
stretch of the v-mosM" sequence corresponding to residues
227-309 and the ASV sequence residues 387-468. In this
region there are 36 identical residues over the course of the
83-residue span, or about 45% identity. The overall degree of
resemblance is approximately the same as that obtained in a
comparison of mouse haemoglobin a-chain and hagfish

Received 21 July; accepted 11 November 1980

1. B~shop.J. M. A. Rev. Biochem. 47, 37 (1978).
2. Spector, D. H.. Varmus, H. E. & Bishop, I. M. Proc. nam. Acad. Scr. U.S.A. 75, 4102
(1978).
3. Duesberg, P., Bster, K. & Vogt, P. K. Proc. narn. Acad. Sci U.S.A. 74,4320 (1977).
4. Hayman, M. I., Roy-Pokora, B. & Graf, T. Virology 92, 31 (1979).
5 Rettenmier. C. W.. Anderson. S. M.. Rieman, M. W. & Hanafusa, H. J. V~rol.32, 749
(1979).
6. Lee, W.-H. eral. Proc. nam. Acad. Sci U.S.A. 77, 2018 (1980).
7. Hanafusa, T, er al. Proc. natn. Acad. Scr. U.S.A.77, 3009 (1980).
8. Reynolds, F. H. Jr, van de Ven, W. J. M. & Stephenson,J. R. J. Virol. 28, 665 (1978).
& Ba1timore.D. Proc. nam. Acad. Scl.
9. Witte.0.N.. Rosenberg.N..Paskind.M..Shields,A.
U S A . 75,2488 (1978).
10. van de Ven, W. J. M., Kahn, A. S.. Reynolds,F. H. Jr, Mason, K. T. & Stephenson, J. R.
I. V~rol.33. 1034 (1980).
11. Hu, S., Davidson, N. & Verma, I. Cell 10, 469 (1977).
12. Chien, Y:H. eral. J. Virol. 31,752 (1979).
13. Hu, S. F., Lai, M. M. C. & Vogt, P. K. Proc. narn. Acad. Sci. U.S.A.76, 1265 (1979).
14. Shields, A., Goff,
S., Paskind, M., Otto, G. & Baltimore, D. Cell 18,955 (1979).
15. Donoghue, D. J., Sharpe, P. A. & Weinberg. R A. I. Virol. 32, 1015 (1979).
16. Sherr. C. 1..Fedele, F. A., Oskarsson,M., Maizel, J. & Van de Woude, G. I. Virol. 34,200
(1980).
17. Van de Woude. G.F. er al. Proc narn. Acad. Sci. U.S.A.76,4464 (1979).
18. Mellon, P., Pawson. A., Bister, K., Martin. G. S. & Duesberg. P. Proc. "am. Acad. Sci.
U.S.A.75, 5874 (1977).
19. Van Zaane. D. & Bloemers, H. P. J. Biochim. biophys. Acta 516, 249 (1978).

h a e m ~ ~ l o b i nor~ of
~ , the P-chain of human h a e m ~ ~ l o b i n ~ ~
and human myoglobin4'. The murine and avian haemoglobin
a-chains, by contrast, have evolved less rapidly than have
the proposed murine and avian viral proteins: the mouse
haemoglobin a-chain has about 70% identity with chicken
a-haemoglobin.
No detectable nucleotide sequence homology was observed
when ASV src-specific cDNA was hybridized to viral RNA
obtained from the Mo-MSV(MLV)
This result was
confirmed by a computer comparison of the nucleotide
sequences of v-mosM" (Fig. 3) and the ASV src gene product36.
However, the amino acid sequence homology between the
oncogene products of ASV and Mo-MSV indicates that there
can be little doubt that the two proteins have evolved from a
common ancestor.

Conclusion
We have studied the biogenesis of the transforming gene of
Mo-MSV. The Mo-MSV was isolated from sarcomas obtained
after passage of Mo-MLV in mice. The Mo-MLV acquired
cellular sequences, and underwent several deletions to yield a
transforming but replication-defective virus. We have determined the entire nucleotide sequence of the Mo-MSV-specific
sequences, referred to as the v-mosMOgene, in the Mo-MSV.
Furthermore, the recombination junctions between the MoMLV and the cellular sequences leading to the formation of the
v-mosMOgene have also been determined. The v-mosM" and
c-mosMOshare an uninterrupted stretch of 1,159 nucleotides
with few base changes. There are n o apparent structural
homologies between the Mo-MLV and c-mosM" sequences at
the point of recombination. However, there seems to be a small
sequence homology, TTCA (Fig. 4), at the 5' junction of MoMLV and c-mosM0, perhaps suggesting homologous recombination. Due to the absence of specific tumour antisera, no
authentic v-mosMogeneproduct has been identified. The 37K in
uitro translation product" is specifically inhibited if the RNA is
hybridized to a v-mosM0gene DNA fragment before translation
(J. Papkoff, T. Hunter and I. M. V., unpublished results). The
predicted amino acid sequence of the v-mosMogene product can
now be used to verify the nature of the 37K protein synthesized
in uitro. Finally, the proposed amino acid sequence of the
v-mosMOgene product has been shown to share significant
homology with the ASV src gene product.
We thank Bart Sefton and Walter Eckhart for reading the
manuscript and Carolyn Goller for preparation of it, Drs
P. Czernilofsky and Hugo Martinez for help with computer
programs, Dr E. P. Reddy for communicating his sequencing
data on Mo-MSY-specific fragment before publication, and
M. Jones and D. Murdock for help with various stages of the
study. This work was supported by grants from the NCI and the
American Cancer Society. D.J.D. acknowledges a Helen Hay
Whitney Foundation postdoctoral fellowship.
20. Aaronson, S. A. & Rowe. W. P. Virology 42.9 (1970).
21. Scolnick,E. M. er al. Proc. narn. Acad. Sci. U.S.A.72,4650 (1975)
22. Moloney, I. B. Narn. Cancer Insr. Monogr. 22, 139 (1966).
23. Ball, J. K., McCarter, J. A. & Sunderland, S. M. Virology 56, 268 (1973).
24. Dma, D., Beemon, K. & Duesberg, P. Cell 9,299 (1976).
25. Verma, I. M rral. Proc. narn. Acad. Sci. U.S.A.77. 1773 (1980).
26. Berns. A. er al. J. Virol. (in the press).
27. Oskarsson, M.. McClements, W. M., Blair. D. G.. Maizel. J. V. & van de Woude. G. F.
Science 207, 1222 (1980).
28. Jones, M. er a/. Proc. nam. Acad Sci. U S A 77, 265 1 (1980).
29 Renr. E. D. Jr & Dina. D. Pror, narn. Acad. Sci. U.S.A.76.3294 (1979).
30. Verma. I. M. Nucleic Acids Rrs. 6, 1863 (1979).
31. PapkoH, J., Hunter, T. & Beemon, K. Virology 101.91 (1980).
32. Moloney. J. B. I. narn. Cancer Insr. 24. 933 (1969).
33. Fan, 1-1. & Paskind, M. J. Virol. 14,421 (1974).
34. Grosschedl. H. & Birnstiel, M. L. Proc. narn. Acad. Sri. uS.A. 77, 1432 (1980).
35. Proudfoot, N J & Brownlee, G. G. Narrue 263,211 (1976).
36. Czern~lofsky.
A. P. eral. Narure 287, 198 (1980).
37. Needleman, S. B. & Wunsch, C. D. J. molec. Biol. 48,443 (1970).
38. Jue, R. A.. Woodbury, N. W. & Donlittle. R. F. J. molrc. Euol. 15, 129 (1980).
39. Liljeqvist, G., Braunitzer, G. & Paleus, S. Hoppe-Seyler's Z. physiol. Chem. 360, 125
(1979).
40. Braunitzer. G. er ul. Hoppe-Seyler'sZ. physiol. Chem. 325.283 (1961).
41. Herrera, A. E. R. & Lehmann, M. Narure new Btol 232, 149 (1971).
42 Stehelin,D.,Guntaka, R . V., Varmus, H . E. & Blshop,J. M. J , moler. Biol. 101,349(1976).
43. Van Beveren, C., Goddard, J. G.. Berns. A. & Verma. I. M. Proc. narn. Acad. Sci. USA
77,3307 (1980).

O 198lNature Publishing Group

Nature Vol. 289 22 January 1981

A model from electron microscopy for the

molecular structure of fibrinogen and fibrin
John W. Weisel, George N. Phillips Jr & Carolyn Cohen
Rosenstiel Basic Medical Sciences Research Center, Brandeis University, Waltham, Massachusetts 02254

The overall architecture of the fibrinogen molecule and aspects of its packing to form fibrin have been derived from a study of
electron microscope images of ordered arrays. The molecule, of length 450A,is seen in more detail than the Hall-Slayter
model and is made up of seven globular domains connected by rod-like segments.
FIBRINOGEN is the soluble precursor of the fibrin clot in the
plasma. O n tissue injury, an enzymatic clotting cascade is initiated leading to thrombin cleavage of the two small pairs of
highly charged peptides from
The molecule then
becomes highly insoluble and rapidly aggregates into the fibrin
clot. This noncovalently bonded polymer is cross-linked by the
fibrin-stabilizing factor enzyme and is later dissolved by
cleavage with plasmin. The processes of clotting and fibrinolysis
both occur continually in the normal functioning of the vascular
system. The complexity of this clotting process is reflected in the
complex structure of the fibrinogen molecule.
Fibrinogen has a molecular weight of 340,000 and an
elongated
but its detailed shape is unknown. Hall
and Slayters obtained electron micrographs of individual
molecules, revealing a protein about 475 A long consisting of a
linear array of three globular domains (Fig. 1A). Despite
considerable controversy, this 'trinodular' model is now
generally accepted5-".
T o account for the small 223 A axial period of fibrin, Hall and
Slayter invoked a 'shrinkage' of the molecule, but their proposal
is inconsistent with results of X-ray diffraction studies, which
show no large-scale changes in molecular structure on the
conversion of fibrinogen to fibrini2.l3.The axial period of fibrin
can be accounted for most simply by a half-staggered arrangement of fibrinogen monomer^'^.
Chemical studies indicate that fibrinogen is a dimer: the
molecule consists of three pairs of different polypeptide chains
(called A a , BP and y) linked by 2 9 disulphide bridges14.". The
complete amino acid sequence of human fibrinogen and the
complex pattern of disulphide cross-links have recently been
e ~ t a b l i s h e d ' ~ -Sequence
~~.
analysis has also been used to locate
certain regions of a - helical coiled ~ o i 1that
~ ~had
. ~been
~ predicted to connect the globular domains of the molecule2s.
However, these studies d o not provide direct evidence on the
overall morphology of the molecule o r its packing in fibrin.
The most powerful approach to a determination of the structure of fibrinogen is X-ray crystallographic analysis. A n
important discovery was that microcrystals can be produced on
limited cleavage of bovine fibrinogen with a bacterial protease26
and that with further digestion, large single crystals, ordered to
about 4 A resolution, can be
Crystallographic and
chemical evidence show that the modified molecule in the crystal
is only 5-10% smaller in mass than native fibrinogen2'.
Moreover, the molecules forming the crystals clot and the fibrin
so produced has a band pattern in the electron microscope very
similar to that of native fibrin, indicating that the modified
protein is largely intact (Fig. 2 0 , E). (Other proteolytic enzymes
have also been shown to modify fibrinogen s o that microcrystals
and crystals can be grown29.)
Our strategy for determining the low resolution crystallographic structure of fibrinogen is the refinement of a model
against X-ray data3'. Here we derive a more detailed molecular
model for fibrinogen than the trinodular Hall-Slayter structure
and describe packing arrangements that account for electron
0028-083618 11040263-05SO 1.00

microscope images in a variety of highly ordered arrays including fibrin.

Development of the model

It has been recognized for some time that the electron microscope band pattern of fibrin cannot b e accounted for by a simple
Hall-Slayter model. Within the repeat of 225 A in negative
contrast there is one bright strain-excluding band about 60 A
wide and three narrow light bands, the middle one being somewhat less prominent o r variable in i n t e n ~ i t y ~ ' -(Fig.
' ~ 2 0 ) . The
size and number of bands in this image show quite plainly that
the globular domains in the fibrinogen molecule must be
considerably smaller than those of the Hall-Slayter model (Fig.
1 A ) and that a trinodular structure cannot give rise to the details
of the staining pattern.
A key to the interpretation of the fibrin band pattern can be
found in the so-called 'orthogonal sheet' s t r u c t ~ r e ~(Fig.
~ . ~ '2A).
A simple, half-staggered arrangement of this highly regular
array gives a band pattern closely resembling that of negatively
stained fibrin2'. An important aspect of the orthogonal sheet
form is its extremely detailed substructure. The features are such
that neither the Hall-Slayter model, nor, indeed, any simple
arrangement of molecules running parallel to the microcrystal
axis can readily account for this image. However, analysis of a
variety of microcrystals of fibrinogen, including the orthogonal

Fig. 1 Models for the fibrinogen molecule. A, Trinodular Hall-Sla ter

model. scaled to a length of 450 A. The end domains are about 65 in
diameter and the central region is about 50 A. Recent micrographsshow that
the end domains seem to be elongated in shape"'3. B. Fibrinogen model
derived from electron micrographs of crystals and microcrystals. Diameters
of the globular regions (starting from the end): 36, 32, 20. 30 A. The
diameter of the rod region has been taken as 10 A. The molecule is not
co-linear, but has a bend of a few degrees. This model from electron
microscopy has about one-third the hydrated volume because of dehydration and stain penetration. C, Hydrated version of B. The dimensions
have been increased so that the molecule has the volume calculated from
crystallographic data (3.8x 10'
This approximate model will be used
in the initial phasing of the X-ray data. D, Predicted a-helical coiled-coil
regions based on analysis of the amino acid
The length of the
region between two 'braces' of disulphide bonds is 111 residues, or about
160 .&;the coiled coil would begin at the central domain. We infer that the
smallest globular domain may correspond to a region predicted to be
non-helical where plasmin cleavage occurs2' (see text).

@ 1981 Macmillan Journals Ltd

Nature Vol. 289 22 January 1981

264

Fig. 2 Simulation of electron microscope images of the orthogonal sheet microcrystals and fibrin. A, Electron micrograph of a negatively contrasted microcrystal of
the orthogonal sheet formz8; the image has been optically filtered to remove background n ~ i s e ~ ~Unit
. ~ ' cell
. dimensions are 900 x 90 A. ~450,000.B, TV display of
computer simulation of A, based on the model below (C) (as described in the text). Note that the models in C and G are drawn to a scale about 1.5 times as great as the
electron micrographs above. D,E, Computer filtered electron micrographs of negatively contrasted fibrin. The axial repeat is 225 A. In D, the fibrin is derived from
native fibrinogen whereas in E the fibrin is from Pseudomonas-modified fibrinogen. Note that the weakest light band, which arises from the smallest globular domain,
is less prominent in the modified fibrin26.28,suggesting that the bacterial protease, like plasmin (see text), cleaves some material from this region. F,TV display of
computer simulation of fibrin, based on the model shown in G. The fibrin band pattern was generated by superposition of arrays of the orthogonal sheet, shown in A,
shifted by 225 A (refs 26, 28). For clarity, two such layers of molecules are shown, separated by a dashed line, in the upper and lower portions of G. The lateral
contacts in both fibrin and the orthogonal sheet form incorporate a half-stagger, where the central globular domain of one molecule fits into the 'pocket' formed by the
end-to-end linkage of two molecules in an adjacent filament. A simple half-staggered array of five molecules is shown as an inset to F (and with the same scale) so that
the relationship of the molecular shape and packing to the fibrin band pattern is apparent. The bright axial striations in both the orthogonal sheet form and fibrin arise
from this juxtaposition of the end globular units of two molecules with the middle unit of a third. In the orthogonal sheet, the pairs of small bright units on either side of
this feature can be seen to arise from the adjacent globular domains in these molecules. To account for the angle of the thin strands in the gap, the molecules are
arrhyed with a slight tilt. This arrangement generates the correct symmetry for the structure and also places the smallest globular units at the centre of the dark gap
where they are seen as bright nodes.

sheet form, has indicated that all these structures seem to be

made up of 450 A-long fibrinogen molecules that bond end to
end2'.
The best characterized macrocrystal form of fibrinogen is the
monoclinic P2, crystal of protease-modified bovine
fibrin~gen~"~'.
A packing model for this crystal was established
using the morphology and special features of the diffraction
patterns29'35.Here too it was shown that the molecules are
450 A long and bond end to end to form filaments. The general
run of the molecules in the lattice was established but not their
lateral relations. This knowledge of the molecular packing
allowed an interpretation of electron microscope images of
crystal fragments. In particular, a characteristic net or mesh
form was observed that could be accounted for as the view along
the unique symmetry axis (Fig. 3A, D ) .The strands of the mesh
seen in the electron microscope were interpreted as pairs of
symmetry-related molecular filaments in a distorted cell. These
strands showed prominent features of macromolecular size, but
again it was not possible using the Hall-Slayter model to define
the lateral packing or to delineate individual molecules.
A form that we found useful in establishing the model was a
'segment' aggregate obtained from human fibrinogen modified
by bacterial protease36 (Fig. 4). The segments displayed a strikingly simple band pattern in negative staining that revealed
where the molecules begin and end. Analysis of this structure
showed that the molecules are in register in layers which are
tilted alternately with respect to the long axis. Here again, the
molecules were found to be -450 A long.
Because there is no molecular overlap, the packing of
fibrinogen in this form yields direct information about the mass

distribution along the molecule. The overall structure seems

qualitatively like the Hall-Slayter model in that there are three
major stain-excluding domains per molecular length. There is,
however, additional fine structure in the segments. Within each
major band there is another thin line of opposite contrast, which
seems to define smaller domains. This evidence again suggested
that the Hall-Slayter model was valid as a very low resolution
image of the molecule, but that a more detailed mass distribution could be derived.
We may interpret the staining of the segments to mean that
the outer domain of the Hall-Slayter model is, in fact, made up
of two units, each about 30-40 A in diameter. Close inspection
of the segment images reveals an additional thin white line
within the broad dark-staining region (Fig. 4). This feature
indicates that another smaller globular unit may be located
between the outer domain and the central nodule. This additional globular unit is, in fact, necessary to account for the
detailed appearance of images of both fibrin and related forms as
well as views of the crystal (see below). The molecule of length
450 A thus appears in all the ordered aggregates as a heptad
made up of seven globular domains. We show below how more
precise dimensions of the globular units, as well as those of the
rod-like regions connecting them, can be derived from the
simulation of a variety of electron microscope images.

Simulation of electron microscope images

Using modifications of the model described above, computer
simulations were carried out to develop a single molecular
model whose substructure and packing could account for the
negatively stained images of a variety of arrays seen in the

O 198lNature Publishing Group

265

Nature Vol. 289 22 January 1981

Fig. 3 Simulation of electron microscope images of the P2, crystal. A, D,

G, Electron micrographs of negatively contrasted crystal fragments which
have been filtered o p t i ~ a l l ~ or
~ ' ,by~ ~computer processing of a digitized
i ~ n a ~ e ~ 'A
. ' ~and
. D are two different mesh images corresponding to crystals
with different amounts of shrinkage viewed along the unique axis. In both
cases, the molecular strands are oriented parallel to the largest face of the
crystal, which is horizontal herez9. Unit cell parameters in A are: a = 132,
b = 97, c = 125, P = 105",whereas in B they are: a = 160, b = 97, c = 130,
p = 65".The image in G corresponds to the state of the crystal seen in D but
viewed in the perpendicular direction. ~ 4 5 0 , 0 0 0B,
. E, H,TV display of
computer simulations of the models in C, F, I below (as described in the
text). Note that the models in C,F and I are drawn to a scale about 1.5 times
as great as the electron micrographs and simulations above so as to display
the molecular packing. In the crystal, the strands correspond to the projection of filaments in different layers staggered by roughly one-third of the
molecular length. In both the mesh images (A, D ) the relative brightness of
the different units of the motif depends on the size of the globular domains
and their degree of overlap. A cluster of three overlapping globular domains
corresponds to the broad light region nearest the dark gap; the very bright
unit at the middle of the group corresponds to two overlapping globular
regions. A small bend accounts in part for the slightly wavy appearance of the
strands. The rod portion of the molecule can be visualized in some micrographs of the mesh, as in D.

electron microscope. Trial models for selected forms were

rapidly generated with the aid of computer graphics and
compared with electron micrographs. The molecule was
represented as a series of seven spheres connected by rods, and
the programs were written so that one set of parameters fixed the
sizes and position of the spheres and rods within one molecule,
and another set located a single molecule in the unit cell.
Additional molecules were then generated according to the
appropriate symmetry elements of the specific form under
analysis. To convert these arrangements to density contours,
which could then be compared with electron microscope images,
the model was represented by sets of points spaced closely
enough to correspond to a continuous structure at this resolution. The projected Fourier transform of this configuration was
then computed, followed by the inverse transform which was
displayed on a television screen. Subjective criteria were used to
alter both molecular and packing parameters in comparing the
simulation to filtered electron microscope images.
A range of molecular models was examined. Slight ditferences
in the size of the globular units or slight departures of the two
halves of the molecule from co-linearity were tested to achieve a
good fit with a particular form. Some differences in the nature of
the end-to-end molecular contacts were also tried with certain

arrays. It was found that all the images could be fitted with the
same heptad model (Fig. 13).

The P21 crystal

Several images from P2, crystal fragments could be simulated.
The net described previously and identified as the view down the
unique axis of the crystal is characterized by strands having
features of macromolecular size running in the horizontal directionZ9(Fig. 3A). The repeating unit along the strands is a group
of seven stain-excluding regions of varying sizes (none larger
middle one being particularly brightthan -40 A)-the
followed by a dark-staining gap. This motif repeats every 450 A
and corresponds to the projection of two symmetry-related

0 1981Nature Publishing Group

Nature Vol. 289 22 January 1981

266
layers of molecular filaments in the unit cell. This form was
particularly useful in establishing the size of the globular
domains and their separation in the model. The small bend in the
molecule accounts in part for the slightly wavy appearance of the
strands. With a shift of about one-third the molecular length
between filaments in the two layers the simulation shows
excellent agreement with the filtered image of the electron
micrograph (Fig. 3B, C).
Electron micrographs of the P2' crystal fragments show a
number of states of the mesh described above. Another commonly observed form has a mesh with a different appearance
(Fig. 3 0 ) . The motif is rather similar to that described above,
although the intensities and positions of the bright units differ
somewhat. Keeping the parameters of the molecular model used
in the previous simulation, only a slight change in the relative
stagger of the molecular filaments within each layer is required
for a satisfactory fit (Fig. 3E, F).
A different view of the crystal could also be identified. Some
preparations contained very thin crystals that tended to lie with
the large face on the grid. This image showed a distinctive
pattern of alternating rows of bright and weak zig-zags repeated
145 A axially (Fig. 3G). The spacings indicated that this was the
view along one of the crystal axes perpendicular to the previous
view and nearly perpendicular to the large face of the crystal (see
Fig. 3 0 ) . Using the same arrangement of molecules as before
(Fig. 3F), and simply taking the appropriate projection (Fig. 3I),
a very satisfactory simulation can be achieved (Fig. 3H).

The orthogonal sheet form and fibrin

One of the most detailed images is that of the orthogonal sheet
form, related to fibrin (Fig. 2A). Here the motif is a very
complex dumbell-shaped bright region about 400 A long with a
dark-staining gap of about 50 A. The motifs in adjacent rows are
staggered by half the lateral period so that the true axial repeat is
900 A. Within the dark-staining gap, thin light strands seem to
connect units in different rows. For the simulation, the dimensions of the molecular model as determined for the P2, crystal
views were maintained, but the molecular filaments in different
layers were half-staggered, as determined previo~sly13.26~28
(Fig.
2C). This arrangement accounts for the very bright axial striation which arises from juxtaposition of the end globular units of
two molecules with the middle unit of a third (Fig. 2B). The
small bright units on either side of this feature can be seen to
arise from the adjacent globular domains in these molecules. To
account for the angle of the thin strands in the gap, the molecules
were arrayed with a slight tilt to the long axis of the unit cell. This
arrangement generates the correct symmetry for the structure.
The smallest globular units can be seen quite clearly as bright
nodes at the centre of the dark gaps.
A half-stagger (225 A) of this arrangement for the orthogonal
sheet form then generates the fibrin image26.28(Fig. 2E, F, G).
The brightest band in fibrin again arises from juxtaposition of
the end globular regions of two molecules together with a centre
domain of a neighbouring molecule and the narrower band on
either side from the adjacent domains. Additional features of
the pattern are also accounted for: the thinnest band in the
period corresponds to the smallest globular domain in the
molecule. In fact, this arrangement is precisely that proposed in
the early half-staggered models for fibrin13, but the size of the
domains in the molecular model now yields correct dimensions
for the band pattern (Fig. 2 F inset). The relatively simple band
pattern of fibrin is thus accounted for by a superposition of
rather complex molecular arrays.

The molecular model

We have now obtained an improved low-resolution model for
the fibrinogen molecule (Fig. 1B). The basic structure is closely
related to the trinodular Hall-Slayter model, but we have been
able to establish a precise length for the molecule of 450 A, and
a more detailed substructure. The outer globular domain,
incompletely resolved by the shadowing technique used by Hall
and Slayter, is now seen to consist of two regions of nearly equal

size. The narrow rod-like connection between the central and

outer units had not previously been visualized, and we have
shown that its regular run is interrupted by a small globular
region displaced towards the outer domains. There is also a
slight bending of the molecule by a few degrees. We consider the
evidence in favour of this model to be compelling: all the
features of the model are directly visible in certain electron
microscope images, and the model can account for the detailed
appearance of a variety of forms, including the complex pattern
of fibrin.
Fibrinogen displays different properties at different stages of
clotting. As shown for other fibrous proteins37, particular
domains may be associated with each of its various functions.
The central and outer globular domains of the molecule are
involved in assembly of the fibrin fibre; the middle region
contains the small, highly charged fibrinopeptides cleaved by
thrombin3' whereas the outer globular domains contain
complementary bonding sites for fibrin f ~ r m a t i o n ~Covalent
~'~~.
cross-linking of the molecules by the fibrin-stabilizing factor has
been shown to involve the outer globular domain4'. Some of
these sites may also be located in the neighbouring globular
domain. The plasmin-sensitive portion of the molecule can be
identified with the smallest globular domain which interrupts the
regular run of the rod-like a-helical coiled-coil regions (Fig. lB,
D)23.24.On the basis of this oversimplified picture, we can begin
to localize a number of the characteristic properties of
fibrinogen in different structural domains.
These results also indicate that a modest degree of molecular
'flexing' is a necessary feature of the packing arrangements in all
the forms we have examined. Hence, the molecule should not be
considered as rigid, but there is no compelling reason to postulate marked departures from colinearity of the two halves of the
molecule.

Fig. 4 Electron micrograph of negatively contrasted 'segment' prepared by

precipitation of human fibrinogen modified by the Pseudomonas p r o t e a ~ e ' ~ .
The apparent axial repeat in the structure is 385 A. The corrugations visible
along the upper and lower edges indicate that the true period is 770 A and
that the molecules are both tilted in layers and bent. The end of the segment
on the right shows where the molecules begin. x135.000.

The model described here represents the first stage in our

approach to the detailed three-dimensional structure of
fibrinogen. The dimensions used in our simulations were derived
from negatively stained electron microscope images, but the
approximate hydrated envelope can be calculated from the
molecular volume in the crystal29 (Fig. 1C). We have now
com leted three-dimensional X-ray data collection to about
20 on the P2, crystal form. To derive truer shapes of the
molecular domains we are refining the hydrated version of the
model against the diffraction data42; solvent flattening techniques may also be useful for phasing reflections at this resolution30. The new model will reveal in even greater detail the
complex architecture of this highly evolved protein molecule.
We thank Cynthia Stauffacher and Nancy Tooney for helpful
discussions, Paul Norton for electron microscopy, William
Saunders for photography, David DeRosier for computer pro-

@ 1981Nature Publishing Group

Nature Vol. 289 22 January 1981

grams and Edward Swinarski for computing. This work was
supported by NIH grant AM17346 to C.C. and NSF grant
PCM78-15913 to D.D. J.W.W. is an Established Investigator
of the American Heart Association; G.N.P. held a National
Research Service Award from the USPHS and is a Medical
Foundation Fellow.
Rccewed 4 July. arceptcd 4 Novembcr 1980
1 Dawe. E W & Fullkawa. T. A RPL..Biochpm 44. 799 8 2 9 (1975)
2 McFarlane. R G In H u m a n BlvodC~x~,o~ulofron.
I i o r m o r r o ~ r s o n dI%romhon< led. Btggq. R1
1 31 1Blackwell. Oxlord. 1976).
3. Shulman. S. L A m . =hem. Soc 75. 5846-5852 (1953)
4. Caspary. E A & Kekw~ck.R A. Biorhrm 16 7 . 4 1 4 8 11957)
5 Hall. C' F B Slaytrr. H S I hmphy,. biorhrm. Tvrol 5, 11 -15 119591.
6 Koppcl. G Narorr 212. lhO8- I609 11966).
7. Krakow. W.. Endws. G F . S q c l . B. M. & Sheraga. H. A. J molrr Rcol 71. 95-107
(1972)
8. Rachmann. L Schmttl-Furnnan. W W Hammrl. R. & Lederer. K Mokromolcr. Chem
176,2603-2618 (1975).
9. Marguerle. G & Stuhrrnann. H B 1 molrr. Bml 102. 143 156 (19761.
LO. Slaytcr. H. S Ulfromtcrorcupy 1,341-357 (1976)
11 Fowlcr. W. E . & Enukson, H. P. J. mnlcr Htol. 134, 241 2 4 9 (IV7YI
12. Batley. K . Asthury. W T. & Rudall. K. M Morurr 151.716-717 (1943)
13. Stryer. L.. Cnhen, C & Langndgc. R. Norurt- 197, 793 794 (1963).
1 4 Hcnrchen. A. A r k Kcmr 22, 3 5 5 3 7 3 11964).
I 5 Hrnschcn. A. 7hromh. Hucmurrusrr 42, 14 (1979).
I 6 Blomhack. R.. Hcssel. B & Hogg. D. 7 h m m h RPT 8. 6 3 9 4 5 8 (19761
17 Henschen. A. & L o t t w e ~ c h .F. H o p p ~ ~ S ~ y l7
p rphymd
'~
C h o n 357. 1801- 1803 (197h)
1 8 Dool~ttle,R. F. Horrron> Bmrhcm Btuph>, 3. I 6 4 191 11977)
' r phyrrol T h e m 358. 935-938 (19771
1 9 I.ottspr~ch.F & Hrnschcn. A I f o p p e - S ~ ~ l r r%
20 Watt. K. W. K.. 1-ayakl, T. & Doolntlc. R 1; Brorhpmlrrrv lR, 6 8 7 6 ( 1 9 7 9 )

Cosmic ray antiprotons and

modified closed galaxy model
S. A. Stephens
Tata Institute of Fundamental Research, Homi Bhabha Road,
Bombay 400005, India

Observations suggest that a small flux of secondary antiprotons

(p), created in the interaction of cosmic ray nuclei with inter-

stellar gas, should also be present in the primary cosmic radiation. Calculations based on accelerator data for the production
of p and existing models for the propagation of cosmic rays,
predict too small a flux of p compared with the finite flux
observed recently1. Although the observed excess could be
attributed to the existence of primary cosmic ray antimatter, the
experimental upper limits2" of the ratio of antinuclei to nuclei of
charge IZI z 2 seem to contradict this. Here we use a 'modified
closed galaxy' model to attribute this excess to secondary
antiprotons. In this model, -50% of the observed cosmic ray
nucleons are of recent origin and they propagate according to
the 'nested leaky box' model, while the rest propagate according
to 'closed galaxy' model. This model explains the observations
on p and e', and predicts more D and 'He than do the existing
models.
The secondary ji spectrum in cosmic rays has been calculated
. ~ the
in the past by Gaisser and ~ a u r e r "and
, Badhwar e r ~ lusing
extensive accelerator data available for the production of p.
However, these calculated spectra differ by more than a factor of
two at - 1 0 GeV and by even larger factors at lower energies,
suggesting that cross-sectional representations used by them
must have been responsible for this difference. The interpretation of a finite flux of p depends crucially o n the accuracy of
the estimated flux of p in cosmic rays. Szabelski et al.' showed
that the results of Badhwar ct uL5 are in error. Thereby, they
questioned the 'standard leaky box' model (SLBM) for the
propagation of cosmic rays. I recently obtained a more accurate
parametrization7 of the invariant cross-sections, for the
production of p in inclusive reactions, and calculated the

21. D w l ~ t t l r .R F . Watt. K W K.. Coltrell. B A . Strong. D D & Riley. M Narure 280,

4 6 4 4 6 8 ( 1 979).
22. Henrchen, A . I artspckch. F & Hessel. B I i o p p r ~ S t y l r r ' rZ phyr~ol Chcm 3h0, 19511956 ( 1 9 7 9 )
23 D w l ~ t t l c .R F . Goldhaurn, D M & Dwllttle. I. R I, molut Biol 120. 31 1 3 2 5 (1978)
24 Parry, D . A. D l molsr R d 120, 545-551 (1978)
25 Cohen. C J Po1,mrrSct 49, 144-145 ( 1 9 6 1 )
26 Tooney. N.M & Cohen. C Nurum 237. 23-25 11972)
27 Cohen. C & Tooney, N. M Mafur? 251, 6 5 9 4 6 0 (1974)
28. T w n e y . N M. & Cohen. C. J. mmolcc. R d 110, 363-385 (1977)
2 9 W e ~ s e l J, . W.. Warren. S G & Cohen. C J molrr H d 126. 159-183 11978)
3 0 P h ~ l l ~ pGs . S , lnttrnan, E E , Curnmsns, P Lec. K Y & Cohen. C Nnlure 278.413 4 1 7
(1979)
31 Cohen. C . Rcvel, J P & Kucera. J Science 141.436 4 3 8 (11963)
32 Cohen, C.. Slayter. H S t i o l d s t c ~ n ,L . Kuccrr. I. & Hall. C J molrr Bml 22, 3R5-3RR
(1966).
1 3 . Kay. D & Cuddtgan. B J Hr I H a r m o f 13,341-347 (1967)
34 Karges. H . E & Kuhn. K Eur 3 Bwchem 14, 9 4 9 7 (1970)
35. Tooney, K M & Welrrl. J. W. i n Fdbrous Profcrns Scwnf!fic. Indurrrroland M ~ d i c oArpccfr
l
,
1979)
Vol I (eds Parry. D A . D & Cicamcr. 1.. K ) 3 9 3 4 2 7 ( A c a d e m ~ cI.undon.
36 W e ~ s e l .J . W . , I n n n e y . N M.. Kaplan. I.. Amran,, D. & (:ohen. (' J mniw H i d 143.
329-334 ( 1 9 8 0 )
37 Cohcn. C ~n Molccuior Archffrrrure rn C d l Physzology icds Hayash,, T & Szent-Gyorgy~.A
G 169 190 ( P r e n l u ~ H a l lS. e w Jcrscy. 19661
38 Hlornhack. R ? ! a 1 Nnrurr 218, 130-134 11968)
3 9 Blomhack. B . . Hessel. B HDRR.I1 & Therktldsen. L ' o r u r r 275. 511I~SO5( 1 9 7 8 )
4 0 Olcxa, S A & Rudzynskt. A Z Proc narn A r o d Srt [ I S A 77, 1374-1378 119801
41. Fowler. W E Erlcksnn. H . P , H a n t g m . R R McDonagh. J & Hermans. J . Screnre Iln thc
press1
42. StauHarher. C . P h ~ l l ~ pGs . h' Welsel, J W . , Ford, E D B ('ohen. C
preparallon)
4 3 Slayter. H S In Pnnr,plcr und Technrqurs of Elprrron M,croscop, Vol 9 ( r d Hayat, M )
I 7 5 2 4 5 IVdn Nostrand. New York. 19781.
4 4 . Kluy, A & D e R o s x r , D . J Narurr 212. 29-32 (1966)
45 Fnckson, t { P , Votcr. W A & I x o n a r d . K Mcrh Enzym 49. 3 Y 4 3 (19781
4 6 Mlsell. D L In P r a c r m l .Uuhudr In Ekcrron M~crorropv Vol 7 icd Clauert, A M )
(North-Holland. Amrtcrdam. 1978).

production spectrum of p. This calculation confirmed the

conclusions of Szabelski etaL6.The results from this calculation7
are now used to examine the implication of the observed p in
terms of cosmic ray propagation models.
The equilibrium spectrum of p has been obtained by taking
into account ionization loss, annihilation of p, and inelastic
interactions in which p retains nearly half its energy. The
resultant p / p ratio as a function of the kinetic energy of p is
shown in Fig. 1 for various cosmic ray models. Curve S is the
estimated p / p ratio according to SLBM, in which cosmic rays
traverse 5 g cm-* of interstellar gas independent of energy. This
~~
curve can be compared with other existing c a l c u ~ a t i o n scarried out using SLUM only. The calculations of Gaisser and
Maurer4, and Szabelski et al.' agree within 25% with curve S
over the entire energy region above 2 GeV. This small deviation
arises partly from the differences in the parametrization of
cross-sectional data and partly due to different demodulated
proton spectra used. Hadhwar et al? overestimate the p / p ratio
bv a factor of 2 at -10 GeV and this difference increases with
decreasing energy; however, they also agree within 25% at
energies 2 5 0 GeV.
The SLBM has been abandoned since it was observed that the
abundance ratio of secondary nuclei (like Li, Be and R ) to
primary nuclei (like C and 0)decreased with increasing energy.
This result suggested that high-energy cosmic rays traverse less
amount of matter. Recent observations on the spectra of cosmic
ray nuclei suggest"hat the energy dependence of the matter
/~,
traversal is approximately proportional to ( ~ G V / C / R ) ' where
R is the rigidity in GV/c, for R 2 8 G V / c and is nearly constant
for R < 8 GV/c. This dependence has been explained in two
ways. (1)The confinement o f cosmic rays in the Galaxy is energy
dependentp-'modified
leaky box' model (MLBM). (2) A
certain fraction, f , of the matter traversed by cosmic rays in the
source regions is energy dependent-'nested leaky box' model"'
(NLBM). Curve M in Fig. 1 is the calculated p / p ratio on the
basis of MLBM and the hatched region (N) is according to
NLBM. The energy-dependent escape probability in NLBM
from the source regions is assumed to be similar to that of
MLBM, and that the median energy of interacting nucleons is
-10 times the energy of p produced; the upper and lower
bounds are for f = 0.5 and 0.8 respectively. Figure 1 shows that
-

1981 Macrnillan Journals I td

Nature Vol. 289 22 January 1981

268
NLBM predicts too small a flux of P compared with MLBM in
the energy region below a few tens of GeV.
Figure 1 plots the observed p/p ratio of (5.2* 1 . 5 ) ~
reported by Golden et al.' in the energy region 4.7-11.6 GeV;
this result was based on 28 P events of extraterrestrial origin.
The correct value may be higher by -15% because the expected
flux7of p of atmospheric origin is smaller than that estimated by
Golden etal. The P/p ratio obtained by Bogomolov etal." is not
plotted because it was based only on two p events, and in this
experiment, splash albedo particles in the atmosphere could not
be eli~ninated'~.
Figure 1 shows that popular models of cosmic
ray propagation predict too low a flux of p by a factor 2 4
compared with the observed flux; however, the significance of
this difference is only about 3 a . In the following, we examine the
'closed galaxy' model originally proposed by Rasmussen and
Peters13to understand the apparently high flux of P observed by
Golden et al.'.
In the 'closed galaxy' model, cosmic rays are totally confined
until they are destroyed by interaction. Curve C in Fig. 1 is the
calculated p/p ratio on the basis of total confinement. The
observed ratio falls short of curve C by a factor of -2. It is
therefore proposed that -50% of the observed cosmic ray
nucleons are of recent (-lo7 yr) origin, and are propagated
according to either MLBM or NLBM; the remaining ones are
the result of total confinement. We can now combine 50% of
curve C with either curve M or hatched area N, and the resultant
P/p ratio lies between the two short curves passing through the
data point in Fig. 1. Thus, the 'modified closed galaxy' model
(MCGM) can account for the observed flux of P.
The secondary positron spectrum in cosmic rays can now be
calculated using MCGM. The equilibrium spectrum of e' is

Kinetic energy (GeV)

Fig. 1 The calculated p/p ratio plotted as a function of the kinetic
energy of p, for various models of cosmic ray propagation. The
solid curves passing through the data point are the estimated ratio
by summing 50% of curve C with 50% of either curve M or the
lower bound of the hatched area N. @,Data from ref. 1; S, SLBM;
M, MLBM; N, NLBM; C, CGM.

Energy (GeV)
Fig. 2 The calculated energy spectra of the old and new
components of secondary cosmic ray e' are shown in the left side,
and the combined e + and total electron spectra on the right side;
the flux values are multiplied by E'. Symbols as in Fig. 1; curves C1
and C2 correspond to n~ = 0.1 and 0.2 atom ~ m - ~
the, mean
hydrogen density of the interstellar gas traversed by the old
component. Data are taken from: +,ref. 17; @, ref. 18; A, ref. 19.

calculated by using the recently estimated14 production spectrum and taking into account all energy-loss processes-ionization, bremsstrahlung, synchrotron and inverse Compton
scattering. For convenience, the values of mean magnetic field
B, and the radiation density are assumed to be the same for all
models and are taken to be 4 p G and 0.7 eV c W 3 respectively.
The calculated spectra of e' from the old component are
shown in Fig. 2 by curves C1 and C2, which correspond respecn , is the mean
tively to n ~ = 0 . 1and 0.2 atom ~ m - where
~,
hydrogen density of the interstellar gas traversed by this
component. The flux values in Fig. 2 are multiplied by E2.The
flux of e' is proportional to nH at high energies, where synchrotron and inverse Compton scattering losses dominate, and is
independent of n~ at low energies, where ionization and bremsstrahlung losses dominate15. The dashed curve M is the calculated new component due to MLBM. In this calculation, the
mean lifetime of cosmic rays of R =S 8 GV/c is assumed to be
lo7 yr, as determined by radioisotope measurementsI6, and
during this period they traverse 5 g cm-2 of interstellar gas. The
recent e+ component according to NLBM is calculated as in the
case of p and is shown by the hatched area N.
The sum of the old and new components of the e' spectrum is
shown on the right side in Fig. 2. The solid curves (T) are the sum
of curves M and either C, or C,, and the hatched area corresponds to the sum of the hatched area N with curve C1 or C2.
The observed flux values of e' are plotted as filled data points at
energies 2 3 GeV (refs 17-19), where the effect due to solar
modulation is small. The band representing MCGM agrees with
the observations, which have a large scatter. Note that other
existing models cannot satisfactorily explain e' observation^^,'^.
Also the value of n~ needed for the old component cannot be
very much smaller than 0.1 atom ~ m - This
~ . means that the
effective confinement region for cosmic rays may not include
regions very far away from the galactic disk, say larger than a few
kpc. Also the total (e++e-) spectrum in interstellar space

O 198lNature Publishing Group

Nature Vol. 289 22 January 1981

derived on the basis of curve T can be shown to be consistent
with that deduced from the non-thermal radio spectrum in the
Galaxy.
Half of the observed nucleons, belonging to the old
component, is shared among various nuclear species approximately in proportion to the respective interaction mean-free
paths. The estimated percentages of the old component among
the observed p, He, (C+ 0 ) , (F to Sc), (P to Cr) and (Fe Group)
nuclei are respectively 58, 21, 9, 6, 4 and 3. Note that the new
component constitutes >90% of the observed flux for nuclei
other than p and He. Therefore, one can explain the observed
energy dependence of the abundance ratio of secondary to
primary nuclei using MLBM or NLBM for the new component.
Further, the relative abundances of heavy nuclei of charge 2 6
would not differ from other models by more than a few per cent.
We now need to choose between MLBM and NLBM for the
new component. According to MLBM, the energy spectrum of
nucleons of recent origin above 4 GV/c (=4 GeV/n) would
be steeper than the injection spectrum by a power-law spectral
index of about 0.5; the old component retains the injection
spectrum. This would mean that the old component would
dominate the observed flux beyond a certain energyz0. The
calculated energies at which the intensities from these two
components become comparable are -56,
460 and
1,950 GeV/n for He, (C+ 0 ) and (Fe group) respectively; protons have 58% old component even below 8 GeV. This clearly
indicates that the observed similarity in the spectral shape of p
and He above a few GeV/n cannot be understood, unless
protons are accelerated differently. One should also expect a
flattening of He spectrum beyond about 50 GeV/n, which has
not been observed at least up to a few hundred GeV/n. These
two inconsistencies discredit MLBM, and the natural choice is
for NLBM, because the spectral shape of the new component is
the same as the old.
A reliable estimate of the abundance of deuterium can be
made using this model at relativistic energies. The estimated
D/He ratio from this model is nearly twice as much from other
models at energies ~4 GeV/n a n d decreases slightly at high
energies due to the energy-dependent confinement of the recent
component in source regions. A similar enhancement is expected in the case of 3He abundance. The available measurements
have been restricted to energies S 2 0 0 MeV/n and these results
differ by a large factor2'. However, they can be easily detected at
relativistic energies using magnetic spectrometryz2. The observation of an enhanced abundance of D would not only confirm
the predictions of this model but would also reduce the difficulty
in understanding the observed charge ratio of muons at sea level.
Perhaps the distribution and spectral shape of high-energy
y rays, and the extended nonthermal radio emission in the
Galaxy might also be explained by this model.
I thank Professors M. V. K. Apparao, R. R. Daniel and P. V.
Ramana Murthy for helpful criticisms.
Received 15 September; accepted 10 November 1980

1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.

Golden. R. L. eral. Phys. Rev. Lcn. 43. 1196-1199 11979).

Radhwar. G. D. etai. Nature 274,137-139 (1978).
Smoot, G. F.. Buffington. A. & 01th. C. D. Phys. Rev. Len. 35.258-261 (1975).
Gais6er.T. K. & Maurer. R. H. Phys. Rev. Lctf. 30, 1264-1267 (1973).
Badhwar. G. D.. Golden, R. L., Brown. M. L. & Lacy. I. L. Astrophys. Space Sci. 37,
283-300 (1975).
Szabelski. J., Wdowuyk, J. & Wolfendale, A. W. Nature 285, 386-388 (1980).
Stephens. S. A. Astrophys. Space Sci. (in the press).
01th. C. D.. Buffington. A,, Smwt. G. F. & Mast. T. S. Astrophys. 3. 226. 1147-1161
(1978).
lul~usson,E., Mayer, P. & Muller. D. Phys. Rev. Lett. 2 9 , 4 4 5 4 4 8 (1972).
Cowsik. R. & Wilson. L. W. Proc. 13th inf. Cosmic Ray Conf. 1,500-505 (1973).
Bogomolov. E. A. cl al. Roc. 16th in:. Co.smic R a y Conf. 1. 330-335 (1979).
Daniel. R. R. & Stephens. S. A. Buil. a m . Soc. Ind. (in the press).
Rasmuse.cn. I . L. & Peters. B. Nature 258,412-413 (1975).
Badhwar, G. D. & Stephens. S. A. Prm. l 5 f h in:. Cosmic R a y Con( 1, 3 9 8 4 0 3 (1977).
Badhwar. G. D. & Stephens. S. A. Phys. Rev. D 14,356-358 (1976).
Buffington, A,, Onh, C. D. & Mast, T. S. Asnophys. 1. 226, 355-371 (1978).
Fansclow,I. L., Hanman, R . C . , Hildebrand, R.H. & Maye1.P. Asfrophys. 1.158.771-780
(1969).
Golden, R. L. el al. Preprint PSL-PE00947 (1979).
Buffington, A. & Orth, C. D. Asfrophys. J. 199,669-679 (1975).
Peters, B. & Westergaard. N. J. Asnophys. SpaceSci. 4 8 , 2 1 4 6 (1977).
Daniel. R. R. & Stephens. S. A. Space Science Review 17.45-158 (1975).
Badhwar. G. D.. Stephens, S. A. & Golden. R. L. Proc. 14th inf. Cosmic R a y Conf 9,
3183-3187 11975).

0028-0836181 1040269-03$0 l .OO

269

Expanding haloes in cometary comae

Max-Planck-Institut fur Aeronomie, D-3411 Katlenburg-Lindau 3,
FRG

Observations indicate that cometary activity can be rather

violent. Flare-ups, sunward-pointing jets and periodic formation of concentric haloes expanding outwards can also be
persistent features in some comets. For example, a halo structure was observed in the coma of Comet Halley during its last
visit1. It is argued here that halo formation may be related to the
propagation of shock pairs driven through the cometary atmosphere as a result of spin-modulation of the cometary outgassing
rate. Even though the non-stationary process could be less
dramatic than the example considered here, its effect on the
chemistry and expansion of the neutral gas as well as dynamics of
the cometary plasma must be significant.
Although theoretical models of cometary coma generally
assume that neutral atmospheres expand spherically, symmetrically and without time variation, inhomogeneous structures have often been observed in the cometary comae. These
include the formation of concentric shells (or haloes) moving
outwards from the central source in succession and the
appearance of fans or jets. Such non-stationary behaviour has
long been thought to be related to the spin-modulation of the
outgassing rate from the cometary n u ~ l e u s ~ * ~ .
There are two possiblities. If the surface composition of the
cometary nucleus is uniform, the dependence of the surface
temperature on the angle of insolation ( 8 ) would mean higher
gas emission rate at the point shifted from the sunlit position
with 0 = 0". (The exact position of the 'hot spot' is determined by
the direction and period of the nuclear spin as well as thermal
properties of the surface material; the resultant jet effect has
been used to explain the nongravitational acceleration or
deceleration of comets2.) As the jetting of the volatile gasthough localized-is more or less constant, the expansion of the
cometary atmosphere, except the persistent sunward jets,
should show no marked time variation. On the other hand, if the
surface of the cometary nucleus can be divided into one active
hemisphere and an inactive one (say), there will be more violent
ejection of the volatile gas every time the active region turns
towards the Sun. In the extreme case, there will be regular
outbursts at the centre with a period equal to that of the spin
period leading to large spatial and temporal variations in the
structure of the coma. Clearly, the formation of expanding
haloes as seen in the case of Comet Donati in 1858 (ref. 4) and
Comet ~ a l l e must
~ ' be somehow related to such periodic
outbursts. (From measurements of their respective halo structures, W h i ~ p l e ' . ~
has determined the spin period of Comet
Donati to be 4.6 h and for Comet Halley to be 10.3 h.)
To understand this phenomenon, time-dependent expansion
of the gas outflow must be considered. Here it is assumed that
the cometary atmosphere is an ideal gas and that the radial
expansion is spherically symmetric. Furthermore, the time
variation of the outgassing rate is approximated by a step
function and we are basically interested in the initial propagation of the cometary neutral gas as the strong source is
switched on. If p is the density, u the radial velocity, p the
pressure and y ( = 513) the ratio of specific heats of the gas, then

@ 1981 Macmtllan Journals Lrd

270

Nature Vol. 289 22 January 1981

for the hydrodynamic equations of expansion of the neutral gas

from a point source. Introducing the similarity variable7,
for the expansion of a piston gas with a constant speed cu and
defining
with the density distribution of the cometary coma at t
as:

0 given

(that is, p = po at r = ro for the unperturbed flow), we can find an

equivalent set of equations for V, a and a with q as the sole
variable. As discussed by Simon and Axfords, similarity solutions of V, a and T can be obtained for the motion of such a
piston gas describing a pair of shocks (S, and S-) separated by a
contact surface (C). The characteristics of the strong-shock
solutions are that at the contact surface with q ~ 0 . 8 4 ,V = 1,
a = 0 and a + 00; and that the mass efflux from the source can be
expressed as
where f 5 0 . 4 in the case of strong shocks. With V, a and .rr
known for q 3 0 , the physical quantities u, p and p can be
derived in turn for various values of r and t from equation (5).
Details of the similarity solutions for a blast wave can be found
in ref. 8 and numerical solution of the above adiabatic nonsteady flow equations in ref. 9. For brevity, only radial variation
of the mass density will be discussed here. Figure 1 shows the
profile of the gas density (p) plotted for the case of strong
forward (S,) and reverse (S-) shocks. As u + co at the contact
surface (C),p increases to a large value also. For weaker shocks,
the snow-plow effect of piling up the material between S, and
S-, though less pronounced, should be present. The similarity
solutions of the piston gas apply only to the inner coma region
which is still dominated by collisional effect. For a Halley-type
comet with a nuclear radius (r,) of 3 km and a surface number
~ , nominal dimensions of such
density (no) of 3 x 10" ~ m - the
collision zone (r,) is ~ 1 . x4 lo4 km. As a result of the varying
outgassing rate, r, could be a factor of 3 larger or one-third
smaller as the comet rotates around. Even with the large value of
4 x lo4km, rc is still small compared with the radius of outer
haloes (-3 x lo5km) (see ref. 1).The compression effect by the
piston gas must then stop a long way inside. However, the sharp
definition of the haloes may still be maintained at large distances
if the temperature of the gas at the contact surface is kept
sufficiently low by collisional cooling. The delineation of the
haloes is also helped by the rare-fraction region just behind the
reverse-shock S-. Indeed, the density variations as shown in Fig.
1 may explain why in some cases a dark zone is observed4
immediately behind a halo of enhanced brightness. Another
effect is that the propagation speed of these haloes could be
considerably lower than the value of 1km s-' generally quoted
as the flow speed of the neutral gas in the inner coma close to the
nucleus (r s
km) could be very small.
The dimension of these haloes as observed in the C2-emission
are limited by the scale length of these molecules (=lo5 km at
1AU). Furthermore, the daughter fragments of the parent
molecules (such as H 2 0 and CO) in the compressed shells would
pick up extra kinetic energy during photodissociation or ionmolecular reactions such that the corresponding thin-shell
structures would be dispersed. Therefore, expanding haloes are
not likely to be seen in the density distributions of H, C and 0
atoms, for example. We may then envisage the formation of a
new spherical halo moving outwards every time the active
hemisphere of the cometary nucleus faces the Sun. Partly due to
the pile-up of material between the forward and reverse shocks,
and partly due to the projection effect along the line-of-sight, a
concentric halo will be observed progressing outwards. Note
that the time variation of the outgassing rate could be anything
but a step function as adopted here and the shock velocity a

should have certain time variation. The dynamical process,

however, is not likely to vary much as long as the active
hemisphere of the comet can produce a strong driver gas. The
real issue is, therefore, what is the energy source for such
outbursts if they indeed are explosive.
In connection with the flare-ups of cometary activities sometimes observed, Donn and Ureylo have suggested that explosive
chemical reactions involving free radicals or unstable molecules
on the nuclear surface may provide the answer. The free-radical
reactions studied in the laboratory are observed to be highly
exothermic, yielding a heat of reaction of approximately a few
electron volts per molecule. This would mean that there are
pockets of unstable chemical mixture on the surface of the active
hemisphere with energetic explosions triggered by solar radiation. Another possibility suggested by Greenberg'' in the case of
interstellar grains-is simply that some of the solid grains could
release free energy stored in the mantle of organic molecules
and free radicals through breakup of the cometary grains.
Although highly uncertain, it is, nevertheless, interesting to look
into the energy budget of the coma expansion. Taking the radial
speed to be 1 km s-', the kinetic energy is then only of the order
of 0.1 eV per H 2 0 molecule and about E == 10'' eV s-' in total if
the gas production rate is ==loz9H 2 0 molecules s-I. An
equivalent amount of chemical-energy source could be obtained
if the sublimating ice contains 3% or so of the highly volatile
material capable of exothermic chemical reaction. The energy
budget of the halo expansion is therefore not a serious problem.
In other words, even if the gas production rate were the same
over one spin period, the contamination of one side with exothermic material would result in high-speed ejecta driving a
shock pair periodically.
The hydrodynamics of the neutral coma in the inner region is
not well understood. Various possible flow patterns including
shock formation in steady-state conditions have been considered (see refs 12 and 13). The time-dependent expansion of the

Normalized radial distance ( q )

Fig. 1 The halo formation in the cometary comae through the
compression effect of a piston shock. Radial variations of
the density (6) of the cometary neutral atmosphere normalized to
the corresponding value at thezposition of the forward shock (S,) at
17 = 1. Therefore 6 = u ( q ) / ~for q = 1 and 6 = 1/17' for 17 > 1.
The value of u(q) is obtained by taking a similarity solution to
equations (1)-(6) with strong shocks (see ref. 8). The brightness
enhancement in the haloes is caused by the density compression
between the forward and reverse shocks. The dark zone sometimes
observed immediately behind the haloes could be the manifestation of the rare-fraction region behind S- as indicated. The
density profile not modified by the shock pair is depicted by the thin
curve.

O 198lNature Publishing Group

271

Nature Vol. 289 22 January 1981

Whether any of these mechanisms may provide the energy

required to drive a shock through the cometary atmosphere
remains to be seen. But high-speed HCN jets (u 2 5 km s-I)
have been detected in the coma of Comet Kohoutek (19730 by
Huebner et al.l4, who also favoured explosive chemical reactions as the origin of these iets. Thus the formation of shock pairs
together with compression region in between must beconsidered a viable ex~lanationfor the ex~andinehaloes. For
illustration and simplicity we have considered only the strongshock case for the similarity solutions. Other types of compression processes with weaker shocks must also exist; the present
interpretation, nevetheless, highlights the nature of non-steady
expansion of the cometary atmosphere.

cometary atmosphere should be even more complicated. Some

idea can be derived from Bobrovnikoff's record of Comet
Halleyl where the expansion speed of the halo is estimated to be
~0.25-0.5kms-'. Identifying this value as the propagation
speed of the contact surface and using the relation that the flow
speed u * behind the reverse shock should be larger by a factor of
2S8, we obtain u* =O.65-1.3 km s-'. This would mean that an
additional energy source amounting to E*-- f Q(u *)' 1:
loz8eV s-' (for Q loz9H 2 0 molecules s-I) is needed to drive
the piston gas. This requirement can be accommodated by
introducing a small fraction of exothermic volatile material into
the sublimating gas.

Received 28 July; accepted 14 November 1980.

8. Simon, M. & a n d Axford. W. I. Planet. Space Sci. 14,901 (1966).

9. Hundhausen,
J. I. geophys. Res. 78, 1528 (1973).
10. Donn, B. & Urey. H. C. Astrophys. 1.123,339 (1956).
11. Grecnberg. J. M. Moon & Planets 20, 15
12. Shul'man. L. M. in Physics of Comers
Konoplcva, V . P.)84 (transl. NASA 'IT F-599.

1. Bobrovnikoff, N. T.Uniu. Calif. Publ. Lick Obs. 17, 2 (1931).

2. Whipple, F. L. Asrrophys. 1.111, 375 (1950).
3. Sekanina. Z. Icarus 37.420 (1979).
4. Rahe. J.. Donn. B. & Wurm, K. Atlas ofComcrary Forms (NASA SP-198. 1969).
5. Whipple. F. L. Nature 272, 134 (1978).
6. Whipple. F. L. I A U C i r c . No. 3459 (1980).
7. Courant. R. & Friedrichs. K. 0.Supersonic R o w and Shock Waues (Interscience. New
York. 1948)

A.

(1979).
(ed.

1969).
13. Mendix. D. A. & Ip. W.-H. Astrophys. Space Sci. 39,335 (1976).
14. Huehner, W. F., Snyder, L. E. & Buhl, D. Icarus 23,580 (1974).

