Checkpoints: Controls That Ensure the

Order of Cell Cycle Events

The events of the cell cycle of most organisms are ordered

into dependent pathways in which the initiation of late
events is dependent on the completion of early events. In
eukaryotes, for example, mitosis is dependent on the
completion of DNA synthesis. Some dependencies can be
relieved by mutation (mitosis may then occur before
completion of DNA synthesis), suggesting that the dependency is due to a control mechanism and not an
intrinsic feature of the events themselves. Control mechanisms enforcing dependency in the cell cycle are here
called checkpoints. Elimination of checkpoints may result
in cell death, infidelity in the distribution of chromosomes or other organelles, or increased susceptibility to
environmental perturbations such as DNA damaging
agents. It appears that some checkpoints are eliminated
during the early embryonic development of some organisms; this fact may pose special problems for the fidelity
of embryonic cell division.

distribution is important for cell viability. While executing these

events, the cell must avoid or correct errors that lead to the
production of nonfunctional organelles and those that lead to the
production of or distribution of the wrong number of organelles.
Recent results comparing somatic and embryonic cell cycles (10)
have revealed basic similarities as well as striking differences in how
events are controlled. In the early embryonic cell divisions of
Xet~opus, the initiation and order of events of the cell cycle is
determined by cyclic activation of maturation promoting factor
(Ml'F), and the events occur independently of one another. This
subject is reviewed by Murray and Kirshner in this issue of Science
(11). In the cell cycle of yeast and the somatic divisions of many
metazoan organisms, an additional principle appears to operate.
Although MPF plays an initiating role in the somatic cell cycle, the
order of events is ensured by dependent relationships; the initiation
of late events is dependent on the completion of early events. We
think dependent relationships seen in somatic cells are a key element
in understanding the high fidelity of organelle reproduction and
distribution during cell division. The purpose of this article is to
consider how these dependent relationships are achieved and the
consequences incurred by the cell upon their elimination.

HE CELL CYCLE IS OFTEN CONSIDERED TO BE COMl'OSED

of four phases, the gap before DNA replication (GI), the

DNA synthetic phase (S), the gap after DNA replication
(G2),and the mitotic phase, which culminates in cell division (M).
Although this formulation is useful and can serve as an organizing
principle, the cell cycle is seen on closer examination to be more
complex. A large number of macromolecular components are assembled, activated, or moved; a sequence of events involving one
organelle (such as the centrosome) may occur throughout all four
stages of the cell cycle, and independent sequences are coordinated
with one another.
Biochemical, genetic, and cytological research in the last decade
has greatly increased our appreciation of the structural and functional complexity involved in numerous cell cycle processes such as
DNA replication (1, 2), chromosome organization (1, 3), centrosome duplication and movement ( 4 ) , the dynamic organization of
the mitotic spindle (5), chromosome movement on the spindle ( 6 ) ,
nuclear envelope breakdown and reformation ( 7 ) ,organelle duplication and distribution ( 8 ) , and the establishment of the site of cell
division (9).Each of these processes involves cellular organelles that
are present in small numbers and whose accurate reproduction and

The authors are in the Department of Genet~cs,University of Washington, Seattle, WA

98195.
*Present address. 13eparunent of Mc)lecular and Cellular Biology, Univcrs~ty of
Arizona, Tocson, AZ 85721.

3 NOVEMBER 1989

Dependent Relationships in the Cell Cycle

The existence of dependent relationships is not usually apparent
simply by observing the normal cell cycle. Dependencies are revealed
by perturbations of specific events, that is, by the application of
chemicals, by the study of mutants that specifically inhibit one event
in the cell, or by surgical and cell fusion techniques (12). For
example, specific chemical inhibition of DNA replication prevents
nuclear division and cell division in most cell types including
bacteria, fungi, and vertebrate somatic cells. Mutants that block
specific events in the cell cycle provide additional resolution of
dependent relationships. In the yeasts Sacchavomyces cevevisiae (13)
and Schizosacchavomycespomi~e(14),many mutants have been isolated
and characterized that appear to have defects at specific stages of the
cell cycle. In S. cevevisiae. temperature-sensitive mutations exist that
block bud formation, spindle pole body enlargement and division
(15), spindle pole body separation and migration (16), tubulin
assembly (16); spindle elongation, initiatio~l-ofDNA replication,
DNA elongation ( 17), DNA ligation (18),chromatin assembly (19),
bipolar association of chromosomes (20),sister chromatid decatenation (21), sister chromatid separation (22), nuclear division, and
cytokinesis (13).The phenotypes of the mutants suggest that most
of these events are ordered into a few dependent pathways. For
example, the sequence of events that encompass spindle pole body
duplication and segregation as well as those comprising chromoARTICLES

629

some replication and segregation constitute one dependent pathway.

What principles does the cell use to establish an ordered pathway
of events? Does the existence of such order imply the existence of
control mechanisms that enforce order? Since many of the events of
interest in the cell cycle are those involving the assembly of
macromolecular complexes, it map be informative t o consider the
principles that have been gleaned from the extensive investigations
of another case of macromolecular assembly-the formation of a
bacteriophage particle (23). Bacteriophage T 4 is constructed from
three components-the head, tail, and tail fibers-ach
of which is
assembled by an invariant pathway. All str~ictural proteins are
synthesized at the same time, and unassembled proteins remain
unassociated until the partially assembled structure becomes ready
for their addition. At this stage, reactive sites that accommodate the
assembly of the next component are created by the addition of the
previous protein in the assembly pathway. For example, tail tube
subunits do not associate with themselves ~ultilthe baseplate is
assembled; at this staee
" some tail tube subunits associate with the
baseplate, and these provide a seed for the polymerization of
additional identical subunits to form a long helical polymer. Similarly, assembly of the bacteriophage T 4 baseplate itself occurs by an
invariant pathway where each step utilizes the previous step as
substrate and in turn provides the substrate for the next step (24).
The important lesson from bacteriophage assembly is that a very
complicated series of morphogenetic events can be ordered by a
principle intrinsic t o the components themselves without requirement for extrinsic control mechanisms. We shall refer to this type of
ordered pathway as substrate-product ordered.
The tnaturation of bacteriophage h DNA during packaging into
the phage head provides an example of dependence due to a control
mechanism. Concatameric DNA is not cut by the enzytne terminase,
unless phage proheads are present to package the DNA. This
dependence of DNA cutting on the presence of proheads could have