Blocking highs over Asia

and monsoon droughts over India

middle latitude westerlies and the elongation of mid-tropospheric westerly troughs into India. However, eastward-moving
westerly troughs penetrating into northern India enhance
monsoon activity, and ~ a m a n a t h a nhad
~ visualized that migrating middle latitude ridges influence monsoon onset. Examining
the severe droughts of 1979 (MONEX), 1972,1966 and 1965,
Raman et al.' found that the breaks in these years were associated with stagnant upper tropospheric blocking ridges over the
east Asia along -lOOOE, extending from 3S0N to 70N. In such
situations, westerly troughs do not move beyond 80E and
instead tend to extend southwards.
Cold lows form south-west of the EABR and the lower
tropospheric monsoon trough over India gets displaced northwards into the Himalayas-the most significant characteristic of
breaks. We suggest that an upper trophospheric WABR along
-50E is a necessary precursor to the development of the
EABR. Also that in drought years, a well-marked surface high
appears over USSR centred at 55"N and 50E. This surface high
is linked by a ridge with the Azores High over the central
Atlantic. The surface high over the USSR, overlain by a ridge of
up to 100 mbar, is due to the WABR. In contrast, in good
monsoon years, the surface high and the associated upper ridge
over the USSR are notably absent. Instead, a trough from the
Aleutian low extends into Siberia with several low centres. Such
a configuration is akin to the normal pattern.

C. R. V. Raman
MONEX, New Delhi-3, India

Y. P. Rao
B3/33 Azad Apartments, New Delhi-3, India

Severe summer droughts of the Indian subcontinent have been

found to accompany prolonged 'breaks' in the south-west
monsoon. Data from the years of severe drought suggest that the
associated breaks were-due to upper tropospherk blocking
ridges over East Asia. We present here a model for the evolution
of G o such blocking ridge;: the East Asia (EABR) and the West
Asia (WABR). Apparently an intense WABR-the initiator of
the monsoon break-precedes the formation of the EABR. This
cycle could be a way of predicting years of intense drought or
monsoon activity.
For a long time, studies of monsoon breaks were confined to
changes in circulation in the monsoon area. R a m a ~ w a m ~was
'-~
the first to associate breaks with the low index circulation in

Mg. 1 Evolution of the WABR and

the EABR. Solid lines denote 200mbar contours; H, high; L, low;
dashed-dotted line indicates northsouth orientation of the 200-mbar
blocking ridge. Stage 1, Appearance
of WABR. Stage 2, Maximum phase
of WABR with intense cold lows on
either side. Surface high delineated
by pecked isobar. EABR had commenced formation a few days earlier.
Lower
tropospheric
Monsoon
Trough (MT) in normal position.
Stage 3, EABR had intensified along
with weakening of WABR. Low
upstream of the EABR shifts southwards. Stage 4, Monsoon trough is
displaced significantly northwards
leading to monsoon break.

STAGE-I

STAGE-2

-60'

60"

40"
I

20

. 40'
MT

"

- 20"

@ 1981 Macmillan Journals Ltd

272

Nature Vol. 289 22 January 1981

Table 1 Evolution of WABR and EABR

EABR (date of)

WABR (date of)

Month and year

July 1972

Duration of
monsoon break/
deficient
monsoon rains
18-3 1 July

August 1972

1-9 August

August 1966

23-27 August

August 1965

4-15 August

July 1960

6-21 July

Appearance
(mbar)
4 July
(200)
29 July
(200)
12 August
(300)
25 July
(300)
27 June
(300)

Max intensity
with value
of closed
contour at
300 mbar/
200 mbar/
(gpd*
14 July
(1,240)
4 August
(1,208)
20 August
(944)
30 July
(928)
1 July
(944)

Formation of
surface high
with departure
from normal
(mbar)
14 July
(+14)
8 August
(+IS)
23 Qugust
(+12.5)
4 August
(+12.5)
4 July
(+12.5)

Appearance
(mbar)
8 July
(200)
1 August
(200)
18 August
(300)
31 July
(300)
2 July
(300)

Max intensity
with value
of closed
contour at
300 mbar/
200 mbar
(gpm)*
18 July
(1,200)
5 August
(1,216)
23 August

Formation of
surface high
with departure
from normal
(mbar)
25 July
(+14)
8 August
(+IS)
23 August

10 August
(944)
5 July
(952)

* gpm, Geopotential metres.

The evolution of the WABR and the EABR, during drought
years from 1960, was obtained from Northern Hemispherical
Synoptic charts6. The sequence of events preceding a monsoon
break starts with the appearance of a predominant ridge at
200 mbar, first in the longitudinal belt 40-70E, extending from
the subtropical high pressure cell in the monsoon regime to
almost 80N. Stagnating at times for periods of up to one to two
weeks, the WABR builds up as a warm high, reaching almost to
the surface in a week after appearance. At its intense phase,
even a closed high develops in the whole column, with the
surface pressure departingfrom the long term normal (1931-60)
by as much as +20 mbar. Such a deep meridionally-oriented
blocking ridge configuration over west Asia elongates the westerly troughs both on the upstream (west European) and the

downstream (central Asian) sides. About a week after the

maximum phase, the WABR begins to weaken.
The EABR generally appears in the belt 90-120E some four
days after the WABR forms, and extends from 35 to 70N. The
EABR begins to build up in a similar way to the WABR,
generally when the latter starts to shrink in meridional extent
and weaken. During this phase of the build up of the EABR, the
westerly trough over central Asia between the WABR and the
EABR further elongates to more southern latitudes. (The
temperatures in this trough are -5-8 "C below normal in the
mid-troposphere but warmer than the surrounding at 200 mbar
and 100 mbar.) The superposition of this elongated cold upper
trough/low at southern latitudes is manifested as the displace-

Fig. 3 Mean conditions during August 1975, a year of good

Fig. 2 Mean conditions in August 1972, a year of severe
monsoon drought. a, Surface isobars. Note marked high centre at
55"N 50E and the associated ridge linking with the Azores High
over the central Atlantic. T h e ridge extends t o 90E across Eurasia.
b, 200 mbar. Solid lines denote contours and dashed lines represent
temperature (OC). C, cold; W, warm. Note markedly cold upper
tropospheric ridge along 50E, above the surface high over USSR.

monsoon rains. a, Surface isobars. Note the absence of a high

centre over the USSR. Instead, a trough from the Aleutian low
extends into Siberia with several low centres. Note that the
monsoon low over India and neighbourhood is more intense than
in Fig. 2 a (1972). b, 200 mbar. Solid lines denote contours and
dashed lines represent temperature (OC). Note the circumpolar flow
and absence of ridges along 5O0E.

O 198lNature Publishing Group

Nature Vol. 289 22 January 1981

273

ment of the lower tropospheric monsoon trough (MT) to the

north into the Himalayas-symptomatic of the monsoon break.
The build-up of an intense WABR seems essential for creating
the EABR, which brings about the monsoon break. Thiscycle of
the WABR building up well before the break could be useful in
predicting monsoon breaks and consequent droughts. The
sequence of evolution is presented in Fig. 1. A few examples of
the cycle are given in Table 1.
Certain broad features of the WABR and the EABR are
striking even in the monthly mean charts of 1972, a severe
drought year. Figures 2 and 3 present the surface and upper air
circulation features during August 1972 and 1975 respectively.
In 1972, the WABR had started to build-up by the beginning of
June and continued during the first half of July 1972. The
subsequent build up of the EABR coincided with the
pronounced break over India during the second half of July. The
WABRIEABR configuration continued in August 1972, when
the monsoon rains failed. The WABR prominent in Fig. 26 at
200 mbar, is associated with cold temperatures along -50E.
The surface high vertically below this 200 mbar ridge is quite
prominent (Fig. 2a). The EABR is less marked in the monthly
mean chart (Fig. 26), although it is seen markedly in decadal
mean charts covering break occurrences in this and other years.

In the good monsoon year of 1975, the circulation at 200 mbar

(Fig. 36) is more circumpolar with a weak trough in the location
of the 1972 WABR (50E). The contrasts between the good and
bad monsoon years described above in the surface pressure
distribution and the upper air features from the eastern Atlantic
to east Asia are shown by other years during the past two
decades.
Negative temperature anomalies are noticed at 200 mbar and
above in WABR and EABR, with positive anomalies in the
troposphere. The tropospheric warmth seems to be associated
with the subsidence characteristic of high pressure. The stratospheric cooling must be due to the ascent, and maintains the
high. Hoskins and Simmons have revived interest in downstream development as an important atmospheric scale
phenomenon. It might be useful to apply this study to monsoon
variability over India.

Possible role for metals

in stratospheric chlorine chemistry

metallic constituents of meteors are (by weight): A1(1.7%),

Ca(l%), F e ( l l S O / ~ Mg(12.5%),
),
Ni(1S0/0) and Si(20%). Here
Si is included in the term 'metallics'. Other, less abundant metals
(such as Ti, Cr, Mn) have been observed in micrometeorites
collected in the ~ t r a t o s p h e r e ~ ~ .
We now consider chemical reactions involving these metals or
their intermediates with CI, C10 and HCI, even though a
quantitative evaluation of these reactions is not possible because
of the lack of kinetic and thermochemical data. C1, C10 and HC1
have significant concentrations at altitudes between 15 and
50 km (refs 23, 24). Sodium will probably be in the form of
NaOH (ref. 18) and/or NaHC0, by the time it has descended to
50 km (refs 19, 25). Similar compounds will be formed in the
case of the other metals. The basic mechanism for the formation
of the metal hydroxides (MOH) and the bicarbonates (MHCO,)
seems to be:

Edmond Murad & William Swider

Air Force Geophysics Laboratory, Hanscom Air Force Base,
Massachusetts 01731

Sidney W. Benson
Department of Chemistry, University of Southern California,
Los Angeles, California 90007

Received 4 August: accepted 14 November 1980.

Ramaswamy, C. Geophysics 6,455-477 (1958).
Ramaswamy. C. Tellus 14. 337-349 (1962).
Ramaswamy. C. Bull. Ind. norn. Sci. Acad. 54, 109-132 (1975-76).
Ramanathan. A. S. I. mar. Biol. Ass. Ind. 14, 843-861 (1972).
5. Raman, C. R. V., Rao, Y. P. & Alvi. S. M . A. Curr. Sci. Ind. 49, 123-129 (1980)
6. Synoptic Bull. Nonhem Hcmisphrrc Parts 1-111 (USSR. 1960-1978).
7. Simmons. A. J. & Hoskms. B. J. J. armos. Sci. 36, 1239-1254 (1979).

1.
2.
3.
4.

Meteoroids ablate largely in the Earth's ionosphere (at altitudes

between 80 and 120 km) and the products are seen mostly as
atomic
Atomic neutral atoms have also been observed
MO + H,O -+ MOH + O H (M = Na and, possibly, Al)
(1)
and studied optically, but the data are not extensive for elements
other than s o d i ~ m ~The
' ~ .processes responsible for maintaining
MO + CH, + MOH + CH3 (M = Na, Al)
(2)
the equilibrium of metallic atoms and ions in the ionosphere are
understood generallyd-9. Below 8 0 km, the metal ions are
largely neutralized. The chemistry which occurs during the
MOH + CO, + X + MHCO, + X (for all MOH)
(4)
subsequent transport of the metals to the ground is uncertain;
aerosols collected in the lower stratosphere have been reported
where M is any meteor metal and X is any third-body. Metal
to contain chlorides and sulphates of metals (some of-which
chlorides, MCI, may then be formed by the reactions:
occur in meteor^)'"'^ while aerosols collected in the troposphere have been found to contain many metals among which are
meteor metals13-". It has been suggested that ions of the type
MOH + C10 -+ MC1+ H 0 2 (all M)
Na' . (H20). may be present in the upper stratosphere1', their
presence being plausible in view of a meteor sodium model
(7)
MOH + HCI + MCl + H 2 0 (all M)
which indicated that large amounts of NaOH (almost lo6 ~ m - ~ )
+
HCI
+
MCI
+
(H2C03
or
H
2
0
+
CO,)
MHCO,
would be present at 40 km (ref. 18). This latter conclusion was
(all M) (8)
questioned19 in view of thermochemical data which indicated
Reactions (7) and (8) are expected to proceed with large rate
that NaOH, once formed, should be removed quickly by reaccoefficients (-lo-' cm3 molecule-' sC') and can deplete HCI
tion with CI, CIO and HCI. We consider here reactions of
significantly relative to the reaction
meteor, and other, metals with CI, CIO and HCI to assess the
impact of halocarbons on ozone. Release of 1,000 tonne of
sodium (or similar metal) may significantly reduce the harmful
effects of halocarbons on ozone for several years.
which is an important step in the chain reaction leading to the
depletion of ozone26. Reaction (9) has a rate coefficient2' of
Although sodium comprises only about 2% by weight of the
total metal concentration in meteors2', its reaction with CI, C10
6.6 x lo-', cm3 molecule-' s-' so that even if k7+ k, [rate
and HCI may represent a class of reactions which could be
coefficients for reactions (7) and (8)] were a factor of 1,000 less
than estimated (that is -10-l2 cm3 molecule-' s-'), they would
important in the chemical maintenance of these species in the
still be important competition for reaction (9) for [OH]stratosphere (the condensation of Ga, Ge and Sb in nebulae has
[MOH] + [MHCO,]; the data of refs 18 and 23 indicate this to be
been suggested to proceed through the respective metal hytrue at -20 km. Of the meteor metals, sodium and aluminium
droxide and metal chloride intermediatesz1). The other major
0028-0836/81/040273-02$01.OU

@ 1981 Macmillan lournals Ltd

274

Nature Vol. 289 22 January 1981

can readily form chlorides through reactions (7) and (8). The
possibility of forming hydroxides of other metals (Fe, Mg and
Ca) by three-body reactions or by reactions of the monoxides
with minor constituents (such as HO,) should not be discounted.
Silicon is likely to enter the stratosphere as SiO, as the large
dissociation energy of
the monoxide, D&(Si-0)=
l90rt 2 kcal mol-' (ref. 28), precludes its reduction by 0 atoms.
Si02 might also form weakly bonded compounds with Cl, C10
and HCI; photochemical studies indicate that solid SiOP
catalyses the rate of photochemical decomposition of halocarbonsZ9.Reactions leading to the formation of MOCl may also be
possible, perhaps through

Thermochemical data are available only for AIOCI, and these

indicate that the dissociation energy of AIOCI, D&(AlO-Cl) is
greater than Dk(C1-0) or D&(H-CI) (ref. 30). Indeed, analysis
of the high-temperature reactions among the constituents of
solid-fuel rocket engines predicts large amounts of AlOH and
AlOCl at the expense of A1203and HCI (ref. 31) (note that the
engine of the space shuttle releases considerable amounts of
A1203,Fe and FeCI, (ref. 32)).
A quantitative assessment of the importance of reactions
(I)-@) will have to await the availability of their rate coefficients. However, the strongly ionic character of many of these
metals suggests that reactions of the above type will be fast. For
example, Ca and C102react to yield both CaO and CaCl and the
total reaction cross-section has been estimated to be 10-l4 cm2
at thermal energies33,which corresponds to a rate coefficient of
cm3 molecule-' s-I. Taking k7+ ks to be lo-" cm3
molecule-Is-' (which is a conservative estimate) and using
[MHCOJ + [MOH] 2 x lo6c r f 3 for M = Al + Na at 40 km
[this is equivalent to twice the calculated total sodium compound
concentration at 40 km (ref. la)], then the time constant for
reactions 7 + 8 is 5 x lo4s. On the other hand, reaction (9), the
dominant process for the removal of HCl not involving metals,
has a time constant of -2 x lo6s because [OH] lo6
and
kg = 6.6 x 10-l3 cm3 molecule-' s-' (ref. 27). This shows that
reactions such as (7) and (8) may be important.
The metal chlorides, particularly NaCl, are very stable in the
stratosphere, and would be expected to act as nucleating agents,
eventually being incorporated in aerosols. (Rocket exhaust
which contains metals such as aluminium has been found to
enhance ice n ~ c l e a t i o n ~Observations
~.)
indicate the presence of
Cl (refs 10-12, 16, 3 3 , and Na in some aerosols. At high
temperatures both condensed and gaseous Na,S04, a possible
aerosol constituent, react^^^.^^ with HCI to yield NaCl. NaCl
will, of course, polymerize readily. The condensation of NaCl
will probably be in the form of heterogeneous nucleation of
monomer, dimer or trimer NaCl with other constituents such as
water molecules or sulphur dioxide. A cluster containing 100
molecules, only 1-3 of which are NaCI, will have a concentration
of -lo6 cm-3 and its sodium would be undetectable. On the
other hand, a cluster of NaCI, (NaCl), with n = 100, would have
and would again be undetectable
a concentration of -lo4
in the atmosphere.
How much metal is deposited by meteors and how does this
compare with the total amount of halocarbons deposited in the
atmosphere? Parkin and Tilles3' estimated the total meteor
influx to the Earth to be from 3 . 6 10"g
~
yr-' to 3 . 6 ~
10" yr-I. Gadsden7 estimated the total meteor influx to be
9 x 10" g yr-' (9 x lo4 tonnes per yr), corresponding to a total
metal atom influx of 2 x lo6 metal atoms cm-2 s-I. Hughes39
quoted values in the range 3 x 1O8-5 x loi3g yr-t with a
recommended estimate of 5 x 10' g yr-'. Later, Hughes4'
reviewed the data and estimated the total amount of meteors
deposited into the Earth to be 1.6 x 10" g yr-I. Hunten et aL4'
analysed this problem from the point of view of smoke and dust
particles deposited in the Earth's mesosphere and stratosphere
by meteoroids and preferred the value given by Hughes4',

although the possibility of a higher value was noted, and suggested4' a meteoric origin for the condensation nuclei of sulphate
aerosols. The total weight of halocarbons deposited in the
atmosphere has been estimated to be -7 x 10' tonnes yr-'. If
Gadsden's estimate7 is correct, then meteor metals may be
important in the chemistry and fate of the halocarbons in the
stratosphere.
High-flying aircraft provide two additional sources of metals
in the stratosphere through the combustion of jet fuels and the
erosion of engine parts (the latter being orders of magnitude
greater than the former4'). The principal metal contaminants in
the fuels are iron, copper, calcium, vanadium and sodium, and
these may be released in amounts varying between 100 and
1,000 tonnes y r ' depending on the type of fuel and on the
frequency of flights4'. Engine erosion elements include predominantly aluminium, manganese and chromium.
An intentional or an unintentional release (such as, jet fuels
and engine erosion) in the stratosphere of -lo3 tonnes of Na
and, possibly, A1 might reduce [CI] and [CIO] by a factor of two.
It would require several years for the diffusion cycle to reestablish those concentrations if the release occurred above
30 km. Finally, although the above discussion is speculative, it
may help explain recent controversial results obtained by
Anderson et
who were the first to measure [Cl], [CIO] and
[O,] simultaneously. Their results indicate some difficulties in
current chemical models used to predict the effect of these
species on [03].
We thank Dr R. S. Narcisi for helpful comments.
Received 16 April; accepted 15 November 1980.
I. Narcisi, R. S. Space Res. 8,360-369 (1968).
2. Zbinden, P. A., Hidalgo, M. A,, Eberhardt, P. & Geiss, 1. Planet. SpaceSci. 23, 1621-1642
(1975).
3. Goldberg, R. A. & Aikin, A. C. Scrence 180,294-296 (1973).
4. Hunten, D. M. Space Sct. Rev. 6,493-573 (1967).
5. Megie, G., Bos, F., Blamont, 1. E. & Chanin. M. L. Planet. Space Sci. 26, 27-35 (1978).
6. Swider, W. Planer. Space Sci. 17, 1233-1246 (1969).
7. Gadsden, M. Annls Geophys. 26, 141-150 (1970).
8. Brown, T. L. Chem. Rev. 73, 6 4 5 4 6 7 (1973).
9. Murad, E., I . geophys. Res. 83,5525-5530 (1978).
10. Shedlovsky, J. P. & Paisley, S. Tellus 18,499-503 (1966).
1I. Delaney, A. C.. Shedlovsky. J. P. & Pollock. W. H. I . geophys. Res. 79,5646-5650 (1974).
12. Lazrus, A. L. & Gandrud, B. I . geophys. Res. 79, 3424-3431 (1971).
13. Cadle, R. D. & Grams. G. W. Rev. Geophys. Space Phys. 13,475401 (1975).
14. Charlson. R. I.. Covert, D. S., Larson, T. V. & Waggoner, A. P. Atmos. Enuir. 12.39-53
(1978).
15. Penkett, S. A., Atkins, D. H. & Unsworth, M. H. Tellus 31,295-307 (1979).
16. Parungo, F., Ackerman, E., Caldwell, W. & Weickmann, H. K. Tellus 13,521-529 (1979).
17. Ferguson, E. E. Geophys. Res. Lett. 5. 1035-1038 (1978).
18. Liu, S. C. & Reid. G. C. Geophys. Res. Len. 6, 283-286 (1979).
19. Murad, E. & Swider, W. Geophys. Res. Lett. 6,929-932 (1979).
20. Dube, A. er a / . in Mineral Science tnuesrigarions, 1974-75 (ed. Mason, B.) (Smithsonlan
Contributions to the Earth Sciences No. 19, Washington DC. 1975).
21. Wai, C. M. & Wasson, 1. T. Narure 282,790-793 (1979).
22. Ganapathy, R. & Brownlee, D. E. Science 206,1075-1077 (1979).
23. Logan, J. A,, Prather, M. 1.. Wofsy, S. C. & McElroy, M. B. Phrl. Trans. R . Soc. 290,
187-234 (1978).
24. Abbas, M. M. etal. I . geophys. Res. 84,2681-2690 (1979).
25. Keneshea, T. J., Murad, E., Swider, W. & Zimmerman, S. P. E O S 60, 898 (1979).
26. Rowland, F. S. & Molina, M. J. Rev. Geophys. Space Phys. 13, 1-35 (1975).
27. Baulch, D. L. er a / . I . phys. Chem. Ref. D a t a 9, 2 9 5 4 7 2 (1980).
28. Stull, D. R. & Prophet, H. (eds) IANAFThermochemical Tables 2nd edn NSRDS-NBS-37
(US Government Printing Office, Washington DC, 1971).
29. Ausloos, P., Rebbert. R. E. & Glasgow, L. I Res. natn. Bur, Stds, 82, 1-8 (1977).
30. Chase, Jr., M. W., Curnutt, J. L., McDonald, R. A. & Syverud, A. N. J. phys. Chem. Ref.
D a t a 7 , 3 1 1 4 8 0 (1978).
31. Park, C. Armos. Enuir 10,693-702 (1976).
32. Isaacs, R. G., Burke, H.-H. & Sze, N. D. Environmental Impact Studies for Several Space
Transporration Vehicles (Environmental Research and Technology, No. P-3139.
Concord, Massachusetts, 1977).
33. Engelke, F.,Sander. R K. & Zare. R. N. I . chem. Phys. 65, 1146-1155 (1976).
34. Parungo, F. P. & Allee. P. A. I . appl. Met. 17, 1856-1863 (1978).
35. Lazrus. A. L.. Gandrud, B. & Cadle. R. D. I . geophys. Res. 76,8083-8088 (1971).
36. Stearns, C. A,, Kohl. F. J.. Fryburg. G. C. & Miller. R. A. in High Temperatur~MeralHalide
Chemisrrv (eds Hildenbrand. D. L. & Cub~cciotti. D. D.. 555-573 (Electrochemical
Society, Princeton, 1977).
37. Hancock, P. in High Temperature Metal Halide Chemistry (eds Hildenbrand, D. L. &
Cubicciotti, D. D., 645-669 (Electrochemical Society, Princeton, 1977).
38. Parkm, D. W. & Tilles, D. Science 159,936-946 (1966).
39. Hughes, D. W. Planer. Space Sci. 20, 1949-1959 11972).
40. Hughes, D. W. Space Res. 15,531-539 (1975).
41. Hunten, D. M.,Turco, R. P. & Toon, 0. B. I . armos. Sci. 37, 1342-1357 (1980).
42 C I A P Monogr. 2, Propulsion EfPuenrs in rhe Prarorphere, DOT-TST-75-52 (NTIS.
Springfield, 1975).
43. C I A P Monogr. 3. The Stratosphere Perturbed by Propulsion EfPuents, DOT-TST-75-53
(NTIS. Springfield, 1975).
44. Anderson, J. G , Grassl, H. J., Shelter, R. E. & Margitan, 1.I . geophys. Res. 85,2869-2887
(1980).

O 198lNature Publishing Group

Nature Vol. 289 22 Januarv 1981

Humic soil and coal structure

study with
magic-angle spinning 13CCP-NMR
P. F. Barron
CSIRO Fuel Geoscience Unit, North Ryde, New South Wales 2113,
Australia

M. A. Wilson
Fuel and Atmospheric Chemistry Section, CSIRO Division of
Process Technology, North Ryde, New South Wales 21 13, Australia

In our recent report on using the I3C cross-polarization (CP)

NMR technique to characterize the organic matter in whole
soils', we discussed the limitations of the technique in terms of
resolution and the improvements possible using 'magic-angle'
sample spinning (MAS). W e now report experiments on a soil
and a brown coal in which these techniques have been
combined. The spectra provide information on the chemical
structure of the materials and throw new light o n the nature of
the various types of carbon in coals and humic materials.
Although carbonaceous substances such as coals and humic
materials are among the most widely distributed and important
natural products on the Earth, there have been few techniques
available by which their intimate chemical structure could be
obtained. However, NMR techniques have allowed semi-quantitative estimations of aromatic carbon in coals'-3 and whole
soils' to be made. Recently, high power cross-polarization
techniques coupled with magic-angle spinning4 have improved
quantification, through increased resolution, enabling better
estimates of the aromaticity of coalss and humic substances6.
High field techniques have also been used to obtain structural
information on coals7 and soilss, but research here has had to be
confined to soluble material.
We have combined the I3C CP-MAS technique with the use
of high fields to obtain spectra at 75 MHz for a New Zealand silt
loam and an Australian brown coal xylite fraction. The spectra
(Fig. 1) consist of many broad resonances, each of which gives
information on the structural environments of the carbon atoms
present.
A major problem with obtaining I3C CP-MAS spectra is the
need to spin at speeds sufficient to remove both the chemical
shift anisotropy (CSA) of highly anisotropic carbons (that is,
aromatic and carboxyl carbons) and the spinningsidebands from
the chemical shift range. As the CSA is field dependent, the
required spinning speed increases with increasing field strength
and speeds of the order of the CSA are usually considered
necessary. A t 75 MHz, experiments would have to be performed at speeds in excess of 7 kHz, which are not technologically feasible at present. However, Stejskal et aL9 have shown
that an anisotropy of -3 kHz is effectively removed at spinning
speeds of -1 kHz but at the expense of an increase in spinning
sideband intensities. The spinning speeds we have used at
75 MHz (3.2-3.6 kHz) should b e sufficient to remove large
anisotropies effectively and hence reveal the expected resolution advantage of high fields. The increased problem of spinning
sidebands is difficult to solve but, because their position depends
on the spinning rate, their contribution can be reduced by
multiplication of spectra obtained at significantly different
spinning speeds.
Figure l a shows the
CP-MAS spectrum of Maungatua silt
loam. We have previously obtained a 22.6 MHz 13C C P spectrum' which indicated that the silt loam was largely aliphatic and
highly oxygenated. The high field CP-MAS (3.2 kHz) spectrum
confirms these conclusions and provides greater detail.

275
However, as a consequence of the low carbon content of the soil
(22% C) the spectrum is complicated by resonances from the
rotor (perdeuterated polymethylmethacrylate9).
I3C resonances in the spectrum can be assigned if it is assumed
that the isotropic chemical shifts (6) in such solids are not
significantly (*3 p.p.m.) different from those found in solut i o n ~I .~ The
. ~ spectrum can be divided into four regions: S
10-50 p.p.m. (aliphatic carbon), S 50-1 10 p.p.m. (oxygenated
alkyl carbon), S 100-160 p.p.m. (aromatic and alkenic carbon)
and S 160-220 p.p.m. (carbonyl carbon).
The strong group of resonances at -6 176 p.p.m. arises from
carboxyl carbon in the soil and/or the rotor. Clearly carboxyl
carbon is only a minor contributor to this soil. The 0-alkyl
region exhibits signal over the r a n g e 4 50-110 p.p.m. with the
resolved resonance at S 7 4 p.p.m. indicating oxygen-bonded
carbons in ethers other than methoxy o r ethoxy, o r possibly ring
carbons in polysaccharides. That the 0-alkyl region extends to
greater than 100 .p.m. would be consistent with the presence of
polysaccharides'! Unfortunately, the methoxy and methyl ester
region is complicated by strong rotor resonances (8 555 1 p.p.m.) and therefore we cannot comment on the presence of
these groups.
The alkyl region contains clearly defined resonances at 6
29-32 and 15 p.p.m. The range of signals 6 29-32p.p.m.
presumably arises from methylene carbon P, y, 6 and E from
terminal methyl groups. The signal at S 1 5 p.p.m. arises from
terminal methyl carbon of alkyl chains. A small but probable
contribution from rotor signals to this resonance, means the only

Fig. 1 75 MHz "C CP-MAS spectra obtained using HI fields of

19 and 76 G for 'H and 13crespectively, with a single contact of
1 ms, on a Bruker CXP-300 spectrometer and 232 PE/MAS
probe. The rotor was machined from perdeuterated polymethylmethacrylate. a, Maungatua silt loam. Obtained without
block multiplication and using a spinning speed of 3.2 kHz.
* Indicates rotor signals and sidebands. b, Yallourn brown coal
xylite. Obtained by block multiplication of spectra using spinning
speeds of 3.2 and 3.6 kHz to reduce spinning side bands. The
resulting free induction decay was gaussian resolution enhanced.
These materials have been described in detail elsewhere'. No
special procedures were needed for sample preparation other than
crushing. Assignments were made by comparison with known
c o m p o ~ n d s ~ Chemical
~ ~ * ~ . shifts are relative to external tetramethylsilane and were initially determined relative to external
adamantane and are considered accurate to *3 p.p.m.
1981 Macmillan Journals Ltd

276
comment we can make on alkyl chain length is that polymethylene is an important constituent of this soil. Our results
clearly show, however, the potential of MAS in soil research. W e
are investigating the potential of other rotor materials for
obtaining spectra of whole soils.
Numerous attempts have been made to devise structural
formulae which represent the alkali soluble fractions of soil
organic matter. These m o d e ~ s ' ~ differ
- ' ~ considerably in structure but all involve the idea that soil humicsubstances are highly
aromatic materials.
It has recently been suggested that soil humic substances may
not be as aromatic as previously thought. Anderson and
Russell16 suggested that polymaleic-acid-like polymers are
important contributors to humic substances. Grant" has shown
that polymethylene may constitute as much as 30% by weight of
humic structures. Signals from aromatic carbons contribute only
moderately to the I3C NMR spectra of some soil humic materials in s o l u t i ~ n ' ~ Although
-~~.
it may be argued18 that nuclear
Overhauser and relaxation effects prevent quantitative determination from I3C NMR of the extract aromaticity, gated ' H
decoupling and significant pulse delays have little effectz0 on
relative signal intensities of soil humics recorded so far. O u r
work on whole soils1 has shown that the soil organic matter
present can vary from highly aliphatic to at least moderately
aromatic. The soil investigated in the present study is clearly
highly aliphatic. Hence the structure models proposed by
FuchsI3, FelbeckI4, Dragunov13 and Schnitzer" are not applicable to this soil, though they may adequately represent the
organic matter in extracts from other soils.
Figure l b is the spectrum obtained from a Yallourn brown
coal xylite. Previous examination of this material' by the 13CC P
technique (without MAS) at the relatively low frequency of
22.6 MHz showed that the xylite contained considerable
amounts of aromatic, 0-alkyl and alkyl carbon. The presence of
highly anisotropic aromatic carbon would be expected to lead to
significant sideband problems. Hence, Fig. l b is the result of the
multiplication of spectra obtained at two spinning speeds to
reduce this problem. The 75 MHz 13CCP-MAS spectrum
exhibits considerable detail and confirms the previous
conclusions.
The broad strong resonances present in the spectrum are
clearly not from spinning side bands. The resonance at 8147
arises from aromatic carbons bound to oxygen whereas that at
6112 is indicative of carbon ortho to C,,-0 o r of alkenic
carbons. The major resonance at 8123 p.p.m. is due to C,,-H
carbons.
The region 850-100 p.p.m. can be assigned to the 0-alkyl
carbon of alcohols, ethers and esters. Resonance from 87090p.p.m. can in general be assigned to carbon a to oxygen
where the alkyl group contains more than two carbons. The
spectrum exhibits signals over this whole range. Primary alcohols, ethoxy ethers and ethyl esters resonate over the range
660-65 p.p.m. and small amounts of these materials may be
present. The strong resonance at 8 5 5 p.p.m. can be assigned to
methoxy carbons. The presence of large amounts of methoxy
groups has been confirmed by pyrolysis-gas chromatography
and mass spectroscopy.
The alkyl region extending from 8 1 0 to 6 5 0 p.p.m. is not
dominated by any alkyl types. There is no large polymethylene
peak at 629p.p.m. Zilm et al.' obtained the 15.1 MHz 13C
CP-MAS spectrum of an American lignite and identified a
major peak at 6 2 9 p.p.m. which is readily assigned to polymethylene (CH2)n. There is only a very minor peak at
8 2 9 p.p.m. in the xylite spectrum. It follows that the alkyl chains
in Yallourn brown coal xylite are short and branched whereas
the alkyl chains in the American coal are much longer. Hence
our results indicate the variability of alkyl structures in brown
coal components.
Assigning resonances to carboxyl and carboxylic carbon is
complicated by the uncertainty of the contribution of spinning
side bands to the spectrum but brown coals are known to contain
carboxylic carbon. Although the peak at 165-180 p.p.m. could
0028-0836/81/040276-03$01.00

Nature Vol. 289 22 January 1981

be assigned to carboxylic carbon it may also arise from a residual
spinning side band signal. The peak at 8193 p.p.m. possibly
represents ketone carbon, although residual spinning side bands
may contribute to this region. Ketone and carboxylic carbon
were, however, also identified in pyrograms.
It is generally b e l i e ~ e d ~that
~ . ~brown
~
coals contain
considerable amounts of carboxylic acid, ether and phenolic
functional groups bound to aromatic structures of one or two
rings. O u r results show the important contribution of methoxy
ethers in brown coal xylite. The very strong resonance at
8 5 5 p.p.m. highlights the importance of methoxy ethers in this
brown coal. The unequivocal assignments made here (to those
signals which cannot arise from spinning side bands) are also
those found in ligninZ3. This strongly suggests that Yallourn
xylite is a modified lignin.
W e are not aware of any reported NMR spectra of whole soils
o r solid coal which permit the detailed structural assignments
achieved here. The present results show the potential and
problems of high field high power NMR techniques in soil and
coal chemistry.
W e thank Drs D. Muller and H. Forster, and Bruker Analytische Messtechnik GMBH, Karlsruhe, F R G for obtaining the
CP-MAS spectra. P.F.B. was supported under the National
Energy Research, Development and Demonstration program
administered by the Commonwealth Department of National
Development.
Received 7 July: accepted 3 November 1980.
1. Barron. P. F.. Wilson, M. A,, Stephens. J. F., Cornell, B. A. & Tate, K. R. Narure 286,
585-587 (1980).
2. Retcofsky, H. L. & VanderHart, D. L. Preprints 1976 Coal Chemrstry Workshop, 202-217
(Stanford Research Institute. Menlo Park, 1976)
3. Vanderhart, D. L. & Retcofsky, H. L. Fuel 55,202-204 (1976).
4. Schaefer, J. & Stejskal, E. 0. J. Am. chem. Sor. 98, 1031-1032 (1976).
5. Zilm, K. W., Pugmtre, R. J., Grant, D. M., Wood, R. E. & Wiser, W. H. Fuel 58, 11-16
(1979).
6. Hatcher, P. G., VanderHart, D. L. & Earl, W. L. Org. Geochem. 2.87-92 (1980).
7. Bartle, K. D. Rev. Pure appl Chem. 22.79-113 (1972).
8. Wilson, M. A,. Jones, A. & Williamson, B F. Narure 276,487-489 (1978).
9. Stejskal, E. 0.. Schaefer, J. & McKay, R. A. J. Magn. Reson. 25, 569-573 (1977).
10. Lippmaa, E. T., Alla, M. A., Pehk, T. J. & Engelhardt, G. I Am. chem. Soc. 100,
1929-1931 (1978).
11. Earl, W. L. & VanderHart, D. L. Mucromokcules 12, 762-767 (1978).
12. Voelter, W. & Breitmaier. E. Org. magn Reson. 5, 31 1-319 (1973).
13. Kononova, M. M. Soil Organic Matter 2nd edn, 73-75 (Pergamon, New York, 1976).
14. Felbeck, G. T. Jr Adu. Agron. 17, 327-333 (1965).
15. Schnitzer, M. & Khan, S. Y. Soil Organic Matter, 45-47 (Elsevier, New York, 1978).
16. Anderson, H. A. & Russell, J. D. Narure 260, 597 (1976).
17. Grant, D. Narure 270,709-710 (1977).
18. Wilson, M A. & Goh, K. M. I. Soil Sci. 28,645-652 (1977).
19. Ogner. G. Soil Biol. Biochem. 11, 105-108 (1979).
20. Newman. R. H., Tate. K. R., Barron, P. F. & Wilson, M. A. L Soil Sci. (in the press).
21. Wender, I. Caral. Reu. Sci. Engng 14,97-129 (1976).
22. Given, P. H. Fuel 39, 147-153 (1960).
23. Mikinis, F. P., Maciel, G. E. & Bartuska, V. J. Org. Geochem. 1, 169-176 (1979).
24. Levy, G. C. &Nelson, G. L. Carbon-13 Nuclear Magnetic Resonancesfor Organic Chemists
(Wiley-Interscience. New York. 1976).
25. Stothers, J. B. Carbon-13 N . M . R . Specrroscopy (Academic, New York, 1976).

Was the Laramide orogeny related

to subduction of an oceanic plateau?
Richard F. Livaccari, Kevin Burke & A. M. C. Sengor
Department of Geological Sciences, State University of New York at
Albany, Albany, New York 12222

Numerous models have been presented to explain the late

Cretaceouslearly Cenozoic Laramide orogeny, which affected
the foreland region of the western cordillera within the US1. The
most attractive models invoke low-angle s u b d u ~ t i o n which
~~,
can develop for various reasonsss6. Evidence from South
America indicates that one such reason may be the subduction
of buoyant ocean floor. We here extend the low-angle subduction models by attributing the shallowing of the subduction
angle during the Laramide orogeny to subduction of a large
oceanic plateau of anomalously thick and buoyant oceanic crust.
Pacific plate history indicates that a twin to the Hess Rise may
represent the oceanic plateau that triggered Laramide events.
@ 1981 Macmillan Journals Ltd

277

Nature Vol. 289 22 January 1981

Large regions of anomalously shallow ocean floor, 100,000
km2 o r more in area, are found in the Pacific, Indian and
Caribbean ocean basins. These regions were formed between
the late Jurassic and the late Cretaceous and lie at depths several
kilometres shallower than contemporaneous normal oceanic
crust7. The plateaus have been interpreted as composed of
abnormally thick and more buoyant oceanic crusts.
Regions such as the Ontong Java and Manihiki plateaus, the
Hess Rise and the eastern portion of mid-Pacific mountains
seem to have formed by areally extensive, on ridge-axis, volcanic events between 120 and 9 0 Myr Bp (Barremian to
T ~ r o n i a n ) ~ - " .Conversely, the western portion of the midPacific mountains and the present oceanic crust of the Caribbean
area seem to be a result of thickening of normal oceanic crust by
voluminous off ridge-axis volcanic activity9-". The shallowness
of the Caribbean ocean floor is related to the late Cretaceous
intrusion of large volumes of sills that thickened and made more
buoyant the oceanic crust of this areas. Plateau regions that are
generated by intense on ridge-axis volcanism may form symmetrical objects or 'twins' on two platesI2.
Regions of unusually buoyant ocean floor are difficult to
s ~ b d u c t ~ ~Therefore,
' ~ - ~ ~ . the subduction o r attempted subduction of oceanic plateaus o r linear aseismic ridges can be
expected to have a profound effect on the internal tectonics of
the associated arc8."-15.
Because oceanic plateau regions are composed of crust that
has a buoyancy between normal oceanic crust and continental
crust, the subduction of such plateaus will cause the subducted
lithospheric plate to ride higher in the asthenosphere, considerably lowering the angle of s u b d ~ c t i o n ' ~For
. example, beneath
the South American Cordillera the subducted oceanic plate has
been segmented along strike, resulting in alternating sections of
high- and low-angle subduction of the same plateI6. Segments
characterized by low-angle subduction correspond with portions
of the cordilleran belt that lack arc-related Quaternary volcanism. Above the segments of high-angle subduction, Quaternary volcanism is abundant. An extremely important observation is that the segments of low-angle subduction beneath
Peru and central Chile correspond with places where the eastern
extensions of the Nazca and Juan Fernandez aseismic ridges
have been s u b d ~ c t e d ' ~ - 'This
~ . suggests that the thickened and
more buoyant oceanic crust of both the Nazca and Juan
Fernandez aseismic ridges causes the descending oceanic plate
to ride higher in the asthenosphere, resulting in a lower angle of

FARALLON PLATE
-+

110-135 Myr

FUTURE
6LARAMTDE
FORELAND

80 Myr

LOW ANGLE
SUBDUCTION

65-40 Myr BP
.-F

FARALLON PLATE

Fig. 1 Schematic cross-sections of the western US cordillera

illustratingthe progressive approach of an oceanic plateau towards

the western US during the medial Cretaceous times and eventual
subduction of the plateau during the Laramide orogeny.

BP

Fig. 2 Cretaceous reconstructions of the Kula, Farallon, Pacific

and North American plates (modified from ref. 33): a, The formation of the Hess Rise and Hess Rise 'twin' by an unusually intense
on axis volcanic event during the early Cretaceous; b, the movement of the Hess Rise 'twin' towards the future Laramide foreland
during medial and late Cretaceous. Black arrows show the relative
motion between the North American and Farallon plates33(KP,
Kula Plate; FP, Farallon Plate; PP, Pacific Plate; MFZ, Mendicino
Fracture Zone).

s u b d ~ c t i o n ' ~ .Enhanced
'~.
seismic activity in the upper 5 0 km of
the overriding South American platez0 and extensive high- and
low-angle thrust faulting in the foreland regions of Peru and
Ecuador up to 700 km east of the South American t r e n ~ h ' ~ . ~ ~ . ~
are associated with these sections of low-angle subduction.
The Laramide orogeny was characterized by the development
of basement-cored overthrusts and adjacent synkinematic
depositional basins in the foreland of the western US between
northern Montana and northern New Mexicoz3. Deformation
began 7 0 Myr BP and continued until -40 Myr BP (refs 24,
25). T o t h e north and south of the Laramide foreland, Seviertype retroarc thrusting continued4.". The Laramide orogeny
was a non-collisional, compressional event that resulted from
the collapse and telescoping of the entire foreland region
between Montana and New Mexicoz6. Faults that bound the
ranges seem, mostly, to be low-angle thrusts that d o not steepen
at depth, indicating that horizontal compression and collapse of
the foreland region was responsible for development of
Laramide structure^^^^^^. TO the west of the Laramide foreland,
arc volcanism ceased along the Sierra Nevada2"nd migrated
inland, implying a rapid shallowing in the an le of subduction of
the Farallon plate beneath the western US28..'O.
All these events may b e interpreted as being the result of
subduction of an oceanic plateau adjacent to this region. Subduction of a plateau of approximately the same dimensions as
the Laramide foreland would result in flattening of the subduction angle, cessation of arc volcanism and increased
compressive deformation, resulting in collapse in the foreland
region (Figs 1 and 2). Therefore, tectonic patterns that characterized the Laramide orogeny such as cessation of arc volcanism
and intense foreland deformation may be explained by inferring

BP

O 198lNature Publishing Group

Nature Vol. 289 22 January 1981

the subduction of an oceanic plateau adjacent to the Laramide
foreland in the late Cretaceous/early Palaeogene, analogous
with the situation presently found in South America.
Although it cannot be directly proved that such an oceanic
plateau region ever existed or was subducted beneath the western US, the formation of numerous oceanic plateau regions in
Cretaceous times, such as the Ontong Java and Manihiki plateaus, the mid-Pacific mountains and the present oceanic crust
of the Caribbean, implies that these and perhaps other plateau
regions lay within the Pacific Ocean during the late Mesozoic.
The Hess Rise is an oceanic plateau up to 800 km across lying
just north of the westernmost extension of the Mendocino
fracture zone and just south of the Great Magnetic Bight3' (Fig.
2). The oldest sediments found on the Hess Rise are early Albian
in age''.32, implying that the Rise was formed sometime in the
early Cretaceous. Watts et al." and Vallier et a1." have presented evidence demonstrating that the Hess Rise formed by an
intense on-axis volcanic event between 120 and 90 Myr BP
(Barremian to Turonian). If this was the case, then symmetrical
plateau twins may have been generated on two plates12 (Fig. 2).
That is, a Hess Rise twin could have formed on the Farallon
plate symmetrically with the Hess Rise on the Pacific Plate.
Figure 2 shows reconstructed plate geometries for the
C r e t a c e ~ u sand
~ ~ the approximate positions of both the Hess
Rise and postulated Hess Rise twin. Symmetrical spreading
about the Pacific-Farallon ridge brings the Hess Rise twin into a
position just opposite the future Laramide foreland by the late
Cretaceous (Fig. 2). Continued convergence between the North
American and Farallon plates at the rapid rate of 14 cm yr-l
(ref. 33) would have resulted in subduction of a Hess Rise twin
opposite the Laramide foreland during the late Cretaceous
through the early Cenozoic. Thus, subduction of a Hess twin
beneath the western US may have been the cause of the distinctive Laramide orogeny.
Indirect evidence within the Cordillera for the subduction of
an oceanic plateau beneath the western US during the Laramide
orogeny may exist. Because plateau regions lie at depths 2-3 km
above oceanic crust of the same age, then during the subduction
of an oceanic plateau, uplift would be expected along the entire
fore-arc region as it rose to accommodate oceanic crust riding at
. documented
~ ~
a much higher elevation. Dickinson et ~ 1 have
such an event for the forearc region of California in the late
Cretaceous that lifted the trench slope-break to a position near
sea level resulting in widespread subaerial unconformities and
filling of the fore-arc trough to the east to form a wide fore-arc
platform area. Also, winterer3' has suggested that the Calera
and Laytonville limestones found within the Franciscan melange
may represent obducted slices of an oceanic plateau.
We thank J. F. Dewey, R. H. Pilger and W. R. Dickinson for
preprints, and R. V. Ingersoll, R. H. Pilger, T. A. Cross and W.
S. F. Kidd for reviews. Research at the State University of New
York at Albany on oceanic plateaus has been partly supported
by NSF grants 7614754 and 7803319.

Received 30 May; accepted 11 November 1980.

1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.

Woodward. L. A. N e w M e x . geol. Soc. Spec. Publ. 6, 11 (1976).

Coney, P. I. A m . J. Sci. 272,603 (1972).
Dickinson, W. R. &Snyder. W. S. Geol. soc. A m . M e m . 151.355 (1978).
Burchfiel, B. C. & Davis, G . A . A m . J. Sci. 275A, 363 (1975).
Molnar, P. & Atwater, T. Earth. planer. Sci. Lett. 41, 330 (1978).
Dewey, I. F. Geol. Ass. Can. Spec. Publ. 20, (in the press).
Sclater, J. G., Anderson, R. N. &Bell, M. L. J, geophys. Rex. 76, 7888, (1971).
Burke. K., Fox, P. 1. & Sengor, A . M. C. I. geophys. Res. 83,3949 (1978).
Winterer. E. L. A m . geophys. Un. Monogr. No. 19,269 (1976).
Vallier, T. L. er a / . Nature 286.48 (1980).
Watts, A. B., Bodine, J. H. & Ribe, N. M. Nature 283, 532 (1980).
Sharman, G. F. E O S 60,949 (1979).
Vogt, P. R., Lowrie, A., Bracey, D. R. & Hey, R. N. Geol. Soc. A m . Spec. Pap. 172 (1976).
Hughes, G. W. &Turner. C. C. Bull. geol. Soc. A m . 88,412 (1977).
Kelleher, I. & McCann, W. A m . geophys. U n . Maurice Ewing Ser. 1, 115 (1977).
Isacks. B. L. & Barazangi, M. A m . geophys. U n . Maurice Ewing Ser. 1.99 (1977).
Pilger, R. H. E O S 58, 1232 (1977).
Aleman Ramirez, A . M. & Pilger, R. H. E O S 59,383 (1978).
Pilger, R. H. BUN. geol. Soc. A m . (in the press).
Barazangi. M. & Isacks. B. L. Geology 4,686 (1976).
Burchfiel. B. C. & Davis, G. A. Nature 260,693 (1976).
Brewer. I. A.,Smithson, S. B., Oliver. J. E.,Kaufman, S. & Brown, L. D . Tectonophysics 62,
165 (1980).

23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.

Hamilton, W. Soc. Econ. Paleonr. Miner. Pacific Coasr Paleogeogr. Symp. 2, 33 (1978).
Tweto, 0. Geol. Soc. A m . M e m . 144, 1 (1975).
Armstrong. R. L. Nature 247, 348 (1974).
Coney, P. I. New Mex. geol. Soc. Spec. Publ. 6, 5 (1976).
Smithson, S. B., Brewer, J., Kaufman, S., Oliver, I. & Hurich, C. Geology 6,648 (1978).
Snyder, W. S., Dickinson, W. R. & Silberman, M, L. Earth planer. Sci. Len. 32,91 (1976).
Coney, P. J. & Reynolds, S. J. Nature 270,403 (1977).
Cross, T. A . & Pilger, R. H. Jr A m . J. Sci. 278, 865 (1978).
Chase, T. E., Menard, H. W. & Mammerickx, I. I . M . R . Tech. Rep. Ser. TR-17 (1971).
Scientific Party Init. Rep. D S D P Leg 32, 233 (1975).
Coney, P. J . Geol. Soc. A m . M e m . 152, 33 (1978).
Dickinson, W. R.. Ingersoll, R. V. & Graham, S. A. Bull. geol. Soc. A m . 90, 1458 (1979).
Winterer, E. L. Texas A & M Uniu. Symp. on Oceanic Plateaus, 15 (1979).

Old model Nd ages in

Namibian Pan-African rocks
C. J. Hawkesworth*, J. D. Kramerst
& R. McG. Miller*

* Department of Earth Sciences, The Open University,

Walton Hall, Milton Keynes MK7 6AA, UK
t Department of Earth Sciences, The University,
Leeds LS2 9JT, UK
$Geological Survey, PO Box 2168, Windhoek, 9100, Namibia
The Nd-isotope analytical technique is a powerful tool for
studying many geological processes: particularly the evolution
of the upper mantle as seen through the isotope and trace
element geochemistry of mantle-derived volcanic rocks1", and
magmatie processes along destructive plate margins4.5.
However., .
~reviousstudies of continental crustu have been
largely restricted to Archaean examples and concerned primarily with dating their time of derivation from the upper
mantle. In the present study we investigated the Nd- and
Sr-isotope characteristicsof rocks involved in a relatively young
(800-450 Myr) orogenic event. Such events represent critical
stages in the evolution of most of the Earth's continental crustnew material is added from the mantle, and pre-existing crust is
remobilized by erosion and sedimentation, deformation and
magmatic activity. Nd- and Sr-isotopes were used to outline the
range in age and geochemical characteristics of rock sequences
and provinces within both the upper mantle and the pre-existing
crust which were sampled during the orogeny, and hence provided the major components of what is now a stable segment of
continental crust.
The Damara high-temperature belt of Namibia (Fig. l a ) ,
formed during the widespread orogenic event first categorized
as Pan African by Kennedy9 and it was chosen because it is
arguably the best exposed orogenic belt of its age and has been
studied in detail both geologically and geochemically (refs 1013 and unpublished results). Some aspects of the stratigraphy
and in particular the age and regional significance of the earlier
tectonic events are still poorly understood, but an outline is
presented in Fig. 16.
Radiometric ages on pre-Damara rocks range from
1,100 Myr along the southern
to 1,700-2,000 Myr in
the centre of the belt16 and near the northern margin". Within
the Damara succession, the first major magmatic event is
recorded in a suite of large ionic lithophile (L1L)-elementenriched lavas (Naauwpoort volcanics) and associated syenites
and carbonatites exposed along the northern margin. U/Pb
zircon results are variable, but they seem to cluster around an
age of 800 Myr (ref. 18). In the north, Damara sediments reflect
a well developed shelf facies, whereas in the centre and south,
clastic sediments with only minor carbonates are followed by a
thick flysch-type succession (Kuiseb schists). Within the flysch
there is a thin band of basic rock, the Matchless amphibolite,
which has been interpreted both as a fragment of tectonically
emplaced ocean floor material19, and as just one of several basic
volcanic units within the flysch trough1'. Whichever interpretation is correct an Rb/Sr whole-rock age of 765 37 Myr

@ 1981 Macrnillan Journals Ltd

Nature Vol. 289 22 Januarv 1981

279

from this Matchless amphibolite (A. Kroner, personal communication) is probably a minimum age for most of the sediments still preserved in the southern zone.
Granitic rocks, varying from diorites to granites and highly
potassic alaskites, dominate the centre of the Damara belt (Fig.
1). Some, which intrude Damara sediments and have undergone
extensive regional deformation, yield ages of 700-750 Myr (A.
Kroner, personal communication). They are loosely termed
'early' granites and their relationship to later events is not
understood. 'Late' granites, by contrast (Fig. l b ) , all post-date
the major regional deformation and many have been studied
previously (refs 12, 20, 21 and unpublished results). Initial
Sr-isotope ratios vary from 0.705 to 0.759 (unpublished results)
and many of the younger granites, which tend to have the higher
initial "SrlR6Sr ratios were probably emplaced during late uplift
in the centre of the belt.
The Damara orogenic cycle may therefore be broadly subdivided into four phases: (1) clastic sedimentation and LILelement-enriched magmatism; (2) basic magmatism (possibly
ocean floor) and predominantly flysch-type sedimentation; (3)
early granites and regional tectonism; and (4) late, largely
post-tectonic granites with associated high-temperature
metamorphism. Our samples included volcanic and sedimentary
rocks and a range of lithologies from the late granites (Fig. 1).
143Nd/144Ndratios and the concentration of eight rare earth
elements (REEs) were determined using techniques described
previously4 and the results are presented in Tables 1 and 2.
Except for K211, all samples were selected from suites which
plot on whole-rock Rb-Sr isochrons (unpublished results). Their
initial "Sr/'%r ratios are thus representative of the particular
rock unit. Initial Nd-isotope ratios vary from 0.51196 in the
Matchless amphibolite to 0.5 1117 in the K20-rich Rossing
alaskite. Figure 2 shows chondrite-normalized R E E patterns for
selected samples.
The carbonatite (28132) from the Naauwpoort magmatic
suite contains extremely high R E E concentrations similar to

those reported from carbonatites elsewhere22.Although we are

unaware of other casbonatites with a similar 'concave-up' distribution (Fig. 2), such patterns are predicted by experimental
work on the distribution of REEs between common mantle
minerals and a C0,-rich vapour phase23.
The samples of Matchless amphibolite (AM 62,63) exhibit
higher Sm/Nd and 143Nd/'44Ndratios than the chondritic uniform reservoir (CHUR2, presumed to be representative of the
bulk Earth) at the time of their formation. This indicates that
their source region had been relatively depleted in light REEs
for a considerable time and, because such depletion is characteristic of recent mid-oceanic ridge basalts, these results are
in keeping with the interpretation that the Matchless amphibolite represents a fragment of oceanic crustIg. The Gariep Belt
is another possible ophiolite sequence, probably of similar age to
the Matchless, and whereas the dolerite sample (K211) is enriched in light REEs, it exhibits the same relatively high initial
1 4 3 N d / ' M ~ratio
d (Table 1).
Kh 24 (Kuiseb schist) is a metamorphosed flysch-like sediment which was probably deposited 700-800 Myr ago (Fig. 1).
Rb/Sr whole-rock results scatter about an 'errorchron' corresponding to an age of 548* 56 Myr, which has been interpreted as the time of widespread regional metamorphism
(unpublished results). The REEs exhibit a negative Eu anomaly
which is typical of post-Archaean sediments24 and a Sm/Nd
ratio in the range suggested for average continental crustz5
(0.17-0.21).
The Damara 'granites' sampled are from massive stocks of
diorite and granite and from small bodies of K-rich alaskite. The
diorite (RM586) and two granites (RM662 and SM2) range in
age from 560 to 490 Myr, in initial "Sr/"Sr from 0.7048 to
0.709, and in initial 143Nd/144Nd
from 0.51186 to 0.51146. The
diorite, with its initial Sr- and Nd-isotope ratios similar to
C H U R 550 Myr ago, probably originated in the upper mantle
although its generation must have involved considerable
fractionation of the light REEs (Fig. 2). The granites have higher

UPLIFT
RbfSr
Mineral
Ages

AND
COOLING

REGIONAL
TECTONICS

'L'

l>J

EARLY

li,l

PJ GRANITES

Naauwpoott volcanics

Kamoo and p o s t - K a m

Darnaran 'granitas'

Fig. 1

Damaran metased~mentsand
metavolcan!~~
Pm-Damarsn basement

a, Geological map of the Damara high-temperature belt, Namibia, illustrating sample localities. b, Summary of the major geological

events recognized within the central Damara succession.

O 198lNature Publishing Group

Nature Vol. 289

280

22 January 1981

Table 1 Nd-isotope results

a

Rock no.
AM62
AM63
K211
28/32
Kh24
RM586
SM2
RM662
26159-155.5m
DG58

Lithology
Matchless amphibolite
Matchless amphibolite
Gariep dolerite
Lofdal carbonatite
Kuiseb schist
Salem diorite
Salem granite
Sorris Sorris Granite
Vaiencia alaskite
RBssing alaskite

Age
(Myr)
765*37
765 37
-765
-800
548*56
-550
563*63
495* 15
484* 18
458 8

'43~d/144~d,
0.513124*34
0.513093* 18
0.512440*26
0.512201 *20
0.512219*20
0.512330* 30
0.512153*28
0.511795* 16
0.511735*24
0.511588*28

Sm/Nd
0.366
0.365
0.143
0.181
0.209
0.178
0.189
0.144
0.219
0.204

e
Nd
TCHUR
1 4 3 ~ d / ' 4 4 ~ d o (Ga)
0.51196
0.51194
0.51196
0.51157
0.51171
0.51186
0.51168
0.51146
0.51126
0.51117

2.5
2.4
0.3
0.8
1.O
0.6
1.0
1.2
2.3
2.3

f
87~r/86~ro

0.70566 7
0.70566 7

0.7032* 3
0.715 k 2
0.7048 6
0.707 4
0.709* 1
0.724 2
0.759* 1

a, Rb/Sr whole rock ages (unpublished results and A. Kroner, personal communication), except for the Lofdal carbonatite (U/Pb zircon18) and
Gariep dolerite assumed to be of the same age as the Matchless amphibolite).
b, Measured '43Nd/144Ndratios. Normalized to 1 4 6 ~ d / 1 4 4=~0.7219.
d
Quoted errors are two standard errors on the mean. 14'Nd/14'Nd,
BCR-1=0.51266*2.
c, From Table 2, except AM63 (R.M., unpublished results).
d, Initial 1 4 3 ~ d / 1 4 4normalized
~d
to BCR-1 = 0.51262.

where A = 6.54 x 10-l2 yr-', ' 4 3 ~ d / 1 4 4 ~ d C H=U0.51262,

R
147~m/144~d=
~H
0.1935.
u R The T Eage~
for K211
~ is meaningless as its Sm/Nd
ratio is probably much less than that in its source rock, whereas those for 26/59 and DG58 are probably slightly too old (see text).
f, Initial 8 7 ~ r / s 6ratios
~ r (unpublished results and A. Kroner, personal communication).

initial 87Sr/86Sr and lower initial 1 4 3 N d / 1 4 4 ~ratios

d
and thus
probably contain material with a crustal prehistory. Their much
larger E u anomalies (Fig. 2) show that their generation involved
either separation from a plagioclase-bearing residue, o r plagioclase fractionation. Alaskites in contrast were produced by
partial melting a t high levels in the crust, sometimes apparently
in ~ i t uSamples
~ ~ . 26/59 and D G 5 8 are from 480- and 460-Myr
old bodies with unusually high initial 87Sr/s6Sr (0.724, 0.759)
(ref. 21 and unpublished results) and low initial 1 4 3 ~ d / ' 4 4 N d
(0.51 126,O.S 1117)ratios. Note the low concentrations of Sr and
the medium- and light-REE, especially in the samples from
Rossing ( D G 5 8 and 59) and the 'concave-up' distribution
patterns (Fig. 2). T h e low Sr concentrations are consistent with a
small degree of partial melting in equilibrium with residual
plagioclase. T h e R E E distribution patterns could reflect the
presence of minor phases in the residuum, or the scavenging of
R E E with U in the formation of the adjacent U mineralization.
Rossing, for example, is the site of a major uranium deposit and
monazite, xenotime and sphene (all minerals with very high
crystal-liquid partition coefficients for medium- and lightR E E ) ' ~ are commonly found in association with the uranium
mineralization (refs 4, 28, and Marlow, personal communication).
The initial '43Nd/'44Nd ratios of these Pan African rocks are
plotted against time in Fig. 3, with the evolution of the chondritic
reservoir (CHUR2) shown for reference. The possible oceanfloor volcanics (amphibolites and dolerite) and the carbonatite
and diorite have initial Nd-isotope ratios respectively higher
than and similar to that of C H U R at the time of their formation,
whereas the late granites and alaskites have progressively lower
initial '43Nd/144Ndratios. Moreover, as there is a broad inverse
correlation between initial Sr- and Nd-isotopes (expressed as
EN^ and esr r e s p e ~ t i v e l y Fig.
~ ~ , 4), the lower initial 143Nd/144Nd
ratios in the younger rocks is matched by their progressively
higher 87Sr/86Srratios.
T:hUR agess are estimates of how long a rock (and its source
region) have had different Sm/Nd, and hence '43Nd/144Nd ratios
t o C H U R . They denote the time at which the Nd-isotope
evolution curve of a system with particular present day
'43Nd/144Nd and Sm/Nd ratios intersects that of C H U R (see
Fig. 3). T E u Rages are model ages because it is assumed that a
simple two-stage history applies. T h e first stage is in a (mantle)
province with the isotope and trace element characteristics of
C H U R , and the second stage reflects the ' 4 3 N d / 1 4 4 ~ dand

Sm/Nd ratios in the rocks a t present. If such a model is inapplicable as, for example, when light REE-enriched crustal
rocks are derived from depleted mantle (higher 143Nd/144Nd
than C H U R ) , the time of separation from the mantle may be

28/32 ,.

.....,.... carbonatite

.*...

- diorites and granites

----- alaskites

Fig. 2 Chondrite-normalized REE distribution patterns for

selected Damara rocks.

O 198lNature Publishing Group

Nature Vol. 289 22 January 1981

I.

281

with the uranium mineralization. Thus, rather than assuming

that the Sm/Nd ratios of the individual alaskites were the same
as their source rocks, we have plotted the evolution of material
0.513
A carbonrtlte
lntra
which was derived from a chondritic reservoir (CHUR)
o dioritea and granite
,/
2,000 Myr ago (that is, the oldest age reported from basement
143~d
claatic redlment
rocks in the Damara'") and had Sm/Nd ratios in the range for
'44 Nd
A alasklter
average continental crust (0.173-0.214, see ref. 25 for discussion). Its evolution path in Fig. 3 encompasses the low initial
//
Nd-isotope ratios of the alaskites which strongly suggests that
they were derived from crustal material which had been out of
the mantle since 2,000 Myr.
The initial 143Nd/'44Ndratios of the granites (SM2, RM662)
and the metasediment (Kh24) are intermediate between those of
the alaskites and C H U R at the time of their formation (Fig. 3).
Apart from equilibrium with minor phases (as presumably
happened in the case of the alaskites), equilibration with a
large amount of cumulus o r residual amphibole is the most
likely mechanism capable of altering significantly the Sm/Nd
ratios of a crustal melt3'. This would be reflected in a strong
depletion in heavy R E E s and, because this is not observed in
any of these samples, we conclude that the Sm/Nd ratios of the
granites SM2 and RM662 are likely to be similar to those in their
source rocks. Sample Kh24, being a metasediment, is also likely
to have an Sm/Nd ratio similar to that in its source terrain" The
Fig. 3 Variations in initial 1 4 % d / ' 4 4 ~ dratios in the Damara
rocks. CHUR (chondrite uniform reservoir), 1 4 7 ~ m / 1 4 4 ~ d = T&,, ages for these two granites and the metasediment range
0.1935, present day ' " ' ~ d / ' " ~ N d= 0.51262. Evolution paths; 1,
from 1,000 to 1,200 Myr indicating that, unlike the alaskites,
Sm/Nd = 0.366 (as in the samples of Matchless amphibolite); 2 and
they were not derived from 2,000-Myr old crustal material
3, Sm/Nd=0.214 and 0.173 respectively (that is the range in
alone. Two possibilities remain: the granites (and the metaaverage continental crust2'). The samples between evolution paths
sediment) could simply have been derived from younger crustal
2 and CHUR have T & , ~ ages in the range 550-1,200 Myr
sources (perhaps 1,100-Myr old) o r they could contain vari(Table 1).
able amounts of both mantle-derived and 2,000-Myr crustal
material. For the granites this might reflect contamination of
mantle-derived magmas within the crust, whereas the flyscholder than that implied by their T;!,,, age. Nonetheless, for
mantle-derived volcanics, such model Nd ages reflect how long
type sediment could contain a mixture of young volcanic and
older continental detritus. A t present there are insufficient data
chemical variations have persisted in that part of the upper
to choose between these two hypotheses, however: (1)
mantle, and for crustal-derived rocks (be they magmatic o r
significant quantities of new crustal material was generated
sedimentary) they provide a useful estimate of the age of their
1,100 Myr ago along the present-day southern margin of the
source region. In all cases they require that the Sm/Nd and
Damara'".", and crustal rocks of this age may also occur at
143Nd/144Ndratios of the rocks are similar to those in their
source areas; inevitably this assumption is more valid for some
depth elsewhere within the belt; (2) where contamination of
magmas with continental crust has been convincingly demonrock types than for others.
strated, it tends to result in a flat-lying trend between Nd- and
The basic volcanic rocks probably reflect 10-20% partial
Sr-isotopess. This is not observed for the Damara diorite and
melting of spinel peridotite and thus their Sm/Nd ratios are
granites (Fig. 4).
likely to be similar to those in their source regions3. Thus the
evolution path of material with Sm/Nd = 0.366 (that is, as in
The inverse correlation of esr and EN*observed in the crustal
AM62) has been plotted in Fig. 3 ; T&,, equals 2,500 Myr, and
rocks studied is interesting. The inverse correlation in mantleits present day 14'Nd/14"Nd is similar to those in recent M O R
derived volcanics (also shown in Fig. 4) is generally accepted as
an expression of coherent fractionation of Rb/Sr and Nd/Sm
basalts. This suggests that these rocks were derived from
ratios during melting in the mantle. However, magmatic and
mantle which had been depeleted in light R E E since the
Archaean and supports the interpretation that both the Matchsedimentary processes, particularly in the upper crust, fracless and Gariep Belts represent fragments of oceanic crustI9.
tionate Rb/Sr much more readily than Nd/Sm. Thus sediments
and K20-rich granitic material develop with time relatively high
The anomalously high E,, values of these samples (Fig. 4)
87
probably reflects contamination, either with seawater on the
S T / ' ~ Sratios,
~
resulting in a distinctive, curved inverse correlation between e~~ and esr at higher levels in the continental
ocean f l ~ o r ' ~or, during post-emplacement metamorphism.
crust (Fig. 4). Furthermore, although magmatism in this oroThe low concentrations of R E E s in the alaskites and
'concave-up' distribution patterns (Fig. 2) indicate that they
genic event has sampled material from the upper mantle through
to the upper continental crust, it does not seem to have sampled
have probably been modified-perhaps by processes associated

1won.

basic volcanka

,..'.'

Table 2 Trace elements (p.p.m.)

-

Rock no.
AM62

Rb*

Sr*

Nd

Sm

Eu

* Unpublished results.
O 198lNature Publishing Group

Gd

Nature Vol. 289 22 January 1981

6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.

ARCHAEN CONTINENTAL CRUST

-30

Fig. 4

against

esr

(ref. 2) for Damara rocks (key as for

Fig. 3)

where 1 4 3 ~ d / 1 4 4 ~ d C H
isU
theR1( r4)3 ~ d / 1 4ratio
4 ~ dof a chondritic
uniform reservoir at the time of formation of the rock in question.
.cs, is similarly defined using the relevant 8 7 ~ r / 8 ratios2.
6 ~ r EN^ and
E S , calculated using the ages and initial Nd- and Sr-isotope ratios in
Table 1, but note that the age of the sediment (Kh24)is believed to
reflect regional metamorphism (unpublished results).
any old granulite facies rocks (-E,, and -csr, Fig. 4, refs 1,7).
Such rocks may not be ubiquitous in the lower crust, but it is also
possible that such LIL-element depleted material is not readily
mobilized.
We conclude that Pan African rocks in Namibia were derived
from upper-mantle and continental sources which at the time of
the Damara orogeny had already been respectively depleted and
enriched in light rare earths for up to 2,000 Myr. High initial
143Nd/144Nd
ratios on basic volcanic rocks support suggestions
that they represent fragments of oceanic crust1', and imply that
their source regions had been depleted in light REEs since the
Archaean. T&,
ages of the granitic rocks range from 550 to
2,000 Myr. The older ages are from alaskites which, on field
evidence, were generated at higher levels in the crust. The lower
crust seems to contain younger (?1,100 Myr) crustal material.
More generally this evidence for the remobilization of 2,000Myr old crust highlights an important difference between hightemperature intracratonic Pan-African belts such as the Damara
(Fig. 1) and those in north-east Africa which appear to be less
deeply eroded, contain no evidence for significantly older
continental crust, and are believed to reflect crustal accretion
along convergent plate boundaries3=.The Damara granitic rocks
exhibit an inverse correlation between initial Nd- and Sr-isotopes (Fig. 4), although the alaskites and the metasediment are
displaced to comparatively high 87Sr/86Srratios consistent with
a relative increase in Rb/Sr at higher levels in the continental
crust.
We thank M. P. Coward, C. J. Hartnady, J. Barnes, K.
Downing and A. G. Marlow for discussions and A. R. Gledhill,
J. C. Roddick and A. Kroner for unpublished results and
comments, and also Professor Nicolaysen for his constructive
review. Analyses were carried out at Leeds where isotope and
REE research is funded by NERC and the Royal Society.
Received 9 June; accepted 20 November 1980.
1.
2.
3.
4.
5.

O'Nions, R. K.. Evensen, N. M. & Hamilton, P. I . Phil. Trans. R. Soc. (in the press).
DePaolo, D. P. & Wasserburg, G. J. Geochim. cosmochim. Acta 43,615-627 (1979).
Hawkesworth, C. J., Norry, M. J., Roddick, J. C. & Vollmer, R. Nature 280.28-31 (1979).
Hawkesworth, C. et al. Earth planet. Sci. Lett. 42.45-57 (1979).
Hawkesworth, C. J. in Origin of Granite Batholiths: Geochemical Evidence. 76-89 (Shiva,
Kent, 1979).

22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.

Jacobsen, G. B. & Wasserburg, G. J. Earth planet. Sci. Lett. 41,245-253 (1978).

Hamilton, P. J., Evensen, N. M., O'Nions, R. K. & Tarney, J. Nature 277, 25-28 (1979).
McCulloch, M. T. & Wasserburg, G. J. Science 200, 1003-1011 (1978).
Kennedy, W. 0.8th A . Rep. Res. Inst. Afr. Geol. 48 (University of Leeds, 1964).
Kroner. A. Tectonophysics 40, 101-135 (1977).
Martin, H. & Porada, H. Precamb. Res. 5, 311-338 (1977).
Blaxland, A.,Gohn, E., Haack, U. & Hoffer, E. Neues. Jb. Miner. M h . 11,498-508 (1979).
Coward, M. (in preparation).
Watters, B. R. Nature 259.471473 (1976).
Mall~ng,S. A . Rep. Precamb. Res. Unit Vol. 15, 183-193 (University of Cape Town, 1978).
Jacob, R. E., Kroner, A. & Burger, A. J. Geol. Rdsch. 67,706-718 (1978).
Burger. A. J.. Clifford, T. N. & Miller. R. McG. Precamb. Res. 3.415-431 (1976).
Burger, A. J. & Walraven, F. Annls. Geol. Sum. S. Afr. 11,323-326 (1978).
Burke, K.. Dewey, J. F. & Kidd, W. S. F. Tectonophysics 40,69-99 (1977).
Miller, R. McG. Geol. Surv. S. Afr. Mem. 64 (1973).
Kroner. A. & Hawkesworth. C. J. Reo. Res. Inst. Afr. Geoi. Vol. 20. 14-16 (Universitvof
~ e e d s 1977)
.
Mitchell, R. H. & Brunfelt, A. 0 . Contr. Miner. Petrol. 52,247-259 (1975).
Wendlandt, R. F. & Harrison, W. J. Yh Carnegie Inst. Wash. 77,695-703 (1978).
Nance, W. B. &Taylor, S. R. Geochim. cosmochim. Acta 40,1539-1551 (1976).
Shaw, D. M., Dostal. I . & Keays, R. R. Geochim. cosmochim. Acra 40, 73-83 (1976).
Sawyer, E. W. thesis Univ. Cape Town (1978).
Condie, K. C. Chem. Geol. 21,131-149 (1978).
Berning, J., Cooke, R.,Hiemstra, ?. A. & Hoffman, U. Econ. Geol. 71, 351-368 (1976).
DePaolo, D. I . & Wasserburg, G. 1. Geophys. Rex Lett. 4,465-468 (1977).
O'Nions. R. K., Carter, S. R., Cohen, R. S., Evensen. N. M. & Hamilton. P. 1. Nature 273,
435-438 (1978).
Arth, J. G. & Hanson, G. N. Geochim. cosmochim, Acra 39,325-362 (1975).
Gass, I. G. I. geoi. Soc. Lond. 134, 129-138 (1977).

A model for Plinian

eruptions of Vesuvius
M. F. Sheridan*, F. Barberi, M. Rosi & R. Santacroce
Istituto di Mineralogia e Petrografia, via S. Maria 53,
56100 Pisa, Italy

The term 'Plinian' has been widely

to describe continuous gas-blast eruptions of large magnitude a typical example5,of
which is the AD 79 eruption of Vesuvius which destroyed
Pompei and the surrounding region. We develop a new model
here for the AD 79 event that explains the complete Plinian
eruptive episode including pyroclastic fall, pyroclastic flow, base
surge, laharic and phreatic activity. This model has widespread
implications with regard to volcanic hazard evaluation and
geothermal exploration at Vesuvius and other volcanoes with
similar patterns of activity, such as Mount St Helens.
Three representative sections from archaeological excavationsdemonstrate the principal characteristics of the entire
eruptive sequence, assuming emplacement of pyroclastic flows,
surges and lahars above the pumice bed, during the 30-h eruptive period. Data from more than 30 sections included in the new
model show that factors such as distance from the vent, topography and wind velocity, control the depositional facies.
At the Pompei excavation 4.5-m of section (Fig. 2) includes a
lower white pumice-fall layer that grades into an upper grey,
pumice-fall bed overlain by interbedded surge and grey pumicefall levels and an uppermost surge sequence. There is an upward
increase in lithic xenoliths from 12 wt0/0 for the white part to 20
wt0/0 for the grey5.The character of the xenoliths also changes
with a slight increase in fraction of carbonates and a strong
increase in cognate cumulates in the grey level. The lower group
of surge beds is of the dry (superheated steam) massive and
sandwave bed type9.10. This group includes two prominent
phreatic fall levels rich in scoria and calcareous fragments. The
contact between the grey pumice and the first surge beds is a
prominent wave form. The uppermost part of the surge
sequence probably represents emplacement of very wet
(condensed steam) phreatomagmatic products. These massive
fine-ash beds are full of accretionary lapilli and may have the
widspread dispersal characteristic of phreatoplinian deposits1'.
At the Oplonti excavation the section is 8.5-m thick (Fig. 2).
The grey pumice-fall layer, which is much thicker than the white,
* Present address: Department of Geology. Arizona State University.Tempe. Arizona 85281
@ 1981 Macmillan Journals Ltd

Nature Vol. 289 22 January 1981

283

Fig. 1 Somma-Vesuvius Volcano.

a, Location map. b, Schematic crosssection. Key: 1, volcanic rocks of
Vesuvius; 2, volcanic rocks of
Somma; 3, Tertiary and Quaternary
clastic sediment; 4, Mesozoic limestone; 5, chilled margin and contact
metamorphic halo of the magma
chamber; 6, differentiated magma of
the AD 79 magma chamber that
produced the white (w) and grey (g)
pumice of the Plinian eruption.

is interrupted by three prominent surge beds. The progressive

strength of the surge blasts is indicated by the increase in cast
diameter of destroyed trees from 10 cm for the lower surge to
20 cm for the upper one. Above the grey pumice is a prominent
sandwave horizon that contains a broken tree trunk 25 cm in
diameter, and a mould of a tree 40 cm in diameter. Above the
surge sequence are two pyroclastic flows. The lower one is rich in
grey pumice and lacks stratification, the upper one has

numerous reversely graded laminations, probably indicating

that it is at its distal reaches". Above the pyroclastic flows is a
series of wet surge deposits with planar and massive beds rich in
pisolites.
The Herculaneum section is -20 m thick (Fig. 2). There is no
pumice-fall horizon: at the base are sandwave surge beds.
Above is a pumice-rich pyroclastic flow containing fragments of
houses, carbonized wood and fumarolic pipes towards the top.

Fig. 2

Representative sections of AD 79
pyroclastic deposit from archaeological
excavations. FA, air fall pumice; S, surge
deposits; FL, pyroclastic flows.

600 25 August

1200 24 August

O 198lNature Publishing Group

284

Nature Vol. 289 22 January 1981

The overlying sequence of sandwave surge deposits is followed

by another pyroclastic flow. The remainder of the section is
composed of lahars.
These deposits can be related to eruption phenomenology. A
large volume (-2 km3) of tephritic magma intruded a shallow
(3-5 km deep) magma chamber within the Mesozoic limestone
~ e r i e s ~The
- ~ . residence time of the magma was sufficient for a
reaction skarn to form and a contact metamorphic halo that
extended outwards into the carbonate country rocks to
develop7. The metamorphic halo effectively sealed the differentiating magma system, although pco2 may have been
significantly increased within the chamber as a result of decarbonation reactions in the carbonate wall rocks. A chilled magma
margin may have originally formed, but with slow cooling and
crystallization a coarser-grained syenite layer grew inwards
from the upper part of the chamber as mafic cumulates settled to
the bottom. The long period of differentiation produced a
stagnant compositionally zoned upper zone of evolved magma
rich in volatiles. The basic lower part was more homogenous,
presumably due to circulation.
The eruption began (Fig. 3 a ) either when the gas pressure
within the chamber became sufficient to rupture a conduit to the
surface or, less likely, with the introduction of a new batch of hot
magma from depth13.The initial eruption followed the model of
W i l ~ o n ' ~ -with
' ~ decompression due to exsolving magmatic gas
driving the gas thrust during the opening and widening of the
conduit. Buoyant convective uplift of particles above the
continuous gas thrust zone produced a persistent eruptive
column from which the Plinian gumice-fall deposits were
derived through lateral wind shear . Using the average wind
velocity at 9-11 km elevation (20 m s-') and terminal fall velocity data5, a column height of -17 km is computed. This corresponds well with the height of 16-26 km calculated from the
theoretical formulae of Wilson et a1.16, assuming an average
volume eruption rate of 4 x lo4m3 s-' for an 18-h period and an
F value of 0.3-1.0.
The initial eruption produced the white highly fractionated
grey pumice-fall with -10 wt% lithic clasts in the approximate
ratio of the stratigraphy traversed by the conduit. The eruption
proceeded uninterrupted into the less-fractionated grey
pumice-fall level with an increase in lithics and fraction of
carbonates relative to other lithics and the appearance of felsic
cumulates. This represents the removal of solidified magma and
country rock surrounding the upper part of the magma chamber.
The greater density, larger size and wider dispersal of the grey
pumice indicate a reater column height than for the white
pumice. Lirer et al.$ attribute this to tapping of a long (5 km)
narrow (340 m diameter) magma chamber to provide sufficient
increase in gas pressure to offset the decrease in gas content. The
present data which indicate a magma chamber of -2 km height
and 1 km diameter, cast doubt on the above mechanism. Rather,
the increased explosivity which widens the vent conduit system
above the chamber could result from vaporization of a small
mass ratio of external water to magma in the chamber during the
grey pumice eruption. Minor surge layers within the upper part
of the grey pumice deposit at Oplonti and other locations and an
increase in abundance of lithic inclusions in the grey pumice
support this interpretation.
The initial mass fraction of xenoliths in the white pumice was
small and their ratio was proportional to their relative abundance in the stratigraphic column above the chamber. As the
chamber became partially emptied during the eruption of the
grey pumice, cavitation of the roof and the walls took place, (Fig.
3b) as indicated by the sharp increase in the proportion of deep
inclusion types such as high-grade contact metamorphics, skarns
and cognate plutonic nodules. The eruption may have become
discontinuous at this stage.
Rupturing of the tight metamorphic encasement allowed
interaction with the regional hydrologic system (Fig. 3 c ) .
Because the magma level within the emptying chamber was
below the principal aquifer (Mesozoic carbonate sequence) an
abundant source of water was available to the chamber when the

venting pressure dropped below that of the hydrostatic head

(-300-500 bar). The immediate effect was to generate strong
hydromagmatic or phreatic explosions, depending on the degree
of mixing of water and magma before decompression of the
steam. The products included surges of fine-grained ashes,
widespread phreatoplinian layers, vesiculated tuffs18 and
phreatic breccias.
At this stage, the production of small pyroclastic flows may
have been directly related to contemporaneous hydromagmatic
and magmatic activity occurring in different parts of the chamber. Marginal hydromagmatic explosions would have provided
the fine fragmentation typical of pyroclastic f l o ~ s ' ~ -that
~ ' is
difficult to explain by other processes such as column collapse.
Magmatic eruptions from the centre of the chamber would have
the coarse pumice and provided a strong gas blast to
maintain the erudion column. Homogenization within the
chamber, vent anderuption column accohts for the rounding of
the pumice and uniform texture characteristic of the pumice
flows. Cooling of the system by water vaporization and enlargement of the vent due to strong explosions would have favoured
column collapse.
At first hydromagmatic explosions alternated with purely
magmatic activity; later they dominated. Because the eruptive
system was gradually cooled through water vaporization during

Fig. 3 Model of the

AD 79 eruption of Vesuvius related to

eruption phenomenology described by Pliney the ~oun~er''.a,
Gas pressure in the chamber ruptures a path to the surface producing an eruption column of about 17 km height. The eruptive rate
during this phase averages 4.0 x lo4 m3 s-'. b, A strong reduction
in activity associated with cavitation of the chamber roof produces
intermittent magmatic and hydromagmatic explosions. c, The
terminal phase is dominantly hydromagmatic with pyroclastic flows
and surges as the products. The eruption rate may reach
lo5 m3 s-', close to the maximum theoretical rate for a prolonged
maintained column20.

O 198lNature Publishing Group

Nature Vol. 289 22 January 1981

285

the hydromagmatic phase, the products proceed through early

hot dry surges and pyroclastic flows which are followed by colder
wet surges and lahars and finally by phreatic explosion breccias.
If the chamber had become completely evacuated of magma
before significant hydromagmatic cooling, the final eruption
would have been purely phreatic. These later products included
xenoliths of syenite and mafic cumulates of biotite, olivine and
pyroxene torn from the walls and floor of the chamber.
A similar pattern has been observed in four other Plinian
eruption cycles identified at VesuviusZ2:Avellino (3,500 yr BP),
Gemelle (8,000 yr BP), Verdoline (13,000 yr BP) and Pomici di
Base (17,000 yrsP). All began with a highly fractionated
pumice-fall deposit in which the percentage of lithics increases
with time. Those eruptions characterized by strong hydromagmatic activity (py;oclastic flow and surge) had a strong
concentration of carbonate lithics in the vumice-fall immediately below the surge beds. This indicates an enlargement of the

chamber roof within the regional aquifer by removal of the

surrounding country rock. The surge deposits proceeded
through an early hot dry type associated with pyroclastic flows to
a final cold wet or even phreatic type.
The late-stage behaviour of the eruption depends both on
internal magma dynamics, and on the hydrological framework.
For example, the Verdoline magma chamber was located in the
relatively impervious Tertiary flysch mark above the Mesozoic
carbonate basement. Although there was an increase in amount
of lithic clasts with time in the pumice fall deposit, the hydromagmatic phase was only minor. The other Plinian eruptions
with magma chambers entirely within the carbonate basement,
to the contrary, have produced extensive hydromagmatic and
phreatic deposits.
This work was supported by USA-Italy cooperative research
program of NSF, grant INT-7823984, and by the Italian
Geodynamics Project, publication 319.

Received 6 May; accepted 31 October 1980.

12.
13.
14.
15.
16.

1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.

Walker, G. P. L. & Croasdale, R. I. geol. Soc. Lond. 127, 17-55 (1971).

Walker. G. P. L. Geol. Rdsch 62,431-446 (1973).
Sparks. R. S. J. & Wilson, L.I. geol. Soc. Lond. 132,441451 (1976).
Nairn. I. A. &Sell, S. I. Volcan. georherm. Rcs. 3, 39-60 (1978).
Lirer, L., Percatore, T., Booth, B. & Walker, G. P. L. Bull, geol. Soc. Am. 84, 759-772
(1973).
Barberi. F. et al. IUGG. XVIIGcneral Assembly. Canberra (1979).
Barberi, F. & Lconi. L.Bull. uolcan. (in the press).
Barberi, F. el al. EEG 2nd Seminar on Gcorhcrmal Energy. Strasbourg (1980).
Sheridan,M. F. & Updike, R, G. Bull. ecol. Soc. Am. 86,571-581 (1975).
Wohletz. K. H. & Sheridan.M. F. Gcol. Soc. Am. Spcc. Pap. 180, 177-193 (1979).
Self, S. & Sparks, R. S. J. Bull, uolcan. 41, 196-212 (1978).

Middle Ordovician chondrite in

fossiliferous limestone
from Brunflo, central Sweden

17.
18.
19.
20.
21.
22.
23.

Sheridan, M. F. Geol. Soc. Am. Spec. Pap. 180, 125-136 (1979).

Sparks. R. S. 1.. Sigurdsson. H. & Wilson. L. Nature 267,315-318 (1977).
Wilson. L. Geophys. I R. asp. Soc. 30, 381-392 (1972).
Wilson, L. Geophys. I. R. asrr. Soc. 45,545-556 (1976).
Wilson, L.,Sparks, R. S. J., Huang,T. C. & Watkins, N. D.J. geophys. Res. 83,1829-1836
(1978).
Walker, G. P. L.. Wilson, L. & Boswell. E. L. G. Geophys. I. R. astr. SOC.
22, 377-383
(1971).
Lorenz. V. Sedimenrology 21,273-291 (1974).
Walker, G. P. L. J. Geol. 79.696-714 (1971).
Sheridan, M. F. I. geophys. Res. 76,5627-5634 (1971).
Sparks. R. S. J. Sedimentology 23, 147-188 (1976).
Delibrias. G.. Di Paola, G. M.. Rosi, M. & Santacroce. R. R. Soc. Ital. Miner. Petrol. 35,
41 1-438 (1979).
Radice. 8. Pliny Lsrrers and Panegyricus I (Harvard University Press. Massachusetts, 1972).

present position. However, as no similar ultramafic rocks were

known on the continental (eastern) side of the Brunflo area the
slab was set aside. The Siljan Ring, over 200 km to the south of
Brunflo was one of Sweden's best areas for Ordovician and
Silurian rocks, is an impact structure. In December 1979 the slab

Per Thorslund
Department of Palaeontology, Uppsala University, Box 558,
S-75122 Uppsala, Sweden

Frans E. Wickman
Department of Geology, University of Stockholm, Box 6801,
S-113 86 Stockholm, Sweden

The few discoveries of fossil meteorites have been attributed1to

the ignorance about the appearance of meteorites among
palaeontologists and other geologists who examine limestone
quarries, coal mines, and other excavations in sedimentary
rocks. We now report the first find of a fossil stony meteorite in
Ordovician limestone. The meteorite is a chondrite, possibly an
H-chondrite, and its terrestrial age is -463 Myr. The fine
structural details of the chondrules are often extremely well
preserved, but the chemical composition and the mineralogy
have changed dramatically, chromite being the only primary
mineral preserved. The present major minerals are calcite,
barite, a Cr-V- 'phengite' and a cobaltite-group mineral.
An earlier find reported by Yudin2 consists only of isolated
barred chrondrules in a Mesozoic bauxite. In 1952, a polished
slab of Ordovician limestone was sent to one of us (P.T.) for
examination because it contained an almost black clast about
10 cm across (Fig. 1). Thin sections were prepared of the foreign
body and were examined by a petrographer, who reported that
the clast was a metamorphic ultramafic rock consisting mainly of
clinozoisite, calcite, magnetite and serpentine. This unusual
discovery was explained by algae transporting the stone to its
0028-0836/81/040285--02501.00

Fig. 1 The polished slab of limestone with the chondrite. The size
of the slab is -65 x 65 cm, and the meteorite -10 cm across. The
dark portions contain the structurally best preserved chondrules.
The white portions are now calcite and barite. The cephalopod
(arrowed) seems to have been struck by the meteorite. The
differently coloured zones around the meteorite are visible.
@ 1981 Mscmillan Journals Ltd

286

Nature Vol. 289 22 January 1981

Table 1 Electron microprobe analyses of chondritic chromites

Crz03
A1203

vzo3

Ti02
FeO
MgO
MnO
ZnO
Total

56.8
6.5
0.74
1.73
30.9
1.98
0.50
0.93
100

2
56.9
5.9
0.68
2.33
31.2
2.66
0.94
100.61

3
56.1
5.3
0.72
2.81
33.0
1.99
0.74
100.66

4
54.4
5.7
0.73
3.23
34.5
1.62
0.63
100.81

5
55.9
6.0
0.65
2.72
30.1
2.72
1.03
1.09
100.21

6
60.8
3.9
0.70
1.47
30.5
2.32
0.79
100.48

7
57.2
6.3
0.71
2.12
30.6
2.73
0.86
100.52

8
56.7
5.9
0.68
2.27
31.3
2.51
0.99

9
57.2
5.8
0.69
2.47
30.8
3.0
0.81

100.35

100.77

Analyses 2-9 from ref. 4. Number in parenthesis gives number of meteorites analysed for 2-4, and number of grains analysed for 1, 5 and 6-9.
Analyses: 1, average present find, normalized to a sum of 100 after exclusion of 0.33 wt% CaO (11);2, average H5 and H6 (10); 3, average L5 and
L6 (7); 4, average LL5 and LL6 (6); 5, average Pultusk (4); 6, average H3 (12); 7, average H4 (21); 8, average H5 (28); 9, average H6 (11).
with its clast was re-examined with the idea that it could be a
stony meteorite; the presence of chrondrule-like boundries was
noticed.
The meteorite occurs in a reddish brown limestone from the
Rijdbrottetquarry (lat 63'7' N, long 14O17' E), near Brunflo, not
far from Ostersund, central Sweden. Determination of the
conodonts in the slab shows that the meteorite fell during the
Aseri Stage of the Middle Ordovician (-463 Myr ago).
In a -3 cm-wide border zone, delicate structural details of
chondrules are preserved, although the primary minerals have
been replaced. Two examples are shown in Fig. 2, one barred
chondrule and one radiating chondrule. The chondrules are well
defined in a matrix, giving the impression of the petrologic types
3-5 of Van Schmus and Wood3.

The present mineralogy of the chondrite is quite different

from the original one (and from that given by the petrographer).
Olivine, pyroxene, nickel-iron and troilite no longer exist. In
fact iron and magnesium have been lost almost completely from
the meteorite. Instead, the dominating mineral is calcite, the
other main constituents being barite, a Cr-V-bearing 'phengite'
and a cobaltite-group mineral with more cobalt than nickel.
Electron microprobe experiments reveal that only one primary
mineral seems to have been preserved, chromite.
Table 1 shows some chromite analyses. The present chromite
is seen to be a typical chrondrite chromite, with a homogeneous
composition. The calcium content is interpreted as belonging to
the calcite. The present chromite contains 0.93 wt0/0 ZnO; the
only chondrite found by Bunch et aL4 to contain zinc-bearing
chromite is the H5 Pultusk. It is very difficult to make a definite
classification of the find, but it seems to be most similar to the H
group, most likely the subgroups H4-H5. However, note that
analysis 1 in Table 1 could be interpreted to indicate an
unknown group of chrondrites.
Around the meteorite are concentric rings in different shades
of brown, red, grey and green. They clearly represent reactions
between the carbonate mud or, if later, limestone and the
original meteoritic matter. These zones have not as yet been
studied.
A badly preserved orthocone nautiloid cephalopod, which
occurs in contact with the meteorite, could have been hit and
killed by the meteorite.
Much of the discussion of the sources of chrondritic
meteorites focuses on the part played by Apollo and Amor
objects in their relationships to comets. Wetherill's results'
indicate that for bodies in Earth-crossing orbits one half is
eliminated in the first 10 Myr but that the rate of elimination
then slows down greatly. He has shown that in steady state
conditions and a population of -800 Apollos (15 added per
Myr) their median age will be -200 Myr and -35% of them will
be more than 500 Myr old. Even if these figures were considered
with great caution the present find is not inconsistent with
Wetherill's conclusions.
The concentration of cosmic ray exposure ages around 4 Myr
for modern H chondrites-this indicates that they come from
another source than did the clast we have studied. Of course, it
has been assumed that the composition of the chromite corresponds to those of H chrondrites and not to some unknown
group. A detailed description of this first definite fossil stony
meteorite will be given elsewhere6.
Received 29 September; accepted 2 December 1980.

Ymm

.%.!

..

..i
4 b

Fig. 2 Examples of well preserved chrondrules from the polished

surface of the slab: a, a barred chondrule; b, a radiating chondrule.

Nininger. H. H. Find a falling star (Eriksson, New York, 1972).

Yudin. 1. A. Meteoritics 6.99-103 (1971).
Van Schmus, W. R. & Wood, J. A. Geochim. cosmochim. Acta 31,747-765 (1967).
Bunch, T. E., Keil, K. & Snetsinger, K. G. Gcochim. cosmochim. Acra 31, 1569-1582
(1967).
5. Wetherill. G . W. ~ bCarnegie
.
Instn. Wash. 77.456471 (1978).
6. Thorslund, P. & Wickman. F. E. Lirhos (in preparation).

1.
2.
3.
4.

0 1981Nature Publishing Group

Nature Vol. 289 22 January 1981

Direct use of dissolved organic carbon

by agglutinated benthic foraminifera
T. E. DeLaca*$, D. M. Karlt & J. H. Lipps*
* Dcpartrnent of Geology, University of California, Davis,
California 95616
+ Department of Oceanography, University of Hawaii at Manoa,
2525 Correa Road, Honolulu, Hawaii 96822

Foraminifera are known to obtain nutrients in a variety of ways:

they are omnivores, carnivores or herbivores1-3and some species are known to use the extracellular metabolites of their
photosynthetic e n d o s y m b i o n t ~ " ~None,
~ ~ . however, has previously been proven to utilize exogenous dissolved organic
carbon directly, although this is well known in other marine
species5'". Knowledge of foraminifera1 trophic positions is
important because foraminifera are common in most marine
communities1 and may be the most abundant eukaryotic
organism in the extensive deep-sea
Our studies of
benthic foraminifera from an unusual Antarctic shallow water
embayment now show that certain species utilize both particulate and dissolved organic material in their nutrition.
Our studies were conducted at McMurdo Sound, Antarctica,
located in the southwest corner of the Ross Sea at latitude 77 OS
and longitude 166OE (Fig. 1). The Sound is approximately 5 0 km
wide; however, its eastern and western sides differ greatly in
water column productivity and benthic biomass due to local
oceanographic conditions"' (see Fig. 1). The eastern side of
McMurdo Sound is extremely productive during the austral
summer, whereas the western side at New Harbor has very low
productivity. The western side is rich in nitrate, phosphate and
silica and low in plankton and particulate organic material (ref.
11 and D.M.K., unpublished data). Based on water analyses,
species diversity and faunal affinities, New Harbor is more
similar to the deep sea than to other shallow-water areas around
Antarcticax1.The organisms we discuss here were found only at
New Harbor. Our investigation covered an area of 15-40 m
depth, with a temperature of - 1.8 "C. The sediments were rich
in organic material. Analyses of dissolved amino acids, as shown
by fluorometric analysesI2 of water collected in dialysis bags,
demonstrated maximum concentrations in the sediment
(20-30 p M at 1 cm) and smaller concentrations in the water
column (5 p M at the sediment water interface, 1 p M at 1 0 cm
from the bottom). This trend in the distribution of amino acids is
similar to that of total dissolved organic carbon in the deep sea".
Our studies focused on Notodendrodes a n t a r c t i k o ~ and
' ~ an as
yet undescribed species. Both are arborescent agglutinated
foraminifera in the superfamily Ammodiscacea (Fig. 2). These
large foraminifera are conspicuous components of the bottom
assemblages at New Harbor, with abundances of > 200 individuals per m2 (ref. 14). Our initial studies on the physiology and
trophic positions of these organisms revealed that they use
captured suspended particulate organic material and dissolved
organic materials in their nutrition. As seen by scanning electron
microscopy of specimens which were fixed in situ and later dried
using a critical-point technique, suspended particles are trapped
in sticky cytoplasmic masses which extend from the branches of
the tests. The foreign material caught in pseudopodia consists of
benthic diatoms, bacteria, amorphous flocculent organic and
inorganic material (such as silt, sand, broken diatom frustules
and spongc spicules) with adhering bacteria. This material is
probably suspended by the activities of larger invertebrates.
Marine bacteria, labelled with 'H-adenine (in the laboratory)
were presented to foraminifera in situ, and these foraminifera
were later fixed in situ. Microautoradiography revealed that the
$Present address: Marine Biology Research Division A-002. Scripps Institution of
Oceanography. La Jolla. Cahtornia 92093.

0028-083618 1 /040287-03501

.OO

labelled bacteria are captured, digested and incorporated into

the protoplasm of the organisms. However, the capture of
particulate matter, in natural conditions, seems to occur
infrequently in the population studied (3 of 32 specimens).
The foraminifera were tested for their ability to take up
dissolved organic carbon by exposing them to several
radiolabelled substrates, including amino acids (algal protein
hydrolysate, Amersham), glucose and adenine (Table 1). We
found that they took up the first two but did not absorb adenine.
As bacterial contamination in our experiments could have
produced erroneous results, we used 'H-adenine (rapidly taken
up by most bacterial3) in some of our experiments as a control in
dual-label experiments with '4C-labelled substrates. When tritium was detected in liquid scintillation analyses, those samples
were assumed to be contaminated and were eliminated. 'Time
zero' and NaCN- killed controls were used in all experiments.
Experiments measuring the uptake of dissolved organic
carbon were conducted in the laboratory. Foraminifera cleaned
in filter-sterilized sea water (0.22 pm) were placed in 10 ml of
filtered (0.22 pm) sea water (1.0 p m free amino acids in solution) containing 2.5 pCi of 14C-UL-protein hydrolysate

Fig. 1 Antarctica, showing the location of the study area. [During

past investigations in McMurdo Sound we have surveyed several
localities (+).]The two new species of foraminifera discussed in the
present report were found only at New Harbor. Arrows indicate
the probable direction of water movement. In the eastern and
central parts of the Sound currents move south, advecting large
standing crops of plankton from more northerly waters. The annual
break up of sea ice on the eastern side allows local productivity to
increase dramatically. In contrast to the highly productive eastern
side, the western side of the Sound is bathed by northward moving
currents12which probably originate under the Ross Ice Shelf and
are poor in organic nutrients.
@ 1981 Macmillan Journals Ltd

288

Nature Vol. 289 22 January 1981

(0.0165 pmol free amino acids). The final concentration of free

amino acids in solution was 2.65 JAM,and incubation times of
0.58, 1.08, 2.25, 3.25 and 4.17 h were used, all experiments
being conducted at - 1.0 OC. Following incubation, the
foraminifera were washed and: (1) solubilized in hot and cold
10% trichloroacetic acid (TCA), (2) placed in sealed reaction
flasks with 1 0 ml of filtered sea water for measurement of 14C02
evolution, o r (3) fixed for autoradiography.
Cellular contents of solubilized foraminifera were analysed
using liquid scintillation. Uptake rates of 6.14 x
pmol h-'
n = 20) were found. The tests of individual
(s.d. = 4.26 x
organisms were subsequently removed, washed three times in
hot and cold 10% T C A to remove any remaining soluble
material and re-analysed. U p to 41.3% of the label was
incorporated into non-soluble material that remained in the test.

Fig. 2 a, Notodendrodes antarctiko~'~

drawn in life position. The
branching upper half extends above the sediment-water interface

and the bulb and root system are buried in the sediment (average
length 20 mm). b, The protoplasm of this species as would be seen if
the test was removed. Cytological studies have shown that approximately one-third of the protoplasm (containing the nucleus)
resides within the bulb. About one-third of the cytoplasm extends
up the stem and occupies the branching system. Small pseudopods
extend from the branches and upper portions of the stem. Particulate organic material is entrapped in these web-like projections.
The remaining one-third of the cytoplasm resides in the extensive
root system. Autoradiographic studies have shown that although
dissolved organic material is taken up on all exposed cytoplasmic
surfaces, the highest uptake occurs in the root system where the
surface area is greatest. c, An unnamed agglutinated foraminfer
(Notodendrodidae gen. et sp. n.) drawn in life position (average
size 18 mm).

Table 1 Summary of uptake

Protein hydrolysate
Total uptake
Incorporation into TCA-insoluble form
14c02
liberated in 12 h

6.14 wmol h-'

41.3% h-'.
10%

Glucose
Total uptake '4~-glucose
14
C 0 2 liberated in 12 h
Adenine
Total uptake

Undetectable

Experiments were conducted at - 1 'C. Final concentration of free

amino acid in laboratory experiments was 2.65 pM. Concentration of
free amino acid in the sediment where the organisms are found was
found to be 20-30 bM. See text for further explanation. Concentrations
of glucose in the environment and laboratory experiments were not
adequately measured.
T h e labelled foraminifera placed in reaction vials were
incubated for 1 2 h at - 1C. The incubation was terminated with
the addition of 0.2 ml of ethanolamine (placed on filter paper in
the reaction vessel) and 1 ml H2S04 (0.5 M) was injected into
the seawater. After 2 h at room temperature (23OC), the
ethanolamine and filter paper were analysed by liquid scintillation. The results demonstrated that 10.0% of the total 14C
taken up was respired.
The results of autoradiography showed that the foraminifera
took up dissolved organic carbon through any cell surface;
consequently, uptake was a function of surface area. The 'root
system', with its large surface area, was the principal site of
uptake in the organisms analysed.
These two species of foraminifera seem to have ecological and
physiological adaptations which are well suited to a highly
seasonal oligotrophic environment. A t New Harbor, primary
productivity is very low and limited to a period of 1 o r 2 months
during the year, during which time the growth of a diatom flora
on the undersurface of the 3-4-m thick sea ice or on the
sediments of very shallow water ( < l o r n ) provides sparse
organic input. The under-ice flora rains down to the bottom,
where a portion is directly utilized while the remainder enters
the detritus food web. Decomposition of this sedimented
organic material generates relatively high concentrations of
dissolved organic carbon in the sediment and supports a large
and metabolically active microbial population (D. M. K.,
unpublished data). These foraminifera can capture and digest
this particulate material either as it settles to the bottom or as it is
resuspended by deposit-feeding invertebrates, although capture
of particles seems to occur only infrequently. Dissolved organic
carbon may contribute significantly to the nutrition of these
foraminifera. During periods of no primary productivity, -10
months of the year, they may sustain themselves on the high
dissolved organic carbon reserves which are continually
generated by the activities of heterotrophic microorganisms in
the sediment.
Our results demonstrate that foraminifera are able to use both
saprozoic and holozoic (phagotrophic) modes of nutrition. The
amount and type of organic material available to foraminifera
are therefore both much greater and more diverse than was
previously thought. Re-evaluation of current ideas of the trophic positions of these extremely abundant organisms in benthic
ecosystems is necessary, especially in deep-sea communities
where they have been found to represent the highest test-free
biomass of all eukaryotic organismsg.
Present studies are analysing the metabolic rates of these
organisms and will determine the proportion of their total
energy requirements supplied by dissolved organic carbon.
W e thank Professor A. E. V. Haschemeyer for her assistance
in experimental design and encouragement, Drs W. L. Stockton
and R. R. Hessler for reading the manuscript and Ms E. Bailey
for drawing the figures. This research was funded by NSF grant
D P P 76-17231.

O 198lNature Publishing Group

Nature Vol. 289 22 January 1981

Rece~ved14 April; accepted 24 October 1980
1. Boltovsky, E. & Wright, R. ReceniForaminifera (The Hague. 1976).
2. Lee, J . J. in Foraminifera (eds Hedley, R . H. & Adams. C. G . ) 207 (Academic, London.
1974).
3. Lee, J . J. e r a / . I. Protozoal. 13,659-670 (1966)
4 Smith. D. F. & Wiehe. W. J. Ausr. J. Firh. Fresh. Res. 28, 311-319 (1977).
5. Jorgensen. C. B. Biol. Reu. 51.291-328 (1976).
6 . Stevens, G. C, in Symp. Niirogen Merabolism and rhe Enuironmeni (eds Campbell. J W. &
Goldstein. C.) 155-184 (Academic, New York, 1972).
7. Hesslcr. R. R. in The Biology of the Oceanic Pacific led. Miller, C. B.) 79-93 (Oregon State
University Press. Corvalis, 1974).
8. Tendal, 0. & Hessler, R. R. Galathea Rep. 14, 165-194 (1977).
9 . Smith, K. I.. Jr M a r . Biol. 47. 337-347 (1978).
10. L~ttlepage.J. L. Anmrcr. Res. Ser. 14, 1-37 (1971).
l I . Dayton. P. K . & Oliver, J. S. Sctencc 197, 55-58 (1977).
12. North. B. B Limnol. Oceanogr. 20.20-27 (1975).
13. Karl. D. M. Appl. envir. Microbtol. 38. 850-860 (1979).
14. DeLaca. T. E.. Lipps. J . H . & Hessler. R . R. I. Linn. Soc. Lond. Zool. 69.205-224 (1980).

Lichen recolonization
in London's cleaner air
C. I. Rose* & D. L. Hawkswortht
* Department of Applied Biology, Chelsea College, University of
London, Hortensia Road, London SWlO OQX, UK
*Commonwealth Mycological Institute, Ferry Lane, Kew, Surrey
TW9 3AF, UK
Mean sulphur dioxide levels have fallen markedly in London,
UK during the past 15 yr. Although lichens growing on trees are
varticularlv sensitive to this air vollutant. there are few studies of
iichen recolonization following pollutio~episodes. A survey of
29 sites in the north and west of Greater London, reported here,
demonstrates that several species extinct or very rare in the area
in 1970 have extended their ranges considerably. Studies on
growth rates of the species concerned suggest that many of the
sites discovered have been recolonized within the past 3-7 yr. If
current trends continue, further improvements in the lichen
flora can be expected in the next few years. However, it is
unlikely that London will regain in the forseeable future many of
the species lost during the past two centuries.

Northwood

In the British Isles, smoke and sulphur dioxide (SO2) levels

have fallen by 80% and 50% respectively in urban areas since
1960'; in Inner London, the mean annual level of SO2declined
from 200-250 pg m-3 to < 130 kg m-3 in the same period2.
Recorded changes in mean annual and mean winter SO2 values
at six stations in north and west London are shown in Fig. 1.
Lichens have been used extensively as indicators of air pollution
and have proved to be of particular value in estimating levels of
SO2 pollution in temperate regions3-'. Most studies have been
concerned with changes in conditions of increasing levels of air
pollutants but there are a few scattered reports of improving
lichen vegetation in ameliorating conditions from Germanyg,
wede en'^." and several urban areas of the British Isles~ i r m i n ~ h a m " , ~ e e d s ' ~ - " , ~ewcastle-upon-Tyne16 and
Sheffield". Some increases in the abundance of the SO2-tolerant
Lecanora conizaeoides between 1953 and 1967 at Holland
Park, London, have been reported1' and there are a few recent
records of other more sensitive species from London, including a
single plant of Hypogymnia physodes at Brent Reservoir in 1972
(ref. 19),and Parmelia sulcata on a wall at Brook Green in 1976
by P. W. James (personal communication). No detailed survey of
the recolonization of lichens on trees in a large urban and
suburban area has, however, been previously documented in the
British Isles.
To determine what, if any, improvements of the lichen flora of
the London area had taken place, 29 sites in the north and west
of Greater London were studied between December 1979 and
February 1980. The number of trees available for study varied
but a minimum of five was examined at each site. Lichen
abundance was estimated on a 1-5 scale (Table 1) and particular
attention was paid to H. physodes, Evernia prunastri, Parmelia
caperata, Parmelia subaurifera, P. sulcata and Usnea subfloridana, as these species have different ranges of tolerance to SO2
(ref. 3) and are easily identified even when only a few millimetres
in size. Additional interesting species were encountered in some
sites, especially the Ruislip Local Nature Reserve, where 25
lichens not recorded from there in 1963 were found, including
nine which are probably recent colonists (these results are
presented in detail elsewherez0).Other notable records include
Buellia punctata and Xanthoria polycarpa on dead elms near
Northwood Hills (national grid ref. TQ 104898), the former also

Ruislip

Harrow

\
Fig. 1 Changes in mean annual
(black) and mean winter (unshaded)
SO2 values at six recording stations
(see Fig. 3) in north-west London.
Based on data compiled b r the
Warren Spring Laboratory ".

0
E

3-

1
I

Westminster

Hampstead

@ 1981 Macm~llanJournals Lld

290

Nature Vol. 289 2 2 January 1981

Table 1 Abundance of six lichen species at 29 sites in north-west London in 1979-80 compared with their status in Inner London in 1970$
Site

Ruislip LNRt
Ruislip Lidot
Copse Woodt
Park Woodt
Brent Reservoir
Hampstead Heath
(Wildwood Road)
Hampstead Heath
(Vale of Health)
Harrow Weald Common
(south)t
Mad Bess Woodt
Bayhurst Woodt
Pinner Parkt
Horsenden Hill
Harrow Hill Pond
Harrow Weald Common
(north)t
Ickenham Marsht
Hanwell Golf Course Pond
Osterley Park
River Thames at Kew
River Thames at Isleworth
River Brent at Perivale
Lyttleton Fields
Coldfall Wood
Old Park Woodt
West Brompton Cemetry
River Crane at
Twickenhamt
Holland Park
Wormwood Scrubs
Hyde Park
Regents Park
Status in London in 1970 (ref.

National grid
ref.

Hypogymnia
ph ysodes

Parmelia
sulcata

Lichen
zone*

TQ 090901
TQ 088906
TQ 084901
TQ 093905
TQ 217884
TQ 262886

3
3
3
3
1
2

5
4
2
2

6
5
5
5
5
5

TQ 278874

TQ 139934

TQ 075906
TQ 068891
TQ 119895
TQ 163854
TQ 155870
TQ 143933

3
1
1
1
1
1

1
1

4
4

1
1
2

Extinct; last
seen 1800

Rare or
extinct; last
seen 1957 at
Mill Hill

4
4
4
3
3
3
3
3
3
3
3

4
4
4

1
2

TQ 088869
TQ 142813
TQ 147795
TQ 175777
TQ 173777
TQ 170833
TQ 262891
TQ 276918
TQ 045926
TQ 257778
TQ 133733
TO 246802
TQ 214823
TQ 270800
TQ 280830
18)

Extinct; last
seen 1800

Extinct; last
seen late
1800s

Rare; a single
record on
concrete in
1968 at Wood
Green

3
3
2
2

Genus last
seen 1800

Sites are arranged according to their rating on the Hawksworth and Rose3scale. Most species occurred on Quercus and Salix, but records from Betula and Carpinus are
also included; nutrient-rich bark is not considered.
* Zone to which the richest part of the site is referred on the 0-10 scale of Hawksworth and Rose3.
t Outside the 16-km-radius circle centred on Trafalgar Square used in the study of Laundonl8.
S Abundance estimated on a 1-5 scale: 1. cover under 1% (1-5 thalli Der tree); 2, 1-5% (5-20 thalli); 3, 5-20% (20-100 thalli); 4, 20-50% (100-1,000 thalli); 5,
50-100% (over 1,000 thalli).

in Pinner Park (TQ 135906) on dead elm stumps and Rinodina

exigua on poplar in Eastcote (TQ 107889).
Comparison of the data presented in Table 1 with that of the
same species in London in 1970 (ref. 18) leaves no doubt that
there has been a marked incursion into north-west London in
the past decade. To estimate the dates at which recolonization
might have occurred, growth rates were studied by making
direct area measurements, tracing 30 thalli of each of H. physodes, P. sulcata and P. subaurifera between April 1979 and
March 1980 in the Ruislip Local Nature Reserve. These studies
(to be reported in detail elsewhere) showed that the increases in
area per unit time were significantly different (t-test between
P = 0.5 and P = 0.001) between these three species. The annual
increases in area at this site were 100-150% for H. physodes,
215-230% for P. subaurifera and 160-200% for P. sulcata, but
no correlation between starting size and percentage increase per
unit time was obtained (H. physodes, R = -0.072; P. subaurifera, R = 0.019; P. sulcata, R = -0.002; at P = 0.05 not
significantly different from zero). In general, relative growth of
circular lichen thalli expressed on a percentage basis is expected
to be constant during the exponential growth phase, but to
decline thereafter2'; the exponential phase tends to last 1015 yr and the Ruislip data therefore conform to the expected
pattern. Reports of the effect of air pollutants on the growth rate
of lichens are contradictory and the effect may vary both with the
species and pollutant levels involved. However, a study on
Parmelia saxatilis (a species closely related to P. sulcata) in
north-east England indicated that growth was not significantly
affected by substantial increases in ambient SO2 levels22.
Furthermore, in species where the growth rate is depressed, the

amount of depression involved would not materially affect the

results reported here23.Variations in the growth rate of a species
can also arise from differences in humidity or aspect, or whether
specimens are growing on vertical as opposed to horizontal
surface^^^,^^. Based on the mid-value between the mean of

.,

Time (yr)

Fig. 2 Calibration curve used to estimate the minimum age of

selected lichens from the diameter of their thalli. Based on the
observed rates of increase in area per unit time at Ruislip Local
Nature Reserve.
P. subaurifera; e, P. sulcata; 0,H. physodes.

O 198lNature Publishing Group

291

Nature Vol. 289 22 January 1981

North-west London showing the minimum age of foliose

lichen colonies on trees (expressed in yr) calculated from observed
thallus sires and growth rates (Fig. 2). The arc indicates the limit of
the survey by ~aundon'' (16-km radius from Trafalgar Square).
Scalc bar, 5 km.

Fig. 3

readings for each species and the median of the range of values,
the expected rate of increase in area per unit time was used to
construct a calibration curve relating thallus diameter to age for
these species (Fig. 2). This curve is designed to place a minimum
age on individual colonies of the species concerned (where many
specimens are present) rather than provide unequivocal dates
for each individual thallus.
The results of the growth-rate study have been incorporated
into Figs 3, 4, which show that recolonization by the foliose
lichens considered has, subject to the above qualifications, taken
place in the past 3-7 yr at most of the sites investigated, although
this process was already in operation 8-10 yr ago in the RuislipNorthwood and Harrow Weald areas. There is some collaborative data for these latter two areas: in Harrow Weald, populations of H. physodes declined between 1969 and 1971 but
were still present at the latter datez5,while in the Ruislip Local
Nature Reserve, foliose lichens were well developed in 1963
(ref. 20) (unfortunately there is no information on fluctuations in
this site between 1963 and 1979).The results also show that the
ranges of foliose lichens in the Ruislip area have extended
considerably within the last 6 yr (Fig. 4).
As there is much evidence of the sensitivity of lichens to SOz
p o l l ~ t i o n 'and
~ ~ of the declines in SO2 levels in the study area
during the period when most of the lichen recolonization seems
to have taken place, it is reasonable to suggest that the lichen
recolonization now being seen is a direct response to falling
mean SO2 levels. Changes in levels of smoke would not be
expected to cause responses in the lichen floraz6. Air pollutants
such as nitrogen oxides and ozone, which may have increased in
the London area over the past two decades, would not have any
adverse effects judging from experimental studies2'-*'. We
therefore consider that these data demonstrate that London's
lichens are responding to improvcments in the air quality in the
manner forecast by studies on deteriorating lichen floras. They
also demonstrate that a cover of L. conizaeoidrs is insignificant in prcventing the establishrncnt of foliose lichens; indeed
H. physodes and P. srdcnra can establish directly on L. conizaeoides. Further, thc lichenicolous fungus Afhelia arachnoidea,
which can kill many lichens and is especially common on
L. conizaeoides in suburban areas, d o e s not seem to oppose
significantly re-establishment, as has been suggestedm. It is-also
of interest that the sites in which recent recolonization has been
found are not all contiguous. Recolonization has been more
rapid in areas already with a nearby source of appropriate
propagules, but dispersal has evidently been possible over

considerable distances. Lichen diaspores are widespread in the

air spora but soredia can also be carried on the feet of birds; as
demonstrated at Brent R e ~ e r v o i r s ' ~a, site included in this
survey.
The most suitable sites for recolonization are ones with well lit
sheltered trees, in humid sites (for example, Ruislip Local
Nature Reserve, Harrow Hill Pond, Wildwood Road), and often
with wiIlows overhanging a pond o r marshy area surrounded by
dense shrubs. In woodland sites, the earliest colonization tends
to be at ground level, where air movements are less and the
microclimate more favourable to growth; this would be expected as foliose lichens remain longest on tree bases in conditions
of increasing SOz (ref. 3). In larger woodlands where the tree
bases are very shaded and lichen growth is precluded, lichen
communities including foliose species tend to develop on
horizontal branches higher in the canopy where they can be
easily missed. O n willow, the first foliose lichen to colonize
is H. physodes, followed by P. subaurifera and P. sulcata; P.
subaurifera then tends to become more common (frequently
growing where other species have died off) and P. sulcata
dominant. O n dry oak o r birch, H. physodes seems better able to
withstand competition from other species.
The investigations reported here concern only trees and
studies in selected sites. We have not been able to undertake a
tree-by-tree survey throughout north-west London and some
recolonization sites will almost certainly have been overlooked.
The sites listed in Table 1 are consequently representative rather
than definitive as to the present position. Although the present
survey was restricted to north-west London, the current
improvement may be more general. For example, specimens of
Hypogymnia physodes to 5 mm long were discovered on an oak
stump near Pembroke Lodge in Richmond Park by F. S. Dobson
in November 1980; thisspecimen was last recorded here in 1929
(ref. 17).
Lichens on man-made substrates, such as brick and asbestoscement, can also be used as indicators of air pollutants but
relatively little information is available on the way in which they
respond, which is complicated by variations in the chemical
nature of the substrate. However, our superficial observations
suggest that they are responding in a parallel manner to the
results presented here. In particular, Lecanora muralis, a species
which changes its substrate range as SOz levels i n c r e a ~ e *can
~,
now grow on less basic substrates in the Kew area than it did in
1970; increases in this species have even led to comments in the
local press arising from the general public's concern about this
'green fungus' on p a v e r n e n t ~ ~ ' . ~ ~ .

Northwood

\*

Copy W'

A*

Ruislip

Ruislip-Northwood Woods showing the minimum age of

foliose lichen colonies on trees (expressed in yr) calculated from
observed thallus sizes and growth rates (Fig. 2). Scale bar, 500 m.
Fig. 4

O 198lNature Publishing Group

Nature Vol. 289 2 2 January 1981

The lichen flora of London has been studied since the early
seventeenth century3%nd the historical records were combined
into the survey of ~ a u n d o n ' ~ . "These
.
records show a serious
and progressive decline starting at least by the early nineteenth
century, but which was particularly pronounced in the later
decades of that century (for example, in Epping Forestz5)and
continued into the 1950s-60s. The loss of 129 lichen species
from the area within a 16-km radius of Trafalgar Square has
been attributed to increases in air pollution between 1800 and
1970 (ref. 18). Although the slight improvements reported here
are encouraging, more drastic reductions in mean SO2levels are
required before the environment is again suitable for many of
the eliminated species. Indeed, a reduction in Inner London

from the current mean of about 130 ygm-3 to around 4050 yg m-3 would be necessary for survival of many of the more
attractive foliose and fruticose species; transplant experiments
have shown how hostile the Inner London environment is to
some of these3'. It has been estimated that mean SO2 levels in
urban areas will tend to fall until 1985, but thereafter return to
1975 levels due to changes in fuel-mix'. If this is true, only a
limited rather than a substantial improvement in the lichen flora
of the area seems likely.
We thank Mr P. W. James, Mr M. P. Jones, Mr J. R. Laundon
and Drs C. J. Muskett and F. Rose for help. This work was
carried out while C.I.R. was in receipt of a Chelsea College
(University of London) Research Studentship.

Recemed 2 September, accepted 12 November 1980

19. Bailey, R. H. & James, P. W. Lichenologisr 11, 105-106 (1979).

20. Hawksworth, D. 1 . & Rose, C. I. I. Ruislip Disrr. nar. Hisr. Soc. 22, 23-29 (1979).
21. Topham, P. B. in Lichen Ecology (ed. Seaward, M. R. D.) 31-68 (Academic, London,
1977).
22. Gilbert. 0 . 1 . Lichenologisr 5, 11-17 (1971).
23. Seaward, M. R. D. in Lichenology: Progress and Problems (eds Brown, D. H., Hawksworth,
D. 1.& Balley. R. H.) 323-357 (Academic. London, 1976).
24. Armstrong. R. A. thesis, Univ. Oxford (1974).
25. Hawksworth, D. L., Rose, F. & Coppins, B. J. in AirPollurion andLichens (eds Ferry, B. W..
Baddeley, M. S. & Hawksworth, D. 1.) 330-367 (Athlone Press of the University of
London, London, 1973).
26. Hawksworth, D. 1.in Air Pollurion and Lichens (eds Ferry, B. W.. Baddeley, M. S. &
Hawksworth, D. 1 . ) 38-76 (Athlone Press of the Unwersity of London, London, 1973).
27. Brown, D. H. & Smirnoff, N. Lichenologisr 10.91-94 (1978).
28. Nash, T. & Sigal, 1 . Blyologisr 82,280-285 (1979).
29. Nash, T. Bryologisr 79, 103-106 (1976).
30. Arvidsson. 1.Suensk bor. Tidskr. 72. 285-292 (1979).
31. Brenrford & Chiswick Times, 11 January (1980).
32. Gilbert, J. 1 . Bull. Br. Lichen Soc. 46,7 (1980).
33. Hawksworth, D. 1.& Seaward, M. R. D. Lichenology in the British Isles 1568-1975
(Richmond Publishing, Richmond, 1977).
34. Laundon. J. R. Lichenologist 3, 277-327 (1967).
35. Ferry, B. W. & Coppins, B. J. Lichenologisr 11.63-77 (1979).
36. Warren Spring Laboratory The Invesrigarion of Air Pollurion. National Survey. Smoke and
Sulphur Dioxide (Department of Trade and Industry, Stevenage, 1962-80).

I . The Ilnued Kmgdom Envlrunmenr 1979: Progrrhj of Pollurion Conrrol, Pollut. Pap. no. 16
(HMSO. London, 1979).
2. Sanford. H. Iond. Nor. 58, 89-92 (1979).
3. Hawksworth, D. 1 . & Rose, F. Narure 227. 145-148 (1970).
4. Ferry. B. W , Baddeley, M. S. & Hawksworth, D. 1 . (eds) Air Pollurion and Lichens
(Athlone Press of the University of London, London, 1973).
5. Hawksworth, D.1 . Lichenologisr 6, 122-125 (1974); 7.62-66, 173-177 (1975); 8.87-91,
179-182 (1976); 9.77-82, 147-151 (1977); 10.95-100 (1978).
6 . Hawksworth, D. 1 . & Henderson, A. L~henologisr10,227-230 (1978); Henderson, A. &
Hawksworth. D. I. Lichenologisr 11, 91-95 (1979).
7. Henderson, A. Lichenolog~sr11,313-319 (1979); 12, 145-148, 397-402 (1980).
8 . Hawksworth, D 1 . & Rose, F. Lichens as Pollurion Monirors. Stud. Biol. no. 66 (Arnold.
London. 1976).
9. Jiirging, P. Blblrhra lick., Lehre 4, 1-164 (1975).
10. Skye, E. & Hallberg, I. O~kos20, 547-552 (1969).
11. Westman, L. Wahlenbergia 2, 1-146 (1975).
12 Lindsay, D. ('. Pror. Bgham nar. Hisr. Soc. 23, 234-249 (1978).
13. Seaward, M. R. D. Proc. Leeds lit, phil. SOC..Sci. 10, 141-208 (1976).
14. Henderson, A. Naruralisr, HUN 1977, 141-144 (1977).
15. Henderson-Sellars, A. & Seaward. M. R. D. Environ. Pollur. 19, 207-213 (1979).
16 Gilbert, 0 . 1 . Lichenologisr 12, 325-395 (1980).
17. Hawksworth, D. 1 . Lichenologisr 4, 105-193 (1969).
18. Laundon, J. R. Lond. Nar. 49, 20-69 (1970).

Vacuoles as storage compartments

for nitrate in barley leaves
Enrico Martinoia, U r s H e c k & Andres Wiemken
D e p a r t m e n t of G e n e r a l Botany, Swiss Federal Institute of
Technology, Sonneggstrasse 5, CH-8092 Ziirich, Switzerland

Nitrate, the principal nitrogen source of most plants, can accumulate in large quantities in certain crop plants, notably
members of the Chenopodiaceae (spinach and beet),
Gramineae, Cruciferae (radish and kale) and Compositae
(lettuce). Concentrations may exceed 2% fresh weight (1724% dry weight) in extreme physiological conditions1. This is
alarming because nitrate is readily reduced in organisms to the
toxic nitrite, which may react with amines to form very potent
carcinogenic nitrosamines2. Large stores of nitrate can be maintained in plant cells even in the presence of high nitrate reductase activity3-'. Incoming nitrate does not seem to mix with
existing storesqds9and a steady influx of nitrate into the cells was
found to be necessary to keep nitrate reductase stably
inducedJv6.Such observations have been explained by postulating a small 'metabolic' pool of nitrate accessible to nitrate
reductase4 and presumably also responsible for the induction of
the enzyme3, and a large 'storage' pool separate from the sites of
m e t a b ~ l i s m ~ - ~Repeated
"~.
attempts to measure these pools by
an indirect method4 have given equivocal re~ults'~'~,
but storage
pools of nitrate have been generally thought to be located in
vacuoles, which are difficult to isolate and analyse. However, we
were able to isolate and purify the large central vacuoles of
barley mesophyll cells and found most of the accumulated
nitrate in the vacuoles.
Vacuoles were isolated from protoplasts as described in Fig. 1
legend and purity was confirmed by phase contrast microscopy
(Fig. lc). Contamination with intact protoplasts never exceeded
5%. Purity was further tested using marker enzymes. The
0028-0836/81/040292-03f01.00

specific activity of a-mannosidasell,'* was much higher in the

vacuolar fractions than in the homogenate of whole protoplasts
(Table 1).Activities of a-mannosidase measured per lo6 protoplasts and per 106 vacuoles were about the same (Table I),
indicating that almost all the enzyme was located in the vacuoles,
of which each protoplast has only one (Fig. lb). The vacuolar
fraction contained 13% of the total activity of a-mannosidase
present in the homogenate of protoplasts. On a basis of number
protoplasts lysed the yield of a-mannosidase (that isof vacuoles)
was 26%. The activities of two other hydrolytic enzymes, P-Nacetylgiucosaminidase and protease, were also mostly restricted
to the vacuoles (Table 2), supporting results obtained with other
p l a n t s ~ - ~ 3..However,
~s
enzymes attributed to the chloroplasts
(aldolase) and the cytosol (glucose-6-P isomerase, nitrate
reductase16) were hardly detectable in the vacuolar preparation
(Table 2).
We measured the contents of nitrate and amino acids in our
pure preparation of vacuoles. As soluble substances can be
easily lost from the vacuoles by leakage during isolation, we
were surprised to find that almost all the nitrate in the protoplasts was contained in the vacuoles, together with about half of
the soluble amino acids (Table 2). Vacuolar pools of amino acids
have been reported e l ~ e w h e r e ' ~ - ' ~ .
This demonstration of the vacuolar localization of nitrate
raises the question of the value of nitrate storage pools to the
plants. These pools could be nitrogen reserves. The rapid
accumulation of nitrate from the soil could also help to prevent
Table 1 a-Mannosidase as vacuolar marker
a-Mannosidase activity
(nmol min-')
Per (mg) protein
Per lo6 protoplasts or vacuoles

Protoplasts
6.8
8.6

Vacuoles
138'
8.3t

* T h e specific activity of a-mannosidase in the isolated vacuoles is 20.3 times

higher than that in the homogenate of protoplasts.
t Assuming that each protoplast has only one vacuole, 97% (i12%) amannosidase was found in the vacuoles. Procedures described in Table 2 legend.
@ 1981 Macmillan Journals Ltd

Nature Vol. 2 8 9 22 January 1981

293
Table 2 Subcellular localization of enzymes and substances
Activities or contents
Per lo6
Per lo6
protoplasts
vacuoles
Vacuolar marker
enzymes:
a-Mannosidase*
0-N-Acetylglucosaminida!
Proteaset
Extravacuolar marker
enzymes:
Aldolase*
Malate dehydrogenase*
Glucose-P isomerase*
Nitrate reductaset
Protein5
Amino acids))
Nitratel]

% In
vacuoles

100
89
85

<1
3
5
<1
4.7
52
99

Protoplasts were counted in a haemocytometer. The number of vacuoles was

calculated from the a-mannosidase activity (the activity of a-mannosidase per 10'
vacuoles (Table 1) was known and remained fairly constant (i8%) from one
experiment to another). Per cent in vacuoles was calculated by assuming that
100% of the a-mannosidase is located in the vacuoles. The calculation was as
follows: % activity or content in the vacuolar faction divided by % activity of
a-mannosidase in the same fraction multiplied by 100. Enzyme activities were
measured by standard methods: aldolaseZz, malate dehyd~ogenase*~,
nitrate
reductasez3, a-mannosidase and N-acetylglucosaminidase with the corresponding
nitrophenol substratesI2,and acid protease with haemoglobin as substrate in 0.1M
sodium acetate buffer, pH 3.9, as described elsewhere". Amino acids were
measured with ninhydrin and protein by a modified method of Lowry". Nitrate
concentrations were determined potentiometrically with an ion-selective liquid
membrane electrode as described e l s e ~ h e r e ~The
~ ~ ~membrane
'.
composition was
6% (by weight) methyl-tridodecyl-ammonium-chloride, 65% dibutylphthalate
and 29% polyvinylchloride (by weight). The calibration for measurements in
solution with protoplast lysate were done in solutions of the composition given in
Fig. 1 legend (step 2) and in
and lo-' M sodium nitrate. For the
determination of nitrate in vacuoles the solution of step 3 was used as background.
The detection limit for this experiment was checked using barley which had not
been irrigated with nitrate.
* nmol min-'.
t pg bovine serum albumin (BSA) equivalents per min.
t nmol NO; h-'.
5 kg BSA equivalents.
(1 nmol.

Fig. 1 a. Barley mesophyll protoplasts; b, protoplasts releasing their single

large vacuole; c, isolated vacuoles. Phase contrast micrographs x 300. Protoplasts were Prepared from the mesophyll of primary leaves of barley
seedlings (Hordeurn oulgareL. cv Nymphe) grown for 8 days on vermiculite,
irrigated with 30mM KNO, (27 "C,12 h lightldark cycle, 60% humidity).
The leaves were cut into pieces (1 mm2) which were incubated in medium
containing 2% Onozuka R-10 cellulase (supplied by Welding & Co.
Hamburg, Germany), 1% Pectolyase (Seishin Pharma, Noda. Chiba, Japan),
0.4M sucrose, 2OmM citrate-Na' buffer, pH 5.6 (2 g fresh weight of leaves
per 10 ml in a Petri dish). After 3 h a t 20 "Cthe protoplasts, which came only
from the mesophyll, were filtered through a 80 pm nylon net and purified hy
flotation (7 min at 1,400g in a swingout rotor) through a three-step dcnsity
gradient: (step 1), 10 ml suspension of protoplasts diluted with 10 ml of 10%
Ficoll400 (Pharmacia Uppsala, Sweden), 0.4M sucrose, and 'supplements'
(15mM phosphate buflcr (pH 7.6). 2mM EDTA-Na (but not for nitrate
determination), lOmM cystein-HCI (only in nitrate reductase determination)); (step 2), 10 ml 2.5% Ficoll. 0.4M sucrose and 'supplements';
(step 3). 4 ml 0.4M sorbitol and 'supplements'. The protoplasts recovered
after centrifugation from the interface above step 2 were diluted with an
equal volume of 10% Ficoll, 0.4M sucrose plus 'supplements' and Rotated a
second time through an identical density gradient. Vacuoles were liberated
from the purified protoplasts by pressing the suspension (3 x lo6 protoplasts
per ml, 4 "C) through a long needle (10 cm x 0.2 mm) using a syringe. After
this treatment about half of the protoplasts were ruptured and about two
thirds of these released their single large vacuole intact. The vacuoles were
immediately purified by flotation (2 min at 200g, followed hy 3 min at 1,000g
in a swingout rotor) through a three-step density gradient: (step 1). 5 ml
lysate in 0.4M sucrose, 2.5% Ficoll and 'supplements'; (step 2), 5 ml 0.2M
sucrose, 0.2M sorbitol and 'supplements'; (step 31, 2 ml 0.4M sorbitol and
'supplements'. Purified vacuoles were recovered from the step 3 layer. The
whole isolation procedure from the time of rupturing of the protoplasts, was
no longer than 8 min.

leakage, o r uptake by neighbouring plants. It may also play a

part in ionic and osmotic balance of the cells. This is particularly
likely in plants in which nitrate reaches high concentrationsof up
to 300 m M (ref. 1) (generally halophytes) and in which nitrate
may be accumulated as an alternative to chloride2'. The answer
to the question on the control of transport of nitrate through the
vacuolar membrane could shed light on the relationship
between the nitrate pools and the highly complex regulation of
nitrate reductase activity2'.
We thank D. Ammann and P. Anker for help with nitrate
measurements and S. Tiirler for reading, and D. Furrer for
typing the manuscript.
Received 1 September; accepted 6 November 1980
1. Hewitt. E. 1. Huckleshy. D. P.. Mann, A. F., Notton. B. A. & Rucklidge. G. J. in Nirmgen
Asrirnilalion of Plants (eds HewitI, E. J. & Cutting, C. V.) 255-287 (Academic,London.
1979).
2. Walters, C. L & Walker. R. in Nirrogen Assirnilarron of Plants (eds Hewitt, E. J. & Cutting,
C. V.) 637448 (Academic. London. 1979).
3. Heimer. J. & Filner. P. Biochim. biophyr. Acra 230, 362-372 (1971).
4. Ferrari. E. I'.. Yder. 0. C. & Filner. P. PI. Physiol. 51,423431 (1973).
5. Aslam. M., Oaks. A. & Huffaker. R. C PI. Physiol. 58.588-591 (1976).
6. Shaner, D. L. & Boyer, J. S. PI. Physiol. 58,499-504 (1976).
7. Jones, R. W. & Sheard, R. W. in Nitrogen Assimilarion ofP1ns (eds Hewitt. E. J. &Cutting,
C. V.) 521-539 (Academic. London, 1979).
8. Ashley. D. A,, Jackson, W. A. & Volk, R. J. PI. Physiol. 55. 1102-1106 (1975).
9. Martin. P. Z. Ppanzenphysiol. 70, 158-165 (1973).
10. Hageman, R. H., Reed, A. J. Femmer. R.A.,Sherrard. J. H.&Dalling,M. J. PI. Physiol.65,
27-32 (1980).
11. Wiemken. A., Schellenberg. M. & Urech. L. Archs Microbiol. 123.23-35 (1979).
12. Boller, Th. & Kende. H. PI. Physiol. 63, 1123-1132 (1979).
13. Nishimura. M. & Beevers. H. PI. Physiol. 62,4448 (1978).
14. Oaks, A. & Bidwell. R. G. S. An. Rev. PI. Physiol. 21.4346 (1970).
15. Matile. P. An. Reu. PI. Physiol. 29, 193-213 (1978).
16. Wallsgrove, R. M., Lea. P. 1. & Milfin, B. J. PI. Physiol. 63, 232-236 (1979).
17. Wiemken, A. & Diirr, M. Arch Microbiol. 101.45-57 (1974).
18. Durr. M.. Urech, K., Boller.Th.. Wiemken, A,, Schwcncke. J. & Nagy. M. Archr Micmtiol.
121,169-175 (1979).

O 1981Nature Publishing Group

294
19. Sasse, F., Backs-Hiisemann. D. & Barz, W. Z. Narurf. 34c, 848-853 (1979).
20. Stewart. G. R. in Nitrogen Assimilarion of Planrs (eds Hewitt, E. J. & Cutting, C. V.)
651-652 (Academic, London. 1979).
21. Srivastava. H. S. Phyrochem. 19,725-733 (1980).
22. Bergmeyer, H. U. Methods of Enzymatic Analysis 2nd edn (Verlag and Academic. New
York and London, 1974).
23. Hageman, R.H. & Hucklesby, D. P. Merh. Enzym. 23,491-503 (1971).
24. Hartman, K. er al. Mikrochim. Acra 11.235-246 (1978).
25. Nielsen, H. J. & Hansen, E. H. Analyrica chim. Acra 85, 1-16 (1976).

Subpopulations of cultured chick

sympathetic neurones differ in
their requirements for survival factors
David Edgar, Yves-Alain Barde & Hans Thoenen
Abteilung Neurochemie, Max-Planck-Institut fiir Psychiatrie,
D-8033 Martinsried, FRG

Peripheral autonomic and sensory neurones derived from the

neural crest will survive In oftro only if the culture medium is
supplemented with specific factors (for review see ref. 1). For
example, the nerve growth factor (NGF), although supporting
the survival of sympathetic and spinal sensory neurones", is
ineffective on parasympathetic neurones3, whereas medium
conditioned by chick heart cells (HCM) supports the survival of
all three neuronal types4. We showed previously that the
requirements of cultured spinal sensory ganglion neurones for
survival factors changed during development, and so provided a
basis for the classificationof such factors5.We now demonstrate
that cultured post-mitotic neurones from chick paravertebral
sympathetic ganglia respond differentially to NGF, HCM and
medium conditioned by C6 glioma cells (GCM). Thus, these
three survival factors are functionally distinct in that they
support the survival in culture of discrete subpopulations of
sympathetic neurones. Those subpopulations responding to
HCM and NGF are shown to differ not only in their requirements for survival factors but also in their contents of the
cholinergic and adrenergic marker enzymes choline acetyltransferase (ChAT) and tyrosine hydroxylase (TH).
Cultures were established using methods similar to those
previously de~cribed'.~,the age of the chick embryos being
checked b staging according 6 the criteria of ~ a m b ; r ~ ean;r
Hamilton:
Lumbar and sacral paravertebral sympathetic
ganglia were taken from embryos incubated between 8 and 18
days, dissociated and the cells pre-plated to remove most of the
non-neuronal cells'-6. The few remaining non-neuronal cells in
the cultures were easily distinguishable from neurones, the latter
having compact, round cell bodies with thin neurite processes a t
least five cell diameters in length. NGF, G C M and anti-NGF
antibodies were prepared as previously described, G C M being
used in the present experiments after 10-fold concentration5.
HCM was prepared by the method of collins8, except that 5%
heat-inactivated horse serum was substituted for fetal calf serum
to cultivate the heart cells, and the H C M was concentrated
two-fold by ultrafiltration (Amicon U M 10) before use.
After removal of most of the non-neuronal cells by preplating and in the absence of added factors, less than 0.1% of the
sympathetic neurones taken from 12-day old embryos survived
after 2 days in culture (Table 1). Addition of NGF, H C M o r
GCM at the start of the culture period resulted in the survival of
differing proportions of neurones, although each of the three
factors was shown to be at saturating concentration: doubling
the concentration of a given factor did not increase the number
of surviving neurones. Culture with a combination of factors,
however, resulted in an additive effect (Table I), a maximum of
90% 12-day old neurones surviving in the presence of both N G F
and HCM. Because such a high survival is possible in culture,
0028-0836/81/040294--02$01.00

Nature Vol. 289 22 January 1981

Table 1 Survival of cultured 12-day chick sympathetic neurones
% Neuronal survival in response to:

Factor
shown
Control
medium
0.09*0.10
NGF (5 ng ml-') 37.1 k4.5
HCM
45.0*3.0
GCM
6.1*2.1

Double
Factor
concentration plus NGF Factor
of factor
(5 ng ml-I) plus HCM
39.O* 5.0
41.0+6.0
8.0*2.5

88k9.5
55*4.5

71 H . 8

Mean values *s.e.m. are given (n = 3). Sympathetic neurones were

taken from paravertebral ganglia of 12-day old embryonic chicks. After
trypsinization and dissociation, the cells were pre-plated in 10-cm
plastic tissue culture dishes for 150 min as previously de~cribed'.~.The
culture medium was F14 (ref. 14), supplemented with 10% heatinactivated horse serum, penicillin (100 U ml-') and streptomycin
(100 pg mi-'). The neuronal suspension was then removed, those cells
excluding trypan blue (>90%) counted using a haemocytometer and
plated in poly-DL-ornithine-coated dishes8 at a density of 5,000-8,000
cells per 35-mm dish (equivalent to one ganglion) in 1.5 ml medium
containing the factors shown. Where NGF was not added, the media
were supplemented with 500 ng ml-' anti-NGFantibodies to inhibit any
endogenous NGF, as found, for example, in GCM'. Neurones were
counted after 48 h culture, the number being checked to remain
constant on two subsequent days.
Table 2 Enzyme levels of 12-day chick sympathetic neurones

Specific activity
(pmol per min per mg)*
s.e.m. (n = 3)
ChAT
After 48 h culture with:
NGF
102*44
HCM
842*77
NGF+ HCM
855*45
After 48 h culture with NGF,
followed by 48 h with both
NGF and HCM
Paravertebral sympathetic
chains
680*28

TH

Ratio
ChAT/TH

2813~75
137*31
310*40

0.36
6.07
2.76

266*20

2.56

Cultures were set up as described in Table 1 legend, the number of

cultured neurones being increased to 50,000-80,000. After culture for 2
days, neurones were collected and ChAT activit was measured1' using
the ChAT inhibitor naphthylvinyl pyridinel' (5OpM) to confirm
specificity of the assay. Tyrosine hydroxylase was measured by a
modification of the method of Levitt et al.17, the difference being that a
tetrahydrobiopterin cofactor (640 pM) was used1', in the presence of
20 mM ascorbate,
M ferrous ammonium sulphate and
24,000 U ml-' catalase19. The tyrosine concentration was 23 pM and
specificity of the assay was checked using the TH inhibitor a-methylM)~'. Initial reaction rates were linear with time and
tyrosine (w4
amount of sample. Protein was estimated by the method of ~ r a d f o r d ~ l
using bovine y-globulins as the standard.
and because there was no great loss of cells during the dissociation procedure which led to a complete disruption of the
ganglia, this system provides an opportunity to observe in vitro a
representative picture of the whole neurone population present
in uivo. The differing proportions of neurones surviving in
response to the different factors and the additivity of the effects
of those factors point to the presence of discrete subpopulations
of sympathetic neurones, the survival in culture of each being
selectively supported by a given factor. This conclusion supports
and extends the previous observation that not all chick sympathetic neurones respond to N G F in cultured ganglion explants9.
T o analyse further the relationships between the survival
factors and their responsive neurones, we examined the effects
of NGF, H C M and G C M on the survival of neurones taken from
sympathetic ganglia of different embryonic ages. Figure 1shows
@ 1981 Macmillan Journals Ltd

Nature Vol. 289 22 January 1981

295

that N G F supports the survival of a maximum of 3 7 % cultured

neurones taken from 12-day old embryos, whereas H C M is
maximally active on 14-day old neurones (57% survival) and the
effect of G C M increases with age throughout the time period
studied. The developmental patterns of sympathetic neuronal
survival in response to N G F and GCM are strikingly similar to
those previously obtained for dorsal root ganglion neuroness.
Thus, N G F and GCM act as functionally distinct factors for both
sensory and sympathetic neurones, and HCM may be differentiated from the other two on the basis of the age of the neurones
on which it is maximally active. We have not demonstrated
whether those neurones which are maximally dependent on
N G F for their survival at day 12 become dependent on either
HCM o r G C M later during development. Furthermore, a relationship remains to be established between the various subpopulations defined here by their factor requirements in culture,
and those subpopulations defined by morphological criteria in

situ ' O .
T o approach these problems, we characterized biochemically
the two major subpopulations of 12-day old neurones which
survived in culture in response to either HCM o r NGF. Table 2
shows that the HCM- and NGF-responsive neurones differ in
their contents of T H and ChAT, necessary for the synthesis of
the sympathetic neurotransmitters noradrenaline and acetylcholine, respectively. Thus, the cultured HCM-responsive
neurones have a C h A T j T H ratio of 6.1, some 17-fold greater
than that of those neurones surviving with NGF. Cells cultured
with both NGFand HCM have an intermediate C h A T j T H ratio
of 2.8, similar to that determined for freshly dissected 12-day
old sympathetic chains.

We d o not know whether the high ChAT levels found in the

HCM-responsive neurones is a consequence of a selection of
those cells in culture which had pre-existing high ChAT activity
in vivo o r whether this enzyme was induced in the chick
neurones by H C M in a manner similar to that shown for
neonatal rat sympathetic neurones (for review see ref. 11). Note
in this connection that HCM maintains the ChAT activity in
cultured neurones at the levels measured in the embryonic
sympathetic chains (Table 2, see also ref. 12) at a time when
preganglionic
(cholinergic)
innervation
is
largely
undevelopedt" that is, when most of the C h A T may be in
ganglionic neurones rather than innervating axons. Furthermore, H C M does not induce ChAT activity in that subpopulation of neurones previously selected by culture with NGF
(Table 2). These observations are consistent with the hypothesis
that N G F and HCM selectively support the survival of distinct
subpopulations of embryonic chick sympathetic neurones which
differ in their contents of adrenergic and cholinergic marker
enzymes.
W e thank Veronika Bohm for technical assistance.
Received 26 August; accepted 19 November 1980.

I. Varon, S S. & Bunge, R. P. A . Rev. Neurocri 1, 327-361 (1978).

2. Levl-Montalcini. R.& Angelctti. P. U. Deul Biol. 7, 653-659 (1963).
3. Helfand, S. L., S m ~ t hG
, . A. & Wessels, N. K. Devl Biol. 50, 541-547 (1976).
4. Helfand. S. L..Riopelle. R T. & Wessels, N . K. Expl Cell RPS.113. 39-45 (1978).
5. Barde, Y. A.. Edga;. D . & Thoenen. H. Proc. narn. Acad. Scr. U.S.A. 77, 1199-1203
(1980).
6. Greene, L.A . D e ~ , I B i o l58.96-105
.
(1977).
7. Hamburger. V. & Hamilton. H. L. I. Morph. 88.49-92 (1951).
8. Coll~ns.F. Devl Biol. 65, 50-57 (1978).
9. Chamley, J. H., Mark, G . E., Campbell, G. R. & Burnstock, G. Z. Zellforcch. 135.287-314
(19721
10. Edds. L. L. & van Horn. C. I. camp. Neurol. 191.65-76 (1980)
11. Patterson. P . H. A . Rev. Neurosci. 1, 1-17 (1978).
12 Marchisio, P.C. & Consolo. S. J. Neurorhem. IS, 759-764 (1968).
13. Ross. L. L.,Cosio, L., Witczak, L. & Smolen. A. Anar. Rec. 190, 525 (1978).
14. Vogel, Z..Sytkowski, A . J. & Nirenbcrg, M. W. Proc. nam. Acad. Sci. U.S.A. 69,
3180-3184 (1972).
15. Fonnum, F. I. Neurochem. 24,407-409 (1975).
16. White, H. L & Cavallito, C. J. I. Neurochem. 17, 1579-1589 (1970).
17. Levitt. M..Gibb, J. W., Daly. J. W., 1.ipfon. M. & Udenfriend, S. Biochem. Pharmac. 16.
1313-1321 (1967).
18. Black, I. B. Brain R P S95, 170-176 (1975).
19. Lerner, P., Nose, P., Ames, M. P. & Lovenberg, W. Neurochem. Res. 3,641451 (1978).
20. Udenfriend, S., Nirenberg, P. Z. & Nagatsu, T. Biochem. Pharmac. 14, 837-845 (1965).
21. Bradford, M. M. Analyr. Biochem. 72,248-254 (19761.

The number of polarizing region cells

required to specify additional
digits in the developing chick wing
C . Tickle
Department of Biology as Applied to Medicine, The Middlesex
Hospital Medical School, Cleveland Street, London W1P 6DB, UK

Embryonic age (days)

Fig. 1 Neurones were dissociated from paravertebral sympa-

thetic ganglia taken from chicks of embryonic ages shown. After

pre-plating, 5,000-8,000 neurones were plated on p o l y - ~ ~ - o r n i thine as described in Table 1 legend. Neuronal survival was estimated between days 2 and 5 after the start of culture, and it was
checked that the percentage neuronal survival did not decrease
significantly on two subsequent days. The dashed line shows the
low basal levels of neuronal survival without added factors. An
increased neuronal survival seen in control cultures of neurones
from 8- and 18-day old chicks is a consequence of the slower death
rate of the younger cells, and of the presence of more non-neuronal
cells which support neurones in the cultures from older embryos.
Parallel cultures were set up with supplements of: 5 ng ml NGF
(0),HCM plus 500 ngml-' NGF antibodies (e)and GCM plus
500 ng ml- NGF antibodies (A). Results are expressed as
percentages of surviving neurones + s.e.m. ( n = 3).

'

0028-0836/X 1 /04029544$01 .OO

A group of mesenchyme cells at the posterior margin of the

developing wing-the polarizing region--can have a dramatic
effect on the pattern of structures which develop across the
antero-posterior axis of the limb. If one grafts a polarizing
region to the anterior margin of a wing bud so that this bud now
has one polarizing region at the anterior margin and one at the
posterior margin, the wing that develops has duplicated structures across the antero-posterior axis in mirror-image symmetry'-'. That is, the normal pattern of digits 2 3 4 (reading
from anterior to posterior) becomes 4 3 2 2 3 4. The ability of
the polarizing region to specify additional digits from adjacent
tissue can be progressively attenuated by y-ray radiation4.If this
attenuation is caused by progressively fewer cells remaining
viable to signal, there should be a quantitative relationship
between the number of polarizing cells used and the digit
specified next to the graft. Here, two tests are reported which
confirm this idea. The results suggest that the apical ridge
cooperates with a small number of polarizing region cells in a
monolayer to specify structures across the antero-posterior axis
of the wing.
@ 1981 Macmillan Journals Ltd

296
,

Nature Vol. 289 22 January 1981

"

fl

Cell pellet

,
Cell pellet held
In place wlth
wlre pm

mesenchyme
plus ectoderm

<

Cut along base of

I ridge

Table 1 Effect of diluting polarizing cells with anterior margin cells on

the digit formed next to the graft
O/OOf polarizing
cells in pellet

No. of cases

Extra digit adjacent

to graft
4
3
2 Knob None

a, Standard grafting procedure shown in Fig. l a

,

Cell pellet

,
Cell pellet in
place beneath

stretched away

Fig. 1 Diagram to show how grafts of pieces of cell pellet were

made to host wing buds (stage 20-21). The grafts were cut from a
centrifuged pellet of cells which had been incubated for 1 h at
37 "C. The grafts were -150-200 p,m per side and contained about
6,000 cells (estimated from serial sections: average for 4 pieces =
5,979, s.d. 464). The cell suspensions were prepared from polarizing regions and equivalent-sized pieces of anterior mesenchyme
tissue dissected from the wing buds of chick embryos (stage 22:
Hamilton-Hamburger) following removal of the ectoderm with
2% trypsin (Difco). Various mixtures of polarizing region cells and
inactive cells from anterior margins were made. For each suspension a total of 20 pieces of tissue was used. The standard grafting
procedure, shown in a, involved removing a piece of mesenchyme
plus overlying apical ridge from the anterior margin of the host
limb bud opposite somite 16 or 17 and placing the piece of pellet in
the hole. The graft was held in place by a pin of platinum wire. In
the other grafting procedure, shown in b, a cut was made along the
base of the apical ectodermal ridge over the anterior part of the
host bud with a sharpened tungsten needle. The apical ectodermal
ridge was raised up from the underlying mesenchyme and the piece
of pellet was placed between. No mesenchyme tissue was removed.
The apical ridge was stretched away from the mesenchyme and
once the graft was in place it contracted back to hold the pellet in
position.

100
100 (50150 wing/leg)
Total

5
9
14

3
4
7

2
2
4

0
3
3

0
0
0

0
0
0

55 (quail)
50
50 (quail)
Total

4
11
6
21

0
0
0
0

4
4
1
9

0
5
3
8

0
0
2
2

0
2
0
2

33
25
20
10
5
0

1
13
4
6
7
9

0
0
0
0
0
0

1
1
0
1
0
0

0
7
2
4
2
0

0
2
0
0
2
0

0
3
2
1
3
9

b, Grafts made to a slit beneath an intact apical ridge as shown in Fig. l b

100
50
20
9-1 0
0

11
7
3
13
4

1
2
0
0
0

0
4
3
7
0

0
1
0
6
0

0
0
0
0
0

0
0
0
0
4

In the first set of experiments, cells from the polarizing region

were mixed with inactive cells from the region along the anterior
margin of the wing bud. By progressivel~dilutingldisassociated
polarizing region cells with these inactive cells and grafting the
mixture after re-aggregation to an anterior position in another
bud, a clear relationship between the number of polarizing
region cells and the digit formed next to the graft can be shown.
The results of grafts carried out by the standard procedure
(Fig. l a ) , but with varying proportions of re-aggregated
polarizing region cells and inactive anterior cells, are listed in
Table l a . As the proportion of polarizing region cells in the
grafts decreased, the digit formed next to the graft changed
progressively: first a digit 3 (Fig. 2 c ) instead of a digit 4 (Fig. 2a)
was produced, then a digit 2 (Fig. 2 d ) instead of a digit 3, and
finally no additional digits at all (Fig. 2 b ) were specified. A

Fig. 2 Dorsal views of whole mounts of wings

obtained following grafts of re-aggregated cells
made as shown in Fig. l a . a, 100% polarizing
region cells. Pattern of dieits is 4 3 2 2 3 4. b.
1 6 % anterior margin cells. Pattern of digits is
2 3 4 . c, 50% polarizing region cells and 50%
anterior margin cells. Pattern of digits is
3 2 2 3 4 . d, 5% polarizingregion cells and 95%
anterior margin cells. Pattern of digits is 2 2 3 4 .

O 198lNature Publishing Group

297

Nature Vol. 289 22 January 1981

Table 2 Comparison of independent estimates of the number of
polarizing region cells, placed under an intact apical ridge, which are
required to specify each additional wing digit

a
b

Digit 4

Digit 3

Digit 2

100(1)
164

79(10)
64-32

35(12)
32 or less

a, Mean numbers of polarizing cells on pieces of plastic film grafted

under the apical ridge with the cells facing the ridge; b, number of
polarizing cells on a face of a piece of pellet of re-aggregated polarizing
and anterior mesenchyme cells grafted under an intact apical ridge. In h,
data were calculated from the average number of cells on a face of a
pellet and '10 polarizing cells in mixed pellet required to specify each
digit. Values in parentheses indicate number of cases.

similar effect has also been obtained when polarizing region cells
are diluted with BHK cells (unpublished observations). Even
when only 5% of the cells were from the polarizing region, an
additional digit 2 was obtained in one-third of the cases (Fig.
2d). Because the same order of loss of duplicate digits is found
when the polarizing region is treated with increasing doses of
y-ray radiation4, it seems likely that the effect of the irradiation
is to diminish the number of viable cells in the graft.
To investigate whether close contact with the apical ridge
leads to more efficient signalling, as suggested by the recent
work of saunders5, another series of grafts, in which the polarizing region cells were progressively diluted with anterior mesenchyme cells, was carried out (Table I b). In this series, however,

0
2
3
4
Additional digit specifled next to graft

Fig. 4 Data on the digit which formed next to the graft after
implantation of pieces of plastic film carrying varying numbers of
polarizing cells. The plastic film was grafted to a slit beneath the
apical ridge with the cells facing the ridge.

Fig. 3 a, Piece of plastic film (Melinex, ICI) after 3 h incubation

on Parafilm (American Can Company) with adropof suspension of
polarizing cells prepared from at least eight polarizing regions. This
piece of plastic (scale bar is 100 r*m)has been fixed and stained with
Leishman's to show attached cells, some of which have spread. To
count the numbers of cells on such pieces of plastic (100-250 pm
on a side) before grafting, the pieces were transferred with a spatula
to a dish and viewed in an inverted phase-contrast microscope. The
piece of plastic with a known number of polarizing cells was then
carried into position over the host wing bud on a spatula and
manoeuvred into the grafting site. h, Wing which developed after
grafting of a piece of plastic film carrying 28 polarizing cells. The
graft was made to a slit beneath the apical ridge and the film was
oriented so that the cells faced the ridge. Digit pattern is 2 2 3 4.
Note that the piece of plastic has ended up at the elbow of the
developed wing.

the apical ridge was left intact (Fig. lb). It is clear that these
grafts placed under an intact apical ridge are much more
effective. For example, if the effects of grafts containing 50%
polarizing region cells are compared, digit 4 was still specified in
some cases when grafts were placed under the ridge, whereas an
extra digit 4 has never been obtained with a standard graft.
From the total numbers of cells in pieces of pellet the following estimates can be made: with the standard grafting procedure,
at least 6,000 polarizing region cells (100% polarizing region
cells in the graft) are required to obtain digit 4 next to the graft;
to obtain digit 3, 3,000-6,000 cells are required, and for digit 2
between 3,000 and as few as 300 are needed. For grafts made
under an intact apical ridge, a digit 4 requires about 3,000
polarizing region cells, but only about 600 cells are required for
a digit 3 and even fewer for a digit 2.
In the second set of experiments, small numbers of polarizing
cells stuck to pieces of plastic film (Fig. 3 a ) were implanted just
beneath the apical ridge. Figure 4 shows that remarkably small
numbers of polarizing cells can now specify additional digits in
the chick wing (Fig. 3h). Grafts of pieces of plastic with small
numbers of anterior mesenchyme cells (five cases) always resulted in normal wings. Figure 4 also shows a clear relationship
between the number of polarizing cells grafted on the piece of
film and the'digit which developed next to the graft. The
estimate of the mean number of cells required to specify each
wing digit is shown in Table 2. Thus, to specify digit 4 at least 100
polarizing cells must be grafted beneath the apical ridge,
whereas specification of digit 3 requires 79 cells and digit 2, 35
cells. In one case, the transfer of only nine polarizing cells led to
a wing with an additional digit 2.
The two independent estimates of the number of polarizing
cells in contact with the ridge that are required to specify each
additional wing digit are apparently very different. However, if
the number of polarizing cells on only the face of each mixed
pellet, used for grafts under the apical ridge, are considered, the
number of cells required for signalling agrees well with the

O 198lNature Publishing Group

Nature Vol. 289 2 2 January 1981

298
estimates obtained from implanting known numbers of cells on
pieces of plastic film (Table 2). Thus, it seems likely that the
signalling ability of the polarizing region involves only a
monolayer of cells, the rest of the cells in a grafted pellet being
superfluous or inactive. Both this finding and the observation
that the mixed aggregates signal best when grafted under an
intact apical ridge suggest that cooperation between the
monolayer of polarizing cells and the apical ectodermal ridge is
essential for positional signalling.
This work was supported by the MRC. I thank Professors L.
Wolpert and B. Alberts for their enthusiastic encouragement. I
also thank Professor Alberts for two visits to his laboratory
supported by NIH grant GM23928, and M. Goodman and A.
Crawley for help with photography and histology.
Received 14 August; accepted 21 November 1980
1. Saunders. J. W. & Gasseling, M. T. in Epirhelial-Mesenchymal Inreracflons (eds Fleishmajer, R. & Bill~ngham,R. E.)78-97 (Williams & Wilkins, Baltimore, 1968).
2. Tickle, C., Summerbell, D. & Wolpert, L. Narure 254, 199-202 (1975).
3. Tickle, C. in Deueloprnenr in Mammals Vol. 4 (ed. Johnson, M. H.) 101-137 (NorthHolland. Amsterdam. 1980).
4. Sm~th,1. C., Tickle, C. & Wolpert, L. Narure 272,612-613 (1978).
5. Saunders, J. W. in Vertebrare Limh and Somite Morphogenes~s(eds Ede, D. A,, Hinchliffe.
J . R. & Balls, M.)1-24 (Cambridge University Press, 1977).

Flow cytometry analysis of T cells

and continuous T-cell
lines from autoimmune MRL/l mice
D. E. Lewis, J. V. Giorgi & Noel L. Warner

Table 1 Comparison of cell-surface markers on in uiuo MRL/lpr

lymph node cells and in vitro MRL/lpr cell lines
--

--

--

MRL/lpr MLN cells

MRL/lpr T-cell line

Moderate density

Lyt-3
T-30

Moderate density
(lower than MRL/n 1)
Increased % positive
Lower density
Marked reduction in
oh positive
As for Lyt-2
Lower density

Th-B
FcR

Negative
Negative

Surface marker
Thy-1.2
Lyt-1
Lyt-2

Most positive
Low density
Most lines variable
As for Lyt-2
Most lines positive
Low density
Negative
Negative

Surface marker was detected by flow cytometry with monoclonal

antibodies. Results for MRL/lpr are intensity and per cent cells
relative to age matched control MRL/n mice. The MRL/lpr T-cell
line is representative of most MRL/lpr cell lines. MLN, mesenteric
lymph node.

Attempts were made to establish continuous cell lines from

MRL/l, MRL/n and C57BL/6 mesenteric lymph node cells by
plating 5 x 10' to 1x lo6 cells ml-' in 5-ml volumes of Dulbecco's modified Eagle's medium (DMEM) supplemented with
non-essential amino acids (1 ml 100 ml-', 100x, Gibco),
M
2-mercaptoethanol, 10% fetal calf serum and 25% T-cell
growth factor (TCGF) (v/v), which was prepared from rat spleen
cells stimulated with 5 pg ml-' concanavalin A (Con A) for 46 h
(ref. 5).
The results of the analysis of Lyt phenotypes of lymph node
cells from MRL/I and MRL/n mice are shown in Fig. 1. The

Immunobiology Laboratories, University of New Mexico School of

Medicine, Albuquerque, New Mexico 87131

The study of spontaneous autoimmunity in mouse models

affords an opportunity to determine the cellular basis of the
immune dysregulation observed in this disease. Recently, a new
mouse strain, MRL/Mp-lpr/lpr (MRL/I) has been developed'
which carries an autosomal recessive gene (lpr) that results in
massive lymph node enlargement concomitant with the
development of several autoantibodies14. The interest in this
strain lies in the possibility that the defect in T-cell regulation of
the immune response is manifested at a different level from that
in the NZB mouse3. It has been reported that the proliferating
population of lymphoid cells in the nodes of these mice are T
cells, but that many of them are devoid of Lyt surface antigens4.
We have accordingly initiated several lines of research with
these mice, including quantitative flow cytometry characterization of Lyt antigen expression of cells in the lymph nodes
of the mice. In an approach to isolate and study the properties of
these cells, we have also established continuous cell lines from
the lymph node cells of MRL/I mice, using techniques similar to
those used to establish continuous lines of antigen-activated
cytotoxic T cell^^.^ and helper T-cell populations7.
Lymph node cells from MRL/1 and MRL/n mice at various
ages, and T-cell lines which were established from MRL/l lymph
node cells were examined for Lyt-1, Lyt-2 and Thy-1.2 cell
surface density using indirect immunofluorescence. Xenogeneic
(rat) hybridoma-derived monoclonal antibodies which have
specificities for Lyt-1, Lyt-2 (both framework determinants) and
Thy-1.2, were provided by Drs Ledbetter and Herzenberg of
Stanford Universitys. The lymph node cells were stained for
20min on ice, washed three times in Eisen's balanced salt
solution (BSS) and then fluorescein-conjugated (Fl) goat antirat immunoglobulin was added as a second step reagent. The
cells were incubated for an additional 20 min, washed three
times in BSS and examined using the flow cytometry systems at
Los Alamos Scientific L a b o r a t o r i e ~ ~or. ' ~a Becton Dickinson
FACS-111 cell sorter.
0028-0836/81/040298-03$01.00

Fluorescence channel no. (3 decade log amp)

Fig. 1 Flow cytometry analysis of Lyt-1 and Lyt-2 expression on
mesenteric lymph node cells of MRL/lpr and MRL/n mice. In each
analysis cell suspensions were stained with the hybridoma-derived
supernatants followed by fluorescein-conjugated SJL anti-rat
immunoglobulin. The left column of panels (a, c, e, g,) shows cell
suspensions from 2-month old mice, and the right-hand column of
panels (b, d, f, h ) those from 5-month old mice. a, b, MRL/n cells
stained with anti-Lyt-1; c, d, MRLJlpr with anti-Lyt-1; e, f,
MRL/n with anti-Lyt-2; g, h, MRL/lpr with anti-Lyt-2. The key
distinction between MRL/lpr and MRL/n mice is the presence of a
low density Lyt-1 peak around channel 110, and a selective
reduction in Lyt-2 cells in older MRL/lpr mice.
@ 1981 Macmillan Journals Ltd

Nature Vol. 289 22 January 1981

25"/ TCGF,

299

Time (days)
Fig. 2 Dependence on T-cell growth factor (TCGF) of MRL-111
continuous T-cell line. Four aliquots of MRL-111 cells were seeded
at 10'ml-' in the presence of Dulbecco's medium with the
indicated additive. Growth occurs in the presence of 25% TCGF
either with or without a-methylmannoside. Growth of the line is
not supported by the addition of Con A alone.

MRL/I lymph node population at 8 weeks of age is characterized by a shift in the surface density of Lyt-1, in that,
compared with MRL/n lymph node cells, a population of cells
with lower Lyt-1 expression is evident. By 5 months of age this
low surface density population of Lyt-1 cells has become the
predominant population and by 5 months of age there are very
few Lyt-2-positive lymph node cells (that is, either Lyt-1'2' o r
Lyt-1-2'). The Thy-1.2 stainingcharacteristics indicate that the
MRL/l mouse lymph node cells, which are all Thy-1.2 positive,
have lower amounts of Thy-1.2 antigen than d o MRL/n lymph
node cells. As the cell population is also somewhat larger, the
actual cell-surface density is considerably lower than in the
MRL/n mice.

Seven MRL/I cell lines have now been established and several
of the lines have been in culture for up to 8 months. The lines
were all established from mesenteric lymph node cells from
3-5-month old MRL/l mice. The cells will not grow in DMEM
without TCGF, nor will they grow in D M E M supplemented with
5 pgml-' Con A (Fig. 2). Furthermore, addition of 6 0 m M
a-methylmannoside to the growth medium does not inhibit the
effect of the TCGF. These data indicate that the growth of the
MRL/I cell lines is dependent on a growth factor in the TCGF
medium which is not Con A itself.
Attempts made to establish continuous cell lines from MRL/n
and C57BL/6 lymph node cells were unsuccessful. The cells
multiplied much more slowly than the cells from MRL/I mice,
and never reached high cell densities in uitro. These lymph node
cells were occasionally maintained in culture for periods of up to
2 months.
Because it is important to know whether these in vitro
continuous T-cell lines are representative of the abnormal proliferating cells in vivo, we have compared the in vivo and in vitro
cell lines using a range of surface markers. After several weeks in
culture, the in uitro lines show a complete absence of B cells, and
all cells are Thy-1.2 positive. The Lyt phenotype of the cells has
fluctuated, however, and as shown in Fig. 3, includes both
Lyt-1'2- and Lyt-1'2' lines. The predominant phenotype of
most of the lines has been Lyt-1'2-, with fluctuations primarily
among cells that are Lyt-2'. Some lines thus contain many
Thy-1' Lyt-null cells. These may, however, have lower
undetectable amounts of Lyt-1.
Using a broader range of cell-surface markers, we show in
Table 1 that the in vitro lines, as studied here, are generally
similar to the predominant population of cells in vivo.
O u r results indicate that the massive lymphoproliferation
observed in MRL/l mice is associated with a decline in the
proportion of Lyt-2-positive T cells. This population includes
both Lyt-123- and Lyt-23-positive cells. The majority of the
cells present in these large nodes are Lyt-1 positive, but have a
distinctly lower density of Lyt-1 on their surfaces than normal
cells. These results are in contrast to those of Theofilopoulos et
aL4, who reported that about 5 0 % of the cells in the MRL/I
lymph node are Lyt-null T cells, although they did suggest that
the 'Lyt-null cells' may not be completely devoid of Lyt
antigens. With flow cytometry we find only approximately 10%

VI stained, respectively, with antiLyt-1 and anti-Lyt-2, and panels c

and d cell line MRL-VII stained,
respectively, with anti-Lyt-1 and
anti-Lyt-2. In each panel the number
of cells is on the ordinate and the
relative fluorescent intensity on the
abscissa.

0 1981Nature Publishing Group

Nature Vol. 289 22 January 1981

of the T cells in the MRL/I lymph node to be of Lyt-null type
(but Thy-1.2 positive). Thus, this quantitative analysis clearly
indicates that the proliferating T-cell population in the MRL/l
lymph node may be a relatively homogeneous T-cell subset.
Further analysis of these cells with other differentiation antigens
is proceeding.
The present report also extends the findings of other investigators which indicate that continuously proliferating lines of T
cells can be readily established in long-term culture as long as
T-cell growth factors are present. Our results indicate that it is
possible to establish cells from MRL/I mice (but not cells from
normal mice) in continuous culture without deliberate exogenous antigenic stimulation. Thus, it might be proposed that
antigenic stimulation for these lymph node cells from MRL/I
mice is from an autoantigenic source, probably derived in uivo

before culture, as the lines d o require continual T C G F for

maintenance.
It is interesting that the cell lines generated from the lymph
nodes of MRL/I mice are variable in their Lyt expression,
although all are T cells. Cloning and cell sorting experiments are
in progress to resolve this problem and to determine whether
this variability truly represents instability in Lyt expression in a
continuously proliferating clonal line. The fact that cell lines can
be established at all in the absence of antigenic stimulation may
allow the development of pure subpopulations of specific Lyt
antigen-bearing cells for further dissection of the functional role
of Lyt subsets in the aberrant immune response observed in
these autoimmune mice.
This work was supported by NIH research grants AM19841
and fellowships A105779 and CA05921. W e thank Mrs Jerri
Davis for maintenance of the M R L mouse colonies.

Received 7 May; accepted 24 October 1980

1. Murphy, E. D. & Roths, J. B. Fedn Proc. 36, 1246 (1977).
2. Murphy, E. D. & Roths. J. B. in Generic ControlofAuroimmunity (eds Rose, N. R., Bigazzi,
P. E. & Warner. N. L. 1207-221 (Elsevier, New York. 1978)
3. Gershon, R. K., Horowitz, M., Kemp, J. D . Murphy, D. B. & Mruphy. E. D. in Generic
Control of Auroimmuniry (eds Rose, N. R., Bigazri, P. E. & Warner, N. L.) 223-227
(Elsevier, New York, 1978).
4. Theofilopoulos, A. N., Eisenberg, R. A., Bourdon, M.. Crowell. J. S. Jr & Dixon, F. J. I.exp.
Med. 149, 516-534 (1979).

Cyclosporin A promotes spontaneous

outgrowth in vitro of
Epstein-Barr virus-induced B-cell lines
A. Graham Bird*$, Sandra M. McLachlanl& Sven Britton*

* Department

of Immunobiology, Wallenberglaboratory,
Karolinska Institute, S-104 05 Stockholm, Sweden and
Department of Infectious Diseases, Danderyd Hospital,
S-182 0 3 Danderyd, Sweden
t Department of Medicine, University of Newcastle-upon-Tyne,
Newcastle-upon-Tyne N E l 7 R U , UK

Cyclosporin A (CSA) is a fungal metabolite which exerts profound effects on the immune system and has potential as a
selective immuno-suppressive agent'. Clinical trials with human
renal allograft recipients have confirmed this potential but there
have been disturbing reports of lymphoma in a significant
number of patients2". Despite extensive animal studies, the
specificity of this drug for human lymphocyte subpopulations is
largely unknown4. We demonstrate here that in uitro CSA has
no apparent effect on Epstein-Barr virus (EBV)-induced Blymphocyte activation, but totally inhibits the T-cell dependent
pokeweed mitogen (PWM) B-cell response. In addition, CSA
markedly facilitates the outgrowth of B-lymphoblastoid cell
lines from both EBV-infected and non-infected lymphocytes of
EBV immune donors cultured in oitro. These results indicate
that CSA can interfere with the lymphocytes normally responsible for maintaining the life-long carrier state initiated by
primary infection with EBV, allowing outgrowth of the persistently infected cells circulating in the peripheral blood.
W e have investigated the effects of CSA o n human polyclonal
B-cell activation in vitro. Figure 1 demonstrates the differential
effects of CSA on activation induced by two mechanistically
different activators, E B V and PWM. The results clearly show
that CSA completely inhibits PWM-induced B-cell activation
which is T-cell and monocyte d e ~ e n d e n t "whereas
~
the polyclonal activation produced by the T- and monocyte-independent EBV6 is actually enhanced early in the culture period at 6
days. CSA did not influence cell viability in the PWM cultures
but markedly reduced morphological blast transformation.
*To whom correspondence should be addressed at the Department of Immunology, The
Medical School, Vincent Drive, Birmingham B15 ZTJ, UK.

5. Gillis, S. & Smith, K. A. Narure 268, 154-156 (1977).

6. Nabholz, M.. Engers, H. D., Callovo, D. & North, M. Curr. Topics Microbial. Immun. 81,
176 (1978).
7. Watson, J. A., Aarden, L. A. & Lefkovitz, I. I: Immun. 122, 209 (1979).
8. Dennert, G. Narure 277,476-477 (1979).
9. Shreier. M. H. &Tees, R. Int. Archs Allergy appl. Immun. 61,227-237 (1980).
10. Ledbetter, J. A. & Herzenberg, L. A. Immun. Rev. 47.63-90 (1979).
11. Sleinkamp, J. A. er al. Devs scient. Insfrum. 44. 1301-1310 (1973).
12. Warner, N. L., Daley, M. J., Richey, J. & Spellman, C. Immun. Rev. 48, 197-243 (1979).

W e next studied the effects of CSA on the generation of

lymphoblastoid cell lines from individuals of known EBV
immune status whose lymphocytes were either infected in vitro
with E B V or uninfected. Table 1 summarizes the results
obtained from a number of individuals. In cultures of lymphocytes not treated with CSA, lymphoblastoid cell lines were only
obtained from cultures to which E B V had been added. In
agreement with the results of
cell lines were easily
obtained from lymphocytes from a seronegative donor infected

20,000

20,000

10,000

10,000

0 PWM EBV

0 PWM EBV
0 PWM EBV
Cyclosporin 0.2pg ml-1 Cyclosporin 2pg ml-

Fig. 1 The effects of cyclosporin A o n the polyclonal activation of

human blood lymphocytes by EBV and PWM. Lymphocytes from
seven adult volunteers of known EBV immune status were
cultured at a final density of lo6 ml-' in RPMI 1640 medium
supplemented with 10% human AB serum (anti-VCA positive)
and antibiotics. Cultures were incubated for 6 days in 50-1111 plastic
flasks (Falcon 3013) in the presence of 5%CO2. EBV infection was
achieved by 1 h pre-incubation at 37 "C of a suspended pellet of
5 x lo6 lymphocytes with 1 ml of B95-8 cell line supernatant.
PWM was added to appropriate cultures at a final concentration of
20 p g ml-'. Cyclosporin A (Sandoz) was initially dissolved in a
small volume of 950h ethanol and then diluted in Earle's balanced
salt solution to give a concentration of 100 p g ml-'. The drug was
added t o cultures t o give a final concentration of 2.0 or 0.2 p g ml-'.
Cultures were assayed for the presence of antibody-secreting cells
by use of the protein A plaque assay which has been described
elsewhere6. Results are expressed as IgM plaque-forming cells
(PFC) per lo6 viable cells at the end of culture. A similar relation
was found for IgG PFC (data not shown).
1981 Macmillan Journals Ltd

Nature Vol. 289 22 January 1981

Table 1 Effect of cyclosporin A on the

outgrowth of lympoblastoid cell lines from individuals of known EBV-immune status

Donor

Serological
status
(anti-VCA)

1-6

in vitro

CSA
addition
(2 pg mg-')

EBV
infection

+
+

+
+
-

in vitro

Cell lines
established

+
+
-

+
+

+
-

+
+

.-

+
+
+
+

Immune status of the donors was examined by determination of the

antibody titre to EB-viral capsid antigen (VCA). Cultures were
established as in Fig. 1 and were regularly examined for cell line
formation. This was determined visually by the appearance of progressively growing foci of transformed cells7. The scores above are from
4-12 weeksof culture. In many cases the presence of lines was confirmed
by transferring the cells into secondary cultures. Cell lines arising in
non-EBV-infected lymphocytes were examined for the presence of
EBNA by the method of Reedman and ~lein'.

with EBV, but, despite initial transformation in all seropositive

donor cells, regression of cell lines occurred in eight out of nine
donor cultures. Addition of CSA allowed cell lines to be
established by 4 weeks with ease and without regression from all
seropositive donors so far tested. More significantly, B-cell lines
have also been obtained from cultured lymphocytes from six out
of nine seropositive donors exposed to CSA but not infected
with EBV. These cell lines contained the EB-determined
nuclear antigen (EBNA), a characteristic of E B V transformationg and they were of similar morphology to those
derived from cells exposed to EBV (K. Nilson, personal communication).
These results clearly demonstrate that CSA exerts profoundly
selective effects on human lymphocyte subpopulations in oitro.
The enhancement of the polyclonal immunoglobulin response
induced by EBV and abrogation of that by PWM strongly
suggest that at least the fraction of B cells activatable by EBV is
refractory to the action of CSA. We interpret the abolition of
PWM responsiveness to mean that the site of action of CSA is
the T lymphocyte or, less likely, the monocyte-both essential
for a PWM-induced polyclonal immunoglobulin response.
Furthermore, CSA seems to promote the early and complete
outgrowth of EBV B-lymphoblastoid cell lines from immune
donors in vitro. It has been shown that at a cell density of
106 ml-', lymphoblastoid cell line generation from EBVinfected lymphocytes from immune donors is inhibited by the in
vitro appearance of immune T lymphocytes7. It seems that CSA
specifically deletes or functionally inactivates these T cells,
allowing the rapid appeqrance of EBV-transformed cell lines
from all immune donors tested. In six out of nine seropositive
individuals, EBV-bearing cell lines have been directly
established from the blood merely by the addition of CSA alone.
It is known that EBV-infected B cells persist in the lymphoid
system of immune donors and can grow out into lines in appropriate conditions if T lymphocytes are removed7. CSA seems to
facilitate this outgrowth, probably by its specific action on the T
lymphocyte.
The experimental conditions we have used are certainly
unfavourable for de nouo infection of B cells by in uitro released
infectious particles1', because cells were grown in human A B
0028-0836/81/040301-43f0l00

anti-VCA positive serum. In the presence of specific antibody

EBV cannot activate B lymphocytes11. Although we cannot
exclude transfer of virus between cells in intimate contact we
suggest that the lymphoblastoid cells growing out from CSAtreated blood of EBV immune donors really represent the
progeny of the latently infected cell.
There have been recent disturbing reports of lymphomas
presenting in renal allograft recipients treated with CSA and
there is a general tendency for anti-EBV titres to rise in patients
on treatment with CSA. In one reported case lymphoma
followed asymptomatic infectious m o n o n ~ c l e o s i s and
~ ~ in
another EBNA was present in the lymphomatous B cells" On
the basis of these in vitro experiments we suggest that in the
presence of CSA, T-lymphocyte surveillance of EBV-infected B
cells is no longer effective. The result is a polyclonal proliferation of EBV-transformed B cells which in uitro from Blymphoblastoid lines but in uiuo manifest as the polyclonal
immunoblastic proliferation characteristic of immunosuppressed patients13.
We thank Kicki Hild for technical assistance, Kenneth Nilson
for characterizing the lines and G. Klein for supplies of B95-8
supernatant. This work was supported by the Swedish MRC and
in part by a grant from the North of England Cancer Research
Campaign.
Received 2 Supternher; accepted 7 November 1980.
1.
2.
3
4
5.
6.
7.
8.
9.
10.
11.
12.
13.

Borel, I F Feurer, C., Mognee. C. & Stahelin. H. Immunology 32, 1017-1025 (1977).
Calne. R Y. ur ul. Lanrur ii, 1033-1036 (1979).
Crawford. D. H. pr ul. Lancer i, 1355-1356 (1980).
Leoni. P., (iarc~a.R . C. & Allison. A . C. I, clm. l a b . I m m u n 1 , 67-72 (1978).
1anossy.G.. Gomepde la Concha. E., Waxdal. M J. & Platts-Mills.T. Chn. ~ x p Immun.
.
26.
108-1 17 (1976).
Bird, A. G . & Hritton, S. S c u t ~ dJ. Immun. 9, 507-510 11979).
Moss, D. J., Rickinson, A . B & Pope, J . H . Inr. I Cancer, 22, 622-668 (1978).
Thorley-Lawson. D. A,. ('hers. L. & Strorninger. 1. L. I. exp. M e d . 146,495-598 (1977).
Reedman, B. M . & Klein. G . Inr. J. Cancer 11, 499-520 (1973).
R ~ c k ~ n s o A.
n , 8 . . F~nerty,S. & E p t e i n , M. A. Inr. J. Cancer 19, 775-782 (1977).
Bird. A. G. & Britton. S. Immun. Rrv. 45, 541-567 (1979)
Nagngton. J. & Gray, J . Lancer i , 536-537 (1980).
Purtdo. D. T . Lanrrr i , 300-303 (1980).

Inhibition of P. falciparum
growth in human
erythrocytes by monoclonal antibodies
L. H. Perrin, E. Ramirez, P. H. Lambert
& P. A. Miescher
WHO Rescarch and Training Centre, Department of Medicine.
HBpital Cantonal, 1211 Geneva 4, Switzerland

Malaria is increasing in incidence and prevalence in most tropical areas and is a major problem for both individuals and
communities'. Current malaria research is aimed at developing
vaccinesz4 and, for this, it may be useful to define Plasmodium
antigen@) related to the development of a protective immune
response in the host. Monoclonal antibodies have recently been
shown to interfere with rodent malaria infection (Plasmodium
berghei) at the sporozoite5 or merozoite6 stage. We have now
raised monoclonal antibodies against single antigenic determinant(~)of Plasmodium falciparum and report that some of
them inhibit the growth of erythrocytic forms of P. falciparum In
vitro.
Red blood cells (RBCs) parasitized with P. fulciparum were
obtained from a European patient o n his return from Senegal
and were adapted and maintained in in vitro culture according to
@ 1981 Macrndlan Journals I.td

Nature Vol. 289 22 January 1981

302
Table 1 Inhibition of P. falciparum growth by monoclonal antibodies

Culture medium +
10% ascitic fluid
from various
hybridomas
Hb28 cll
Hb26 c13
Hb31 c12
Hb12 c14
Hb9 cll
Medium + 5% NMS
Medium + 10% NaCl
P3-X63-Ag8 (Fo)

Ig class

IgG3
IgG2a
IgG3
IgG 1
IgM

OO
/ Parasitaemia
Intact
Total
parasites

% Growth

inhibition

100
96
95
19
18
17
0
19

The inhibition of P. falciparum growth in vitro by various monoclonal

antibodies was analysed by a modification of the technique described by
Wilson and ~ h i l l i ~ s 'Synchronized
~.
cultures of ring forms of P.
falciparum were diluted with uninfected human RBCs to provide a final
suspension of 1.5% RBCs and 0.5% parasitaemia. Aliquots (150 pl) of
this suspension were placed in 96 flat-bottomed wells (3040 Falcon
plates). Each microculture was done in triplicate. The medium used for
the culture (RPMI 1640 with 10% normal human serum)7 was supplemented with either 10% ascitic fluid produced by various hybridomas or
as control with 5% serum of normal BALB/c mice (NMS),10% isotonic
NaCI, or ascitic fluid induced by myeloma cells used for fusion. The
culture medium was removed from the settled cells each day and
replaced with fresh medium. The plates were incubated at 37 "C in a
candle jar9. Parasitaemia was determined on Giemsa-stained thin blood
smears after 80 h of incubation. In these conditions, more than two
cycles of replication took place in the wells containing the medium
+lo% NaCI. At the end of the experiment, some of the parasites were
pyknotic, especially those in cultures supplemented with Hb28 cll,
Hb26 c13 and Hb31 c12. Therefore, the percentage inhibition was
determined over the number of intact parasites. The results reported
here represent three hybridomas with inhibitory activity and two out of a
group of eight hybridomas without any significant inhibitory activity
(10-19%).

the method of Trager and Jensen7. These cultures were not

synchronized and contained all the various developmental
erythrocytic stages (rings, trophozoites, schizonts, merozoites).
Synchronization was induced by two treatments with 5 %
mannitol at an interval of 34 h (ref. 8); this treatment lysed the
mature forms (trophozoites and schizonts), leaving only the ring
forms. Approximately 30 h after treatment with mannitol,
merozoites were released from schizonts and non-infected
RBCs were invaded.
In the first series of experiments, 8-week-old female BALB/c
mice were immunized intraperitoneally (i.p.) three times at
2-week intervals with 100 k g of merozoite and schizont antigen
in Freund's i n c o m ~ l e t eadiuvant. Four davs before the fusion
experiment a booster of 5 k g of the antigenic preparation was
administered intravenously (i.v.). Merozoites and schizonts
were purified from a synchronized culture of P, falciparum by
Plasmagel sedimentation9 and contained < 1 0 % normal RBCs.
In a second series of experiments, BALB/c mice were infected
i.p. with 1 x 10' erythrocytic forms of P. berghei, treated with
c h l o r ~ q u i n e and
' ~ 3 weeks later injected i.v. with 5 x 10' erythrocytic forms of P. berghei 4 days before the fusion procedure.
For each experiment, 2 x lo7 immune spleen cells in suspension
were fused by polyethylene glycol with 1x lo7 non-secreting
mouse myeloma cells derived from P3-X63-Ag8 (clone F O
given by D r Fasekas, Basel Institute of Immunology). Conditions for fusion were essentially as described by Galfre et al.".
Positive hybrids were clone purified twice on mouse macrophage monolayers by limiting dilution and further grown in T75
Falcon flasks. 5 x lo6 cells were injected i.p. into BALB/c mice
which had been primed 1 week previously with 0.5 ml Pristan
i.p.

A n indirect immunofluorescence technique was used to assay

hybrid supernatants. Acetone-fixed P. falciparum-parasitized
human RBCs o r P. berghei-parasitized mouse RBCs were used
as antigenic preparation and rabbit fluoresceinated anti-mouse
F(ab')2 as the developing second a n t i b ~ d ~ ' ~ Antibodies
.'~.
specific for the surface antigen o r merozoites were similarly
detected by reacting hybrid supernatants o r ascitic fluid with a
suspension of unfixed schizonts and merozoites (-lo5 per 25 ~ 1 )
purified by ~ l a s m a ~ e l ~ .
Spleen cells from nine mice immunized with P, falciparum
merozoites and schizonts and two mice immunized with P.
berghei were fused in 11 experiments: 4 7 of 978 supernatants
tested reacted with P. falciparurn-parasitized RBCs; two of
these supernatants obtained from mice immunized with P.
berghei reacted with both P. berghei and P. falciparurn. From
each fusion experiment, hybrid cells from one o r two wells (total
15) were selected, clone purified twice on macrophage
monolayers and grown in large amounts. The cells were injected
i.p. into mice, and the ascitic fluid was collected. When the 15
ascitic fluid samples were tested by indirect immunofluorescence
against the different developmental stages of acetone-fixed
parasites, 1 0 samples (including Hb28 cll and Hb31 c12) reacted
with trophozoites and schizonts only, two (including Hb26 c13)
reacted with schizonts only, and three reacted with all stages.
The ascitic fluid samples were positive by indirect fluorescence
at dilutions of 1/5,000-1/100,000.
The culture medium used for P. falciparum culture7 was
supplemented with 10% of the various ascitic fluids and the
mixtures were tested for their ability to inhibit the multiplication
of erythrocytic forms of P. falciparum. A n 80-h assay corresponding to more than two cycles of multiplication in normal
culture conditions was used. Three out of 11 ascitic fluids tested
had an inhibitory activity, two of them, Hb26 c13 and Hb31 c12,
were produced with spleen cells from mice immunized against P.
falciparum antigens, and the third one, Hb28 cll, with spleen
cells from mice infected with P. berghei (Table 1). Sera from

+ Hb 28 cl 1
Medium + Hb 26 c l 3
Medium + 5% NMS
Medium + saline
Medium

16-30
40-54
64-78
Time of 'H-hypoxanthine incorporation (h)
Parasitaemia
(% total)

2 2 22

t - t - -

i v ;

Fig. 1 Incorporation of 3~-hypoxanthinewas determined in P.

falciparum cultures supplemented with 10% ascitic fluid from

Hb26 c13 and Hb28 cll hybridomas, 50h BALB/c normal serum
and 10% isotonic NaCI, three times during an interval of 14 h. The
experimental design was identical to that described in Table 1
legend, except that lop1 of 3~-hypoxanthine(1 pCi) were added
to each culture at the beginning of the labellingperiod (18.5-55.5 x
Bq mmol-' specific activity, NEN). The percentage parasitaemia was determined at the end of each labelling experiment on
Giemsa-stained thin blood smears.

0 1981Nature Publishing Group

303

Nature Vol. 289 22 January 1981

Table 2 Specificity of the inhibition of P. falciparum growth

Culture medium
supplemented with:
1. Hybridoma Hb26 c13
10% ascitic fluid
IgG-depleted 10% ascitic
fluid
Purified IgG fraction
(0.5 mg ml-')
2. Mouse immunoglobulins
IgG2a (LPC-1) (1 mg ml-I)
IgG2b (MOPC-195) (1 mg ml-')
Normal mouse IgG (1 mg ml-')
3. Medium + 10% NaCl0.9%

O/O

O/O Growth
Parasitaemia inhibition

0.8
7.2

91
15

0.9

89

7.8
9.1
7.7
8.4

7
0
9
0

The specificity of the inhibition of P. falciparum growth was analysed

in experimental conditions identical to those described in Table 1

legend. The IgG fraction from Hb26 c13 ascitic fluid was purified by 50%
ammonium sulphate precipitation and DEAE cellulose chromatography and shown to be devoid of contaminants when analysed by
immunoelectrophoresis. As a control, Hb26c13 ascitic fluid was
depleted of IgG by passage on a Sepharose-protein A column (Pharmacia) and reconstituted to its original volume by negative pressure
dialysis. Mouse immunoglobulin was purified from normal BALB/c
sera and the IgG2a and IgGZb from the ascitic fluid of BALB/c mice
bearing LPC-1 and MOPC-195 myeloma tumours. These immunoglobulins were prepared by precipitation in 50h ammonium sulphate
followed by DEAE column chromatography. The concentration
indicated in parentheses for IgG is the final concentration in the
medium.
mice injected with hybrid cells of Hb26 c13 and Hb28 cll were
also tested and gave, respectively, 78'10 and 99% inhibition of P.
falciparum growth when added at a final concentration of 5 % to
the culture medium.
Hb26 c13, Hb28 cll and Hb31 c12 monoclonal antibodies
inhibited the development of schizonts and the invasion of
normal RBCs by merozoites. Normal development of ring forms
into trophozoites was observed when differential counts of the
developmental stages were monitored on synchronized cultures
supplemented with either normal medium o r medium containing 1O0/0 ascitic fluid from each of the inhibitory monoclonal
antibodies. Subsequent maturation of trophozoites into
schizonts occurred; however, the schizonts frequently showed
morphological signs of degeneration in cultures supplemented
with any one of the three monoclonal antibodies. (At the end of
the first cycle, these degenerated schizonts represented 15-35O/0
of the expected number of schizonts.) In normal culture, 7.6 new
ring forms were generated in the second cycle from each late
trophozoite and schizont present in the previous cycle.
However, in cultures supplemented with Hb26 c13, Hb28 cll o r
Hb31 c12, the number of ring forms per trophozoite and schizont
fell to 0.5, 0.2 and 0.4, respectively.
Incorporation of 3H-hypoxanthine during a 14-h pulse was
also measured in synchronized cultures supplemented with
ascitic fluid. In the absence of ascitic fluid, there was a low rate of
incorporation from 16 to 30 h, increased incorporation from 40
to 54 h and the highest rate of incorporation from 64 to 7 8 h.
The increase in incorporation paralleled the increase of parasitaemia. In cultures supplemented with Hb28 cll o r Hb26 c13
there was a low rate of incorporation from 1 6 to 3 0 h followed by
a decrease in incorporation for Hb28 d l and a slight increase for
Hb26 c13 from 6 4 to 7 8 h. Here again, the rate of incorporation
paralleled the level of parasitaemia (Fig. 1).
The IgG fraction of Hb26 c13 ascitic fluid was shown to have
full inhibitory activity (Table 2); the activity of Hb26 c13 ascitic
fluid was lost after depletion of IgG. Similarly, IgG fraction
purified from Hb28 cll at a final concentration of 0.25 mg ml-'
showed an inhibitory activity of 9 5 % . In such cultures inhibited
by IgG fractions, schizont degeneration was also present.
The amount of monoclonal antibodies needed to cause 50%
inhibition of the culture growth after two multiplication cycles of

P. falciparum was 8 pg mi-' of culture medium for Hb28 cll

(1180 dilution of ascitic fluid) and 16 CLgml-' for Hb26 c13.
P. falciparum-infected cultures were also internally labelled
with 35S-methioninefor 4 h and lysed by 0.5% Nonidet P40 (ref.
13). The lysate was used as antigen in double immunoprecipitation experiments using monoclonal antibodies (Hb26 c13, Hb28
cll and Hb31 ~12)". The washed immune precipitates were
analysed on 1O0/0 SDS-polyacrylamide gel electrophoresis gels.
By autoradiography it was shown that Hb28 cll and Hb31 c12
monoclonal antibodies reacted with a polypeptide of molecular
weight (MW) 41,000 whereas Hb26 c13 reacted with two polypeptides of MWs 96,000 and 36,000. These polypeptides are
probably present on the merozoite surface. This was indicated
by results of the indirect fluorescence test when an unfixed
suspension of P.falciparum was reacted with monoclonal antibodies. Intense surface staining was seen using monoclonal
antibodies Hb26 c13, Hb28 cll and Hb31 c12. In contrast, no
surface staining was observed with Hb12 c14 o r H b 9 cll.
Previous investigators have shown that some sera from
individuals living in areas endemic for malaria inhibit the growth
~ , they have not identified the
of P. falciparum ~ u l t u r e ' ~ - 'but
specificity of the 'protective' antibodies. In the present experiment, antigens recognized by monoclonal antibodies inhibiting
the growth of P. falciparum have been characterized. Interestingly, monoclonal antibodies with inhibitory activity were
obtained from the Hb28 cll hybrid after fusion of spleen cells
from mice infected with P. berghei, and reacted by
immunofluorescence with P. berghei (trophozoites and
schizonts), suggesting the presence of a common antigenic
determinant on the surface of P. falciparum and P. berghei
merozoites. Several hypotheses can be raised to explain the
inhibitory activity of the monoclonal immunoglobulin. First,
these antibodies may prevent the merozoite invasion of RBCs
by binding to the merozoite surface and interfering directly o r
indirectly with the parasite ligand-RBC receptor. Second, they
may bind to parasite antigens at the surface of infected RBCs at
the schizont stage and thus modify the intracellular parasites'
metabolism. Third, because schizonts containing RBCs are
known to have an increased permeability, one cannot exclude
the possibility that some antibody molecules would reach
intracellular parasites at the last stage of development.
These monoclonal antibodies provide new tools for the
purification of potentially protective antigens of P.falciparum,
although the large-scale production of Plasmodium antigens is
still limited by technical problems. An alternative approach is to
use D N A recombinant technology to produce defined malarial
antigens. In this respect, the protective monoclonal antibodies
may be useful for the identification of Escherichia coli colonies
secreting putative protective antigens".
This work was supported by WHO, the UNDP/World
Bank/WHO Special Program for Research and Training in
Tropical Diseases and the Swiss National Foundation (grant no.
3.890-0.79).
Received 2 May; accepted 21 November 1980.
1. 3rd Annual Rerwrr on Special Program for Rerearrh and Training in Tropical Diseases
( W t i O . Geneva. 1979).
2 . Cohen, S.. Butcher. G. A. & Mitchell. G. H. Adv. exo. Med. Biol. Y3.89-112 (1977).
3. Reese. R. T..Tragcr, W . . Jcnsen. 1. B .Miller.D. A. &~nntravaki.
R Proc. narn. Acud. Sci.
U.S.A. 75,5665-5669 (1978).
4. Siddiqu~.W. A. Science 197. 388-389 (1977).
5. Yosh~da.N., Nussenzweig. R. S., Potocnjak, P Nussenzweig. V. & Alkawa, M. Science
207,71-73 (1980).
6 . Freeman, R. R., Trejdos~ewicl,A. J & Cross, G. A. M. Narure 284. 366-368 (1980).
7 . Trager. W . & Jensen, J. B. Narure 1 7 3 . 6 2 1 4 2 2 (197R);Science 1 9 3 . 6 7 3 4 7 5 (1976).
8 . Larnbros, C. & Vanderberg, I. P. I. Paraslr. 6 5 , 4 1 8 4 2 0 (1979).
9 . Reese. R. T., Langreth. S. G. & Trager. W. Bull. W l d H h h Org. Suppl. 1. 5 3 4 2 (1979).
10. June. C H., Contreras. C. E..Perrin, L. H.. Lamber1.P. H. & Mischer. P. A. I. Immun. 111.
2154-2160 (1979).
11. Galfre. G.. Hove. S. C. & Milstein. C. Nafure 266.550-552 (1977).
12. O'Neill. P. & Johnson. G. D. J. din. Parh. 23, 185-189 (1970).
14. Wilson. R. J. M. & Phillips. R. S. Narure 263. 132-134 (1976).
13. Perrin. L. H.. Ramirez, E., Liu. E.-H. & Lambert. P. H. Clin. erp. Immun.41,Yl-96(1980).
15. Cohen, S.. McGregor, I. A. & Carrington, S. P. Narure 192.733-737 (1961).
16. Phillips, R. S.. Trim, P. I.. Scott-Finnigan, T. J . & Bartholomew, R. K. Parasirology 65. 525
11
9771
~
.
. -,.
.
17. Broome. S. & Gilbert, W. Proc. nam. Acad. Sci. U.S.A. 75, 2746-2749 (1978).

0 1981Nature Publishing Group

Nature Vol. 289 22 January 1981

Respective roles of laminin and

fibronectin in adhesion of
human carcinoma and sarcoma cells

ing on the growth stage (sparse actively growing versus confluent

resting cultures) the medium conditioned by corneal cells
contains 20- to 100-fold less laminin than medium conditioned
by vascular endothelial cells.

Israel Vlodavsky* & Denis Gospodarowiczt

* D e p a r t m e n t of Radiation a n d Clinical Oncology, S h a r e t t Institute,
Hadassah University Hospital, Jerusalem, Israel
-1 C a n c e r R e s e a r c h Institute a n d t h e D e p a r t m e n t s of Medicine a n d
Ophthalmology, University of California, Medical C e n t e r ,
S a n Francisco, California 94143

In a previous study1, Ewing's sarcoma cells and colon carcinoma

cells2failed to attach to plastic and to dishes coated with various
collagen types I, I11 and IV, but adhered rapidly to extracellular
matrix (ECM) produced by cultured corneal endothelial cells.
However, although carcinoma cells required ECM and did not
adhere to fibronectin, the sarcoma cells adhered and flattened
almost equally well on either substrate1. Thus, different
adhesive proteins may mediate the attachment of sarcoma- and
carcinoma-derived cells to extracellular matrices, the most
likely being fibr~nectin"~
and lamininsV6.Fibronectin stimulates
the adhesion of fibroblasts, but not epidermal cells, to collagen
type IV (ref. 7) and could mediate the attachment of sarcoma
cells. Laminin is confined to the lamina lucida region of basement membranes6 and has been localized to cellular adhesion
sites8. Studies of the attachment of epidermal cells in oitro (V. P.
Terranova, personal communication) suggest that it is adhesive
for epithelial cells, and so could have the same role for
carcinoma cells as that played by fibronectin for sarcoma cells.
Cultured vascular endothelial cells secrete fibronectin9-l1 and
laminin12 into both the ECM and the culture medium, and we
report here that $re-exposure of plastic dishes to such conditioned medium induces the attachment and flattening of both
human colon carcinoma and Ewing's sarcoma cells. While
laminin mediates the attachment and spreading of the former
fibronectin is responsible for the attachment and flattening of
the latter.
Figure 1 shows the time-course of attachment of carcinoma
cells to various substrata. Less than 30% of cells seeded on
plastic or fibronectin-coated dishes became firmly attached and
no flattening was observed, regardless of time in culture (up to
10 days) and whether or not serum (10%) was present. In
contrast, 70 and 80% of the seeded cells firmly attached and
flattened within the first hour of incubation on dishes preexposed to conditioned medium or coated with ECM, respectively (Figs 1,2).This was observed regardless of whether or not
carcinoma cells were seeded in the presence of serum (10%).
The best results, in terms of adhesion and percentage of firmly
attached and flattened cells, were obtained with medium from
endothelial cell monolayers cultured for more than 8 days.
Serum-free conditioned medium collected 5-7 days after the
cells had reached confluence also promoted obvious cell
attachment and flattening. Similar results were obtained with
dishes exposed overnight for 2 h (Fig. 2) to the conditioned
medium and seeded with either trypsinized single carcinoma
cells (Fig. 2b, c ) or cell aggregates (Fig. 2). In the latter case
flattening and migration of peripheral cells were observed within
1 h and led to complete dispersion of aggregates 6-8 h after
seeding (Fig. 26). When similar tumour cell aggregates were
seeded on ECM-coated dishes, flattening and migration were
observed within 30 min (Fig. 2a). Pre-exposure of dishes to
medium conditioned by human skin fibroblasts, vascular smooth
muscle cells, or corneal endothelial cells (Fig. 3), or to fresh
medium containing 10% serum supplemented or not with
fibroblast growth factor (FGF), did not promote cell adhesion,
flattening and migration. Studies of the production of laminin by
vascular and corneal endothelial cells have shown that depend-

Time (min)
Fig. 1 Time-course of attachment of carcinoma cells to various substrates.
Stock cultures of colon carcinoma cells grown on ECM-coated dishes were
dissociated by treatment with STV (0.05% trypsin, 0.02% EDTA), and then
2 x l o 5 cells were seeded on plastic dishes (0).
dishes coated with 10 ~ g e a c h
of affinity purified bovine plasma fibronectin (0,
ref. 1). dishes coated with
ECM (A, ref, I), or dishes coated with vascular endothelial cell conditioned
medium (A). To coat them with conditioned medium, dishes were exposed to
a medium (DMEM, H-16 supplemented with 10% calf serum) taken from
vascular endothelial cells maintained for 8 d in culture (from the day of
seeding until 3-4 d after reaching confluence). This medium was centrifuged
(1,500g for 5 min), incubated (14 h, 37 "C)on the dishes, and replaced with
fresh conditioned medium before cells were seeded. Serum-free conditioned
medium was collected from subconfluent endothelial cultures 7-9 d after
they had been supplied with DMEM (H-16) containing no serum. Because
the production and secretion of proteins by endothelial cells is greatly
reduced when they reach c ~ n f l u e n c e " ~ 'medium
~,
conditioned by confluent
cultures was less active than medium conditioned by actively growing
vascular endothelial cells. At various times after seeding unattached
carcinoma cells were removed by gentle pipetting and rinsing of the dishes
(twice) with phosphate-buffered saline (PBS). The remaining firmly attached
cells were dissociated with STV and duplicate cultures were counted with a
Coulter counter. The variation in different determinations did not exceed
+ l o % of the mean.

Fig. 2 Morphological appearance of carcinoma cells seeded on dishes

pre-exposed to endothelial cell-conditioned medium. Floating cell aggregates (a, d ) from stock cultures maintained on plastic or trypsinized single
cells (6, c ) from stock cultures maintained on ECM-coated dishes were
seeded in DMEM containing 5% fetal calf serum on plastic (d) or on dishes
pre-exposed (2 h, 37 "C) to endothelial conditioned medium as described in
Fig. 1 legend (a, 6, c). a, Cell spreading and migration from cell aggregates
8 h after seeding. 6, Single cells 1 h after seeding. c, Confluent culture 6 d
after seeding. The cells proliferate and form a layer of firmly attached and
flattened cells.
1981 Macmillan Journals Ltd

Nature Vol. 289 22 January 1981

305
with purified fibronectin was either more o r as effective in
promoting cell attachment and subsequent flattening.
The nature of the agents mediating attachment and flattening
was analysed by subjecting the conditioned medium to specific
double immunoprecipitation with either rabbit anti-lamininh or
anti-fibronectin antiseraY, followed by goat anti-rabbit IgG
antiserum". Experiments with 35S-methionine-labelled vascular endothelial cells had shown that these antisera, as determined by SDS-polyacrylamide gel electrophoresis, precipitate
specifically and quantitatively the laminin'* and fibronectin"
present in endothelial cell conditioned medium. As Fig. 4b
shows, carcinoma cells no longer became attached to dishes
pre-exposed to conditioned medium if laminin had been
removed bv immunopreci~itation. O n the other hand,
immunopredipitation of fibronectin (Fig. 4 a ) or collagen type IV
(not shown), bv* s~ecific
antibodies had no effect on the attach.
ment and flattenine of carcinoma cells. In contrast. when the
same immunoprecipitated fibronectin-free conditioned medium
was used to coat dishes, sarcoma cells no longer became
attached (Fig. 4d). The induced attachment and flattening of
sarcoma cells were not affected by subjecting the conditioned
medium to immunoprecipitation with anti-laminin (Fig. 4e),
anti-collagen type IV o r non-immune rabbit serum (Fig. 4f)
followed by coating the dishes. These results show that the
human carcinoma- and sarcoma-derived cells show a specific
response to different adhesive glycoproteins, such as laminin
and fibronectin, respectively.
Immunofluorescence has shown that the in vivo distribution
of laminin is identical to that of collagen type IV (refs 6,8) and so
we compared the attachment of carcinoma cells to dishes coated
with collagen types I, 111, IV o r V before being coated with
conditioned medium. (Collagen would not promote cell
attachment unless dishes were subsequently exposed to conditioned medium.) Collagen type IV gave the best results in terms
of adhesion, percentage of firmly attached cells and prevention
of retraction of flattened cells within days of initial attachment.
Other collagens had either a small effect (type I) or were
inhibitory (tipe I1 and in particular type V, ~ i g . - 5 dcompared
)
with direct exuosure to conditioned medium. When conditioned
medium with a relatively low cell attachment-promoting activity
was used (Fig. 5 a ) , its effect was potentiated by precoating the
dishes with collagen type IV (Fig. 56). These results suggest that
laminin may mediate the attachment of human carcinoma cells
v

Fig. 3 Effect of conditioned media from various cell types on carcinoma cell
attachment and spreading when cells are seeded and maintained in conditioned medium or seeded and maintained in fresh medium. Tissue culture
dishes were pre-exposed (14 h, 37 "C) to media conditioned for 8 d by
vascular endothelial cells (a*) or corneal endothelial cells (d). Carcinoma
cell aggregates taken without trypsinization from a culture maintained on
plastic were seeded directly into the conditioned medium and observed 6 h
later ( a ) . Cell aggregates attached loosely to the substrate and only a small
degree of cell flattening and migration was observed, even after 1 week in
culture. When the conditioned medium was replaced with fresh medium,
good cell attachment and spreading were observed after 2 h (6).This effect
was then long-lasting (c, 1 week in culture). This demonstrates that, when
carcinoma cells are seeded into conditioned medium-coated dishes, the
conditioned medium has to be removed and replaced with fresh medium to
promote cell attachment and spreading. Because poor cell attachment and
flattening are observed when dishes contain conditioned medium (a), it is
suggested that, although coating the plastic surface with conditioned
medium induces cell flattening, soluble factorb) which can adversely affect
that process is also present in the conditioned medium, d shows cell
aggregates 4 dafter seeding in fresh medium into dishes coated with medium
conditioned by corneal endothelial cells. Similar results were obtained with
medium conditioned by vascular smooth muscle cells or human skin fibroblasts. In either case, and as was the case with uncoated tissue culture dishes,
cell aggregates attached poorly and cell flattening was not observed.

Pre-exposure of dishes to the endothelial cell conditioned

medium (either serum-free o r containing fibronectin-depleted
calf serum) also induced attachment of sarcoma cells. However,
in this case and in contrast to carcinoma cells, coating of dishes

Induction of tumour cell attachment promoted by mating dishes with vascular endothelial cell conditioned medium from which fibronectin or laminin had
been removed by double immun~preci~itation.
vascular endothelial cell conditioned medium (1 ml) was preincubated (1 h, 37 'C) with 10 r l of rahhit antisera
against fibronectin (a, d )or laminin (b,e).or with 10 kl of non-immune rabbit serum ( c , f ) .Antibodies to laminins.6andbovine pla~mafibronectin~.~'
were the same as
those used p r e v i o u ~ l ~ ~ .The
~ . ~specificities
.'~.
of the antisera and lack of cross-reactivity between the various antigens have been established in these previous reports.
The conditioned media were then incubated (2 h, 3 7 T , followed by 14 h at 4 "C)with 30 ~1 of goat anti-rabbit IgG antiserum (35% ammonium sulphate cut),
centrifuged (12,00013 for 5 min) and the supernatant incubated in 35-mm tissue culture dishes for 12 h a t 37 "C. The conditioned media were then removed and 2 ml of
DMEM supplemented with 5% fibronectin-depleted calf serum and containing either 2 x l o s colon carcinoma cells (a+) or Ewing's sarcoma cclls (d-f) were then
added to the dishes. Phase micrographs of cell attachment and spreading were taken 3 h after seeding. Conditioned medium pre-ex~osedto anti-collagen type IV (refs
5 . 6 ) gave results similar to those observed after pre-incubation with non-immune rabbit serum. Similar results were obtained with serum-free vascular endothelial
cell conditioned medium and with tumour cells seeded in DMEM containing 0.1% bovine serum albumin rather than 5% bovine serum.

O 198lNature Publishing Group

306

Nature Vol. 289 22 January 1981

Fig. 5 Effect of precoating with types IV and V

collagen on colon carcinoma cell attachment
induced by a subsequent exposure to the vascular endothelial cell conditioned medium. Tissue
culture dishes (35mm) were coated with
purified preparations of types IV (human
placenta) or V (bovine cornea) collagen (10 kg
per dish) as previously described7. Plastic and
collagen-coated dishes were then exposed (12 h,
37 "C) to conditioned medium taken 5 d after
replacing the growth medium of subconfluent
endothelial cultures. Carcinoma cells ( l o 5 per
dish) were seeded in fresh medium (DMEM,
10% FCS) and phase micrographs taken 3 d
later. a, Pre-exposure to conditioned medium
alone. b, Precoating with collagen type IV
followed by exposure to conditioned medium. c,
Dishes coated with collagen type IV alone. d,
Precoating with collagen type V, followed by
exposure to conditioned medium.

to type IV collagen which is characteristic of basement

membranes.
Both laminin and fibronectin have been shown to be secreted
underneath the endothelial cell layer and to be part of the ECM
produced by these cells1z. In the ECM, they may function as
conditioned medium does, and may promote the attachment
and rapid spreading of carcinoma- and sarcoma-derived cells1.
The production of both laminin and fibronectin by vascular
endothelial cells may facilitate the arrest of blood-borne metastatic cells1' that produce little or no such adhesive glycoproteinsI4. Metabolic labelling with 35S-methionineof the Ewing's
sarcoma and colon carcinoma cells (cultured on ECM) followed
by specific immunoprecipitation had shown that these cells
secrete no detectable laminin and small amounts of fibronectin
(less than 3% and 0.1% of the newly synthesized proteins,
respectively). Interaction of highly metastatic cells with the
luminal surface of the vascular endothelium is followed bv
endothelial retraction and exposure of the underlying ECM, t;
which the tumour cells adhere strongly13, possibly through
specific attachment-promoting molecules such as laminin and
fibronectin. Cell attachment is then followed by migration,
degradation and invasion of the tumour cells through the endothelial basement
The presence of specific
attachment factors in basement membranes, the ability or
inability of cells to produce and/or interact with such factors, as
well as the preferential interaction of these adhesive factors with
a specific type of collagen may determine the localization of a
given cell type to specific matrices7. Such specific cell-matrix
interactions may direct the growth, migration and differentiation of normal cell^^^.'^, as well as the invasiveness and
metastasis of tumour cellsI3.".
The colon carcinoma cell line (HS703T) was provided by Dr
R. B. Owens, and the Ewing sarcoma cell line by the laboratory
of J. E. Fogh (Sloan Kettering Institute). We thank Dr J. M.
Foidart for affinity purified antibodies against laminin and
collagen type IV, Dr S. Tseng for collagen types, G. M. Lui for
technical assistance, Mr Harvey Scodel for assistance with the
manuscript, and Professor Zvi Fuks for support and discussion.
This work was supported by a grant from NIH to D.G.
Received 30 June; accepted 6 November 1980.
1. Vlodavsky, I., Lui, G. M. & Gospodarowicz, D. Cell 19,607-616 (1980).
2. Nelson-Rees, W. A., Hunter. L., Darlington, G. J. & O'Brien. S. J. Cyrogener. CellGenet. (in
the press).
3. Klebe. R. J. Nature 250, 248-251 (1974).
4. Yamada, K. M. & Olden, K. Narure 275, 179-184 (1978).
5. Timpl, R., Rohde. H., Robey, P. G., Rennard, S. I., Foidart. 1.-M. & Martin, G. R. J. bid.
Chem. 254,9933-9937 (1979).
6. Foidart, J. M. er el. Lab. lnvesr. 42, 336-342 (1980).
7. Murray, J. C.. Stingl. G.. Kleinman, H. K.. Martin, G. R. & Katz, S. I. J. Cell Biol. 80,
197-202 (1979).
8. Alitalo, K.. Kurkinen, M.. Vaheri. A.. Krieg, T. & Timpl, R. Cell 19, 1053-1062 (1980).
0028-0836/81/04030643$01.00

9. Birdwell, C. R.. Gospodarowicz, D. & Nicolson, G. Proc. narn. Acad. Sci. U . S . A . 75,
3273-3277 (1978).
10. Jaffe, E. A. & Mosher, D. F. J. exp. Med. 147,1779-1791 (1978).
11. Vlodavsky, I.,Johnson, L. K., Greenburg. G. & Gospodarowicz, D. J. Cell Biol. 83,
4 6 8 4 8 6 (1979).
12. Gospodarowicz, D., Greenburg, G., Foidart, 1.-M. & Savion. N. J. Cell Physiol. (in the
press).
13. Kramer, R. H. & Nicolson, G. L. Proc. narn. Acad. Sci. U.S.A. 76,5704-5708 (1979).
14. Sm~th,H. S., Riggs, J. L. & Mosesson, M. W. Cancer Res. 39,41384144 (1979).
15. Liotta, L. A., Tryggrason, K.,Garbisa, S.. Hart, 1. & Fo1ts.C. M. Narure 284,67-68 (1980).
16. Grobstein, C. Cancer Insr. Monogr. 26,279-299 (1967).
17. Wessels, N. K. in Tissue Inreracrions and Development, 213-229 (Benjamin, Menlo Park,
California, 1977).
18. Gospodarowicz, D., Delgado. D. & Vlodavsky, I. Proc. narn. Acad. Scl. U.S.A. 77,
4 0 9 4 4 0 9 8 (1980).
19. Gospodarowicz, D. & Ill, C. R. Proc. narn. Acad. Sci. U . S . A . 77,2726-2730 (1980).

cis-Interacting genes in the

S region of the murine
major histocompatibility complex
James Michaelson, A r t u r o Ferreira
& Victor Nussenzweig
Irvington House Institute and Department of Pathology, New York
University Medical Center, 550 First Avenue, New York,
New York 10016

Insight into the control of gene expression may be gained by

analysing genetic systems marked by both regulatory and structural variants. In such systems one can determine whether a
regulatory element controls structural genes on both
chromosomes or only on the chromosome to which it is linked.
The latter may be detected in individuals heterozygous at both
the regulatory and structural loci, in which case the effect of each
regulatory allele is seen to be exerted only on the cis-located
structural allele. In prokaryotic organisms, the identification of
cis interaction of this sort has allowed elucidation of many
features of genetic regulation, first for the lac operon1 and
subsequently for a variety of other systems2. In higher
organisms, however, there have been few opportunities to
observe cis-interacting genes3-13. The most thoroughly characterized mammalian system in this regard is the murine Pglucuronidase locus described by Paigen and his ~ollea~ues"'~,
in which cis interaction has been shown to occur between two
closely linked genetic elements-the P-glucuronidase structural
gene itself and an androgen-activated regulatory gene which
controls the quantity of P-glucuronidase expressed. We report
here that cis-interacting genetic elements are also found in the S
region of the mouse major histocompatibility complex H-2.
@ 1981 Macmillan Journals Ltd

Nature Vol. 289 22 January 1981

Table 1 S-region y-chain variants detected by IEF
p1 of y chain
7.4
6.9
6.5
6.7
6.5
Results are summarized from refs 25-27

Mix
The S region controls two structurally similar serum protein~'~.'',SS(which has been shown to be C4 of the complement
system1g-2z)and Slp (sex-limited protein). Ss(C4) and Slp are
each composed of three chains, a (molecular weight (MW)
lOO,OOO), p (MW 75,000) and y (MW 35,000); in both cases the
three subunits are derived from 200,000-MW precursor^'^^^'.
Ss(C4) may be found in two phenotypes-high and low quantity
in plasma (a 20-fold difference)-with the S region containing
the genes determining phenotype17.24.We have analysed the
Ss(C4) and Slp proteins for biochemical variants which would
serve as markers of the structural genes themselves. Recently,.
we described such variants (herein called p i
identified by the isoelectric point of the y chain (Table 1). The
genetic determination of isoelectric point was found to map to
the S region, where the previously described regulatory genes
are located. Although the most straightforward explanation of
the pZ polymorphism is that of allelic variants of the structural
genes, this has not been proved directly and other explanations
are possible. Peptide mapping experiments to resolve this question are now underway. In any event, this need not hinder the
formal argument of cis interaction between pI genes and the
genes controlling expression.

Genotype

Haplotype

High :pI7.4

High : p16.9

Low : p16e9

(H-2$x ~ - 2 ~ ) ' ) ~
=1

High:p1 7.4
Low :PI
6.9

Fig. 1 Cylindrical IEFgelsof the y chain of Ssprotein from H-Z',

H-2d,
H-2k
and (H-2s H-2k)F,
plasma. Ss precipitation was
performed after removing Slp from plasma. Analogous results
were obtained when Ss was analysed by two-dimensional separation (slab IEF. followed bv SDS-oolvacrvlamide ael electro-

* I
(Basic)

Q (Acidic)

Slab IEF ( p H gradient 7.0-3.7) showing Slp y chain of

(DBA/2) mice, and a mixture of the
two samples. The identification of the pI 6.7 and 6.5 spots as y
chains of MW 35,000 was made by applying individual strips to the
top of an SDS-polyacrylamide gel, thus separating the components
according to molecular weight (not shown) (see refs 25-27 for
technical details).
Fig. 2

H-zW7(C3H.W7) and H-2d

In our analysis of Ss(C4) expression, we examined the interaction between two genetic elements: that determining the pI of
the y chain (H-2' = pI 7.4, H-Zk = pI 6.9) and that determining
the quantity of Ss(C4) (H-2' =high, H-2k = low). (H-2' x
H-2&)Fl mice (high: 7,4/low: 6.9) were constructed. It was
reasoned that, if the regulatory and pZ genes interacted in trans
(that is, through a diffusible control substance), equal quantities
of 7.4 and 6.9 would be seen. If the interaction was cis only,
however. a high cluantitv of 7.4 would be seen, with a low
quantity of 6.9, and indeed this was what was found (Fig. 1).
We realize, however, that even though cis interaction is
formally demonstrated, this might be due to differential labelling or breakdown of the two types of molecules. That the two
genes are not precisely the same is seen by the fact that strains of
both Ss high :pl7.4 and Ss high : pl6.9 type have been detected
(Fig. 1). However, we do not know whether the S-region
regulatory and pI genes are separate or overlapping entities (in
terms of genetic fine structure).
We then performed the same type of analysis for the control of
Slp expression. In most Slp' strains of mice (such as the H-2d
strain described here), Slp is found in the sera of males but not
femalesz8.This sex limited expression occurs because androgens
are necessary to induce the Slp protein, as has been shown by
Passmore and Shrefflerz9 (castrated males are Slp-, whereas
females which are injected with testosterone propionate become
Slp' (ref. 29)). Mice of an exceptional haplotype (H-2"'),
however, are Slp' in both males and females because expression
is independent of androgen, as has been demonstrated by
Hansen and Shreffler (such mice continue to produce Slp after
castration, and indeed after being made genetically insensitive
to testosterone by the Tfm mutation3'). Taken together, these
observations identify a hormone-driven regulatory gene
controlling the expression of Slp. It has two allelic forms:
androgen dependent ( A D ) (that is, H-Zd) and androgen
independent (AZ) ( H - P 7 ) . We have found that these two
haplotypes also differ at their pZ genes (H-Zd = pI 6.7, H-Zw7 =
pI 6.5)27(Fig. 2). We constructed (H-Zd x H-2w7)F1mice (AD:
6.7/AI: 6.5). If the interaction between the regulatory and the
p l genes occurs by a trans mechanism, both males and females
would be positive for both pI 6.7 and 6.5 Slp's. If, however,
interaction occurred in a strictly cis fashion, males would be
positive for the pI 6.7 and 6.5 forms, b u t females (lacking
androgens to turn on cis-located pI 6.7 product) would express
only PI 6.5 SIP. Comparison of the isoelectric
(IEF) gels
of Slp from males and females suggests that this is indeed the
case (Fig. 3). Thus, for both Ss(C4) and Slp, the interaction of
regulatory and pZ alleles occurs only when they are located cis.

O 198lNature Publishing Group

Nature Vol. 289 22 January 1981

308
Slab IEF
pl6.7

(Basic)

p1 6.5

(Acidic)

Fig. 3 Phenotype of the y-chain Slp of (C3H.W7 x DBA/2)FI

(H-z~/H-z"') = (AD: pI 6.7)/(AI: pI 6.5), seen on slab IEF gels
( p H gradient 7.7-4.0). (Components other than the y chains are
nonspecific contaminants.)

22.
23.
24.
25.
26.
27.
28.
29.
30.
31.

Curman, B. et al. Nature 258, 243-245 (1975).

Hall, R. E. & Colten, H. R. Proc. nam. Acad. Sci. U . S . A . 74, 1707-1710 (1977).
Hansen, T. H.,Krasteff, T. N. & Shreffler, D. C. Biochem. Genet. 12,281-293 (1974).
Ferreira, A., Michaelson. J. & Nussenzweig, V. Z. Immun. (in the press).
Ferreira, A., Michaelson, 1. & Nussenzweig, V. I, Immun. 125, 1178-1182 (1980).
Ferreira. A,. Michaelson, J. & Nussenzweig, V. Immunogenefics (in the press).
Passmore. H. C. & Shreffler, D. C. Biochem. Genet. 4,351-365 (1970).
Passmore, H. C. & Shreffler, D. C. Biochem. Genet. 5, 201-209 (1971).
Hansen, T. H. & Shreffler. D. C. J. Immun. 117,1507-1513 (1976).
Saunders, D. & Edidm, M. I. Immun. 112,2210-2218 (1974).

Detection of CML determinants

associated with
H-2 controlled E p and E, chains
Antonio Juretik*?, Zoltan A . Nagy* & Jan Klein*

* Max-Planck-Institut fiir Biologie, Abteilung Immungenetik,

Tiibingen, FRG
t Department of Physiology, Faculty of Medicine, University of
Zagreb, Zagreb, Yugoslavia

As seen in these experiments, the use of IEF to detect

S-region structural polymorphism permits the simultaneous
measurement of both allelic gene products in heterozygotes.
From such data, inferences can be drawn about the properties of
The serologically detectable molecules encoded by the H-2
the genes which control the expression of the variants, as, for
complex, the major histocompatibility complex (MHC) of the
example, cis restriction of S-region regulatory genes. Further
mouse, fall into two c l a s s e s ~ l a sIsand class I1 (ref. 1).The class
experiments of this sort should reveal in greater detail the
I molecules encoded by the K and D loci have a molecular
properties of these regulatory genes. Two features of S region
weight of 44,000 and are noncovalently associated with flz
expression are noteworthy in this regard. First, because
androgens control the serum levels of Ss(C4) ( q ~ a n t i t a t i v e l ~ ) ' ~ microglobulin which is not controlled by the MHC. The class I1
and Slp (absolutely), the analysis of these two proteins may well
molecules are of two kinds, A and E, each consisting of two
noncovalently associated polypeptide chains, cu (M. -34,000)
elucidate some features of the nature of hormone regulation of
and p (M, -28,000)"3. Three of the four chains, A,, A, and E ,
gene activity. Second, Ss(C4) and Slp have been shown to be
are controlled by loci in the I-A subregion, whereas the locus
synthesized in macrophages and liver parenchymal cells, but not
controlling the E, chain is located in the I-E subregion of the
in a variety of other cell types3'. Presumably, there are genetic
H-2 complex. Thus the loci coding for the E, and E, chains are
elements which control this pattern of synthesis, but where they
separated by at least one (J) and perhaps more
It has
are or how they exert their effects is unknown. These regulatory
been shown that the E, and E, chains are synthesized
elements would be identifiable by tissue-specific genetic polyindependently, and that the E, chain is required for the
morphism of Ss(C4) and Slp, if indeed such variation exists. The
insertion of the E, chain in the plasma membrane4. We demonpossibility of such polymorphism is suggested by the observation
strate here that the E,En complex (the E molecule) can evoke
of Roos et a1." that in vitro synthesis of Slp is seen in macroin vitro cell-mediated lymphocytotoxicity (CML) without prephages of H-2"' but not H-2' or H-Zd genotype, although
vious sensitization in vivo, and that the strong CML reactivity is
mice of all three types have the protein in their sera. The detailed
directed against allele-specific determinants on the highly
analysis of this sort of polymorphism, including the use of IEF to
polymorphic E, chain.
test for cis and trans interaction with pI genes, should be
To generate CML against the E molecules, we used three
informative both for the specific case of the S region and with
strain combinations selected from the H-2 recombinant pool
respect to the general question of the control of gene expression
such that the responders differed genetically from the stimuladuring differentiation.
tors only in the region between the A and S loci (Table 1). The
We thank Drs E. A. Boyse, L. Silver and D. Bennett for
region may contain, in addition to the loci controlling E, and E,
critical review of the manuscript, and Dr Yuri Bushkin for help
chains, the B, J, C and perhaps other, as yet unidentified loci.
with the IEF technique. We also thank Ms Holly Yudkowitz, Mr
Four lines of evidence support the conclusion that the CML is
Victor La Regina and Mr Daniel Eichinger. This work was
directed against E rather than against other molecules encoded
supported by NIH grants AI-13224, AI-00270 and CA-09149.
by the A-S interval. (1) When tested on a panel of H-2 recombinant strains, reactivity of some of the effector cells with some
Received 23 June; accepted 16 October 1980.
of the strains excludes the B, J and C loci from coding for the
target-cell determinants. (2) The same panel test indicates that
1. Jacob, F. & Monod, J. I molec. Biol. 3, 318-356 (1961).
2. Miller, E. (ed.) The Operon (Cold Spring Harbor Laboratory, New York, 1979).
the strain distribution pattern of the determinants detected by
3. Phillips, R. J. Genetics 5 1 , 4 8 5 4 9 5 (1966).
the effector cells correlates with the distribution pattern of
4. Glekeler, W., Faversham, I. & Cohn, M. I. exp. Med. 148, 1122 (1978).
5. Dickinson, W. J. & Carson, H. L. Prac. nam. Acad. Sci. U . S . A 76,45594562 (1979).
alleles at E,- and E,-encoding loci. (3) The CML typing data fit
6. Dickinson. W. J. Devl Biol. 42, 131-140 (1975).
well with what is known about the expression of E,- and
7. Boubelik, M.,Lengerova, A,, Bailey, D. W. & Matousek, A. DeuIBiol. 47,206-214 (1975).
8. McCaron, M. e t a / . Genetics 91,275-293 (1979).
E,-encoding loci3s4.(4) Cytotoxicity generated across disparities
9. Daniel. W. L. Biochem. Genet. 14,1003-1018 (1976).
from the I-B to S region is specifically blocked by a monoclonal
10. Carlson, P. Genet. Res. 17, 53-81 (1971).
antibody against the Ia.7 determinant on the E, chain5.
.
113-122 (1974).
11. Jarry, B. & Faulk, D. Molec. gen. G ~ n e r 135,
12. Rawls, J. W. & Frtstrom, F. W. Nature 255, 738-740 (1975).
The CML was generated by incubating unprimed spleen cells
13. Lewis, E. B. Nature 276,565-570 (1978).
of the responder strain in a standard mixed-lymphocyte culture
14. Paigen, K., Meisler, M., Felton, I. & Chapman. V. Cell 9, 533-539 (1976).
15. Swank. R. T. et al. Recent Prog. Horm. Res. 3 4 , 4 0 1 4 3 8 (1978).
(MLC) with y-irradiated spleen cells of the stimulator strain for
16. Paigen. K.. Labarca. C. & Watson. G. Science 203, 554-558 (1979).
5 days. After incubation, the effector cells were tested on a panel
17. Shreffler, D. C. & Passmore, H. C. Genetics 48.9 (1963).
18. Roos, M., Atkinson, I. P. & Shreffler, D. C. J Immun. 121, 1106-1115 (1979).
of lipopolysaccharide (LPS)-stimulated, 5'~r-labelled target
19. Lachmann, P. J., Grennan, D. & Mart~n,A. Nature 258, 242-243 (1975).
cells and specific release of radioactivity was taken as a measure
20. Meo, J., Krasteti. T. & Shreffler, D. C. Proc. natn. Acad. Sci. U.S.A. 72,4536-4560 (1975).
21. Ferreira, A,, Nussenzweig, V. & Gigli, I. I exp. Med. 148, 1186-1197 (1978).
of cell lysis. In all three strain combinations, reproducibly high
0028-0836/81/040308-03$01.00

0 1981 Macrnillan Journals Ltd

Nature Vol. 289

309

22 January 1981
Table 1 Panel testing of primary CML generated against EgE, molecules

Haplotypes
k
a, a1
h l , h2, h3
m, m l

Target cell H-2

Regions
K
I-A
(A,&)
(Ep)

aq 1
yl

k
k
k
k
k
4

tl

a 14

an 1
as 1
h4
i3, i5

s
k
k

k
k
k
k
k
k
k
k
k
k
k
k

by 1

6
b
b
d
d
d
d

b
b
h
d
d
d
d

i7
b
g l , ~3
01
d
g2
13, 14

k
k
k
k
k
k
k
k
k
k
(k)
(k)
h
(6)
(6)
d
d
d
(dl

t2

(s)
b)

I-E
(Em)

7
7

k
d
b

7
7
7
7
7
7
7
7
0

0
7
0
0
7
7
7
0
7
0
0

1
'0 Net "Cr release by effectors
B1O.GDaBlO.
BlO.A(4R)
DZ(R101)
uB lOaA(2R)
(anti-~g7)
(anti- EL^)
12
12
8(4-10)4
10
11
8
NT
30(27-33)2
9(6-12)'
NT
NT
4(3-5)2
3(0-6)'
8(6-1
4
I(-2-3)'
21(16-26)6
15(12-16)'
20(16-24)'
06
11(10-12)2
5(4-5)3
3'

d
b
d
b

f
F

d
d
h
b

k
d
b
d
d
S

37(25-45)3
41(34-52)~
38(28-47) "
39
36(27-44)2
40(34-44)3
33(24-43)'
39(3840)2
40(35-45)'
3(-2-7)4
07
3 l(2 1-44)6
11(7-14)3
8(5-1
19(16-22)*
16(15-16)2
17(15-18)'
2(-3-6)4
19(11-34)16
5(-1-12)'
3(1-6)"

CML were set up with l o x lo6 responder and 8 x 10' irradiated stimulator spleen cells in 4.5 ml of Alpha minimal essential medium (Gibco)
M 2-mercaptoethanol. After 5 days in culture, the cells were tested for lysis of S'~r-labelledLPS-blasts
containing 10% fetal calf serum, and 3 x
at different effector: target (E:T) ratios. For designation of target-cell haplotypes see ref. 8. H-2 regions coding for potential CML determinants are
shown. Alleles of the EDlocus in brackets mean failure of cell-surface expression of the relevant polypeptides. Only plateau levels of killing, obtained
at E:T 100: 1 to 160: 1, are shown. At high E : T ratios, the OO
/ net release, that is, the difference between OO
/ specific killing of various targets and %
specific killing of responder haplotype targets, showed little variation between individual experiments, and was characteristicof the CML combination
tested. (Calculationof OO
/ specific release was as in Fig. 1 legend.) This fact allowed pooling of data from several experiments. The data are expressed as
mean and range of OO
/ net release obtained in 2-16 experiments. Superscript numbers show the number of individual experiments (not shown where
target haplotypes were tested once).
specific lysis was observed. However, in each combination the
reaction was unidirectional-reaction
was obtained in the
direction BlO.A(4R) anti-BlO.A(2R) (hereafter 4Ra2R),
BlO.S(7R) anti-BlOS(9R) ( 7 R a 9 R ) and B1O.GD antiBlO.D2(R101) ( G D a R 1 0 1 ) ; all attempts t o generate effector
cells in the opposite direction have failed. This unidirectional
response agrees well with present knowledge concerning the
expression of E molecules in the plasma membrane. Jones et aL4
demonstrated that H-2 haplotypes b, q, s and perhaps also f
failed t o produce functional E, chains and that this failure
prevents the E, chains from being inserted into the plasma
membrane. Thus cells of these four haplotypes bear n o E
molecules o n their surface. In the strain combinations used, one
of the strains in each pair always expresses E molecules o n the
cell surface and the other does not, and we obtained C M L only
when the responder was the nonexpressor and the stimulator the
expressor-never in the opposite direction. In the combination
4 R a 2 R , unidirectional response was previously observed in
MLC, and a complicated explanation involving unequal crossing-over and gene deletion was proposedh. Such explanations
are n o longer necessary; the nonexpression of the E molecule in
the membrane fully explains the data-the
&encoding gene
cannot be deleted in the BlO.A(4R) strain because the E p chain
is present in the cytoplasm of this nonexpressor.
When tested o n a panel of strains carrying independent and
recombinant H-2 haplotypes (Table I), the three effector-cell
types gave two kinds of reactivity, one that was comparable in
strength with that obtained against the stimulator strain, and one
that yielded approximately half as much specific 51Crrelease as
the lysis of stimulator-type targets (Fig. 1). W e shall refer t o
these two types of response as 'strong' and 'intermediate',
respectively. The distinction between the strong and intermediate reactions was also confirmed by cold-target inhibition
studies (example shown in Fig. 1).
Table 1 also shows the distribution of alleles at the E,- and
ED-encoding loci among the tested strains. T h e E, locus is
shown to possess only two alleles, one designated '7'because it

I
.
1
7

30:l
60:l
120:l
Elkctors. targets

80:l
&I
8:l
Competitors: targets

Fig. 1 Reactivity patterns of BlO.S(7R) anti-BIOS(9R) effector

cells. a, Direct cytotoxicity;6, cold-target inhibition of cytotoxicity
against BlO.S(9R) targets. Targets or competitors: A, t1 (A.TL);
0, t2 (BlOS(7R)); V, t3 (B1O.HTT); 6, t4 (BlO.S(9R)); A, i3
(BlO.A(3R)); 0,
a(A/J); W, a(B1O.A); V, a ( ~ . ~ l ax,~ both
);
targets and competitorsare t2. Per cent "Cr-release was computed
using the formula:
c.p.m. supernatant X 2
x 100
2 x c.p.m. cell lysate +c.p.m. supernatant
where the denominator represents total counts per well, when the
sample size is half-volume supernatant and half-volume zaponinlysate. Per cent specific "Cr release was calculated by subtracting
per cent release in control groups from that in experimental groups.
The controls consisted of labelled target cells mixed with normal
spleen cells at ratios corresponding to E:T ratios in the experimental groups. Spontaneous release (control groups) ranged from
17 to 35%. depending on the target strain.

O 198lNature Publishing Group

Nature Vol. 289 22 January 1981

codes for the serologically detectable antigen Ia.7 (ref. 7) and
another designated '0' because it codes for the absence of Ia.7 in
the plasma membrane (however, in reality, there may be only
one allele at the E,-encoding locus and the expression as
opposed to nonexpression of the E, chain in the membrane may
be controlled by another locus). The distribution of the 7 allele is
that determined by biochemical methods and serological typing
for Ia.7 (refs 7, 8).
The E,-encoding locus is shown to possess at least as many
different alleles as there are independent H-2 haplotypes. This
conclusion is based on biochemical data which demonstrate that
the E, chains controlled by independent H-2 haplotypes differ
in their peptide maps3 (the E, alleles b, s and k have not yet been
distinguished serologically, probably because correct strain
combinations for the production of the relevant antibodies have
not been used).
The CML combinations 7Ra9R and GDaRlOl gave strong
CML reactivity only against target cells carrying the 7 allele at
the E,-encoding locus and, at the E,-encoding locus, the same
allele as the stimulator cell (Table 1). As the E, chains of most
Ia.7-positive strains tested derive from H-2k, the data strongly
indicate that these two types of effector cell recognize allelespecific (private) determinants on the E, chain. The third
combination 4Ra2R gave strong CML against target cells carrying the k and b, but not the d and s alleles at the E,-encoding
locus, together with the 7 allele at the E,-encoding locus. The
panel test in Table 1also demonstrates that recognition of the E,
determinants is not restricted by class I antigens, as target cells
that share the E molecule with the stimulator cells, but differ
from them at the K and D region, are lysed with equal efficiency.
We conclude that the allelic forms of the E, chain are not only
chemically distinct, but also differ functionally, in terms of
recognition by cytotoxic T lymphocytes.
The intermediate CML reactivities in all three combinations
depend on the presence of E molecules in the target cells' plasma
membrane, and thus correlate with the expression of Ia.7 (E,
chain) by the cells (Table 1). Our most recent results suggest that
the determinants detected by intermediate CML reactivity are
also found on the E, chain. So far, direct contribution of the E,
chain to the formation of CML determinants could not be
demonstrated5. Target cells which carry the 0 allele at the
E,-encoding locus, and thus do not express the E molecule on
the membrane, are not lysed. The marginal killing of b and i7
target cells by 4Ra2X effectors is an apparent exception;
however, the absence of this reactivity on g2 target cells maps
the determinant(s) recognized to the K or I-A region.
We have also obtained CML reactivity against the E molecule
in the strain combination BlO.D2(R107) anti-BlO.A(3R). This
CML combination involves multiple genetic disparities at the E
and C regions, the Tla locus and possibly also at other loci to the
right of the S
This effector-cell type seemed to preferentially recognize target cells of the i3 and i5 hapolotypes,
which express the b allele at the E,-encoding locus together
with the 7 allele at the Em-encodinglocus. In addition, they
exhibited cross-reactivities independent of the recognition of
the E molecule. The genetic basis for these cross-reactions is
being investigated.
In conclusion, our results, together with other data1',
demonstrate that the two gene-controlled E molecule serves as a
target antigen for recognition by cytolytic T lymphocytes. The
CML determinants seem to be localized primarily, if not
exclusively, on the highly polymorphic E, chain. The implications of these findings with regard to the function of the E
molecule in immune responses will be discussed in detail elsewhere". But as the E, and E, chains also bear serologically
detectable determinants12, participate in Ir-gene control of the
immune response and immune suppression", and stimulate
mixed-lymphocyte reaction14, there is obviously no need to
postulate separate loci for all these functions.
This work was supported in part by grant Wa 139/10/AI.5
from the Deutsche Forschungsgemeinschaft. We thank Ms
Gudrun Labib for technical assistance.
0028-0836/81/0403l0--03$01.00

Received 25 July; accepted 21 November 1980.

1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.

Klein. 5. Science 203, 516-521 (1979).

Cullen, S. E., Freed, J. H. & Nathenson, S. G. Transplanrn Rev. 30,236-239 (1976).
Uhr, I. W., Capra, J. D., Vitetta, E. S. & Cook, R. G. Science 206,292-297 (1979).
Jones, P. P., Murphy, D. B. & McDevitt, H. 0. J. exp. Med. 148,925-939 (1978).
JuretiC, A. er al. Immunogenerics (submitted).
Lozner, E. C., Sachs, D. H., Shearer. G. M. & Terry, W. D. Science 183,757-759 (1974).
David, C. S. & Shreffler, D. C. Transplanrarion 18, 313-321 (1974).
Klein, I., Flaherty, L., VandeBerg, J. L. & Shreffler, D. C. Immunogenerics 6, 489-512
(1978).
Egorov, J. K., Apt, A. S. & Vedernikov, A. A. Immunogenerics 5,285-290 (1977).
Fischer Lindahl, K. & Hausmann, B. Immunogenerics (in the press).
Klein. J. er al. Narure (submitted).
Lafuse, W. P., McCormick, J. E.& David, C. S. I. exp. Med. 151, 709-715 (1980).
Benecerraf, B. & Germain. R. N. Immun. Rev. 38,70-119 (1978).
Peck, B. A,, Klein, J. & Wigzell, H. I . J. Immun. 125, 1078-1086 (1980).

Nonenzymatic hydroxylations
of proline and lysine
by reduced oxygen derivatives
Robert L. Trelstad, Karen R. Lawley
& Lewis B. Hoimes
Experimental Pathology and Embryology Laboratories, Shriners
Burns Institute,
Departments of Pathology and Pediatrics, Massachusetts General
Hospital, and
Harvard Medical School, Boston, Massachusetts 02114

The biosynthesis of trans-3- and trans-4-hydroxyprolines and

5-hydroxylysine in animal cells requires polypeptide proline or
lysine, enzymes and cofactors including iron, and possibly
involves peroxidatic intermediates1. Several laboratories have
reported the presence of low-molecular-weight hydroxyproline
and hydroxylysine peptides in cell and organ culture^^-^. We
found that these small peptides contained the trans-3 and cis-4
isomers of hydroxyproline as well as trans-4 ones and that their
production was not completely inhibited by a,a-dipyridyl, an
iron chelator and effective inhibitor of enzyme-mediated hydroxylations5. It is known that oxygen or hydrogen peroxide
in the presence of metal can hydroxylate proline and other
aromatic compoundsb1'. We show here that reduced oxygen
derivatives can hydroxylate both free and polypeptide-bound
proline and lysine, and that scavengers of hydroxyl radicals
suppress, but do not completely inhibit, this reaction. Reduced
oxygen derivatives can be generated in normal and pathological
circurn~tances~~,
and some of the low-molecular-weight hydroxyproline and hydroxylysine peptides found in cell and organ
cultures might be derived from these derivatives and therefore
do not reflect collagen turnover, but rather some other cellular
activity.
Free proline and lysine and polypeptide-bound forms of each,
as well as three isomers of hydroxyproline, were exposed to
several different reaction conditions (listed in Table I), which
contained or could generate superoxide anion, hydrogen peroxide or hydroxyl-free radicals12.
The reaction products were identified using the following
techniques: high-voltage electroph~resis'~,
thin-layer chromatography14,chemical derivatization and partitioning by solvent
extraction followed by chromatophore generation (including
characterization of the spectral absorption pattern of standards
and the derivatives)'', nitrous acid resistance16, and behaviour
on the amino acid analyser, including the reaction product
absorption ratios at 440 and 570 nm and co-chromatography
with known standards in at least two different elution conditions
(Fig. 1).
The products of the reactions involving proline include at least
four different hydroxyproline isomers-trans-3-hydroxy@ 1981 Macmillan Journals Ltd

311

Nature Vol. 289 22 January 1981

Table 1 Nonenzymatic hydroxylations of proline and lysine in free and polypeptide forms
Reaction products*
Substrate
Free forms:
L-proline
L-lysine

Polypeptide forms:
Pro-Gly
Pro-Gly + HNOz
Gly-Pro
Gly-Pro-Ala
Lys-Lys
Lys-Gly
Albumen

trans3-~YP

Reaction conditions
Standardt
Standard except at 20 OC
Standard except at 0 O C
Standard except ~ e ' +replaced at same
concentration by
zn2+
cu2+
~ g " , ca2', ~ n ' + M~'',
,
pb2+
Standard except proline increased fourfold
Standard except ~ e increased
~ '
10-fold
Standard with the addition of free radical
scavengers:
50 pM mannitol
50 pM sodium formate
5 pM thiourea
10 pM sodium benzoate
50 pM potassium chloride
50 pM dimethyl sulphoxide
100 pM dimethyl sulphoxide
Standard with the addition of iron chelators:
10 pM a,a-dipyridyl
300 pM diethylenetriamine penta-acetic acid
Standard without HzOz and with
hypoxanthine/xanthine oxidase:
with no other additions
with added catalase or
superoxide dismutase
Standard
Standard
Standard
Standard
Standard
Standard
Standard
Standard
Standard
Standard

trans4-~YP

cis3-~YP

3.2
1.5
1.3

2.7
1.9
1.5
2.0
f
2.3
7.3

cis4-~YP

2.2
4.6
(rrans-3 + unidentified)
f
f
3.1
6.4
15.3
22.8

3.0
f
3.3
30.6

2.4
1.9
2.2
1.9
3.1
1.3
0.4

1.8
1.5
2.2
1.6
3.3
0.4
0.2

3.8
3.4

3.3
3.4

2.3
0.08
0.05

5.2
0.3
0.3
100.0

TR

10.5
2.0
5.7
2.3
TR

TR, trace. trans-3-hyp, trans-3-hydroxyproline; trans-4-hyp, trans-4-hydroxyproline; cis-3-hyp, cis-3-hydroxyproline; cis-4-hyp, cis-4-hydroxyproline; 5-hylys, 5-hydroxylysine.
* Data were determined by amino acid analyses without hydrolysis (free forms) or after hydrolysis in 6 M HCl (polypeptideforms).The presence of
the metal ions did not interfere with the analyses. The values represent the moles of product formed per mole of proline or lysine (or cis-4 or mans-4)
remaining at the end of the reaction x100. Overall recovery of proline/lysine was less than 10%. The values represent averages of two to six
determinations.
t Standard conditions were: 20 pM substrate, 0.2 mM ascorbate, 10 pM FeS04.7H20, 1.0 mM H202,all in 0.1 M NaH2P04,pH 4.4, at 37 OC for
1 h, followed by reduction with 1 mM NaBH4 at pH 8.0.
f These metals with proline gave no reaction; with lysine, 50% of ~ e " .
proline, trans-4-hydroxyproline, cis-3-hydroxyproline and
cis-4-hydroxyproline. No differences were noted when D- o r
L-proline was used in free form. When free proline was the
substrate, the relative amounts of each of these isomers
produced were nearly equal in the conditions used for most of
the experiments. However, with peptide-linked proline, there
were significant changes in the distribution of the isomers
produced. For example, free proline yields approximately equal
amounts of trans-3-hydroxyproline and trans-4-hydroxyproline, whereas peptide-bound proline is predominantly
converted to trans-4-hydroxyproline (Table 1). The proline
sequence in peptide linkage also affected its relative reactivity
and, interestingly, the configuration which is normally hydroxylated in collagen-Pro-Gly-was
approximately five times more
hydroxylated on a molar basis of starting substrate concentration than Gly-Pro o r Gly-Pro-Ala. The proline o r lysine
recovered at the end of the reaction was less than 10% of the
starting material. In addition to the products listed in Table 1,
other products were noted and of particular interest was a
substance which eluted from the amino acid analyser 7 min

before trans-4-hydroxyproline, which we have tentatively

identified as 4-ketoproline". This material was separated from
trans-4-hydroxyproline o n the amino acid analyser by adjustment of the pH of the first buffer to 2.68 (Fig. 1). Several other
products, similar in chromatographic properties o n the amino
acid analyser to the several isomers of the 3,4-dihydroxyprolines, were detected, but not studied f ~ r t h e r " " ~ .
T h e reaction products generated with free lysine as substrate
included 5-hydroxylysine and a second substance which eluted
1 0 min later, which has not been identified (Fig. 1).With lysine
as substrate in polypeptide form, the major product was 5hydroxylysine.
W e suspect that 3- and 4-ketoproline are intermediates in the
reaction. The 3-keto form apparently reduces readily in the
presence of ascorbate as trans-3-hydroxyproline consistently
appeared without the addition of sodium borohydride. The
4-keto form was variably present in the initial reaction conditions, but could be quantitatively eliminated by the addition of
the sodium borohydride. The 4-keto intermediate was also
noted when free cis-4-hydroxyproline was used as substrate.

O 198lNature Publishing Group

Nature Vol. 289 22 January 1981

312
SUBSTRATE PPROLINE
ELUUON BUFFER pH268

-440nm

7 4-Keloprollne

-- 570nm

Trans-4 HYPO

Received 30 April; accepted 16 October 1980

II
I I

SUBSTRATE LYS/M
ELUTlON BUFFER,pH528

11

5 - ~ ~ d r o x ~ l ~ 1 n e f Ii I
II

cis-4-hydroxyproline, may be true biosynthetic products.

Because cis-4-L-hydroxyproline is an analogue of L-proline and
effective inhibitor of stable collagen triple-helix formation28,its
production in vivo would provide a novel mechanism for
regulating net collagen production and add to the growing list of
biological effects of oxygen radicalsz9.
This work was supported by NIH grants HL 18714,
HL 23591, GM 21700 and H D 09689, The March of Dimes
Birth Defects Foundation and the Shriners Burns Institute. We
thank Mary G. Kirby for technical assistance and Dr Charles
Cintron for help with the thin-layer chromatography and highvoltage electrophoresis.

,I

\\

40

50

60

70

80

90

Elution time (min)

Fig. 1 Elution patterns of reaction products on the amino acid
analyser of non-enzymatic hydroxylations of free proline (upper
tracing), polypeptide-bound proline which was hydrolysed to free
amino acids and then treated with nitrous acid (middle tracing) and
free lysine (lower tracing). The solid tracing is the absorbance at
440 nm, the dotted tracing that at 570 nm. The pH 2.68 buffer was
0.18 M in ~ a + the
, pH 5.28 buffer was 0.35 M in ~ a ' . The
identification of the eluting material by means other than cochromatography with standards is detailed in the text. The two
peaks with ? have not been characterized, but that with proline is
probably 4-ketoproline.

When this reaction mixture was reduced, trans-4-hydroxyproline was generated. The experiments with hydroxyl radical
scavengers, coupled with the fact that no hydroxylations occurred in the absence of added oxygen derivatives, indicate that
some form of a reduced oxygen derivative is essential for the
hydroxylation reaction to occur. The dependence of the reaction
on the presence of metal ions is consistent with a reaction
mechanism involving hydroxyl radicals, although the precise
nature of the reactants in such mixtures is not ~ n d e r s t o o d ' ~ .
Reduced oxygen derivatives have been reported in widely
differing circumstances in nature. Of particular interest to
matrix biology is the finding that a major effect of these oxygen
radicals is to alter extracellular macromolecules. For example,
covalent cross-linking via dityrosines in sea urchin fertilization
membranes2', marine mussel byssal fibres2', collagenz2,elastin23
and r e ~ i l i n have
~ ~ been reported. Depolymerization of
hyaluronic acid has been observed by viscometry following
exposure to oxygen radicals2'. Recently, it has been shown that
plasma exposed to superoxide becomes chemotactic for neut r o p h i l ~ our
~ ~ ; data suggest that the chemotactic response of
fibroblasts and monocytes to hydroxyproline-containing
pep tide^^^ may occur in vivo in the absence of collagen breakdown at, for example, sites of inflammation where reduced
oxygen derivatives, which we have now shown can hydroxylate
non-collagenous peptides, are being p r o d ~ c e d ' ~ . ' ~ . ~ ~ .
The present data indicate that hydroxyproline- and/or hydroxylysine-containing peptides could be produced in vivo
by a nonenzymatic reaction involving reduced oxygen
derivatives. If this is supported by further studies, then some of
the isomers or hydroxyproline which are generated, such as the

1. Myllyla, R., Schubotz, L. M., Weser. U. & Kivirikko, K. Biochem. biophys. Rex. Commun.
89.98-102 (1979).
2. Steinberg, J. L CeNSci. 12, 217-234 (1973).
3. Nourse, P. N., Nourse, L. D. & Botes, H. S. Afr. I. Sci. 70, 231-234 (1974).
4. Bienkowski, R. S., Cowan. M. I., McDonald. J. A. & Crystal, R. G. I. biol. Chem. 253,
43564363 (1978).
5. Holmes, L. B. & Trelstad. R. L. Devl Biol. 72.41-49 (1979).
6. Chvapil. M. & Hurych, J. Nature 184, 1145 (1959).
7. Lamport, D. T. A. Narure 202,293-294 (1964).
8. Yip, C. C. Biochim. biophys. Acfa 92,395-396 (1964).
9. Gruber, H. A. & Mellon, E. F. Anolyf. Biochem. 66.78-86 (1975).
10. Udenfriend,S.,Clark, C. T., Axelrod, J. & Brodie, B. B . 1 biol. Chem. 208,731-739 (1954).
l I. Breslow, R. & Lukens, L. N. I. biol. Chem. 235,292-296 (1960).
12. Hayaishi, 0. & Asada, K. (eds) Biochemical and Medical Aspects of Acfive Oxygen
(University Park Press. Baltimore. 1977).
13. Cintron, C., Peczon, B. D. & Kublin, C. L. Analyf. Biochem. 87,622-630 (1978).
14. Villanueva, V. R. & Lederer, E. FEES Letf. 52,308-31 1 (1975).
IS. Kiv~rikko,K. I. In!. Rev. Connecrrue Tissue Res. 5,93-163 (1970).
16. Dziewiatkowsk~,D. D., Hascall, V. C. & Riolo, R. L. Analyf. B~ochem.49,550-558 (1972).
17. Mauger, A. B. & Witkop, B. Chem. Rev. 66.47-86 (1966).
18. Irreverre, F. & Witkop, B. I. Chromar. 43, 127-128 (1969).
19. Michelson, A. M., McCord, J. M. & Fridovich, I (eds) Superoxide and Superoxide Dismurases (Academic, London, 1977).
20. Foerder, C. A. & Shap~ro,B. M. Proc. nam. Acad. Sci. U.S.A. 74,42144218 (1977).
21. DeVore, D. P. & Gruebel, R. J. Biochem. biophys. RPS.Commun. 80,993-999 (1978).
22. Waykole, P. & Heidemann, E. Connecfiv~Tissue Res. 4, 219-222 (1976).
23. LaBella. F., Keeley, F., Vivian. S. & Thornhill, D. Biochem. biophys. Rer. Commun. 26,
748-753 (1967).
24. Andersen, S. D. B~ochrm.biophys. A r f a 93,213-215 (1964).
25. McCord, I. M. Science 185, 529-531 (1974).
26. Petrone, W. F., English. D. K., Wong, K. & McCord, J. M. Proc. nam. Acad. Sci U.S.A. 77,
1159-1163 (1980).
27. Postlethwaite, A. E., Seyer,J. M. & Kang, A. H. Proc. nam. Acad. k i . U.S.A.75,871-875
(1978).
28. Rosenbloom. J. & Prockop. D. J. I. biol. Chem. 245, 3361-3368 (1970).
29. Fridovitch, I Science 201,875-880 (1978)

Biosynthetic labelling of platelet

activating factor from radioactive
acetate by stimulated platelets
H. Chap*, G. Mauco*, M. F. Simon*, J. Benvenistet
& L. Douste-Blazy*
* INSERM Unite 101, HBpital Purpan, 31059 Toulouse, France
t INSERM Unite 200, 32 rue des Carnets, 92140 Clamart Paris,
France

Platelet activating factor (PAF-acether) is a chemical mediator

released by basophils in in uitro and in uiuo conditions of
immunolo~icalchallenge'-5. PAF-acether induces aggregation
and secretion of ~Iateletsfrom different s ~ e c i e s ~ and
. ~ " could
thus be implicated in various immunological forms of tissue
Other sources have been described for this mediator,
including macrophages10 and platelets". As the amounts of
PAF-acether released in the latter case are sufficient to support
the platelet aggregation1' and as platelet aggregation by PAFacether does not require ADP secretion or thromboxane
biosynthesis12, PAF-acether was proposed as the mediator of
the putative third pathway of platelet aggregation". Structural
analysis of PAF-acether by lipases indicated that it could be a
glycerophospholipid devoid of an ester linkage at the 1-position
and bearing a fattv acid esterified at the 2-po~ition'~.
This has
provided the first-evidence for a phosphoiipid displaying the
properties of a powerful chemical mediator. Based on several
-

@ 1981 Macmillan Journals Ltd

313

Nature Vol. 289 22 January 1981

lines of evidence, l-O-alkyl-2-acetyl-sn-glyceryl-3-phosphorylcholine (l-alkyl-2-acetyl-GPC) was recently proposed as
the chemical structure of the mediator, hence the term PAFacether14'15. A s PAF-acether synthesis occurs rather early in
activated platelets, one of the most efficient pathways of PAFacether formation could involve a rapid acetylation of the
corresponding lysoderivative (l-alkyl-GPC). W e now present
evidence that platelets activated by ionophore A23187 are able
to incorporate acetate into a mixture of l-alkyl-2-acetyl-GPC
and l-acyl-2-acetyl-GPC; the former supports the biological
activity of PAF-acether.
Rabbit blood was drawn on to E D T A (5 mM) by intracardiac
puncture and washed platelets were prepared according to
Ardlie et a1.16. Final suspension ( l o 9 platelets ml-') was in
Tyrode's
buffer
( p H 7.35) containing
1 mM MgC12,
1.25 mM CaC12and 0.35% (w/v) fat-free bovine serum albumin
(Sigma). Platelets were incubated at 37 "C for 1 0 min under
stirring in the presence of 250 bCi ml-' 3H-acetate
(12 Ci mmol-', CEA) and 2.5 FM A23187 (Lilly). Lipids were
extracted according to Bligh and Dyert7. The lower chloroformic phase was washed three times with methanol/water ( 1 : 1
by vol.) to remove excess free acetate. An aliquot of the lipid
extract from both control and A23187-treated platelets was
subjected to TLC on silicagel F 254 precoated plates (Merck)
using chloroform/methanol/water (70 :35 :7 by vol.) as a
solvent. The chromatogram was divided into 1 7 bands which
were directly scraped into scintillation vials containing 1 0 ml of
Picofluor (Packard) and radioactivity was determined by scintillation counting. Figure 1 shows that A23187 induced the
appearance of a radioactive lipid migrating very close to sphingomyelin. This chromatographic behaviour is identical to that
previously reported for PAF-acether".
T o characterize it further, the newly formed radioactive
compound was then purified from a similar lipid extract
by two steps of HPLC with a Lichrosorb Si 6015 column
(25 cm x 0.9 cm) at a flow rate of 5 ml min-'. Using acetonitrile/methanol/water (60:30: 10 by vol.) as a first solvent, a

Fig. 1 Distribution of radioactivity after TLC of 3~-acetate-

labelled lipids from platelets. The lipid extracts from control (e)
and A23187-treated (0)
platelets were chromatographed as
described in the text. The lower part of the graph represents the
position of major, iodine-stainable lipids and the localization of the
different bands scraped into scintillation vials. Each point
represents the radioactivity measured in the corresponding band.
Results are representative of four identical experiments. 0,origin;
SP, sphingomyelin; PC, phosphatidylcholine; PE, phosphatidylethanolamine; NL neutral lipids; F, solvent front. The
arrow indicates the position of a pure standard of PAF-acether.

Retention time (min)

Fig. 2 Identification by HPLC of acetic acid as the hydrosoluble
compound released by phospholipase A2 treatment of purified
3~-acetate-labelledplatelet phospholipid. Preparation of 'Hacetate-labelled platelet phospholipids has been reported in the
text. An aliquot (40,000 d.p.m.) was incubated at 37 "Cfor 15 min
with 5 IU N. naja phospholipase A2 as previously reported1'.
After partition according to Bligh and ~ ~ e rto' 80
~ ,~1 of the
aqueous upper layer were added 5 ~1 of pure formic acid (a), 5 p1
acetic acid (6) and 10 ~1 propionic acid (c). The mixture was then
subjected to HPLC using a column of RP 18-10pm (25x
0.46 cm). Elution was performed according to Farinotti etai.19 at a
flow rate of 0.5 ml min-' and 0.5-ml fractions were collected.
Carrier acids were detected by absorbance at 206 nm (. . . . . .) and
radioactivity (-)
was determined by liquid scintillation
counting.

radioactive peak was eluted at 23 min. The pooled fractions

(which corresponded to PAF-acether on TLC) were rechromatographed with hexane/propanol2/water (39 : 52 : 9 by vol.). In
the latter system, a radioactive peak was detected at 30 min and
was clearly separated from other UV-detectable lipids. O n TLC,
the purified compound exhibited the same RF as previously
observed and it co-migrated with l-alkyl-2-[3H]acetyl-GPC and
l - a ~ ~ l - 2 - [ ~ H ] a c e t ~ l - G(see
P C Table 1 legend for their
synthesis).
The biological activity of the purified compound was checked
on rabbit platelets suspended in Tyrode's buffer (pH7.35)
containing gelatin (0.2S0/0, w/v) as previously described3. The
compound displayed a strong aggregating activity, which still
occurred in the presence of creatine phosphate/creatine phosphokinase (4 m M and 1 0 U ml-', respectively) o r when using
platelets incubated for 15 min with 1 m M aspirin before washing
(in these conditions, platelets became unresponsive to 1 0 FM
sodium arachidonate).
These results strongly suggest that stimulated platelets
synthesize radioactive PAF-acether from labelled acetate.
Further biochemical identification was performed by subjecting
the purified compound to various treatments. After basecatalysed m e t h a n ~ l ~ s ifollowed
s'~
by neutralization with HCI
and phase partition according to Bligh and Dyer17, over 90% of
the radioactivity was found in the aqueous upper layer. The
same result was obtained on incubation with Naja naja phospholipase A, in previously reported conditions". In these
conditions, it was verified by HPLC" that the aqueous radioactivity coincided with free acetic acid (Fig. 2). These data
confirm that acetic acid was esterified in the 2-position of the
isolated phospholipid. However, incubation of the compound

O 198lNature Publishing Group

Nature Vol. 289 22 January 1981

Table 1 Percentage of hydrosoluble radioactivity after lipase treatment of 3~-acetate-labelledphospholipids
- - -

No. of
experiments Controls

Lipase

Platelet 3 ~ - p h o s p h o 6
16.2*7.1 60.0k9.6
lipid
~ - ~ C ~ I - ~ - [ ~ H ] ~ C ~5 ~ ~ I -16.2*
G P C8.0 79.5 k4.9
~ - ~ I ~ ~ I - ~ - [ ~ H ] ~6C ~ ~ 29.4k9.7
~ I - G P C3O.9* 8.6

P
<0.001
<0.001
NS

The 3~-acetate-labelledphospholipid isopolar to PAF-acether was

purified from a platelet lipid extract as described in the text. 1-AcylGPC was obtained by phospholipase A2 treatment ( C r o t a l u s adamanteus) of phosphatidylcholine isolated from eggz6 and then purified by
TLC. This and synthetic 1-hexadecyl-rac-glyceryl-3-phosphorylcholine
(Medmark) were acetylated as described by Demopoulos et a1.I4 in
chloroform containing 3 ~ - a c e t i canhydride (Amersham) diluted with
cold anhydride to a specific radioactivity of 1 mCi crmol-l. The two
products were then purified by HPLC. Lipase treatment was performed
at 37 "C for 60 min in 0.01 M Tris-HC1 buffer ( p H 7.4) containing 0.15
M NaCl and 0.01 MCaCI2. The incubation medium contained in
0.25 ml :0.02 ml of labelled phospholipid in ethanol (20,00040,000 d.p.m.) and 0.025 ml of either lipase from R. a r r h i z u s (Boehringer) suspended in 3.2 M (NH4)zS04 or 3.2 M (NH4)2S04 in the
control tubes. Lipids were then partitioned according to Bligh and
~ ~ e and
r " an aliquot of both lower (organic) and upper (aqueous)
phase was determined for radioactivity by liquid scintillation counting
using Picofluor (Packard). Results (mean *1 s.d.) represent the percentage of radioactivity recovered in the upper phase. P, probability of
significance according to Student's t-test. NS, not significant.

with 4 M HCI at room temperature for 3 h (ref. 5) did not modify

its phase partition, excluding the occurrence of a vinyl-ether
linkage at the 1-position of the phospholipid.
More striking results were obtained following treatment of
the radiolabelled compound by lipases. As shown in Table 1,
60% of the radioactivity was water soluble after incubation with
lipase from Rhizopus arrhizus. This value was close to 80%
when incubating l-a~~l-2-[~H]acetyl-GPC,
whereas no change
in phase partition occurred with 1-hexadecyl-2-['H]acetylGPC. Identical behaviour was obtained when using a lipase with
high phospholipase A, activity purified from guinea pig
pancreas (donated by J. Fauvel and M. J. Bonnefiszo). These
results indicate that the radioactive phospholipid that we isolated from A23187-stimulated platelets could correspond to a
mixture of I-acyl-2-acetyl-GPC and 1-alkyl-2-acetyl-GPC, the
former being converted to hydrosoluble 2-acetyl-GPC through
lipase hydrolysis. Indeed, TLC results have shown that the
water-soluble compound released by lipase treatment of the
3H-labelled platelets was the same as that obtained from l-acyl2-[3H]acetyl-GPC. It remained at the origin and was clearly
separated from PAF-acether and from free acetate, the latter
migrating with the solvent front. In contrast the chromatographic behaviour of 1-alkyl-2-acetyl-GPC was unchanged
after lipase treatment, thereby illustrating the specificity of this
enzyme towards the ester bond at the 1-position of phosphoglycerides2'. However, the lower organic phase contained
all the original biological activity after such treatment,
confirming previous data obtained on PAF-acether from pig
leukocytes13.We therefore conclude that only 1-alkyl-2-acetylGPC supports the PAF-acether activity released by A23187stimulated platelets. This agrees with the finding that the 1-acyl
derivative exhibits a much lower activity than the 1-alkyl
deri~ative'~.".
The present study is an attempt to characterize the biochemical pathways of PAF-acether synthesis in a living cell.
More recent results indicate that a similar incorporation of
radiolabelled acetate occurs in stimulated murine macrophages2'. However, several aspects of acetate incorporation in
platelet PAF-acether remain unclear: (1) Are the concentrations of acetate used (20 pM) relevant to available intracellular acetate? Our study does not provide quantitative information; we simply expected the radioactive precursor to label the
intracellular pool of acetate, the precise amount of which is
0028-0836/8l/~40314--O3$0I.OO

unknown. (2) Is the acetylation reaction an enzymatic process

involving, for instance, acetylcoenzyme A as an intermediate?
This should be the object of further studies. (3) What is the
origin of the phospholipid acceptors? 1-alkyl-GPC is released
from hog leukocytes23and various other cell sourcesz4,including
platelets, but we know of no accurate published report on the
putative precursor 1-alkyl-2-acyl-GPC in these cells. The 1alkyl- and 1-acyl-lysophospholipidscould then act as acetyl
acceptors for acetate, leading to the formation of a molecule
with high or low activity. This would agree with the suppression
of PAF-acether synthesis by bromophenacyl bromide, a phospholipase A, inhibitor2'. However, further studies are necessary
to characterize the mechanism and the specificity of acetate
incorporation into these phospholipids.
This work was supported by INSERM (CRL 78.5.148.3). We
thank Dr L. McGregor for correcting the manuscript.
Received 23 July; accepted 31 October 1980.
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.

Siraganian, R. P. & Osler, A. G. J. Immun. 106, 1244-1251 (1971).

Henson, P. M. & Cochrane. C. J. I. exp. Med. 133, 554-571 (1971).
Benveniste, I., Henson, P. M. & Cochrane, C. G. J. J. exp. Med. 136, 1356-1377 (1972).
Benveniste. J. Nature 249, 581-583 (1974).
Pinckard, R. N., Farr, R. S. & Hanahan, D. J. I. Immun. 123.184771857 (1979).
Benveniste, J., Le Couedic, J. P. & Kamoun, P. Lancer i, 344 (1975).
Fesus, L., Csaba, B. & Muszbek, L. Clin. rxp. Immun. 27, 512-515 (1977).
Henson, P. M. & Pinckard, R. N. Monogr. Allergy 12, 13-26 (1977).
Benveniste, I., Camussl, J. & Polonsky, J. Monogr. Allergy 12, 138-142 (1977).
Mencia-Huerta. J. M. & Benveniste. J. Eur. I. Immun. 9, 409-415 (1979).
Chignard, M., Le Couedic, 1. P., Tence, M., Vargaftig, B. B. & Benveniste, 1. Nature 279,
799-800 (1979).
Cazenave, J. P., Benveniste, 1. & Mustard, J. F. Lab. Invest. 41, 275-285 (1979).
Benveniste, I., Le Couedtc, J. P.. Polonsky, J. & Tence, M. Nature 269, 170-171 (1977).
Demopoulos, C. A., Pinckard, R. N. & Hanahan. D. J. J. biol. Chem. 254,9355-9358
(1979).
Benveniste. J. e t a l . Cr. hebd. Acad. S c i , Paris 289,1037-1040 (1979).
Ardlie, N. G., Packham, M. A. & Mustard, J. F. Br. J. Haemat. 19.7-17 (1970).
Bligh, E. G. & Dyer, W. J. Can. I. Biochem. Physiol. 37,911-918 (1959).
Tence, M., Polonsky, J , Le Couedic, J. P. & Benveniste, J. Biochimie 62,252-259 (1980).
Farinotti, R., Caude, M., Mahuzier, G. & Rosset, R. Analusis 7 , 4 4 9 4 5 3 (1979).
Fauvel. J. & Bonnefis. M. I. (In preparation).
Slotboom, A. J.. de Haas, G. H., Bonsen. P. P. M., Burbach-Westerhuis. G . J. & van
Deenen. L. L M. Chem. Phys. Lipids 4. 15-29 (1970).
Mencia Huerta. J. M., Roubin. R. & Benveniste. J. (in preparation).
Polonsky. J.. Tence, M., Varenne, P., Das, B.. Lunel. 1.& Benveniste, I. Proc. nam. Acad.
Sci. U . S . A . (in the press).
Benveniste. J. (in preparation).
Mencia-Huerta. J. M., Akerman, C. & Benveniste. 1. Fedn Proc. 39,691 (1980).
Papahadjopoulos, D. &Miller, M. Blochim hiophys. Acra 135, 624-638 (1967).

Effects of ATP and vanadate on calcium

efflux from barnacle muscle fibres
Mark T. Nelson & Mordecai P. Blaustein
Department of Physiology, University of Maryland, School of
Medicine, Baltimore, Maryland 21201
Department of Physiology and Biophysics, Washington University,
School of Medicine, St Louis, Missouri 63110

Calcium ions carry the inward current during depolarization of

barnacle muscle fibres and are involved in the contraction
process1. Intracellular ionized calcium ([Ca2+],) in barnacle
muscle, as in other cells, is kept at a very low concentration,
againsta large electrochemical gradient2.'.~hislarge gradient is
maintained bv Ca2+ extrusion mechanisms. When ICa2+li is
below the contraction threshold4, Ca2+ efflux from &ant
barnacle muscle fibres is, largely, both ATP dependent and
external Na+(Naf) dependent (see also refs 5,6). When [Ca2+],
is raised to the level expected during muscle contraction (25 @M)', most of the Ca2+ efflux from perfused fibres is Nao
dependent; as in squid axonss, this Naf -dependent Ca2+efflux is
'K
ATP independent. Orthovanadate is an inhibitor of (Na++)
ATPase9 and the red cell Ca2+-ATPaselO.We report here that
vanadate inhibits ATP-promoted, Nai-dependent Ca2+ efflux
from barnacle muscle fibres perfused with low [Ca2+Il (0.20.5 pM), but has little effect on the Naf-dependent, ATPindependent Ca2+ efflux from fibres with a high [Ca2+], (2Q 1981 Macmillan Journals Ltd

315

Nature Vol. 289 22 January 1981

5 pM). Nevertheless, ATP depletion or vanadate treatment of

high [Ca2+],fibres causes an approximately 50-fold increase of
Ca2+efflux into Ca2+-containinglithium seawater. These results
demonstrate that both vanadate and ATP affect Ca2+extrusion,
including the Na:-dependent Ca2+efflux (Na-Ca exchange), in
barnacle muscle.
The intracellular solute composition of giant muscle fibres
from the barnacle, Balanus nubilus, was controlled by internal
perfusion". Intracellular [Ca2'] was maintained with Ca-EGTA
buffers containing 2 o r 8 m M E G T A and appropriate concentrations of CaCI, (see figure legends). The perfusion fluids also
contained caffeine, and the mitochondrial u n c o u ~ l e r ,carbonylcyanide p-trifl~oromethoxyphenyl hydrazone -(FCCP), to
minimize Ca2' seauestration in sarco~lasmicreticulum and
mitochondria, respectively. Caffeine anb FCCP d o not directly
affect Ca2' efflux from internally perfused barnacle muscle
fibres (unpublished data). A T P levels are controlled with a
regeneratkg system consisting of phosphoenolpyruvate and
~hosohokinase. Perfused fibres d o not contract even when
[ca2'], is raised above the normal contraction threshold
(-0.9 F M ) ~ .Thus, Ca2' efflux measurements can be made on
fibres containing Ca2' levels that would normally be expected to
cause contractures (> 1 p,M)'.
As in squid axonsR,Nai-dependent Ca2' efflux from barnacle
muscle fibres is promoted by A T P when [Ca2'Ii< 1 p M (ref. 5).
Ca efflux from ATP-depleted fibres perfused with 0.2-0.5 FM
[Ca2'], was -0.3 pmol cmW2s ' (n = 3) and insensitive to the
removal of external Na'. Similarly, Ca2' efflux from ATPfuelled, vanadate-treated fibres perfused with 0.2-0.5 p M
[Ca2'], was -0.2 pmol cm-2 s-I (n = 3) and was insensitive to
the removal of external Na'. However, as shown in Fig. 1, in a
representative fibre perfused with A T P and 0.2 p M [Ca2'],,
Ca2' efflux was about 0.75 pmol cm-2 s-I. This efflux declined to
about 0.35 pmol
s-I when external Na' was replaced by

Na+,
Li,

ATP.

SOr

Time (h)
h
Na,

Li,
Ca,
ATP,

Time (h)

Fig. 2 a, Effects of external ~ a and

' Li', and internal ATP on

4SCa2+

efflux from a barnacle muscle fibre perfused with high

[ca2'Ii. b, Effects of external Na' and Li' on 45~a2'effluxfrom the
same fibre in the presence gf ATP (last part of Fig. 2a expanded).
Internal perfusion with fluid containing 45~a2',2 mM caffeine,
25 pM FCCP, 2 mM EGTA and 1.95 mM Ca ( [ C ~ ~ ' ] ~ pM)
=S
was begun 200 min before data shown here were obtained. After
250 min of perfusion, 2 mM ATP and an ATP-regenerating system
were added to the perfusion fluid. The fibre was exposed to
456 mM Na' seawater, except during the periods indicated, when
the seawater contained 456 mM ~ i ' . Resting potential, -47 mV.
Temperature, 16 OC. Fibre diameter, 1.4 mm.

Time (h)

Fig. 1 Effects of external Na+ and Tris, and internal vanadate on

45

ca2' efflux from a fibre perfused with low [ca2'Ii. The composition of the perfusion fluid was: 189 mM K glutamate, 10 mM Na
glutamate, 38 mM KCI, 375 mM sucrose, 8 mM MgCI2, 8 mM
EGTA, 4.1 mM CaC12 ([~a"]~=0.2&M), 3mM Na2 ATP
(Sigma), 3 mM phosphoenolpyruvate (Sigma),8 wg ml-I pyruvate
kinase, and 40 mM HEPES. The perfusion fluids were buffered to
pH 7.3 with Tris base. A Ca-EGTA stability constant of 7.6x
I O ~ M - ' was used to calculate free calcium concentrations
([ca2+li).The composition of the standard external solution (Nai)
was (in mmol I-'): 456 NaCI, 6 Tris, 10 KCI, 11 CaCI, and 32
MgS04. All external solutions were buffered to pH 7.8 at 20C
with maleic acid. Osmolalatiesof all internal and external solutions
were 930-980 mosmol kg-'. Tris or Li' (see Figs 2 and 3) was
substituted isosmotically for Na' in the Na-free seawaters
(indicated by Tris of Lio in the bars at the tops of the figures).
Internal perfusion with 4SCa,2 mM caffeine, and 25 kMFCCP was
begun 80 min before the data shown here were obtained. After
150 min of perfusion, 1 mM vanadate was added to the perfusion
fluid. Resting potential, -56 mV. Temperature, 16C. Fibre
diameter, 1.4 mm.

Tris. When 1 m M vanadate was added to the perfusion fluid,

Ca2' efflux into Na' seawater fell to -0.2 pmol
s-'. These
data demonstrate that Ca2' efflux from fibres containing low
[Ca2'Ii is Na: dependent, and can be inhibited by vanadate.
In fibres containing a high [ca2+li, bathed in Ca2'-free
media, Li' and Tris (and choline) could b e used interchangeably
as external Na' replacements (compare with Fig. 2). However,
in fibres with a low [Ca2'],, Li' seems to substitute partially for
external Na' in supporting Ca2' effluxI2. Therefore, Tris was
used to replace Na' in fibres with a low [Ca2'], (Fig. 1).
In squid axons containing high [Ca2'], (>2 FM), most of the
Ca2' efflux is Na; dependent, but ATP independent8. In addition, in high [Ca2'Ii, ATP-depleted axonss, the replacement of
external Na' by Li' increases Ca2' efflux two- to fourfold. This
Lii-stimulated efflux is also completely dependent on external
Ca2'. Ca2' efflux from perfused barnacle muscle fibres exhibits
similar properties. Namely, when ATP-depleted fibres are perfused with hi h [Ca2+Ii,the replacement of external Na' by Li'
increases C 8 + efflux about 10-fold (from -5 to -5Opmol

O 198lNature Publishing Group

Nature Vol. 289 22 January 1981

316

demonstrate that the Na:-dependent Ca2' efflux (presumably

Na'-Ca2' exchange) is blocked by vanadate when it is ATP
dependent (in low [Ca2'], fibres), but not when it is ATP
independent (in high [Ca2'], fibres)". Clearly, vanadate cannot
be used to distinguish Nai-dependent from Nai-independent
Ca2' efflux (contrast with ref. 13).
Resting barnacle muscle fibres have a [Ca2'Ii of about 0.050.2 pM (ref. 14). The threshold for contraction is about 0.50.9 pM (ref. 4), and maximal isometric force is developed with a
10-fold higher [Ca2'], (ref. 7). Our data demonstrate that the
Ca2' efflux is Nai-dependent throughout the normal physiological range of [Ca2'], values for these fibres. When [Ca2'Ii is
below the contraction threshold (< -0.5 pM)4, the Ca2' efflux is
largely ATP dependent and vanadate sensitive. A substantial
fraction of this efflux (-6O0l0) depends on Na: when [Ca2'Ii
0.2 pM (Fig. 1). At the [Ca2'], levels expected at the peak of
contraction (-5 pM)', about 9O0/0 of the Ca2' efflux is Nai
dependent; but this (Nai-dependent) efflux is neither ATP
dependent nor vanadate sensitive.
The mechanism responsible for the residual Ca2' efflux into
Na'-free media is uncertain. This efflux could, in part, be a
passive Ca2' and Ca-EGTA leak. However, Ca2' efflux from
ATP-fuelled fibres into Na'-free media was greater than Ca2'
efflux from ATP-depleted or vanadate-treated fibres (even with
external Na' present; compare with Fig. 1). This difference in
efflux may represent an ATP-dependent, vanadate-sensitive
'uncoupled' Ca2'~pump13.Alternatively, this efflux may simply
represent an exchange that is activated by other monovalent
cations (for example, K', Li', choline or Tris)" and/or by the
residual external Na' located in the spaces enclosed by surface
membrane invaginations.
Our experiments do not show how ATP affects Na'-Ca2'
exchange. However, as vanadate inhibits a number of
ATPasesl', our data may infer that the influence of ATP on Naidependent Ca2' efflux involves a phosphorylation step. This is
consistent with evidence that the non-hydrolysable P - y
methylene analogue of ATP cannot substitute for ATP in
promoting Ca2' efflux from the squid axons16. Unfortunately,
ATP hydrolysis possibly associated with Na'-Ca2' exchange
would be difficult to detect in barnacle muscle fibres which
contain many ATPases.
Our data are consistent with the view8 that ATP increases the
affinity of the Na'-Ca2'
exchange system for external Na'
and/or internal Ca2'. Moreover, the effects of ATP and
vanadate on the Ca2' efflux in high [Ca2'Ii fibres may indicate
that ATP increases the Na'/Li' selectivity ratio of the carriers,
thereby inhibiting Li; + Gag'-dependent Ca2' efflux. Regardless of the uncertainties, our data demonstrate that Na' ions,
probably acting via an Na'-Ca2' exchange mechanism, are
important for regulating Ca2' in barnacle muscle fibres in physiological conditions.
We thank Dr E. Santiago for technical assistance, Dr L. Reuss
for a supply of orthovanadate, and Mrs B. Brooks for preparing
the typescript. This research was supported by MDAA,
NINCDS, and NSF. M.T.N. was supported by an NIH training
grant.

Time (h)
Nao
Li,

Ca,
ATP,

01

Time (h)
Fig. 3 a and b, Effects of external Na', ~ i ' ,and ca2' on 4 5 ~ a 2 '
efflux from an ATP-fuelled fibre perfused with high [ca2'],. The
end tail of the experiment of a is shown on an expanded scale in b.
Internal perfusion with 45~a21,3mM ATP and an ATPregenerating system, 1.2 mM vanadate, 2 mM caffeine, 8 m M
EGTA and 7.8 mM ca2' ([ca2'Ii=5 FM) was begun 160 and
210 min before data shown in a and b, respectively, were obtained.
The fibre was bathed in 456 mM ~ a ' ,except for the periods noted,
when the seawater contained 456 mM ~ i ' . The seawater also
contained 11 mM ca2', except for the periods indicated, when
M~'' replaced ca2' in the seawater. Resting potential, -45 mV.
Temperature, 16 "C. Fibre diameter, 1.5 mm.

s -1., see Fig. 2a). This Li,'-stimulated efflux is also entirely

Cag' dependent-the removal of Caz' reduces Ca2' efflux into
Li' seawater (Li SW) to -1 p m o l ~ m - ~ s - '(not shown).
Moreover, the addition of ATP to the perfusate reduces Li,'+
Ca;'-dependent Ca2' efflux (Fig. 2a), and the subsequent
replacement of Li,' by Na,' increases Ca2' efflux to about the
levelseen in the absence of ATP (-4 pmol cm-2 s-'; Fig. 2a, b).
Vanadate mimics the effect of ATP depletion in ATP-fuelled
fibres perfused with high [Ca2'],: it promotes a large Ca2' efflux
in Li SW (Fig. 3 a : compare with Fig. 2a). Lig'-stimulated Ca2'
efflux from vanadate-treated, ATP-fuelled fibres, like Lig'stimulated Ca2' efflux from ATP-depleted fibres, is completely
dependent on Cai' (Fig. 3a).
Figure 3a, b also shows that Ca2' efflux from vanadatetreated, high [Ca2'Ii fibres can be promoted by Nai in the
absence of Cag'. In summary, the data from Figs 1 and 3a
cm-2

Received 25 August; accepted 30 September 1980.

1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.

Atwater. 1. el al. I. Physiol. Lond. 243,523-551 (1974).

Baker, P. F. Ann. N . Y . Acad. Sci. 307, 250-268 (1978).
Blaustein, M. P. Rev. pkysiol, biockem. Pharmac. 70, 34-82 (1974).
Ashley. C. C. A m . Zool. 7,647-659 (1967).
DiPolo. R. & Caputo. C. Biockim. biopkys. Acta 470,389-394 (1977)
Russell, J, M. & Blaustein, M. P. I. gen. Physiol. 63, 144-167 (1974).
Ashley, C. C. A n n . N . Y . Acad. Sci. 307,308-329 (1978).
Blaustein. M. P. Biopkys. 3. 20. 79-111 (1977).
Cantley, L. C. etal. 3. biol. Chem. 252,7421-7422 (1977).
Bond, G. H. & Hudgins. P. M. Fedn Proc. 37. 315 Abstr. (1978).
Nelson. M. T. & Blaustein, M. P. I. gen. Physiol. 75, 183-206 (1980).
Dipolo, R. eral. I. gen. Physiol. 67, 433-467 (1976).
DiPolo. R. el 41. Narure 281,228-229 (1979).
Hagiwara, S. & Nakajima, S. I. gen. Physiol. 49, 807-818 (1966).
Simons, T. J. B. Narure 281,337-338 (1979).
DiPolo, R. 3. gen. Physiol. 69, 795-813 (1977).

O 198lNaturePublishing Group

Nature Vol. 289 22 January 1981

Co-release of enkephalin and

catecholamines from
cultured adrenal chromaffin cells
Bruce G. Livett, D e a n n e M. Dean, Larry G. Whelan,
Sidney Udenfriend* & Jean Rossier* t
Division of Neurology, T h e M o n t r e a l G e n e r a l Hospital a n d McGill
University, Montreal, Quebec, C a n a d a H3G 1A4
* R o c h e Institute of Molecular Biology, Nutley, N e w J e r s e y 07110

The opioi&peptides Leu-enkephalin and Met-enkephalinl'z are

stored intraneuronally in the brain3 where they are thought to
act as neurotransmitters and/or neuromodulators4. Evidence for
their release from nerve terminals has come from biochemical
and pharmacological studies in uitro with brain tissue slicess"
and synaptosomes6. Enkephalins also exist in the peripheral
nervous system in nerve cell bodies and axon terminals in the
gastrointestinal tract7, sympathetic ganglias-" and adrenal
gland1'. In the adrenal gland, high levels of enkephalins are
present both in axon terminals of the splanchnic nerve1' and in
the adrenal medullary chromaffin
where they are
stored to ether with the catecholamines in the chromaffin
granules' 16. Stimulation of the adrenal gland in u i ~ o ' ~ or
. ' ~the
perfused gland in uitro14 causes release of catecholamines and
enkephalins into the adrenal vein. However, it is not clear
whether the origin of the released enkephalins is the adrenal
medullary chromaffin cells or the enkephalin-containing
splanchnic nerve terminals that innervate the medulla. We now
show that enkephalin and catecholamines are released together
from primary cultures of bovine adrenal medullary chromaffin
cells by nicotine in a Caz+-dependent manner.
Primary cultures of bovine adrenal medullary chromaffin cells
provide a useful experimental system for investigating the
release of catecholamines uncomplicated by diffusion barriers
present in intact tissue. For example, this cell culture system has
been used to characterize the pharmacological release of catecholamines from the cells by nicotinic agonistsI8, to investi ate
the role of substance P".", somatostatin" and enkephalinsk as
modulators of these nicotinic receptors, and to demonstrate the
presence of Leu- and Met-enkephalin in high concentrations in
the processes and terminal varicosities that characterize these
adrenal paraneurones2'. For the present study, primary cell
cultures of adrenal medullary cells were established as
monolayers on collagen-coated plastic tissue culture disheslg
(for further details see Fig. 1 legend). After 8 days in culture, the
cells were processed for immunohistochemical localization of
Leu-enkephalin as previously described2'. Preimmune rabbit
serum, rabbit anti-Leu-enkephalin serum (R164B) preabsorbed
with Met-enkephalin and fluorescein-conjugated goat antirabbit serum used in this study were obtained from Dr R. P.
Elde. The specificity of the Leu-enkephalin antiserum has been
characterized by radioimmunoassay and immunofluorescence21.Strong immunofluorescence specific for leu-enkephalin
(Fig. 1b) was observed in the perikarya and varicose processes of
approximately 30% of.the adrenal paraneurones whereas over
80% of the cells stained positively for catecholamines (Fig. l c )
by the glyoxylic or formaldehyde-glutaraldehyde techn i q u e ~ ' ~ . ~The
' . immunoreactive Leu-enkephalin had a fine
granular disposition within the cytoplasm and processes and
exhibited an increasing proximal to distal ratio of fluorescence
consistent with synthesis in the perikarya with subsequent
transport, processing and storage in the terminals2'.
Catecholamines are normally released from the adrenal gland
in vivo following activation of acetylcholine (ACh) receptors on
the chromafin cells by ACh released from splanchnic nerve
terminals innervating the medulla. In a similar manner, phar-

t Present

address: Laboratoire de physiologic nerveuse. CNRS.91 190 Gif sur Yvette. France.

Table 1 Calcium-dependent release of Leu-enkephalin and catecholamines from

adrenal paraneurones
% Cellular content released

Secretory stimulus

Noradrenaline

Nicotine 5 x
M
Nicotine 5 x lo-' M+Ca2',
2.2 mM

1.4k01
20.9** 1.4

Adrenaline
1.0k0.2
10.4tr0.8

Leuenkephalin
ND
8.2*t* 1.3

The release studies were carried out on 10 culture plates (6 days old) each
containing 1 x lo6 chromaffin cells grown in the presence of Bacitracin (see Fig. 1
legend) in 1 ml medium. Five plates served as controls (nicotine but no Ca2') and
the other five contained nicotine and 2.2 mM CaZ'. Each plate was washed at 37 "C
for I 5 min in 1.0 ml of a HEPES-buffered Krebs-Ringer saline (KRH) containing
0.2% bovine serum a l b ~ m e n and
' ~ the plates were then incubated in the release
experiments for three successive 5-min intervals in: 1 ml of KRH medium alone
(0-5 min); KRH medium containing nicotine (5-10 min); KRH medium alone,
'washout' sample (10-15 min). At the end of each 5-min period, the incubation
media were removed from the cells, placed on ice and the new incubation medium
pre-equilibrated to 37 "C added to the cells. A 100-p1 aliquot of the medium
removed from the cells was acidified with 25 p1 2 M perchloric acid for catecholamine estimation and the remaining 900 pI acidified with 900 p1 2 M acetic
acid for subsequent extraction, chromatography and analysis of Leu-enkephalin.
After 15 min the cells were scraped off the plate in 0.5 ml of ice-cold KRH with a
rubber policeman and the extract acidified as above for assay of catecholamines
and Leu-enkephalin. Assay of endogenous catecholamines: Endogenous catecholamines (both within the cells and released into the media) were assayed by the
two-wavelength differential spectrofluorimetric procedure2%f Renzini er ~ 1 . ' ~ .
Noradrenaline and adrenaline levels were obtained by reference to standards
after solving simultaneous equations for fluorescence of the samples at 410
and 455 nm with a program (written by Mr Serge La Fontaine) for a digital
PDP 11/40 computer. The variability of triplicate assays of the same sample was
*5%. These day-6 cultures contained 7.6k0.8 (n = 10) kg adrenaline per lo6
cells and 2.9+0.5 (n = 10) fig noradrenaline per lo6 cells (molar ratio
adrenaline :noradrenaline = 2.4). Assay of endogenous Leu-enkephalin: The
freeze-dried extracts of cells and media were reconstituted in 10 mM sodium
phosphate buffer pH 7.5, 145 mM NaCl supplemented with thiomersal 0.01%,
gelatine 0.1% and crystalline bovine serum albumen 0.01%. The extracts were
assayed by a radioimmunoassay using a C-terminal-directed Leu-enkephalin
antiserum. The characteristics of the C-terminal Leu5-enkephalin antiserum (RB
92) have been described elsewhere23.This antiserum has a cross-reactivity of 3%
with Met5-enkephalin, and is quite specific for the C-terminus of Leu-enkephalin.
In terms of molar ratio it has a cross-reactivity of 3% with Met-enkephalin2' and
<1/10,000 with Leu-enkephalin-Arg6 and Met-enkephalin Argb. Note that
although the radioimmunoassay is highly selective for Leu-enkephalin, Metenkephalin is generally present in amounts several times greater than Leuenkephalin, and is also released" and will therefore necessarily be detected in the
radioimmunoassay. However, the Leus-enkephalin immunoreactivity in the cell
extracts and in the media was further characterized by gel filtration and HPLC and
was found to correspond to authentic Leucenkephalin. The amount of endogenous catecholamines and Leu-enkephalin contained in the cells at the beginning
of each experiment was estimated for each culture plate by summing the ahounts
in the media at the three time periods and adding it to that remaining in the cells at
15 min. The release of endogenous catecholamines and Leu-enkephalin from the
cells into the medium during the 5-min stimulation period is expressed as a
percentage of the total content in the cells at the beginning of the experiment. The
results give the mean *s.e.m. (n = 5) where the percentage of catecholamines and
Leu-enkephalin released by nicotine stimulation has been compared by Students
r-test. ND, not detected.
*P<0.001; +not significant at P c 0 . 2 .

macological studies on bovine adrenal paraneurones have

shown that 'H-noradrenaline taken up and stored in reserpinesensitive sites can be released from the cells by activation of
nicotinic (but not muscarinic) ACh receptorsIg. The results in
Table 1 show that in the presence of nicotine (ED5,, 5 x
M)
the cells released 13.2 1h0/0 (n = 5) of their endogenous total
catecholamines over a 5-min period. Nicotine produced a preferential release of endogenous noradrenaline (20.9*1.4%)
over adrenaline (10.4 0.8%). It also released enkephalins from
the cells, and the fraction of the cellular Leu-enkephalin
released (8.2* 1.3O/0) was similar to that of the endogenous
adrenaline released. The release of both catecholamines and
Leu-enkephalin by nicotine was Ca2' dependent (Table 1).
Although Met-enkephalin and enkephalin congeners and precursors contained in the chrornaffin granule^'^-'^ are also
the combination of immunoassay gel filtration and
HPLC method^^^.^^ used in our study were highly selective for
Leu-enkephalin (see Table 1 legend).
A characteristic feature of the nicotinic receptor response is
that it shows activation at low levels of agonist (10-'-10-~ M
nicotine) and inhibition at higher levels ( >
M). In accord

@ 1981 Macmillen Journals Lfd

318

Nature Vol. 289 22 January 1981

ing, 35 mm, lo6 chromaffin cells) or Aclar 33C plastic

'hats' (Allied Chemical 15 mm, 5 x 10' chromaffin cells), and incubated in a humid atmosphere at 36 "C for 6 days before use. The media were replaced at days 2 and
M. Immunofluorescence stainingof the cells (b)with the anti-Leu-enkephalin
4;and on days 5 and 6 Bacitracin (Sigma) was added to a final concentration of 5 x
Furthermore, cultures incubated
serum (R164B,1/90) was shown to be specific by its total inhibition when preabsorbed with synthetic Leu-enkephalin (Penin~ula)~'.
with preimmune serum or with the fluorescent conjugate alone showed no specific fluorescence2'. The cells also contained high concentrations of catecholamine in
fluorescence histochemical technique. The final preparations
their perikarya and varicose terminals as demonstrated by the formaldehyde-glutaraldehyde ('Fagl~')'~
were observed unmounted (for immunofluorescence, in 0.1 M bicarbonate-buffered glycerol; for catecholamine, dry) by incident fluorescence illumination with x 10
or X20 dry Fluortar (Wild-Leitz) objectives using a Leitz Ortholux-I1 fluorescence microscope fitted with a Ploem incident illuminator and dichroic mirror primary
and secondary filters optimal for fluorescein isothiocyanate or for catecholamine fluorescence. The results were recorded on Tri-X (Kodak) film rated at 400ASA.
Exposure times were in the range 5-10 s.

with this general property, the dose-response curve for release

of endogenous catecholamines and enkephalins by nicotine (Fig.
2) showed similar responses with inhibition at high concentrations of agonist. At day 6, the adrenal paraneurones
contained 699* 51 (n = 7) pg Leu-enkephalin per lo6 cells and
20.7* 1.7 (n = 7) pg catecholamines per l o 6 cells: the molar
ratio of catecholamines to Leu-enkephalin released by 5 x
lo-' M nicotine (105 x lo3) was similar to that present in the
cells (94 x 103).
A recent study1' showing that Met-enkephalin circulates in
human plasma as the intact pentapeptide and that its concentration in the adrenal vein is twice that in the internal jugular or
in the peripheral circulation, raises the possibility that enkephalins are secreted together with catecholamines from the

adrenal medulla. This is strongly supported by the parallel

release of catecholamines and enkephalins recently demonstrated in perfused dog adrenal1' and by the present study.
Further, the close correspondence of release of adrenaline and
Leu-enkephalin (Table I), and the previous immunohistochemical evidence13 for co-storage of noradrenaline with
Met-enkephalin in adrenal medullary gland cells may be of
physiological significance. There is increasing evidence that
amines and peptides may co-exist not only in adrenal chromaffin
cells but also in several nekonal systems24.The present demonstration that amines and peptides are released simultaneously
from adrenal chromaffin cells in response to a stimulus raises the
possibility of co-secretion of catecholamines and enkephalins
from adrenergic neurones and the question of the role of this
co-secretion.
This study was supported in part by a grant to B.G.L. from the
Canadian MRC. J.R. was supported by the Institute National de
la SantC et de la Recherche Medicale (France) and D.M.D. by a
pre-doctoral fellowship from the Muscular Dystrophy Association of Canada. We thank Mr Joe Donahue for photography,
Mrs Gwen Landrigan for typing the manuscript, and Dr R. P.
Elde for providing the antisera used for the immunofluorescence
studies.
Note added in proof: We have recently shown that Leu- and
Met-enkephalin are co-stored within adrenaline (but not within
noradrenaline) cells in the bovine adrenal medulla (B.G.L., R.
Day, R. P. Elde and P. R. C. Howe, unpublished). A 'preliminary note' by S. M. Stine, H.-Y. T. Yang and E. Costa
(Neuropharmacology 19, 683-685 [1980]) indicates that Metenkephalin-like immunoreactive material is released from isolated bovine chromaffin cells both spontaneously and in
response to high concentrations of K' (56 mM) and acetylcholine
M).
Received 30 June; accepted 9 October 1980.

Fig. 2 Dose-response curves for release of endogenous catecholamines

and Leu-enkephalin by nicotine. Adrenal paraneurones cultured on 14
Aclar hats were equilibrated for 15 min in KRH medium (containing
2.2 mM Ca2' and 5 x lo-% Bacitracin), and this medium was then
replaced by the same medium but containing the concentrations of nicotine
indicated. At each concentration of agonist, two hats containing 5 x 10' cells
per hat were pooled for analysis. After 5 min at 37 "C, the media were
removed, acidified and chilled on ice and the cells collected for analysis of
endogenous catecholamines and Leu-enkephalin as described in Table 1
legend.

1. Hughes, J. e t a ! . Nature 258,577-579 (1975).

2. Simantov, R. C. & Snyder, S. H. Proc. naam. Acad. Sci. U . S . A . 73, 2515-2519 (1976).
3. Bloom, F.. Battenberg, E.. Rossier. I., Ling, N. & Guilleman, R. Proc. nam. Acad. Sci.
U . S . A . 75,1591-1595 (1978).
4. Frederickson. R. C. A. Life Sci. 2 1 , 2 3 4 2 (1977).
5. Bayon, A . . Rossier, 1.. Mauss, A,. Bloom, F. E. & Iverson. L. L. Proc. natn. Acod. Sci.
U . S . A . 75,3503-3506 (1978).
6. Henderson, G.. Hughes. J. & Kosterlitz, H . W. Nature 271, 677-679 (1978).
7. Alumets, I., Hakansson, R., Sundler, F. & Chang, K. J. Histochemistry 56,187-196 (1978).
8. Hughes, I., Kosterlitz, H. W. & Smith, T.W, Br J. Pharmac. 61,639-647 (1977).
9. Schultzberg. M. et a / . Acra physiol. scand. 1 0 3 , 4 7 5 4 7 7 (1978).

O 198lNature Publishing Group

3 19

Nature Vol. 289 22 January 1981

10. Schultzberg. M . er a/. Neurorcrence 4. 249-270 (1979).
lI.D i Giulio, A. M . el ul. Ncu~opharmarology17.989-992 (1978)
12. D i Giulio. A. M.. Yang, H . Y., Fratta, W. & Costa. E. Narure 278. 646-647 (1979).
13. Schultzberg. M . er al. Neuroscience 3. 1169-1186 (1978).
14. V~veros,0.
H.. Dil~berto,E. J., Hazum, F. & Chang, K J. Mulec. Phurmar. 16, 1101-1108
(1979).
15. Viveros. 0. H., Diliberto. E. J.. Hazum. E. & C'hang. K. J. P,ychupharmacologra 22,
191-204 (1980)
16. Stern. A. S. eta1 Pruc. narn. Acad. Sci. U S A . 76, 6680-6683 (1979).
17. Clement-Jones. V.. Lowry. P. I..
Rees. L . H . & Besser,G. M . Narure 283.295-297 (1980).
18. Livett, B. G., Kozousek. V., Mizohe. F. & Dean, D . M . Narure 278,256-257 (1979).
19. Mizohe, F., Ko7ousek. V., Dean, D . M. & I.tvett. B. G. Brain Res 178. 555-566 (1979).
20. Lemalre, S., Lemaire. T., Dean. D . M . & Livett, B. G Nature 288, 303-304 (IY80)
21. Livett, B. G. & Dean, D. M . Nruropeprides 1. 3-13 (1980).
22. Rossier, J.. Dean. D . M., 1.ivett. B G. & Udenfriend, S. Life Sri. ( ~ the
n press).
23. Rossier, 1. r r a l Pror. narn. Acud. SCI. U.S.A. 74, 5162-5165 (1977).
24. Hbkfelt, T.. Johansson. D.. Ljungddhl. A,, Lundherg, J M & Schultzberg. M . Nature 284.
515-521 (1980).
25. Miura, Y .Campese, V., DeQuatro, V. & Meijer, D . J. Lah. clm. Med. 89,421-127(1977).
26. Renz~ni,V., Brunon, C. A. & Valori. C. Clm. chrm. Acra 3U, 587-594 (1970).
27. Fenw~ck.E. M., Fajdiga. P.8.. H0we.N. B. S. & Livett, B G. I. CellBiol. 76.12-30 (1978).
28. Tr~faro,J. M.. Palmer, A. M . & Douglas, W. W. Biorhem. Pharmac. 16,2095-2100(1967).
29. Dean, D . M., Watson. L.. Bray. G M . & L~vett.B. G. Ncurotoxmr (eds Chuhh, I.W. &
Geffen. L. B.)261 (Adelatde Un~versltyUnion Prers, 1979).
30. Wood, P. M . Brain Res. 115,361-375 (1976).
31. Bornstein. M. B Loh. Invrsr. 7. 134-137 (1958).
32. Furness. J. B..Costa. M & Blessing. W. W. Hisrochem. J 9, 745-750 (1977).

artificial vesicles7. Further, and most significantly, it labels only

those residues of the protein which reside within the outer
monolayer (ref. 10 and unpublished data). Chowdry and Westheimer" have reviewed membrane photolabelling studies. We
have previously shown that cholera toxin can be photolabelled
after binding to NDV charged with probe'*,I3. Here we analyse
the kinetics of specific subunit insertion into the membrane.
Figure 1 shows the separation of proteins derived from
cholera toxin, NDV and NDV with bound cholera toxin. The
difference between the amount of toxin bound to NDV and the
total toxin present (60 pg per sample) can be estimated by
comparing lanes 1 and 3 and lanes 4 and 5. Typically, 3-5 kg of
the toxin bound to NDV in the conditions used. Next to each
stained gel the corresponding fluorogram shows proteins
radiolabelled by the photoreactive probe (Fig. 1).Irradiation for
15 s allowed as much labelling of membrane proteins as irradiation for 6 min (data not shown); thus we conclude that most of
the probe was photoactivated. Gel bands designated VP
represent viral proteins whose photolabelling has been
described elsewhere6. About 15-25% of the protein in band

Photolabelling of cholera toxin

subunits during membrane penetration
Bernadine J. Wisnieski & John S. Bramhall*

proteins

D e p a r t m e n t of Microbiology a n d The Molecular Biology Institute,

University of California, Los Angeles, California 90024

There has been much speculation about the mechanism by

which cholera toxin exerts its effect on the cytoplasmic side of
the membranes with which it interactda. After the pentamer of
B subunits (5B) binds to membrane receptors, particularly the
monosialylganglioside GM1, the disulphide-linked dimer
AlSSA, (which together with 5B constitutes the comdete toxin)
is thought to
the membrane, perhaps thro&h a than:
nel formed by 5B and become reduced so that AISH units reach
the cytoplasm and stimulate adenylate cyclase. Evidence for this
mechanism is circum~tantial'.~".If it is correct, a compound
which will specifically label intramembranous sections of the
toxin should label the channel-forming B subunits but not the
channel-contained A, subunit. We have tested this prediction
with a photoreactive glycolipid compound6" and have obtained
the opposite result. Therefore, we propose that only the A,
subunit enters the membrane and we provide here data on the
kinetics of that process.
The experimental protocol described is a direct and nonperturbing method for studying the dynamics of membrane
penetration by proteins and other molecules. For simplicity, we
used the Newcastle disease virus (NDV) system, earlier characterized with the photoreactive glycolipid probe", as a membrane
target for cholera toxin because the viral envelope contains only
three proteins, designated HN, F and M (ref. 8); surface
neuraminidase activity resident in the HN protein makes NDV
especially rich in GM, (ref. 8). The glycolipid probe 12-(4azido-2-nitrophenoxy)-stearoyl[l-'4C]glucosamine (12APSGlcN) is unique in that it will spontaneously insert into any
membrane, and studies with a spin-labelled (nitroxide) counterpart suggest that the glucosamine moiety restricts it to the
surface monolayer for an appreciable time after insertionY.The
photoreactive group of 12APS-GlcN is located on the stearic
acid chain of the amphiphilic compound. It reacts with vicinal
compounds only when samples are exposed to UV light. Photolabelling is relatively nonspecific7.The probe labels coliphage
Ml3(f coat protein only after incorporation into the bilayer of
Present address: Department of Neurobiology, Stanford University School of Medicine.
Stanford, Califorma 94305.
0028-083618 1/0403 1 9 - 4 3 5 0 1 00

I
Viral
proteins

-I

Viral
proteins

4M

-HSA,A2SH
-

Fig. 1 Radiolabelling of NDV and cholera toxin proteins with the

photoreactive probe 1ZAPS-GlcN. Stained gel lanes are 1-6; corresponding
fluorogram lanes are lF-6F. Unreduced samples are in lanes 4 and 5. Lanes
1 and 4 contain cholera toxin (60 pg) labelled in solution and acid precipitated; lane 2 contains NDV (30 pg protein) labelled in suspension and
pelleted. Lanes 3.5, and 6 contain cholera toxin labelled after attachment to
NDV for 1 min at 37 "C.Each sample was irradiated at 360 nm for 15 s. The
amount of probe (50 mCi mmol-') per sample was 4 x lo-' pCi. Total
equilibration time with probe was 16 rnin at 37 OC for toxin alone and for
NDV alone, and 15 min for NDV to which toxin was added 1 min before
irradiation. Procedures as described in Tahle 1 legend.
@ 1981 Macmillan Journals L t d

320

Nature Vol. 289 22 January 1981

Table 1 Photoreactive radiolabelling of cholera toxin subunits and viral proteins as a function of time and temperature of incubation

Net c.p.m.
Unreducedprotein bands 0 'C, 15 min

37 "C, 30 s

37 "C, 1 min

37 "C, 3 min

37 'C, 15 min

NDV alone

Toxin alone

209
6
100
315

174
91
100
365

196
136
105
437

210
86
99
395

192
71
100
363

205
0
102
307

0
21
0
21

0 "C, 15 min

37 "C, 30 s

37 OC, 1 min

37 "C,2 min

37 "C, 3 rnin

200
2
109
5
316
7

194
23
108
69
394
91

191
35
101
107
434
142

223
27
107
84
44 1
111

208
23
110
71
412
94

VP-M
AlSSA2
VP-HN-fragment
Total c.p.m.
Reduced protein bands
VP-M
HSA1A,SH
VP-HN-fragment
AISH
Total c.p.m.
AISH +HSA1A,SH

37 "C, 10 min 37 'C, 15 min

195
20
103
56
374
96

190
19
99
45
353
64

NDV alone

Toxin alone

201
0
102
0
303
0

0
4
0
16
20
20

Experiments were carried out under red safety lights until completion of electrophoresis to avoid activation of probe in unirradiated controls. Cholera toxin, purified14
and stored at 6 mg ml-I in 0.1 M sodium phosphate buffer, pH 7, was thawed and diluted to 1 mg ml-' with 5 mM sodium phosphate-buffered saline containing 1 mM
EDTA, pH 7.4 (PBSE). Two aliquots (0.6 ml) were spun at 149,000g for 30 min to remove any aggregates. The supernatant was used throughout. NDV strain HP-16,
isolated as described elsewhere6, was diluted to 450 pg protein per 9.4 ml PBS. The photoreactive probe 12APS-GlcN (50 mCi mmol-I), synthesized as described
elsewhere1', was diluted to 90,000 c.p.m. per 10 p1 of ethanol. In tube A, 150 p1 of probe was dried and 50 p1 of ethanol added and in tube B, 40 pl of probe was dried and
13 &I of ethanol added. The NDV suspension (9.4 ml) was added to tube A and 240 p1 of toxin solution plus 2.5 ml PBSE added to tube B. Both tubes were incubated at
37 "Cfor 15 min. During this time, toxin solution was dispensed in 60-p1portions into 11 borosilicate tubes (6 x 50 mm), then 0.6 ml of NDV from tube A was added to each
of these tubes, which were incubated for periods of time from 0.5 to 15 min before irradiation for 15 s at 366 nm with a mineral lamp (UV-Products, San Gabriel,
California). Usually two samples were irradiated at each time point to provide reduced and non-reduced samples for gels. One unirradiated sample provided a control for
background radioactivity. Two samples (0.6 ml) of NDV from tube A were incubated for an additional 15 min at 0 OC and at 37 "C before irradiation. Another two 0.6 ml
samples of NDV were cooled to 0 "C before being added to two small tubes, each of which contained 60 p1 of chilled toxin solution. One of these was irradiated after 1 min at
0 *C, the other after 15 min at OC. The contents of tube B were divided into four 0.6-1111 samples which were incubated at 0 OC for 1 min, 0 OC (15 min), 37 "C (1 min) and
37 "C for 15 min before irradiation. The entire experiment was repeated to provide a duplicate set of samples for analysis. Each time, 60-pI portions of toxin solution were
added to 0.6 ml of PBSE and incubated for 30-60 min at 37 "C and centrifuged to determine how much toxin aggregated and could be pelleted without acid precipitation.
This routine check was performed because it was found that commercial lyophilized toxin did not give reproducible results when used as a substitute for freshly isolated
toxin (probably because it proved to be extremely difficult to completely solubilize). All samples were pelleted at 149,000g for 30 rnin on a Beckman SW50.1 rotor in
buckets fitted with 0.7 ml adaptors. To each sample containing toxin with probe alone (from tube B), 50% trichloracetic acid (60 pl) was added after irradiation and the
samples cooled to OC before centrifugation. Pellets were boiled for 2 min in 40 ~1 of reducing or non-reducing SDS-bufferI6. Electrophoresis and fluorography were
carried out as described elsewhere13. Gel slices were oxidized to "CO,, which was collected and counted6. Top, data from non-reduced samples; bottom, reduced samples.
Columns designated NDV alone and toxin alone refer to samples incubated independently with probe for 15 min at 37 "C, before irradiation. The sample of toxin alone
(60 ~ gwas
) acid precipitated. The other columns represent components of toxin which pelleted with virus, estimated to be 5% of the 60 pg added. VP, viral protein; HN-f,
HN-fragment. The H S m S H peptide is the reduced but uncleaved A,A,peptide. Normally A,SSA, completely reduces to A,SH and A,SH; however, in rapidly isolated
cholera toxin preparations, not all the linear A,A, peptide is nicked between the two cysteines which form the disulphide bridge between A, and A,.

10

15

Incubation time (min)

Fig. 2 The kinetics of photoreactive labelling of cholera toxin A subunits.

Open symbols, ratios calculated from Table 1 (top); solid symbols, ratios
calculated from Table 1 (bottom). Zero time points represent ratios calculated from On, 15' samples in Table l. Solid symbols, ratio of radiolabel in
AISH+HSA,A,SH (unnicked) to radiolabel in viral protein M (0)and
HN-f (W). Open symbols, ratio of radiolabel in AISSA, (nicked plus
The ratio of
unnicked) to radiolabel in viral protein M (0)and HN-f (0).
radiolabel in HN-f to radiolabel in M was relatively constant (average 0.51;
range 0.47-0.57).

AlSSA2 represents &

A
linear peptide which was not proteolytically cleaved during toxin purification12 (the underlining
of AlA2 is meant to distinguish this linear (uncleaved) peptide
from the cleaved peptide A1A2).The rest of the band represents
peptide which was cleaved between the disulphide bridge into
discrete Al and A, subunits. The B subunit contains an internal
disulphide bond and migrates with a larger apparent molecular
weight when reduced. Reduced toxin samples show a faint band
representing the reduced, but uncleaved, linear peptide
HSA,A,SH (ref. 12), as well as AISH and B. The AzSH subunit
does not stain effectively. It migrates ahead of B and can be
detected when '251-labelled toxin is e l e ~ t r o ~ h o r e s e dIn
' ~ .the
fluorograms shown, the A2SH peptide could not be resolved
with any confidence from the diffuse band of radiolabelled
lipids.
Other samples, including a series irradiated at defined times
after the addition of toxin to NDV, were similarly electrophoresed. Bands representing viral proteins VP-M and VP-HNfragment and toxin subunits (except A2SH) were cut from
stained gels after fluorography (because VP-HN-fragment does
not stain (Fig. I)), and oxidized to 14C02which was trapped and
counted. The level of radioactivity in the B subunit was not
significantly above background in conditions tested (Fig. 1 and
ref. 13). As previously demonstrated6.", unirradiated samples
did not contain any radiolabelled proteins. Radioactivities of the
pertinent bands are shown in Table 1. The labelling of A, plus
A2, that is, the net radioactivity associated with unreduced
AISSA, (cleaved plus uncleaved), is expressed as a function of
time and temperature of incubation with NDV before the 15 s
irradiation. To standardize the data between gel lanes, raw data
were later expressed as ratios within a particular lane.
Radiolabelling of the two viral proteins chosen for ratio calculations did not vary with time or temperature or the presence
of reducing agent, and therefore accurately reflect the amount of
sample in each gel lane.

O 198lNature Publishing Group

Nature Vol. 289 22 January 1981

Fig. 3 Transmembrane kinetics of the
active A , subunit of cholera toxin. Step 1,
B subunits bind to ganglioside receptors
(forked structures). Within 1 min of binding at 37 "C, A , insertion into the outer
monolayer of the membrane is complete
(Step 2). Within 2 min of insertion, A ,
equilibrates between the outer and inner
monolayers (Step 3). The equilibrium
achieved is undoubtedly a dynamic one;
however, relative to the kinetics of the
photolabelling reaction. the position of
each A , is-fixed. The small arrow
represents the photoreactive group of
12APS-GlcN.

321

Data in Table 1 indicate that the A subunit of toxin can enter

the vicinity of photogenerated reactive groups within seconds of
bindine to the viral membrane. To assess whether the radioactivity inthe A,SSA2 band was due to labelling of Al, A2,or both,
the ratios of counts in this band to the counts in both VP-M and
VP-HN-f were plotted as a function of incubation time at 37 "C
with NDV (Fig. 2, open symbols). The counts in AISH and in
uncleaved HSAIA,SH were summed and expressed as a ratio to
counts in ~ ~ X a ton counts
d
in VP-HN-f (Fig. 2, solid
symbols). Surprisingly, Fig. 2 shows that the ratios obtained as a
function of time using total AlSSA2 (unreduced) counts and the
ratios obtained using a formula in which 75-85% of A, is not
included (due to reduction and dissociation of the cleaved
fraction of AISSA, molecules) are not significantly different.
From this we conclude that the majority of the radiolabel
associated with the A,SSA2 band (unreduced) and the
SHAlA2SH (reduced, uncleaved) is actually attached to the A,
subunit of the toxin and not the A2 portion.
The most striking feature of the data presented in Table 1 and
in Fig. 2 is that the A component of toxin, more specifically the
A, portion (from the rationale above), becomes increasingly
accessible to photolabel within the first minute of incubation at
37C with NDV. During the next 2min of incubation,
radiolabelling of A l decreases by roughly 40% (3441%).
Between 3 and 15 min of incubation, very little change occurs.
The accessibility of A1 to probe after 15 min is 50% (46-54%)
of that observed at 1 min. The rapid penetration and labelling of
A, resemble that of complement component C9 on addition to
vesicles containing C5b-8 (refs 17, 18 and unpublished data).
to label in
The ratio of label in uncleaved toxin (HSAlA2SH)
AISH (calculated from data in Table 1) undergoes a slight
increase with time; the significance of this remains to be determined. The kinetics of labelling of the A, subunit was verified
qualitatively by exposing X-ray film to gels for various time
periods before development. The first A,-containing bands to
produce silver grains in the film emulsion were always from the
very early incubation times. Little or no labelling of any toxin
subunit occurred when toxin and NDV were incubated together
at 0 "C before irradiation, regardless of incubation time, even
though NDV bound the same amount of toxin at this temperature as at 37 "C and displayed no change in its own labelling
profile.
Values for the extent of labelling of the viral proteins and of
the bound cholera toxin Al subunit (at the 1 min time point)
were all at the level of 4 mol%. A similar value was obtained
with M13 coat protein in vesicles7. This strongly implies that
within 1 min of toxin addition and binding, most of the Al
associated with virus entered the membrane envelope. If we
assume that only 10% or 1% of the virally bound A, entered the
membrane, the calculated level of A,-specific labelling needs to
be increased to 40 and 400 mol%, respectively, an unlikely
occurrence unless an equivalent level of preference of probe for
A, is proposed. Thus, we propose that most of the A, associating
with virus actually penetrates the bilayer. We believe this to be
the first such demonstration.
Our data are incompatible with the Gill1 model for toxin entry
because if the B pentamer formed a transmembrane channel, we
would expect to label B but not A (which would be protected

from the probe by residues of the B channel). The reverse was

actually observed, and the detected entry of the enzymatically
active Al subunit was found to be temperature-dependent,
occurring at 37 "C (a temperature compatible with physiological
activity), but not at 0 "C (a temperature compatible with binding
but not activity).
Explanation of the kinetic data shown in Fig. 2 must answer
two questions: (1) where does 50% of the A, entering the
membrane go and (2) why does only 5O0/o of the entering A,
leave? One proposal (see Fig. 3), is based on the assumption that
the probe remains restricted to the outer monolayer. This is a
reasonable assumption because, as previously mentioned, all
available data are consistent with an asymmetric distribution of
such a compound (refs 9, 10 and unpublished data). Within
1 min of toxin binding, the Al subunit penetrates the outer
monolayer of the membrane (step 1). The kinetics of A, entry
between 0 and 1 min would reflect the kinetics of toxin binding
to receptorI9 and subunit penetration. Between 1 and 3 min, the
rapid decrease in A l labelling suggests that about half of the A ,
molecules (or one half of each molecule) leave the vicinity of
probe within 2 min of entry (step 2). We visualize this as an
equilibration of Al between the outer and inner monolayers
until roughly one half is in each monolayer. The calculated
half-time for the A, equilibration process (decrease in labelling)
is of the order of 10-20 s. This time factor is not unreasonable
for a diffusion-limited process in a viscous medium. At any time
after 3 min, only 50% of the A, that entered the membrane
remains accessible to probe, indicating very little change in Al
distribution after 3 min (step 3).
By use of photoreactive probes to investigate the dynamics of
toxin translocation in other systems, we can start to examine the
specific roles of membrane composition, fluidity and asymmetry
on protein traversal through membranes.
B.J.W. is the recipient of USPHS Research Career
Development Award GM-00228. J.S.B. was a Fulbright-Hayes
Postdoctoral Scholar. Research was supported by USPHS Grant
GM-22240, the University of California Cancer Research Coordinating Committee and the Academic Senate.

Rece~ved7 July; accepted 10 Novemher 1980

1. Gill, D. M. Biochrmrsrry IS, 1242-1248 (1976).

2. Sahyoun, N. & Cuatrecasas. P. Proc. narn. Acad. Scr. U . S . A . 72. 3438-3442 (1975)
3. van Heyningen, S. Biochem. J. 168.457453 (1977).
4. Tosteson, M
. T. & Tosteson. D. C. Narure 275, 142-144 (1978).
5 Moss, J. eral. Proc. nurn. Acad. Sci U . S . A . 73,3480-3483 (1976).
6. Bramhall, J. S., Shiflett, M. A. & Wisniesk~,R. J. Biochrm. 1. 177, 765-768 (1979).
7. Hu, V. W. & Wisnieski, B. I. Proc. narn. Arad. Sci, U.S.A. 76, 5460-5464 (1979).
U. Scheid. R. & Chnppin. P. W. Virology 57,475490 (1974).
9. Wisnieski, 8. J . & Iwata. K. K. Biochemisrry 16. 1021-1326 (1977).
10. Simon, P. &
I Wisnieski, B. 1. Fedn Proc. 39, 2190 (1980).
11. Chowdry. V & Westeimer. F. H. A . Rev. Hiochem. 48, 293-320 (1979).
12. Wisnieski. B. I.. Shiflett. M. A,. Mekalanos. J & Bramhsll. J. S. I. supmmoler. Srrttcr. 10,
191-197 (1979).
13. Wisnieski. B. J. & Bramhall, J. S. Biochem. b~ophyc.RES.Commun. 87, 308-313 (1979).
14. Mekalanos, 1. I.. Collier. R. J. & Romig, W. R. Infect. Immun. 20, 552-558 (1978).
15. Bramhall, J.. lshidn. B. & Wisnieski. B. I . supran~olrc.Srrucr. 9. 399406 (1978).
16. O'Farrell. P. H. 3. biol. Chcm. 250,40074021 (1975).
17. Hu, V. W., Esser, A. F., Podack, E. R. & Wisnicski, H.J. Fedn Proc. 39, 1971 (1980).
18. Hu. V. W.. Esser, A. F.. Podack, E. R.& Wisnieski, B.J . Proc. 41h mr. Congr Immun. Abrrr.
No. 35.2.07 (1980).
19. Cuatrecasas. P. (ed.) The Specificily and Acriot~ of An,mul, Racrerial and Planr Toxins
(Chapman & Hall. London. 1977).

O 198lNature Publishing Group

322

Nature Vol. 289 22 January 1981

In equation (2), the diffusive reverse rate constant, k-, is related
to the diffusive dissociation time by

The magnitude of signal

amplification by
ligand-induced receptor clustering

For diffusion-limited reactions (kt >> k-), equation (2) becomes

Charles DeLisi
Laboratory of Theoretical Biology, National Cancer Institute,
National Institutes of Health, Bethesda, Maryland 20205

Lateral mobility of plasma-membrane proteins is now documented for a variety of cell systems. Evidence is accumulating
that the molecular clustering allowed by such mobility may be
important in the transduction of information at the cell surface1-'. Moreover, in a wide variety of systems, only a very small
fraction of receptors need to be bound to induce a responses",
such as in the mast celllbasophii system where receptor clusters
as small as dimers can induce a response1' and only three or four
such dimers are necessary for local degranulationll. As receptors are mobile, some clustering is expected even in the absence
of ligand. As an approximate indication of the numbers
involved, a considerationof constraints on translational entropy
by spontaneous pair formation predictsa2about' 100 pairs on
avera e for a cell with 500,000 receptors (typical of mast
cells1f), if a pair is defined as two receptors whose peripheries
are within 5 di. This estimate need not present a problem as it
can be reduced by several orders of magnitude by postulating
orientational requirements, repulsive potentials andlor more
stringent constraints on tanslational entropy. Moreover, some
cell pre arations do show a slow Ca2+-dependentbackground
releasePand this may, in fact, reflect spontaneous clustering and
suggest that the control of activity is not stringent for those cells.
An alternative view of the regulation of activity, however, which
may have broad implications, is that receptors must be in
proximity for some minimum amount of time for the transmission of biological information. This residency requirement
mar: for example, reflect the time required for the initiation of
Ca flow1' or for recestors to change conforrnati~n~~
or for the
pair to bind an integral membrane protein16. Whatever the
origin, the quantitative predictions of the idea, which I develop
here, suggest, for biologica~~
relevant parameter values, an
exceedingly sensitive control mechanism that can amplify the
probability of an event by a factor of loZoor more.
An estimate of the probability that a pair, having formed, will
be active t s later is easily made if the pair separation times are
exponentially distributed. Thus, if t- is the mean time for a pair
to separate by diffusion to some distance A beyond which it
cannot transmit a signal, the fraction of pairs remaining within A
for t s after pair formation is exp (-tlt-). The fraction that
remains within A long enough to transmit a signal is approximately

where

is the intrinisic equilibrium constant of a ligand site for a

receptor site. The ligand-induced amplification (A) of the
transduction probability is defined as the ratio of the probability
of transduction in the presence of ligand to that in its absence.
Thus

Note that equation (6) is the ratio of two probabilities, each

probability being obtained by division by the actual number of
pairs. Hence the actual number of pairs does not enter equation
(6). Had a different definition of amplification been adopted in
terms of an increase in numbers of pairs (and not just their
duration), the amplification would be somewhat larger than
estimated below, by perhaps two orders of magnitude. It will
become evident that the distinction between the two definitions
is neither qualitatively nor quantitatively important. Equation
(6) was chosen because of its relative simplicity.
KO will be large even if binding is weak. For example, a
binding free-energy change of 5.5 kcal corresponds to an
intrinisic equilibrium constant of lo4 at room temperature. A
plot of ammplification as a function of the ratio tD/t- for
KO= lo4 rises exponentially, reaching 10" at a ratio of 25 and
at a ratio of 75 (Fig. 1). Thus the transduction probability
can be changed from essentially zero in the absence of ligand (for
example, = 10-1 at tD/t- = 25) to close to one in its presence.
The large amplifications are consequences of very low levels of
spontaneous signal transmission in the absence of ligand.

where t~ is the minimumtime the perimeters must be within A

for transduction of a signal.
When receptors are cross-linked by ligand, pair separation
will depend not just on diffusion but on the time required to
break the ligand-receptor bond. If kl and k-l are the reactive
forward and reverse rate constants for ligand-receptor interaction respectively, then the effective rate constant for pair
dissociation

and the effective dissociation time, t, is

Transduction time/separation time (to/?-)

Fig. 1 Probability of transduction in the absence of ligand (- - -)
and the amplification achieved (-)
with a low-affinity
ligand having an intrinsic equilibrium constant for the receptor
of K~= lo4.

O 198lNature Publishing Group

323

Nature Vol. 289 22 January 1981

Table 1 Mean time for a receptor pair to separate to a distance A

A(A)
L(S)

95

1.3~

100

105

5 . 3 ~ 1 0 -1~. 2 ~

110
1.7 x

Diffusion coefficient = 3 x lo-'' cm2 s-',

approach = 90 A.

120

130

4 . 3 lo-'
~ 7.4~
distance

of

closest

How reasonable is the range of ratios for tD//t- shown in Fig.

I ? Transduction times are expected to be very rapid. For
example, conformational changes in proteins in solution
generally occur in lo-'s (ref. 19), though such changes
might take somewhat longer in a membrane. Consequently, if
transduction requires a conformational change, t, would be of
the order of milliseconds or longer. Similarly, experiments in
which the flow of Ca2' into a cell was followed indicate that
uptake is 90% complete within 30 s, suggesting a time constant
of seconds or less. Thus a transduction times of about
s
would seem to be a reasonable lower bound. Amplification
would then, according to Fig. 1, require diffusive dissociation
s.
times of =s
The diffusive dissociation time can be estimated as follows.
Consider circular receptor disks of diameter s with the plane of
the disks coinciding with the plane of the membrane. The
coordinate origin is fixed at the centre of one of the disks. Then,
if the centre of a second disk is at s < p c s + A , the mean time
W ( p )required for its first arrival at s + A satisfiesZ0

where D is the sum of the diffusion coefficients of the disks. The

solution to equation (7) subject to the boundary conditions

The mean time for a particle in the active region to escape is

obtained by averaging equation (8) over all distances between s
and s + A. Thus,

For molecules in solution, some fluctuation in distances

between groups is expected, even if the intervening chain is
short. However, when both groups are constrained to lie on a
cell surface as a result of bonding to receptors, the ability of the
intervening backbone bonds to rotate is severely restricted-a
fact easily verified by examination of molecular models. The
chain is then effectively frozen in whatever configuration it was
in when the second bond formed. From this point of view, the
distances between reactive groups on a ligand, averaged over the
collection of multiply bound ligands, will reflect the distribution
of distances between groups when the ligand is free, even though
the distance between groups on any particular molecule does not
fluctuate. But only that fraction of multiply bound ligands
having configurations constraining the peripheries of the receptors to be within A will be recognized. Thus the reduction in the
concentration of active pairs, at a fixed ligand concentration, will
be the main effect of flexibility and the amplification factor given
by equation (6) remains essentially unchanged. Experiments
using stiff chains or using flexible chains whose dimensions
have-been determined indkpendently, for example by energy
transfer. mav elucidate the validitv of these remarks.
ina all^, note that the form of Guation (6) is not necessarily
limited to systems that rely on clustering. For example, if the
step subsequent to binding were a conformational change, then
t- could be interpreted as the natural lifetime of the active
conformation in the absence of ligand. In the presence of ligand,
the lifetime would be k, the dissociation time of the ligand. The
amplification would then be exp [k, - k-)to] exp (k,tD), and
the fraction of receptors remaining active for t, exp (-k,tD).
As long as k,tD is much less than one, changes in k, will not
affect the fraction of bound ligands that are active. As k,tD
becomes greater than one, however, activity should drop
rapidly. Somewhat similar considerationsz3were used to explain
the precipitous drop in cyclic AMP production as the affinity of
hyponeurophyseal hormones for their receptors decreased,
while the total bound concentration remained constant.
In conclusion, the molecular complex mediating information
transfer, whether it be a receptor cluster or a particular receptor
conformation, is assumed to be present at all times, but intrinsically unstable with a decay time that is short compared with the
time required for information transfer. Signals in these circumstances are too low to be distinguished from background noise.
Ligands, by stabilizing an active complex so that its decay time
exceeds the time required for information transfer, selectively
amplify particular signals. Numerical estimates indicate that
such amplification can be very large.
I thank Drs Henry Metzger and Alan Perelson for reading and
commenting on the manuscript.

Received 14 July; accepted 10 November 1980.

Diffusive dissociation times were calculated according to equation (9) as a function of A (Table l), using a receptor diameter of
90 A and a diffusion coefficient of 3 x lo-'' cm2s-', numbers
typical of the mast cell ~ y s t e m ~ 'The
. ~ ~ calculation
.
is not of
course intended to be precise, but is presented here only to
provide perspective for evaluating the plausibility of the range of
ratios in Fig. 1.
The above analysis assumes that when receptors are crosslinked, they are always within the active radius. However, this
assumption is not compatible with molecular flexibility.
Suppose, for example, that the doubly bound ligand were fully
flexible and its maximum extension exceeded A. Then within the
framework of the residency-time requirement, the ligand would
not be sensed by the cell and the amplification would be one. The
reason is that if the ligand is so flexible that it does not affect the
diffusion of one receptor relative to its partner, the time to
escape from the active region will not be changed; only the rate
of entry into the region will change. A fully flexible molecule is,
however, unrealistic.

1.
2.
3.
4.
5.
6.
7.

Siraganian, R. P., Hook, W. A. & Levin, B. B. Immunochemisrry 12,149-157 (1975).

DeLisi, C. & Siraganian, R.P. J, Immun. 122.2286 (1979).
Demba. M.. Goldstein, B..Sobatka,A. K. & Lichtenstein,L. M. J. Immun. 122,518(1979).
Fanger, M. W.. Hart, D. A. Wells, I. V. & NisonoR, A, J. Immun. 105, 1484 (1970).
Archer. B. G. & Krakauer. H. Biochemisfry 16 (1977).
Schechter, Y., Hernaez, L., Schlessinger, I. & Cuatrecasas, P. Nature 278, 835 (1979).
Kahn. C. R., Baird. K. L., Jarrett. D. B. & Flier. J. S. Proc. narn. Acad. Sci. U.S.A. 75,4209

(1978).
8. Birnbaumer, L.. Pohl, S. L. & Kaumann, A. J. Adv. Cylic NucleorideRes. 4 (1975).
9. Schlessinger, I. in Physical Chemistry of CellSurface Evcnrs in Cellular Regular (eds Deliai.
C. & Blumenthal, R.) (Elsevier, Amsterdam, 1979).
10. Scgal, D.M., Taurog, 1. & Merzger, H. Proc narn. Acad. Sci. U.S.A. 74,2993 (1977).
11. DeLisi, C. & Siraganian, R.J. Immun. 122,2281 (1979).
12. DeLisi. C. Q. Rev. Biophys. (in the press).
13. Metzger. H. & Bach M. K. in Immcdiare Hypersmsiriuiry (ed. Bach. M. K.) 561 (Dekkcr,
New York, 1978).
14. Forman. I. C., Hallet, M. 8. & Morgar. J. L. J. Physiol., Land. 271, 193 (1977).
15. Hiedman, T. & Changeux, J. P, A . Rev. Biochem.47,317 (1978).
16. Bourguignon, L. Y. W. & Singer, S. J. Proc. notn. Acad. Sci. U.S.A. 74,5131 (1977).
17. Eigen, M. in Quantum Sfatistical Mechanics in the Narural Sciences (eds S. L. Minz. &
Weidermayer, S. M.) (Plenum, New York. 1974).
18. Bell, G. I.Science 200,618 (1978).
19. Hammes. G. Adv. Protein Chem. 23, 1 (1968).
20. Berg, H. C. & Purcell, E. M. Biophys. J. 20. 193 (1977).
21. Metzger, H. Immun. Rev. 41. 186 (1978).
22. Schleuinger, I. Webb, W., Elson, E. L. & M+ger, H. Nature 264,330 (1976).
23. Bergman, R. & Hechter, 0. J. biol. Chem. 153,3238 (1978).

O 198lNature Publishing Group

Nature Vol. 289 22 January 1981

324

Evidence for a divided genome

in bean golden
mosaic virus, a geminivirus
S. Haber, M. Ikegami, N. B. Bajet
& Robert M. Goodman
Department of Plant Pathology, University of Illinois, Urbana,
Illinois 61801

Geminiviruses'.', a recently discovered group of plant viruses,

are the only viruses containing circular single-stranded DNA
(ssDNA) found in eukaryotic organisms. Their ssDNA, which is
less than half the size (molecular weight (M,) 7-8 x 10') (refs
3-6) of that of other ssDNA vir~ses"~,
is the smallest yet found
in autonomously replicating viruses. Geminivirus DNA is
encapsidated in paired isometric particles whose monomer units
measure -18 nm. This unique capsid morphology was accounted for by an early hypothesis of a functionally divided genome.
Among the RNA plant viruses, there are several groups that
carry necessary genetic information on more than one nucleic
acid species. In the case of geminiviruses, two nucleic acid
components were postulated to be carried in each paired parti~ l e ~ . Analyses
'~.
of the physical composition of geminiviruses
showed, however, that there is only one DNA molecule in each
paired particle6'"'". We have now investigated the DNA of a
geminivirus using restriction endonucleases and a biological
assay of infectivity and report here evidence for a divided
genome in bean golden mosaic virus (BGMV). Two different
nucleotide sequences are found and both seem to be required for
infectivity.
BGMV was maintained by sap transmission in Phaseolus
uulgaris L. ('Top Crop') kept in simulated humid tropical
conditions in a growth
and purified as described
elsewhere13.14.Viral DNA labelled with 32Pin uiuo was obtained
by collecting systemically infected trifoliolate leaves from plants
5 days after inoculation of primary leaves and feeding with
32P-orthophosphate ( I mCi per leaf) through the cut ends of
petioles. The leaves were then maintained for 4 days in a Petri
dish by feeding a balanced inorganic salt solution through the
petioles.
BGMV DNA is resolved by polyacrylamide gel electrophoresis in 7 M urea into two components which comprise a
covalently closed circular molecule ( M , 8 x 10') and a slightly
shorter linear molecule5; previous evidence suggests that the
linear molecule is derived from (or possibly is a precursor of) the
circular form". We separated 32~-labelled
circular and linear
molecules by gel electrophoresis and recovered the DNA by
electro-el~tion'~
from excised gel slices. The DNA was then
digested separately with restriction endonucleases HhaI and
HaeIII which have sequence-specific activity on the ssDNA of
bacterial v i r ~ s e s ' ~ ~DNA
".
from @XI74 digested with HaeIII
was used as a standard.
Digestion of linear or circular BGMV DNA with HhaI
yielded fragments of the same size and relative electrophoretic
mobility (data not shown), confirming our previous conclusion
that circular and linear forms of BGMV DNA contain the same
nucleotide sequences". Calculation of the length of each
BGMV DNA fragment present in complete HhaI digests of
circular BGMV DNA revealed that the total length of these
fragments (5,380 nucleotides) was about twice that expected
(2,s 10 nucleotides) from electron-microscopic measurements of
the DNA (ref. 5) (Table 1).
We also used a virus-specific double-stranded (ds) DNA,
which we consider to be a replication intermediate, isolated
from 32P-labelled, BGMV-infected leaves. This contains the
complementary strand to viral DNA and both strands are of
0028-0836/81/040324-03$01

00

Fig. 1 Analysis of BGMV DNA by restriction endonucleases. a,

6 pg of @XI74viral ssDNA was incubated with 70 units of HaeIII
for 16 h at 37 "C in 50 mM Tris-HCI, pH 7.5 (120 pl), 5 mM MgC12
and 0.5 mM dithiothreitol. 6, 4 pg of BGMV viral DNA was
incubated as in a, but without restriction enzyme: c, circular DNA;
I, linear DNA. c, 4 pgof BGMV viral DNA was incubated with 120
units of HhaI for 16 h a t 37 "C in 50 mMTris-HCI ( p H 8.0.500 p1)
and 5 mMMgC12. d, 5,000c.~.m.of 3 2 ~ - ~ with
~ ~5 pg~
of '3x174 RF DNA as carrier, was incubated with 20 units of HhaI
as in c. e, 5,000 c.p.m. of 3
2
~ dsDNA
with
~
5 ~pg of ~@XI74
~
RF DNA as carrier was incubated with 20 units of HaeIII as in 6. f,
2,000 c.p.m. of 3 2 ~ - ~RF
1 3DNA, prepared by the procedure of
Messing (unpublished protocol), with 5 pg of @XI74 RF DNA as
carrier, was incubated as in e. After incubation, samples were
phenol/chloroform extracted and the DNA ethanol-precipitatedI5, then redissolved in 10 p1 of sterile distilled water.
Immediately before application to 4% polyacrylamide (acrylamide/bisacrylamide, 22 : 1) 7 M urea slab gels (0.75 mm x
13 cm x 14 cm)15, each sample was added to 6 p1 of a solution
containing 50% sucrose, 30% urea, 0.2% xylene cyanol, 0.2%
bromophenol blue and heated at 90C for 1 min. The
electrophoresis buffer was 50 mM Tris-borate (pH 8.3), 1 mM
EDTA. Electrophoresis was at 200 V until the xylene cyanol dye
reached the bottom of the gel. Gels a, b and c were stained in the
dark for 30 min with 0.005% Stains-All in 50% formamide, then
destained in distilled water under light. For autoradiography,
X-ray film (DuPont Cronex) was exposed to dried gels (d, e and f )
in intensifying screen cassettes at -80 OC for -4 d before developing. Arrows indicate the positions of restriction fragments from
HaeIlI digests of @XI74 viral DNA (left) and HhaI digests of M I 3
RF DNA (right); the numbers indicate approximate number of
nucleotides (QX174) or nucleotide pairs (M13) of the fragments.
Interpolated values for BGMV viral and dsDNA restriction fragments are given in Table 1. The top of the gel is indicated by the
arrow labelled 0.

@ 1981 Macmillan Journals Ltd

Nature Vol. 289 22 January 1981

325

Table 1 Restriction fragment sizes of OX174 ssDNA, BGMV ssDNA, BGMV dsDNA and M13 RF DNA shown in Fig. 1, in order of increasing

electrophoretic mobility
Fragment sizes
(nucleotides or nucleotide pairs)
1,353; 1,078; 872; 603; 310; 281; 271; 234;
194*
2,510 (circular); 2,360 (linear)
(2,510)t; (2,360); (1,650); 1,230; 1,050; (950);
660; 570; 520; 475; 365; 335; 175

DNA
@XI74ss

Treatment
HaeIII

BGMV ss
BGMV ss

Untreated
HhaI

BGMV ds

HhaI

1,350; 1,030; 650; 520; 460; 420; 340; 320

BGMV ds

HaeIII

1,480; 1,380; 750; 585; 510; 290

M13 ds

HaeIII

2,500; 1,500; 800; 340; 300; 220; 200

Remarks
Sizes of these fragments were computed from the nucleotide sequence given in ref. 18
Sizes based on electron microscope measurements5
Sizes based on interpolation using @XI74ssDNA HaeIII
fragments as standards. Sum of length of fragments in
complete digests is 5,380 nucleotides
Sizes based on M13 RF DNA HaeIII fragments as standards. Sum of fragment lengths is 5,280 nucleotide pairs
Sizes based on M13 RF DNA HaeIII fragments as standards. Sum of fragment lengths is 4,995 nucleotide pairs
Sizes of these fragments are from values given in a laboratory protocol (J. Messing, unpublished)

* Fragments 118 and 72 nucleotides long run off the gel in conditions used in our experiments (see Fig. 1 legend). Identity of these fragments was
verified by comparison with HaeIII restriction products of OX174 RF DNA. Other fragments in Fig. l a , some of which occur as discrete bands, are
due to the presence of -20% by weight linear ssDNA in the OX174 ssDNA preparation.
t Values in parentheses are for DNA fragments produced by incomplete digestion of the substrate, as indicated by their absence in digests of BGMV
dsDNA by the same enzyme, and by results of prolonged incubation of ssDNA with larger amounts of enzyme (not shown).
approximately unit length in relation to the circular viral D N A
(M.I. and R.M.G., unpublished). The fragments produced by
HhaI digestion of BGMV dsDNA corresponded exactly in
number and size to those produced by complete HhaI digestion
of circular viral ssDNA (Fig. 1). ,
Digestion of BGMV dsDNA with HaeIII yielded six fragments with a total length of 4,995 nucleotide pairs (Table I),
confirming the results obtained with HhaI.
Attempts to analyse BGMV ssDNA by digestion with HaeIII
were not successful. HaeIII digestion yields the same number
and sizes of fragments from @XI74 ssDNA as are found in
HaeIII digests of @ X I 7 4 replicative form (RF) DNA'". The
HaeIII restriction fragments of circular BGMV viral DNA,
however, were very small (20-100 nucleotides) and did not form
discrete bands (data not shown). The nucleotide ratio of "Plabelled BGMV D N A is 32% A : 3 3 % T : 2 2 % C : 13% C
(unpublished results); thus it is unlikely that the HaeIII recognition sequence (5'GG1cc3') is as numerous in such a small
D N A (with G + C = 35%) as the restriction analysis suggested.
Results with BGMV dsDNA confirm this. We cannot explain
the apparently anomalous behaviour of HaeIII with BGMV
ssDNA.
These results indicated that BGMV may contain a divided
genome with ssDNA components of the same physical size but
different nucleotide sequence. The same result could be expected, however, if two related but distinct strains of the virus were
present. We tested the divided genome hypothesis by investigating the dilution kinetics of infection by BGMV D N A in bean
cell protoplasts. Protoplasts were prepared from leaf mesophyll
cells of healthy beans (N.B.B and R.M.G., unpublished results),
inoculated with different amounts of BGMV DNA, and infected
cells scored after 24 h by immunofluorescence. The slope of the
infectivity dilution plot (Fig. 2) was closer to two than one.
Part of the initial evidence that led to the discovery of divided
genomes in other plant virus groups were infectivity dilution
curves whose slopes deviated significantly from nit^'^.*^. Our
results now indicate that BGMV and possibly other
geminiviruses may also contain a divided genome. This result
extends to ssDNA viruses the interesting prevalence of divided
genomes among the plant viruses and we believe it is the first
example of a divided genome in a D N A virus. We conclude that
the BGMV D N A genome is encapsidated in paired particles of
two classes which, according to all present evidence, differ only
in the sequence of the circular D N A molecule they contain. If
this hypothesis is correct, the infectivity dilution kinetics of the
virus particles should also approach a value of two;
unfortunately, we have not found an experimentally useful local

lesion host for BGMV and attempts to infect protoplasts by

inoculation with whole virus have been unsuccessful.
This discovery is expected to complicate further studies on the
molecular biology of BGMV. T o sequence o r map gene
functions on BGMV DNA. we must be able to resolve two D N A

log [DNA]

Fig. 2 Relationship between concentration of BGMV DNA in

the inoculum and proportion of inoculated bean cell protoplasts
infected. Plots shown are data from one (expt. 3) of five separate
experiments (---) and theoretical plots (-) for a, one-hit and 6,
two-hit kinetics. Protoplasts were isolated from the mesophyll of
Top Crop bean leaves and inoculated with various amounts of
DNA (data are in ~ g m l - ' ) .Infected cells were scored by an
indirect fluorescent-antibody method using anti-BGMV rabbit
serum and fluorescein-labelled anti-rabbit y-globulin goat serum.
Data were corrected for the number of living cells determined by
phenosaffranin stainingz1 and then transformed to logit values2
(log x/(l - x)) where x is the proportion of living cells infected. The
logit transformation was used to correct for the decreased
availability of infectable cells at higher DNA concentrations. Data
from five experiments were plotted and the lines drawn by linear
regression analysis. The slopes, correlation coefficients (r) and the
probability level (P) at which r values are significant are shown
below:

Expt
Slope
r
P

O 198lNature Publishing Group

1
1.99
0.97
0.01

2
1.71
0.96
0.05

3
1.71
0.98
0.01

4
1.78
0.99
0.01

5
1.78
0.97
0.01

Nature Vol. 289 22 January 1981

species which are indistinguishable by size but differ in nucleotide sequence. Perhaps the best approach would be to find
restriction endonucleases that will digest at one site in only one
of the two sequences and then to clone that sequence out of a
mixture of the two using recombinant D N A techniques.
We acknowledge the collaboration and advice of D r K. S.
Browning, and thank Drs C. L. Niblett and R. W. Blakesley for
discussions and D r P. D. Shaw for comments on an earlier draft.
Research was supported by a grant (7800549) from the USDA
Competitive Research Grants Office and from the Illinois Agricultural Experiment Station (project 68-366).
Received 15 August. accepted 13 November 1980.

1. Goodman, R. M. in Handbook of Plant Viral Diseases (ed. Kurstak, E.) (North

Holland/Elsevier. Amsterdam. in the press).
Matthews, R . E. F. Inaruirology 11, 133-135 (1979).
Goodman, R. M. Nature 266,54-55 (1977).
Harrison, B. D . et al. Nature 270,760-762 (1977).
Reisman, D . , Ricciardi, R. P. & Goodman, R. M. Virology 97, 388-395 (1979).
Francki, R. I. B., Hatta, T., Boccardo, G . & Randles, J. W. Virology 101,233-241 (1980).
Ward, D . C. &Tattersall, P. Replication of Mammalian Paruouiruses (Cold Spring Harbor
Laboratory, New York. 1978).
8 . Denhardt, D . T..Dressler, D . & Ray. D . S. 7'he Single-SrrandedDNA Phages (Cold Spring
Harbor Laboratory. New York, 1978).
9 . Bock, K. R., Guthrie, E. J. & Woods, R. D . A n n . appl. Biol. 77,289-296 (1974).
10. Galvez. G . E. & Castano. M. Turrialba 26, 205-207 (1976).
11. Goodman, R. M., Shock,T. L., Haber. S., Browning, K. S. & Bowers,G. R. Jr Virology 106,
168-172 (1980).
12. Hatta, T. & Francki, R. I. B. Virology 92,428-435 (1979).
13. Goodman, R. M., Bird, J. & Thongmeearkom, P. Pkytopathology 6 7 , 3 7 4 2 (1977).
14. Goodman. R. M. & Bird. J. AABICMIDescriptions of Plant Viruses No. 192 (Commonwealth Mycological Institute, Kew, 1978).
15. Browning, K. S. thesis, Univ. Illinois (1980).
16. Blakesley, R. W. & Wells, R. D . Nature 257,421-422 (1975).
17. Horiuchi, K. & Zinder, N. D . Proc, nam. Acad. Sci. U . S . A . 72, 2555-2558 (1975).
18. Sanger, F. e t a / . I. molec. Biol. 125,224-246 (1978).
19. Fulton, R. W. Virology 18, 466-485 (1962).
20. Van Vloten-Doting, L., Kruseman, J. & Jaspars, E. M. Virology 34,728-737 (1968).
21. Widholm, J. Stain Technol. 47, 189-194 (1972).
22. Van der Planck, 1. E. Plant ~ i s e i s e Epidemicsand
:
Control (Academic. New York. 1963).
2.
3.
4.
5.
6.
7.

Table 1 Requirements of the system

Omissions

dCMP incorporated (pmol)

None
MK12
dTTP
ATP
CTP, GTP, UTP
Creatine phosphate
Cyclic AMP
Amino acids

94.9
0.8
5.1
2.1
3.5
1.O
36.7
40.3

Extract was prepared from C600 (pKN177) and supplemented with

exogenous pKN177 DNA (30 pg ml-I). The complete system is the
standard reaction mixture described in Fig. 1 legend with an additional
0.5 mM of each of the 20 amino acids. [(Y-'~P]~CTP
(1,000 c.p.m. pmol-l) was used as radioactive label. Reaction mixtures
(25 p1) were incubated in duplicate at 30 "C for 60 min.

T o develop an in vitro replication system for the R 1 plasmid,

we used the mini-R1 plasmids pKN177 and pKN182, which had
been spontaneously generated from the R 1 Cop mutant
pKN104 (ref. 6). Plasmid pKN182 has a molecular weight of
3.9 x lo6 and carries only the replication region of the parent
plasmid, whereas pKN177, having a molecular weight of -9 x
lo6, also contains the ampicillin-resistance gene of R1. The Cop
phenotype is caused by a single base substitution in the replication region7, which prevents the expression of a trans-acting

Plasmid R 1 DNA replication

dependent on protein
synthesis in cell-free extracts of E. coli
Ramon Diaz & Kurt Nordstrom
Department of Molecular Biology, Odense University,
DK-5230 Odense M, Denmark

Walter L. Staudenbauer
Max-Planck-Institut fiir Molekulare Genetik, Abt. Schuster,
D-1000 Berlin 33, FRG

I/-20

Bacterial plasmids serve as model systems for studying the

regulation of DNA replication1. Elucidation of the molecular
mechanisms involved in plasmid DNA synthesis requires the
development of efficient cell-free plasmid replication systems.
Such in vitro systems have previously only been described for
Col El-type plasmids2" and for the R6K p l a ~ m i d ~Here
' ~ . we
report that extracts of Escherichia coli can carry out the
complete replication of miniplasmids derived from the antibiotic-resistance plasmid R1. This R1 replication system differs
from the previously described ColEl and R6K systems in its
strict dependence on DNA-directed protein synthesis. We
believe this to be the first report of the functional coupling of the
three fundamental reactions of genetic information transfer
(transcription, translation and replication) in a cell-free system.

40

60

Time (min)
Fig. 1 Kinetics of pKN177 DNA synthesis. A cell-free extract
was pre ared from E. coli C600 (pKN177) by the freeze-thaw
method P.In the presence of 1 mM EDTA and 1 mMdithiothreitol.
Standard reaction mixtures contained final concentrations of
50 mM HEPES p H 8.0,100 mM K C 1 , l l mM magnesium acetate,
15 mM creatine phosphate, 0.1 mgml-' creatine kinase, 2 mM
ATP, 0.4 mM each of CTP, GTP and UTP, 0.05 mM NAD,
0.05 mM cyclic AMP and 0.025 mM each of dATP, dCTP, dGTP
and 3 ~ - d T T(500c.p.m.
~
pmol-'). Reaction mixtures (500 p1)
containing 250 p1 extract (7.5 mg protein) were incubated at 30 O C
without (a or with (0)
addition of exogenous pKN177 DNA
(30 pg ml- ). At the times indicated, 25-p1 samples were removed
and assayed for acid-insoluble radioactivity.

@ 1981 Macmillan Journals Ltd

327

Nature Vof. 289 22 January 1981

Table 2 Replication capacity of extracts prepared from plasmidcarrying and plasmid-free strains
-

Source of extract

- -

Plasmid DNA added

None
pKN177
ColE1

Standard reaction
mixtures ( 2 5 ~ 1 ) included
3~-dTT~
(500 c.p.m. pmol-') and 0.5 mM of each of the 20 amino acids. Exogenous pKN177 DNA (30 pg ml-') or ColEl DNA (50 ~g ml-') were
added as indicated. Incubations were performed in duplicate at 30 O C for
60 min. Values shown are dTMP incorporated (pmol).

plasmid D N A . Furthermore, a m a r k e d stimulation of D N A

synthesis was observed o n addition of exogenous p K N 1 7 7
D N A . T h e kinetics of d T M P incorporation with or without
addition of plasmid D N A is shown in Fig. 1. It c a n b e seen that
t h e r e is a lag period of 10 min before t h e onset of D N A synthesis
a n d that t h e incorporation levels off after 40 min of incubation.
Addition of p K N 1 7 7 D N A stimulates t h e total incorporation

Fraction no
Fig.2 Alkaline CsCl gradient centrifugation of in vitro
synthesized DNA. Extracts were incubated for 60 min at 30 "C in
standard reaction mixtures (150 pJ) without or with addition of
exogenous piasmid DNA (30 g g ml-I). Reactions were stopped by
addition of EDTA (50 mM) and the DNA was purified by gel
filtration through a Sepharose 6B solumn. Aliquots were layered
on preformed 4.2-mI alkaline CsCl gradients (density 1.21.4 g ml-' in 0.2 M NaOH) and centrifuged for 35 min in a Beckman SW 50.1 rotor at 5 "C. Fractions (100 p1) were collected from
the bottom of the tubes and assayed for radioactivity. Centrifugation is from right to left. The arrows indicate the positions of
supercoiled pKN177 and pKN182 DNA, respectively. a, C600
(pKN177) extract incubated without (0)or with (0)
addition of
pKN182 DNA; 6, C600 extract incubated with addition of
pKN177 DNA.

EtRr

negative control function6. D u e t o their reduced size a n d

increased copy number (-25 copies p e r chromosome), these
mini-R1 plasmids offer experimental advantages f o r setting u p
a n in uitro system for R l D N A replication.
Plasmid D N A synthesis was first studied in extracts f r o m t h e
plasmid-carrying E. coli strain C600 ( p K N 1 7 7 ) with t h e
assumption that mini-R1 replication may require a plasmidcoded replication protein. W e found that extracts prepared by
freeze-thaw lysis of lysozyme-treated cells c a n catalyse a low but
significant amount of d T M P incorporation into endogenous

Autoradiogram

Fig. 3 Effect of rifampicin on plasmid DNA replication. Extract

from C600 (pKN177) was incubated for 60 min at 30C in standard reaction mixtures ( 1 5 0 ~ 1 )containing [ a - 3 2 ~ ] - d ~ ~ ~
(1,000 c.p.m. pmol-') plus exogenous pKN182 DNA (30 ~g ml-I)
without or with addition of rifampicin (25 p,g ml-I). The reactions
were stopped with EDTA and the DNA was purified as described
in Fig. 2 legend. Aliquots of the untreated control (lane 1) and of
the rifampicin-treated sample (lane 2) were electrophoresed in a
1.0% agarose gel for 3.5 h at 100 V. The gel was stained with
ethidiurn bromide and autoradiographed. The arrows indicate the
positions of supercoiled pKN177 and pKN182 DNA, respectively.

O 198lNature Publishing Group

328

Nature Vol. 289 22 January 1981

-15-fold. The optimal concentration of exogenous plasmid

DNA is about 3 nM; higher concentrations of pKN177 DNA
are inhibitory. Similar results were obtained with pKN182 DNA
(data not shown).
The requirements of the system (supplemented with exogenous pKN177 DNA) are summarized in Table 1. It was found
that the magnesium ion concentration is of critical importance.
Optimal dTMP incorporation requires 9-11 mM MgC12,
whereas almost no DNA synthesis can be detected at 5 or
15 mM MgCI2. In addition, the system requires not only deoxynucleoside triphosphates but also all four ribonucleoside triphosphates and is strictly dependent on an ATP-regenerating
system. A variable degree of stimulation, depending on the
extract preparation, was noted on adding cyclic AMP and/or an
amino acid mixture. The replicative capacity for mini-R1 DNA
is lost if the extract is prepared in the absence of dithiothreitol.
Having optimized the reaction conditions, we investigated
whether exogenous pKN177 DNA can also replicate in extracts
from E. coli cells that do not carry any plasmid. It was found that
extracts from plasmid-free cells can catalyse a significant
amount of DNA synthesis when supplemented with pKN177
DNA (Table 2). However, the replicative capacity for pKN177
DNA of such extracts is an order of magnitude lower than that of
extracts from cells carrying this plasmid. In comparison, exogenous ColEl DNA is replicated by both types of extracts with
similar efficiency.
To characterize the reaction product, the in vitro synthesized
DNA was analysed by alkaline CsCl gradient centrifugation
(Fig. 2). To distinguish between endogenous and exogenous
DNA, extracts from C600 (pKN177) were supplemented with
pKN182 DNA. Comparison of the two superimposed gradients
shown in Fig. 2a indicates that the exogenously added pKN182
DNA not only functions as template for its own replication but
also stimulates the synthesis of the endogenous pKN177 DNA.
This stimulation is R1 specific as it was consistently observed on
addition of mini-R1 plasmids but not on addition of ColEl
DNA (data not shown).
The incorporated label is found in two peaks of fast-sedimenting DNA,.corresponding to closed-circular monomers of
pKN177 and pKN182. Only a minor fraction of the label
sediments in the position of open-circular DNA. Similarly, the
reaction product of pKN177 DNA replication in extracts from
plasmid-free cells consists predominantly of supercoiled
monomeric plasmid DNA (Fig. 2b). No peak of labelled DNA
fragments sediments more slowly than unit-length single
strands, as would be expected for material derived from replicative intermediate^^'^. These results indicate that the in vitro
system carries out the complete replication of both endogenous
and exogenous mini-R1 DNA without detectable accumulation
of partially replicated molecules.
The effects of various antibiotics on mini-Rl replication in
oitro are shown in Table 3. As expected for semi-conservative
DNA replication, the incorporation of dTMP is entirely blocked
by inhibitors of DNA gyrase (novobiocin and oxolinic
An unexpected finding is the complete inhibition of mini-R1
replication by antibiotics which interfere with different stages of
protein synthesis (chloramphenicol, puromycin, streptomycin
and tetracycline). Plasmid replication evidently is coupled
tightly to translation, because DNA synthesis stops within
10 min of the addition of chloramphenicol (data not shown).
Plasmid DNA synthesis is also strongly affected by rifampicin, a
specific inhibitor of RNA polymerase. The residual rifampicinresistant incorporation observed with extracts from plasmidcarrying (but not plasmid-free) cells is probably due to the
presence of endogenous plasmid-specific RNA.
The inhibitory effect of rifampicin on plasmid replication in an
extract containing endogenous pKN177 DNA and exogenous
pKN182 DNA was further analysed by agarose gel electrophoresis of the reaction products (Fig. 3). Comparison of the
ethidium bromide-stained gel with the autoradiogram shows
that the 32P-label is incorporated exclusively into supercoiled
plasmid DNA. Interestingly, only the endogenous pKN177

Table 3 Effects of antibiotics on pKN177 D N A synthesis

Source of extract
C600 (pKN177)
C600

Addition (kg ml-I)

None
Novobiocin (25)
Oxolinic acid (50)
Chloramphenicol(100)
Puromycin (100)
Streptomycin (100)
Tetracycline (100)
Rifampicin (25)

101.7
1.6
2.4
1.1
1.5
1.6
1.4
14.9

13.8
0.2
0.3
0.3
0.2
-

0.1

were supplemented with exoStandard reaction mixtures (25

genous KN177 D N A (30 kg ml-I) and 0.5 mM of each of the 20 amino
acids. %-dTTP (500 c.p.m. pmol-l) was used as radioactive label.
Incubations were performed in duplicate at 30 OC for 6 0 min. Values
shown are dTMP incorporated (pmol).

DNA but not the exogenous pKN182 DNA can function as a

template for rifampicin-resistant DNA replication.
The data presented here demonstrate that mini-R1 replication in vitro requires both transcription and translation. A
requirement for de novo RNA and protein synthesis has previously been observed in vivo for the uncontrolled replication of
a runaway-replication mutant of R1 (ref. 11). It was postulated
that the translation product is an R1-specific initiator protein
that is not re-usable. Indirect evidence that R1 carries a gene
essential for autonomous replication has been presented by
Kollek et al.I2. This gene is considered analogous to the rep A
gene of the closely related RlOO plasmid13. Evidently, mini-R1
replication in vitro is also strictly dependent on the plasmid
DNA-directed synthesis of an essential replication protein. The
R1-specific stimulation of endogenous plasmid replication by
addition of exogenous mini-R1 DNA can thus be explained by
assuming that the presumptive rep A gene product is a transacting protein. The transcription requirement would then reflect
the synthesis of a messenger RNA for this protein. On the other
hand, the differential effect of rifampicin on endogenous and
exogenous plasmid DNA synthesis (see Fig. 3) indicates that a
cis-acting RNA asrociated with the template DNA may also be
involved in R1 DNA replication.
Results similar to those presented here for pKN177 and
pKN182 have also been obtained for other mini-R1 plasmids
(R.D., unpublished data). The plasmids tested include
pKN1562 the copy control of which corresponds to the parent
R1 plasmid14.Therefore, this in vitro system is not restricted to a
particular type of Cop mutant and should prove useful for a
detailed analysis of R1 replication control.
We thank Dr Soren Molin for supplying the plasmids pKN177
and pKN182 and for helpful suggestions, and Dr Heinz Schuster
for providing support and laboratory facilities. R.D. was
supported by a short-term fellowship from EMBO and a grant
from the Danish MRC.
Received 15 August; accepted 21 November 1980.

R.

A . Rev. Gener. 13,355-391 (1979).

1. Kolter, R. & Helinski, D.
2. Tomizawa, 1. in D N A Synthesi.~:Present and Future (eds Molineux, & Kohiyama,
797-826 (Plenum, New York. 1978).
3. Staudenbauer, W. L. Curr. Topics Microbiol. Immun. 83.93-156 (1978).
4. Inuzuka, M. & Helinski. D. R. Biochemistry 17, 2567-2573 (1978).
Proc. nam. Acad. Sci. U S A . 75,5381-5385 (1978).
5. Inuzuka, M. & Helinski, D.
6. Molin, S. & Nordstrom, K. J. Bacr. 141, 111-120 (1980).
Molec. gen. Gener. (in the press).
7. Stougaard, P., Molin. S., Nordstrom, K. & Hansen, F.
8. Staudenbauer. W. L. Molec. gen. Genet. 145,273-280 (1976).
9. Staudenbauer, W. L. J. molec. Biol. 96, 201-205 (1975).
10. Staudenbauer, W. L. Eur. I. Biochem. 62,491-497 (1976).
11. Uhlin. B. E. & Nordstrom. K. Molec. gen. Gener. 165. 167-179 (1978).
Oertel, W. & Goebel. W. Molec. gen. Gener. 162, 51-57 (1978).
12. Kollek,
13. Yoshikawa, M. J. Bact. 118, 1123-1131 (1974).
Uhlin, B. E., Gustafsson, P. & Nordstrom, K. J. Bacr. 138.70-79
14. Molin, S., Stougaard,
(1979).

I.

R.

G.

R..

O 198lNature Publishing Group

P.,

M.)

Nature Vol. 289 22 January 1981

329

Highly conserved lipoprotein

assembly in teleost
and amphibian yolk-platelet crystals
R. H. Lange
Institute for Anatomy and Cytobiology, University of Giessen,
Aulweg 123, D-6300 Giessen, FRG

Ohlendorf et al.' have reported that yolk-platelet crystals in the

frog, Xenopus laevis, were orthorhombic (space group P21221)
and consisted of elongated asymmetric units (-5.5 x 11.5 x
25 nm). The presence of an orthorhombic lattice in amphibian
yolk platelets fulfilling the symmetry requirements of crystals
had been missed in earlier structural interpretationszp3.A
comparative study of teleost, urodele and anuran yolk-platelet
crystals performed in our laboratory has shown the widespread
occurrence among equatic vertebrates of an orthorhombic
lattice with unit-cell sides similar to those of Xenopus. However,
we report here that the space group (P212121) and certain
features of the macromolecular complex are at variance with
those reported for Xenopus'.
The material studied consisted of the amphibians Rana
temporaria (gastrula) and Triturus ~ p(uncleaved,
. ~
two-cell and
blastula stages) and the teleost Pelvicachromis pulcher5 (Cichlidae; immature and mature eggs). Thin-sectioned specimens
were fixed (1% glutaraldehyde/l% Os0,/0.1 M phosphate
buffer) and subjected to two very different embedding procedures (epoxy-resin and glutaraldhyde-urea6). On applying
electron diffraction and tilting, the presence of an orthorhombic
lattice was confirmed in all three cases. This was achieved by
establishing the presence of symmetry mmm for the reciprocal
orthorhombic lattice in each case. Moreover, the unit-cell sides
a, b, c were almost identical among the three species studied
for wet
and-falling in the range derived by Ohlendorf et
and lyophilized platelets-were also very close to those of X .
laevis (Fig. 1 legend). This observation supports our view that
we are dealing with much the same structure in teleost and
amphibian yolk platelets and that our data are also relevant to

Fig. 2 Electon micrograph of a teleost yolk-platelet crystal

viewed along the short axis a of the unit cell (A) and tracings of the
same [loo] projection (B) made approximately to the same scale.
Orientation and size of the unit cell have been unequivocally
defined by corresponding diffraction patterns; the cell may,
however, be displaced. The footprint-like pattern shown can only
be interpreted by assuming 2,-screw-axes parallel to the unit-cell
sides b and c. Note that these two screw-axes have unique positions
in this projection, there are no alternative positions for them. If the
primitive orthorhombic lattice has been correct1 assigned then
only two space group interpretations are possible Y ' (P212121and
P22'21, but not P2'22,); these interpretations are illustrated in B.
P22121 implies a 2-fold symmetry axis ( 0 ) perpendicular to the
plane of the paper in the middle of each pair of dark elliptic bodies
and thus a location of these bodies in the same height along this
axis; PZ12,2, has a 2-fold screw-axis
) at this site, implying that
the height of neighbouring dense bodies differs by a / 2 , which is in
fact shown to be the case (Fig. 3). A is a high-power enlargement of
a raw electron micrograph originally recorded on 35-mm film at
magnification x 25,000 and shows a certain degree of distortion
due to sectioning. Dense particles (probably phosvitin complexes,
see text) are numbered in the same way as in Fig. 3B. International
symbolsM;b 16.5 nm.

(4

Fig. 1 Electron diffraction patterns (tracings) obtained in Triturus

sp. when the crystals were oriented as indicated. In the other two
species studied, only changes in the intensities of single reflections
were noted. The means of unit-cell lengths as determined from
such patterns in the three species had the following rangers (in
parentheses the X-ray data from X. laevis for dried7 and wet'
preparations are given for comparison): a = 8 . 3 . . . 9 . 1
(7.9.. .8.9) nm; b = 15.8.. . 17.7 (15.8.. . 17.2) nm; c = 16.6. .
19.3 (17.6.. . 19.6) nm; the lower limit of our data refers to
epoxy-resin embedding following dehydration in acetone, the
upper limit to glutaraldehyde-urea embedding in the partly
hydrated state.

the findings in Xenopus'~'. The evolutionarily conservative

behaviour of the molecular assembly is underlined by the very
similar appearance of 12 indexed diffraction patterns (for [loo],
[OlO], [OOl], see Fig. 1) as well as seven identified crystal
projections (electron images) in the three species studied.
The absence of reflections (100,300,010,030,050,001,003,
005, 007) in the electron diffraction patterns suggests the
presence of 2-fold screw-axes parallel to all three unit-cell sides,
that is, space group P2,212,. Conclusive proof for the space
group, however, can be obtained in this case from the mass
distribution in electron micrographs (Figs 2, 3). The most
revealing projection with respect to molecular arrangement is
the previously unstudied projection along the a axis ([loo], Fig.
2), the second most revealing that along the c axis ([OOl], Fig. 3).
It is evident from Figs 2 and 3 that there are four asymmetric
0 1981 Macmillan Journals Ltd

Nature Vol. 289 22 January 1981

330

obo
O

0
b

0
O

0
O

0
O

0
O

0
O

produces a reduction in length of some 10% (due partly to

dehydration, especially for epoxy-resin embedding), various
degrees of distortion (due to sectioning) and a resolution limit
between 1.5 and 2 nm. All our findings were reproduced-and
most of our work was indeed done-using glutaraldehyde-urea
embedding in the partially hydrated state6.The original authors
of this procedure have proved that the extremely lipid-rich
myelin sheath does not lose its natural birefringence and
dimensions when subjected to this embedding method6. This
implies that the material studied in our laboratory has been
preserved as adequately as possible. The sectioning process,
furthermore, precludes the bias produced by splitting isolated
platelets along preferential planes1 and thus makes all crystal
projections available with the same probability.
I conclude that teleost and amphibian yolk platelets are quite
similar assemblies of lipoprotein complexes (orthorhombic,
P2,2,2,, almost identical unit-cell lengths for all species studied
so far) and, therefore, contain a macromolecular constituent of
highly conserved tertiary structure. Space group P21221and the
three-dimensional s h a ~ of
e the asvmmetric unit as described in
isolated yolk platelets'of X. laeuisl do not agree with observations made on our comparative material.
I thank Professor S. Haussiihl for crystallographic advice, Mr
G. Magdowski for technical assistance and the Deutsche Forschungsgemeinschaft for support.
Received 1 July; accepted 20 November 1980.

Fig. 3 Electron micrograph of a newt platelet crystal viewed

along the c axis (A) and schematic tracing of the same [OOl]
projection (B) made approximately to the same scale. It can be

seen that only a screw diad (space group P212121)and not a

rotation diad (space group P22121) parallel to unit-cell side a
correctly describes the matter distribution.Note that positioning of
these additional symmetry elements is not deliberate for
geometrical reasons". The rotation diad not consistent with the
observed mass distribution is shown dashed on the right half of B.
Compare A with the negatively stained preparation in Fig. 1 of the
report by Ohlendorf et a/.'. For further details see Fig. 2 legend.

units per unit cell, as in X. laevisl. There is no evidence that the

complex exceeds the length of the b axis1; however, this point
awaits confirmation by further analysis, as does the positioning
of a local 2-fold axis1 of the dimeric structures.
To explain the discrepancies between the findings in X.
laevis1.' and our amphibian and teleost data, we should consider
the methods used. As shown by investigations using negative
staining techniques in electron microscopy of protein molecules9, there is commonly an overlap of negative and positive
staining of such preparations. Such overlap is inevitable and
must be considerable when a highly anionic protein like the
phosvitin moietys of the yolk-platelet subunit is stained with a
cationic dye containing uranyl ions1. The dark spots in the
positively stained crystals in [loo] and [OOl] projection (Figs 2,
3)-unrecognizable in unstained sections but quite evident in
sections stained with uranyl acetate at low pH (3.5)-give the
most probable location of the phosvitin moiety2; in contrast to
previous views2, however, it accommodates both phosvitin
molecules of the dimeric complexs. This significant part of the
complex may largely have escaped representation in the negatively stained preparations of X. laevis platelets1. In fact, the
negatively stained preparation (Fig. 1 of ref. 1) is almost identical to the [001] projection (Fig. 3) in the three species studied
here following positive staining of crystal sections with uranyl
acetate and lead citrate. In a negatively stained preparation, the
space occupied by stain is interpreted as being void of the
macromolecule under study; this makes it very unlikely that the
results of the three-dimensional reconstruction of the amphibian lipovitellin-phosvitin complex1 is reliable.
The procedure used in the present study yields results
comparable with X-ray data of protein crystals10,except that it

1. Ohlendorf, D. H., Wrenn, R. F. & Banaszak, L. J. Nature 272, 28-32 (1978).

2. Wallace, R. A. Biochim. biophys. Acra 74, 505-518 (1963).
3. Honjin, R. in Recent Progress in Electron Microscopy of Cells and Tissues (eds Yamada, E. et
a l . ) 95-108 (Thieme, Stuttgart & lgaku Shoin, Tokyo. 1976).
4. Lange. R. H. Proc. 7th Eur. Congr. Electron Microscopy, Vol. 2 (eds Brederoo, P. & de
Priester, W.) 624-625 (Leiden, 1980).

Lange. R. H. Cell Tissue Res. 209, 511-513 (1980).

Peterson, R. G. & Pease. D. C. I. ultrasrruct. Res. 41, 115-132 (1972).
Ohlendorf, D. H., Collins, M. L. & Banaszak, L. J. I. molec. Biol. 99, 143-151 (1975).
Taborsky, G. A d v . Protein Chem. 28, 1-210 (1974).
Unwin, P. N. T. J. molec. Biol. 98,235-242 (1975).
Lange, R. H., Blodorn, J . , Magdowski, G. & Trampisch, H. J. J. ultrasrruct. Rex. 68.81-91
(1979).
11. International Tables for X - R a y Crystallography Vol. 1, 104-105 (Kynoch. Birmingham.
1969).

5.
6.
7.
8.
9.
10.

Corrigenda
In the letter 'West Antarctic ice sheet fluctuations in the
Antarctic Peninsula area' by D. E. Sugden and C. M. Clapperton, Nature 286, 378-381 (1980), the correction factor of
-750 radiocarbon years for Antarctic Peninsula shells was
added to rather than subtracted from the Holocene 14Cdates of
6,930 60 and 7,200 50 yr BP for the outer and inner fractions
of SRR-1500. The adjusted dates should be 6,180 and 6,450
radiocarbon years BP respectively. The discussion concerning
the possibility of an ice-free George VI Sound in the Holocene
thus refers to conditions 6,000-6,500 radiocarbon years ago and
not -8,000 years ago. This change does not affect the
conclusions of the paper.

In the letter 'Selective killing of mycoplasmas from

contaminated mammalian cells in cell cultures' by Marcus et al.,
Nature 285, 659-661 (1980), the footnote to Table 2 should
refer to a 400-W lamp, not 40-W as published.

Erratum
In the letter 'Expression of H-2, lamin and SV40T and TASA
on differentiation of transformed murine teratocarcinoma cells'
by Knowles et al., Nature 288,615-618 (1980), two lines were
omitted from the bottom of the left-hand column on page 617; it
should read:
. . . . . . differentiates seems to express the H-2Db private specificity. It has been
observed that somatic cell hybrids between F9 and differentiated cells can . . . . . .

O 198lNature Publishing Group

33 1

Nature Vol. 289 22 January I981

. . . and when the cupboard is bare?

Kenneth E. Boulding
THESEtwo books have a somewhat similar
theme, but are about as different in style
and approach as two books could be. The
Krenz volume has a somewhat narrower
scope, dealing primarily with energy,
whereas that by Barnet takes materials into
account; the basic theme of adaptation to
the exhaustion of non-renewable resources
is common to them both. The book by
Krenz is sober, well documented, is based
solidly on the best available quantitative
information, and presents a rounded
picture, even though the author is not
afraid to express value judgments on such
matters as the future of nuclear energy. The
Barnet volume is sparkling, anecdotal in
style, rarely presents a complete, wellrounded picture, and is constantly engaged
in implied, n o t t o o explicit value
judgments.
Krenz begins with a clear general consideration of the energy crisis and energy
conservation, goes on to a thoughtful
discussion of environmental and resource
constraints, analyses the situation in regard
to fossil fuels, then has a key chapter on
moreeffective energy utilization, which is a
better way of describing "conservation".
There follows a discussion of "Nuclear
Energy: A Faustian Bargain?" which
expresses the anxieties and ambiguities
involved without dismissing them. Then
there is an excellent section on alternative
sources of energy, which is both sober and
sobering to the over-optimistic. The book
concludes with a discussion of the need for
a new ethic to deal with the probability of a
more austere future.
The Barnet book starts with a rather
uneven discussion of the "Post-Petroleum
World", and continues with a section on
"Guns, Butter and Oil: Changing the Face
of Power", which also deals with minerals,
food and water, and relates increasing
scarcity to what he calls "the new
international military order", particularly
in the Third World. He concludes with a

The Lean Years: Politics in the Age of

Scarcity. By R. Barnet. Pp.349. (Simon
and Schuster: 1980.) $12.95. T o be
published in the UK, in April, by Abacus,
pbk 2.50. Energy: From Opulence to
Sufficiency. By J . H . Krenz. Pp.264.
.(Hemisphere/Holt, Rinehart & Winston:.
1980.) $24.95, 15.50.
section entitled "The Global Factory: The
Planning of Scarcity", in which he foresees
a world employment crisis resulting from
the disruption of traditional agriculture
and the population explosion. This last
chapter ends with a rather vague plea for
decentralization without any clear
indication as to how to achieve it. Here,
Barnet sees little reason for optimism, and
what emerges is the general feeling that bad
boys will continue to be bad because
whoever stops them will be worse. It is,
indeed, a gloomy end to liberal hopes.
In terms of suggested solutions to the
problem, Krenz, as befits the somewhat
narrower scope of his book, lays great
stress on the value of conservation and the
large technical potential which is latent in
conservation practices, in just getting more
in terms of human values out of the same
input of energy. There are estimates which
suggest that we can maintain much the
samestyle of life that wedonow with 25 per
cent of our current energy input, o r with a
bit of luck and technical change, maybe 50
per cent. This, in a sense, is a common task
which must be shared widely. Barnet, on
the other hand, leans towards conspiracy
theories of what is the matter with us, does
not lay so much stress on conservation, and
ends u p arguing for a rather vague kind of
decentralization as a way of saving us from
the concentrated power either of large
corporations or of socialist states.
The contrast between the two books
illustrates at least one of the virtues of
quantified data. The Krenz book, even
though it is somewhat abstracted from
reality, does give us an overall picture of
the proportions of the problem, which the
qualitative anecdotal approach of Barnet
does not really permit. There is a basic
difference, perhaps, between the scientific
and the journalistic approach.
Both books would be improved by a little
more economics. Although neither of the
authors is an economist, Krenz'recognizes
the importance of economic considerations, though he is perhaps a little overimpressed with purely physical concepts of
efficiency. From the human point of view,
the only significant concepts of efficiency
are those which involve human values.
Thus, a process can be highly inefficient in
O 198lNature Publishing Group

terms of transformation into a particular

form of energy, and even highly inefficient
accordipg to second law efficiency, that is,
transformation of the availability of energy
for work. But if the energy output is
important by human standards and the
input is low-valued, the process can be very
efficient in human valuation terms. Thus,
the rise of electric power, especially after
about 1910, was inefficient by some
physical standards but was extremely
efficient in terms of human valuations. In
commodity output at least, as measured by
gross national product, per unit of energy
input increased sharply in those decades,
largely because of the greater convenience
of subdivisibility of electricity as compared
to alternative means of transporting
energy. Both volumes neglect the problem
of energy transport, which is most important from the point of view of human
valuations. Energy when and where and in
the form in which we want it almost
invariably involves its transportation.
Energy: From Opulence to Sufficiency
by Krenz can be strongly recommended to
those who want an overall survey of the
technological problems of energy in fairly
non-technical language. The Barnet
volume, The Lean Years, would appeal to
those who enjoy liberal-radical rhetoric,
who like to think that everything that is
wrong with us is somebody else's fault or
the result of a defective political structure
of society, which no doubt it often is.
Neither book, however, really comes to
grips with the way in which decision
making and the great network of social
interaction actually move us towards
overall betterment or worsening. But this,
I
nodoubt, is too much toask.

Kenneth E. Boulding is Distinguished Professor

of Economics Emeritus and Director of the
Program of Research on General Social and
Economic Dynamics at the Institute of
Behavioral Science, Universify of Colorado,
Boulder.

332

Nature Vol. 289 22 January 1981

Progress and prospects in tumour virology

Peter W. J. Rigby
DNA Tumor Viruses. Molecular Biology
of Tumor Viruses, Part 2,2nd Edn. Edited
by J. Tooze. Pp.945. (Cold Spring Harbor
Laboratory: 1980.) $65, $78 outside USA.
THEfirst edition of this monograph on the
molecular biology of tumour viruses was
both timely and influential. It was timely
because its publication, in 1973, coincided
with the advent of techniques for the
physical mapping of viral genomes and it
was influential because it acted as an
invaluable source for the many molecular
biologists, including myself, who began
working o n tumour viruses in the hope that
they would, in conjunction with the new
techniques, provide insights into the
mechanisms of eukaryotic gene regulation.
This second edition thus had much to live
u p to and it has, in general, succeeded most
admirably.
Like its predecessor, the book is a
compilation of chapters written by a
variety of authors and brought together
under the editorship of John Tooze. In his
preface Tooze admits to some of the
problems that bedevilled the preparation
of the book; in particular, it proved
impossible to persuade s o many authors to
produce their material a t exactly the same
time. While many of these problems have
clearly been resolved, there remain cases
where it is apparent that closely related
chapters were written, o r finally revised, a t
quite different times and there are thus
inconsistencies of fact or interpretation.
These deficiencies are, however, more than
compensated by the book's strengths. It
aims successfully at a wide audience, being
written in a clear, didactic style quite
approachable by final-year undergraduates, yet it is alsc sufficiently comprehensive and penetrative to be of great value
to established workers in the field.
The first section comprises six chapters
concerned with the smallest DNA tumour
viruses, the papovaviruses and papillomaviruses. Viruses of the former class are now
understood in considerable detail which is
reflected in separate chapters on genomic
organization, the lytic cycle, cellular
transformation and genetics. Each of these
contributions was written by different
authors which leads to some unevenness of
approach due principally, one suspects, to
the writing having been done at different
times. This shows particularly in the
treatment of the middle T-antigen of
polyoma and of the cellular 53K protein
f o u n d in SV40-transformed cells. I
particularly enjoyed the chapter by Nick
Acheson on the lytic cycle of SV40 and
polyoma. This deals in some detail with
experimental problems and gives the reader
an excellent feel for the certainties and
uncertainties of the area. The chapters on
human papovaviruses and on papillomaviruses are quite different in content, being

necessarily much m o r e biological in

approach. However, the molecular biology
of the human papovavirus BK is dealt with
poorly; its DNA sequence, which appears
in one of the appendices, is hardly
discussed.
Adenoviruses a r e covered in three
chapters, two written by Jane Flint alone,
the other by her in collaboration with Tom
Broker. That one author was involved in all
three chapters shows clearly in a much
more uniform approach and consistent
treatment. The chapter on lytic infection is
truly excellent, being both comprehensive
and detailed yet clearly written and easy to
follow. It also deals superbly with some of
the less-known regions of tumour virus
research, i n c l u d i n g a n o u t s t a n d i n g
discussion of adenovirus assembly.
The chapter on herpes simplex viruses,
by Patricia Spear and Bernard Roizman, is
by far the largest in the book and is also the
least satisfactory. It would have been much
more appropriate for a specialized review.
In many places the textual discussions
would have benefited from the provision of
appropriate diagrams while, conversely,
it is the only chapter to use extensive figures
of primary experimental data. These are
discussed in a style so condensed that it will
be impenetrable to all but experts. The
chapter is also somewhat unbalanced. It
deals too superficially with herpes genetics
and is too heavily biased towards the work
and opinions of the authors.
The final pages are occupied by six
appendices giving the DNA sequences of
SV40, polyoma and BK, and of the
transforming segments of adenovirus types
5, 7 and 12. Also presented are lists of

restriction sites, maps of mRNAs and

putative protein sequences. This section
represents an excellent innovation and will
be of the greatest value. I have only one
criticism of this section, albeit a severe one.
The sequence of SV40 has recently been
modified by the addition of an extra, small
restriction fragment. While this is referred
to in the introduction to the appendix, the
sequence used does not include it. It would
have been worth correcting this, even at the
price of a slight delay in publication.
This book is as timely as its predecessor,
detailing the results of one phase of the
work and setting the scene for the studies
that are to come. It will, therefore, be as
influential. I fully share the editor's regret
that the increased size of the book has
necessitated a significant increase in price
and sincerely hope that a paperback edition
will be forthcoming so as to put the book
within the reach of the many students who
ought to own a copy. Economics notwithstanding, this book is mandatory reading
for those wishing to obtain a general view
of tumour virology, for those wishing to
become involved in this most exciting area
of research and for those currently
involved in it. The editor and the authors
are to be congratulated on undertaking so
well what must have seemed to them an
almost impossible task. I can only hope, on
behalf of all in the field, that in five years
time they will have the courage to attempt a
I I
third edition.
Peter W. J. Rigby is a member of the Cancer
Research Campaign Eukaryotic Molec'ular
Genetics Research Group and a Lecturer in
Biochemistry at the Imperial College of Science
and Technology, University of London.

Just beyond the pale

Paul Davies
Gravity, Particles, and Astrophysics. By
Paul S. Wesson. Pp.188. (Reidel: 1980.)
Dfl.65, $34.

COSMOLOGY
and astrophysics attract an
enormous number of speculative theories,
not all proposed by cranks, but many of
them written somewhat tongue-in-cheek.
P a u l W e s s o n is a m a n with a n
encyclopaedic knowledge of those grey
areas somewhere between the canonical
models of orthodoxy, and the manifestly
absurd, that manage to slip into the
reputable journals. His previous book,
Cosmology and Geophysics (Adam Hilger,
1978), dealt lucidly with a variety of
"variable G" theories, in which the
Newtonian gravitational G is assumed to
vary with time. It is a theme continued and
O 198lNature Publishing Group

updated in this latest work, but

accompanied by a range of other, similar,
conjectures: particle creation, anomalous
r e d s h i f t s , t i r e d l i g h t a n d cosmic
numerology, for example.
The author's subjects range ambitiously
from the planets to the primaeval furnace.
Scarcely an unorthodox idea escapes
attention somewhere. The professional
cosmologist will be alternately delighted
and dismayed to see many half-forgotten
theories receiving a n airing. From the
Lyttleton-Bondi cosmology to modern
gauge theories of gravity, Wesson gets
around to them all in some context or
other.
Like it or not, many of these unusual
theories have been extensively, even
infuriatingly, discussed in the scientific
literature for years. Some of them, such as

Nature Vol. 289 22 January 1981

the so-called big number coincidences or
the varying G theories, have occupied the
attentions of eminent physicists a n d
astronomers. One should not blame the
author for attempting a comprehensive
presentation of such issues just because
they command only minority support.
My real criticism of this book is that it
reads rather like a shopping list. In his
enthusiasm for including everything, the
author has only managed to achieve a
compendium of speculations, and not a
coherent discussion. References are s o
liberally used that the text becomes
impossible to read in some parts. On one
page I c o u n t e d o n e h u n d r e d s u c h
interjections. This is fine in a reference
source for non-standard cosmology and
astrophysics, but will not satisfy the
demand by conservative "straight
thinkers" to know something of the work
of their m o r e a d v e n t u r o u s - m i n d e d
colleagues.
Of course, a subject does not become

333
respectable merely by quoting a vast array
of big names. Even Einstein patronized
some odd theories. Nor should one be
persuaded that all unconventional ideas are
in some mysterious way consistent with
each other. This book is written in a style
that leads one to believe there is a sort of
grand, coherent theory of unconventional
physics and cosmology existing alongside
the more publicized one. It gives the
impression of two sciences: the one to
which most professionals subscribe, and
another, almost the same in its predictions,
but with subtle differences. Wesson is fair
in drawing attention to some of the
shortcomings of the latter, but clearly
believes that his several hundred "grey
area" references are probing closer to the
truth than many of us would be prepared to
acknowledge.

Paul Davies is Professor of Theoretical Physics

at the University of Newcastle upon Tyne.

Agricultural prognoses
Kenneth Blaxter
Britain's Future in Farming. Edited by Sir
Frank Engledow and Leonard Amey.
Pp.164. (Geographical Publications:
Berkhamsted, Herts, UK, 1980.) 9.
THE farming industry accounts for less
than 3% of the labour force in the United
Kingdom and contributes about the same
percentage to its gross domestic product.
Yet it is the most obvious of our industries
since it occupies the major part of our
countryside and produces most of the food
we consume. The industry has much
changed with time; it is now capital
intensive and heavily dependent in some of
its sectors on sophisticated inputs from
other parts of the economy. These changes
are relatively recent and, as the authors of
this book state, "a highly urbanized
economy with a romantic view of the
countryside has yet to come to terms with
the unpicturesque aspects of an intensive
and efficient agriculture". Questions
necessarily arise about the future of
farming, in turn inevitably implying
questions about the policy to be adopted
a n d the goals to bc s o u g h t ; these
considerations are basic to this volume.
The book is the result of four years of
s t u d y by a g r o u p o f 15 p e o p l e
knowledgeable a b o u t farming, food
provision, forestry, agricultural science,
banking, and the social and political
history of land use. None of the chapters is
assigned to an author, but the discerning
can detect where initial responsibility
rested for written styles are revealing.
There is thus a collective responsibility and
this at times secms to have engendered
some collective irresponsibility. Surely

some of the statements made cannot have

been agreed by all. Is it reasonable to
suppose that all agreed "that the Ministry
of Agriculture [MAFF] should resume its
role in education a n d research and
s u b s u m e t h e A g r i c u l t u r a l Research
Council [ARC]"? O r that security of
tenure for farmers should be withdrawn
forthwith? Such statements in the text d o
not appear in the conclusions with which
no doubt all agreed. The first becomes
transformed to a plea for better integration
and a suggestion that attention should be
drawn to the relationship of MAFF and
ARC, while the second does not appear at
all.
A few provocative statements, however,
d o not detract from the overall thesis and,
on the contrary, perhaps enhance what is a
sensible account of the ~ r o b l e m srelated to
the management of our resources of land.
It shoulddomuch toshow that theinterests
of agricultural and urban communities in
reality have much in common and that
certain broad principles related to land use
may well be sufficient to provide a basis on
which to develop policies to meet new
eventualities. T h e c o n c l u s i o n s a n d
recommendations are admittedly detailed
- there are 28 of them
but the whole
tenor of the book relates to the theme that
the source of renewable land must be
preserved, that the pressures upon it
whether direct or indirect must be reduced
and farming practices themselves should be
designed t o provide a flexibility t o
accommodate inevitable and continued
change.
0
-

Sir Kennfrh Bluxler IS Direrlor of the Rowett

Rr.warch insl~lutf,A berdren.
--

Ol981Nature Publishing Group

Practical vacua
J o h n Yarwood
A User's Guide to Vacuum Technology. By
John F . O'Hanlon. Pp.402. (WileyInterscience: 1980.) f 13.35, $31.20.

JOHN O'Hanlon is a member of the

research staff at IBM in the USA. As a
result this book is really up-to-date. He has
written the most useful and thoughtful
account of vacuum technology which has
appeared in recent years, and one which
will be of value to the student, technician,
engineer, scientist and technical manager.
As would be expected from an author with
such a backgro"nd, the book (according to
his preface) emphasizes "the operation and
selection of equipment for processes used
in semiconductor, optics, and related
technologies". But he does himself an
injustice - this book is much wider in
scope than such a statement would appear
to indicate.
SI units are used throughout, with the
exception that pumping speed (now called
"volume rate of flow" - not a good term
although scientifically accurate) is given in
I/s or m3/h, depending on the type of
pump, and the diameters of the top flanges
of diffusion pumps are given in inches. The
section on essential gas theory covers gas
properties, kinetic theory, basic relevant
formulae (without proofs), gas flow and
the measurement of pumping speed (where
the work of the International Standards
Organization has not been acknowledged;
all the credit seems to go to the USA).
Succeeding chapters a r e up-to-date,
apposite accounts of real value, dealing
with a wide range of equipment - gauges,
pumps, analysers - and associated topics.
The book is rounded off with six appendices and, overall, can be recommended as
a most useful and comprehensive volume.
John Yarwood wasformerly Head of Physics at
the Polytechnic of Central London, and is the
editor of Vacuum.

UV photobiology
Philip C . Hanawalt
Biological E f f e c t s of Ultraviolet
Radiation. By Walter Harm. Pp.216.
(Cambridge University Press: 1980.)
Hbk f 15, $29.95; pbk f4.95, $9.95.
WF H A V L long appreciated the essential
role that light plays in the maintenance of
life forms on our planet through the processes of photosynthesis. In recent years,
however, we have become increasingly
aware of the deleterious effects of the
shorter wavelengths of light in the Sun's
spectrum. We have learned that ultraviolet

334
light (UV) damages the precious DNA,
resulting in mutations, the death of cells
and sometimes malignant transformation.
The latter effect would be intolerably
frequent were it not for cellular enzymes
that constantly monitor the DNA to
identify and repair the lesions that UV
inflicts. This is dramatically illustrated by
the rare hereditary disease, xeroderma
pigmentosum, in which D N A repair
deficiency renders the victims exquisitely
sensitive to sunlight-induced skin tumours.
Biological effects of UV was an excellent
choice of topic for this first volume in a new
series sponsored by the International
Union of Pure and Applied Biophysics,
since UV is both a n important biophysical
probe for DNA repair systems and an
environmental hazard of some concern.
The books in this series are meant to serve
as specialized texts for advanced students
in particular fields of biophysics. The
author, Walter Harm, was one of the
pioneers in the field of UV photobiology
and he was therefore in a good position to
provide the kind of historical perspective
that could bring the subject t o life and
excite students. Unfortunately this monograph doesn't really gain much vitality as it
plods ponderously through the details of
classical bacterial UV photobiology.
Without first surveying the range of
interesting biological effects, the first
chapter launches into electronic excitation
and the practical nuts-and-bolts details of
experimental UV photobiology. Chapter 2
carries on with more practical advice for
the researcher, including some useful
numerical examples to illustrate important
points. A carefully detailed consideration
of UV interactions with biologically
important molecules is followed by a notso-useful chapter on classical target theory
and survival curve shapes. (The word
"multihitlike" was a new one to me.) I
would have preferred more discussion of
why shoulders o n survival curves may be
explained by existence of repair systems.
T h e terms excision-repair a n d postreplication repair are introduced but not
defined until three chapters later.
T h e subject of enzymatic p h o t o reactivation is treated in great detail,
reflecting the recent research interests of
the author. In contrast, the discussion of
dark repair systems is incomplete and contains a number of inaccuracies, such as a
reference to sequential action of the uvrA
and recA systems (p.107). The book is
nearly devoid of discussion of our current
understanding of the enzymology of dark
repair and its genetic control. Yet this is
perhaps one of the most exciting areas of
research in UV photobiology.
The predominant emphasis in the text is
o n UV inactivation of bacteria and their
phage, even in the chapter on solar UV
radiation effects. Sunlight effects o n
human skin are relegated to less than two
pages followed by a brief chapter o n UV
carcinogenesis and a n even shorter one on
sensitized UV effects on DNA.

Nature Vol. 289 22 January 1981

I must say that I would find it difficult to
teach a course using this book as a text but I
find it quite useful as a reference work. I t
does provide a comprehensive treatment of
the now classical studies on UV effects in
prokaryotes. It might have been more aptly
titled Microbial Effects of UV Radiation;
as Dr Harm emphasizes, much of our

understanding in UV photobiology and

DNA repair has been based upon the solid
foundation of early studies with simple
prokaryotes.
0
Philip C. Hanawalf 1s Professor of Biology and
Dermatology at Stanford University, and
Dire,tor o f [ h e G r a d u a t e proPra,,, in
Blophysccs. *
-

Social science and the stomach

John Yudkin
-

Consuming Passions: The Anthropology

of Eating. By Peter Farb and George
Armelagos. Pp.279. (Houghton Mifflin:
1980.) $12.95.
THEpast 25 years have seen not only a rapid
growth in research in nutrition, but the
increasing recognition that nutrition is as
much a social as an experimental science.
Its wide span is encompassed in the short
definition of nutrition as the study of the
relationship between people and their
food. Thus, there are now fewer academics
who consider nutrition to be a section of
biochemistry or of physiology, and few
among the growing number of university
departments of nutrition that d o not
include in their courses relevant aspects of
social nutrition from economics,
sociology, anthropology and psychology.
This book will certainly be of interest to
nutritionists and dietitians, as well as to
anthropologists, but its appeal is very much
more general since it contains a mass of
fascinating information. Its wide range is
best indicated from this quotation near the
end of the work: "The thesis of this book
has been that to know what, where, how,
when and with whom people eat is to know
the character of their society". For
example, you will learn from the book that
"cannibal" is derived from "Carib" (as in
"Caribbean") and that a "companion" is
the person with whom you eat bread; that
among the people of Trobriand it is
unseemly for couples to eat together before
marriage, just as i t is (was?) for unmarried
Western couples to sleep together; that
cultural customs such as not eating pork or
the reverence of the Hindu for cattle often
have a biological or ecological basis.
The authors deal first with the evolution
of human dietary patterns, how they have
adapted to the availability of local foods,
and the differences called for, and met, at
different phases of the life of t h e
individual. There follows a description and
suggested explanation for the myths and
religious concepts that influence eating
behaviour, and the rules that different
people have adopted towards food in their
relationships with family, clan a n d
strangers. There is considerable discussion
of food preferences and avoidances, and
the development of national cuisines.
Ol981Nature Publishing Group

It would be pleasing to be able to say that

this book had achieved a synthesis between
those aspects of eating that belong to the
social sciences and those that belong to the
natural sciences. The fact is, however, that
until recently at any rate most people who
have become professionally interested in
nutrition have come through the separated
and narrowly confined and demarcated
p a t h w a y s of these t w o g r o u p s of
disciplines. It is understandable then- but
a great pity - that the data on the nutrient
content of foods have not been properly
checked.
It is not true that one person may "burn
calories ten times faster" than another "of
t h e s a m e a g e , sex a n d physical
proportions". It is misleading, too, to
worry about the fact that a portion of
canned peas contains 100 times as much
salt as do fresh peas, when in absolute
terms this would still be about one-fiftieth
of the amount taken in a day; more
importantly, it ignores the fact that people
add very much more salt when cooking
vegetables than that present in the canned
peas. The plausible suggestion that pica
usually has a nutritional significance - for
example, that eating clay provides minerals
lacking in the diet - has been repeatedly
disproven.
One last grumble is that the authors have
adopted the somewhat hysterical current
concern about the safety of modern food
processing. F o r example, they say
categorically that food colours are carcinogenic, a statement t h a t , of all
countries, is highly unlikely to be true in
America with its rigid food laws that are
especially sensitive a b o u t suspected
carcinogenicity of food additives.
Nevertheless, for the general reader this
is a book that is well and clearly written
(except for the use of "human" as a noun!)
and is full of interesting facts about what
and why we eat what we do. For the
specialist reader, too, there is a host of
references. If, therefore, you are dubious
about the theories of Claude Levi-Strauss on
attitudes and behaviour towards food, you
will be directed to the writings of several
other authors, some of whom agree and
some of whom disagree with his views.
John Yudkin is Emeritus Professor of Nutrrtion
at London University.

335

Nature Vol. 289 22 January 1981

Continued from page 218

provisions of the First Amendment of the
Constitution by permitting the teaching of
evolution. This, the suit says, disparages the
religion of creationists and in itself amounts to
the teaching of a religion - the religion of
secular humanism.
The charges in the suit focus on the Science
Frameworkfor California Public Schools,
Kindergarten and Grades One through
Twelve, published by the California State
Department of Education in 1978. Dr Richard
Lemmon and I served on the committee that
wrote this publication, a task that took about
three years. It involved a steady battle with
creationists who resisted the inclusion of
statements concerning evolution. On some
occasions, we obtained their consent, only to
have them change their minds. For example, a
statement that "Darwinian evolution is a
cornerstone of modern biology" was first
permitted, then deleted.
The final document says:
Living organisms have universal properties, including derivation of energy from
outside sources, and reproduction. Evolutionary studies indicate that these organisms are naturally selected from generation
to generation, producing descendants with
different characteristics and producing
variability among populations of living
species. The process has been going on so
long that it has produced all thegroups and
kinds of plants and animals now living as
well as others that have become extinct.
. . . In addition to reproduction, another
characteristic of life is change in its genetic
material with passage of time. This process, termed evolution, takes place
through changes in DNA. Changes in
DNA molecules are produced by mutation, which includes replacement of some
DNA bases by others, and recombination
in which segments of DNA are added to or
subtracted from genes. Duplication of
genes sometimes also occurs. Most mutations are harmful and do not persist; they
are eliminated by natural selection.
Beneficial mutations occasionally take
place a n d a r e responsible f o r t h e
appearance of new characteristics. A third
class is "neutral" or "near neutral"
mutations.
The Creation Research Society (through
Creation-Life Publishers) does not confine
itself to matters of religion but also addresses
scientific questions. A complaint by a parent
in Livermore is that the publication Dry
Bones, published by Creation-Life in 1979, is
being used in lower grades of an elementary
school. It contains a dialogue between children
and their father, including:
Q: "Did dinosaurs live when Noah lived,
Dad?"
A: "That's what I think. The way it looks,
dinosaurs were drowned in the Flood,
too."
Q: "Did Noah have to take them on the
Ark?"
A: "He surely did". . .
Q: "I guess[Darwin] thought fossils were
proof for evolution."
A: "No, Darwin said that fossils were
'perhaps the most obvious and serious
objection' to his theory."
A: ". . Did your friend also tell you about

scientists at Oak Ridge National

Laboratories?"
Q: "No, what about them, Dad?"
A: "They used uranium dating on wood in
rocks of the dinosaur group, and got
ages of only thousands of years."
H. JUKES
THOMAS
University of California, Berkeley

Genes and race

SIR- Some time ago you published a letter

from me concerning the use made by the
National Front of apparently "scientific"
justification of racist arguments. I asked,
through your columns, Professors Eysenck
and Jensen publicly to dissociate themselves
from the use made of their names in apparent
support of these views.
I now wish to bring an analogous issue to
the attention of readers of Nature. Ever since
the publication in 1975 of E.O. Wilson's
Sociobiology: The New Synthesis, the danger
has been clear that in due course racists would
deploy sociobiology in support of their views.
The "New Right" in France has been doing so
for some time, and now the breakaway journal
from the National Front in Britain, New
Nation, has done so too. Its first two issues,
dated summer and autumn 1980, carry articles
entitled: "Nationalism, racialism: products of
our selfish genes" and "Science is
championing our creed of Social
Nationalism", authors respectively John
Thornton Bannerman and Richard Verrall (the
latter is also the author of a pamphlet entitled
"Did Six Million Really Die?").
The first cited of these comes complete with
a picture of "Richard Dawkins, author of The
Selfish Gene", and cites him, Wilson,
Maynard Smith and one Travers (presumably
Trivers) amongst others. The final section of
this article appears alongside an advertisement
from "Nationalist Books" which includes
such choice titles as Debunking the Genocide
Myth and Anne Franck's diary - a hoax.
Opponents of sociobiology, it is therefore
scarcely surprising to note, are categorized as
"Marxist" and "Jewish".
The concluding paragraph reads as follows:
For us, as racial nationalists, this is an
important vindication of our position. For
it is increasingly clear that it isn't just "bad
luck" that our genes don't permit us to live
in a Marxist-Rousseauesque egalitarian
communist utopian World State of universal altruism. It was an inevitable result of
the way evolution works that our genes
would not permit us so to live. What the
evolutionary theoreticians have shown us
is that, with the system of genetic inheritance shared by all vertebrates, the only
type of social organization which can
evolve, let alone work, is one based upon
kinship, upon the ties of blood and of race.
Nationalism is not only an integral part of
our genetic inheritance, it is an inevitable
end product of the evolutionary processes
which shaped that inheritance.
May I suggest that it would be in tl.re public
interest that John Maynard Smith and Richard
Dawkins should clearly dissociate themselves
from the use of their names in support of this
neo-Nazi balderdash.
STEVEN
ROSE
The Open University,
Milton Keynes, UK
O 198lNature Publishing Group

Energetic consensus

SIR- It was disheartening to learn from your

correspondent that the First European
Bioenergetics Conference had shown a
"general acceptance of the chemiosmotic
concept of energy transduction" and that
interest was now moving on from this assured
base to the finer molecular details of the
problem (Nature 4 December 1980, p.432). If
there is such consensus on this matter there is
very good reason to believe that it could be
mistaken.
The chemiosmotic hypothesis is founded
upon a very simple and invalidating
illogicality: no evidence has ever been given
that protons move into the bulk-phase during
synthesis. Oxygen-pulse studies which
supposedly support the hypothesis are carried
out under entirely different experimental
circumstances; and to have used the results of
such studies to generalize upon the nature of
proton movements in ATP synthesis was to
introduce an inferential non sequitur into the
energy transduction argument from the outset.
The basic chemiosmotic equation:
A p =A t&-zApH
(1)
has no meaning if protons are not translocated
to the bulk-phase.
We have attempted to state this view since
1974' and the accumulating experimental
evidence we have obtained has led to the
Q-zoneinterpretation of mitochondria1 energy
transduction in which the proton-motive force
has been regarded as a localized and surface
electrical force, with a minimal osmotic
involvement:
a p Q = Awe
(2)
Recent experiments showing the generation of
negative fixed charges on the inner membrane
specifically associated with the energization
process2 have fulfilled the main prediction of
the Q-zone interpretation and given a further
compelling reason to suspect the adequacy of
the chemiosmotic view; they also call into
question our own formulation of the protonmotive force given in equation (2).
It now seems probable that during synthesis
the proton-motive force should be regarded as
a coulombic rather than a chemiosmotic force,
and that its assessment should be made in
terms of charge-charge relationships taking
place in a surface region of variable dielectric.
Some of the more recondite thermodynamic
implications of such a system have already
been anticipated by Ashcroft and CosteP;
meanwhile an unsophisticated estimate of the
intercharge distance, I, may be gained from
the relationship:

w =2.%! &I = 15 kcal (A@)

and the speculative use of a now dubious
chemiosmotic H + /ATP stoichiometry of 3.
Substituting the dielectric constants for the
"limit" media (tq = 75; rlipid= 21
intercharge distances of 0.9 and 33 are
obtained; the functional value may reasonably
be expected to lie between these extremes and
to be in accord with the requirements of the
Q-zoneinterpretation.
F.H. MALPRESS
Department of Biochemistry,
The Queen's University of Belfast, UK
I . Archbold. G.P.R. el a/. Blochem. Soc. Trans. 2,751-754
(1974).
2. Archbold, G.P.R.etob Biochem. Internet. l,(in the press).
3. Ashcroft, R.G. & Costei, H.G.L. Bioelectrochem.
Bioenerp. 5.3742 (1978).
--

336

Awards
Drs J.-P. Mach and S. Carrel, (Lausanne
Branch of the Ludwig Institute for Cancer
Research) have received the 1980 Cancer
Prize of the Swiss Cancer League.
The Charles Frederick Chandler Medal,
presented by Columbia University f a
achievement in pure or applied chemistry
has been awarded to Frank H. Westheimer
(Morris Loeb Professor of Chemistry at
Harvard University).
The NASA Group Achievement Award
has been presented to Dr Sam Gulkis
(JPL); Dr David Jauncey and Dr MichaelBatty (CSIRO) for the development of a
radio astrometric interferometer in the
Southern Hemisphere.
The Council of The Institute of Physics has
made the following awards for 1981; Bragg
Medal and Prize to G.E. Foxcroft (Rugby
School); Charles Chree Medal and Prize to
Dr K.A. Browning (Meteorological Office
Radar Research Laboratory, RSRE
Malvern); Charles Vernon Boys Prize to Dr
P.N. Pusey (Royal Signals and Radar
Establishment, Malvern); Duddell Medal
and Prize to Dr B.A. Joyce (Philips
Research Laboratories, Redhill);
Glazebrook Medal and Prize to Dr G.H.
Stafford (Rutherford a n d Appleton
Laboratories, Chilton); Guthrie Medal and
P r i z e t o D r J . C . Ward ( P h y s i c s
Department, M a c q u a r i e University,
Australia); Maxwell Medal and Prize to Dr
J.M. Kosterlitz ( D e p a r t m e n t o f
Mathematical Physics, Birmingham
University); Thomas Young Medal and
Prize to N.J. Phillips (Department of
Physics, Loughborough University of
T e c h n o l o g y ) ; P r i z e f o r t h e 1980
Graduateship Examination of the Institute
to James Cooke (Dublin College of
Technology).
The Council of the Royal Society invite
nominations for the 1981 Royal Society
Esso Energy Award which should be sent
by 6 February 1981. The Award is given for
outstanding contributions to the advance
of science or engineering or technology
leading to the more efficient mobilzation,
use and conservation of energy resources,
and consists of a gold medal and a prize of
1,000. Further information from: Dr
R.W.J. Keay, The Royal Society, 6 Carlton
House Terrace, London SW I , UK.
A fellowship in human anatomy, to be
called the Wellcome-Franks Medical
Fellowship, has been endowed at St Hugh's
College, Oxford, by The Wellcome
Foundation Limited. Enquiries to: Ron
Champion, The Wellcome Foundation
Ltd, 183 Euston Rd, London N W l , UK.

Nature Vol. 289 22 January 1981

Environmental researchers and scientists

- either from university or industry - are

invited to compete for Chemviron's fifth
'Award' a $10,000 cash prize to be given in
1982. It is awarded for outstanding work in
the area of physical-chemical treatment of
potable and waste water, or of sludge
resulting from treatment of such waters.
Copies of the regulations can be obtained
from Dr J. Melbourne, c/o Chemviron,
Chaussee de Waterloo 1135, B-1180,
Brussels, Belgium. Only entries originating
in Europe will be accepted.
The Roussell Prize to beawarded in 1982, is
given every two years to a chemist o r a
biochemist whose work on steroids in
therapeutic medicine has been chosen as
the most outstanding by an International
Committee. Further information from:
Professor J . Mathieu, Centre d e
Recherches, Roussel Uclaf, 102 route de
Noisy, 93230 Romainville, France.
The Trustees of the Lady Tata Memorial
Trust invite applications for awards for
p e r s o n a l s u p p o r t f o r research o n
leukaemia, in t h e Academic Year
beginning October, 1981. Candidates with
programmes of research on any aspect of
malignant disease which may throw light
on problems of leukaemia will be eligible
for consideration. Further information
from: Secretary of t h e (European)
Scientific Advisory Committee, Lady Tata
Memorial Trust, MRC Leukaemia Unit,
Royal Postgraduate Medical School, Du
Cane Road, London W12, UK.
The Paul-Martini-Prize, was initiated by
the Medizinisch Pharmazeutische
Studiengesellschaft e.V. (Mainz) in
association with the Deutsche Gesellschaft
Fiir M e d i z i n i s c h e D o k u m e n t a t i o n ,
Informatik und Statistik e.V. to promote
the development of scientific methods for
evaluation in clinical pharmacological and
therapeutical fields. Papers should be
submitted by 28 April 1981 either in
German or English with six copies to the
Medizinisch Pharmazeutische Studiengesellschaft e.V. Bilhildisstr. 2, D 6500
Mainz.
The International Union Against Cancer,
with funds provided by the Cancer
Research Campaign (UK), will award
fellowships for research on cancer to
persons o n the staff of universities,
teaching hospitals, research laboratories or
similar institutions. These enable
investigators to work abroad to gain new
experience in clinical or basic research
laboratories o r similar institutions.
Application forms and additional
information from: International Union
Against Cancer, rue du Conseil-General, 3,
1205 Geneva, Switzerland.
O 198lNature Publishing Group

Meetings
9-10 February, New Concepts in Drug and
Nutrition Delivery Systems, Paris (Roberts
S. First, Inc. Avenue Marnix 19A, 1050
Brussels, Belgium).
9-13 F e b r u a r y Fundamentals of
Microprocessors, London (Bleasdale
Computer Systems Ltd, Francis House,
Francis St,London SW I , UK).
9-13 February, Protein Structure and
Function, Lorne, Australia (Dr R.J.
Simpson, St Vincent's School of Medical
Research, Victoria Parade, Fitzroy,
Victoria, Australia).
12 February, 4th Interdisciplinary Cancer
Research Workshop, New Orleans (Dr P.
Politzer, Chemistry Dept, University of
New Orleans, Louisianna).
15-17 February, Proprioception, Posture
and Emotion, New South Wales (Dr D.
Garlick, Committee in Postgraduate
Education, P O Box 1, Kensington, New
South Wales, Australia 2033).
Idiotypes
8-15 ~ e b r u a r y lmmunoglobulin
,
and their Expression, Salt Lake City (ICNUCLA Symposia, Molecular Biology
Institute, University of California, Los
Angeles, California 90024).
15-20 February, Differentiation and
Function of Hematopeitc Cell Surfaces,
Keystone (ICN-UCLA, Molecular Biology
Institute, University of California, Los
Angeles, California 90024).
20 February, Ulcer Therapy, Frankfurt
( G D C h , P o s t f a c h 9 0 4 4 0 , D-6000
Frankfurt/M8, West Germany).
25-27 February, Recombinant DNA
Research, San Francisco (Edward Ruffig,
C / O Scherago Associates Inc, 1515
Broadway, New York 10036).
22 February - 1 March, Mechanisms of
Chemical Carcinogenesis, Keystone (ICNUCLA Symposia, Molecular Biology
Institute, University of California, Los
Angeles, California 90024).
1-8 March Cellular Recognition, Keystone
(ICN-UCLA Symposia, Molecular
Biology Institute, University of California,
Los Angeles, California 90024).
4-7 M a r c h , Chemical Indices and
Mechanisms o f Organ-directed
Toxicology, Barcelona (Secretariat,
Toxicology 81, 142 Oxford Rd, Cowley,
Oxford, UK).
8-21 M a r c h , Basic C o n c e p t s and
Techniques in Research on Biological
Membranes, Ibadan (Dr Enitan A .
Bababunmi, Dept of Biochemistry,
University of Ibadan, Nigeria).
9-1 1 March, Drug Delivery System
Development, London (A.S. Goldberg,
Powder Advisory Centre, P O Box 78
London NW 11, UK).
1 0 - 1 3 M a r c h , P O W T E C H '81.
Birmingham (Clive H.R. Hopkins, Green
Dragon House, 64-70 High St, Croydon,
Surrey, UK).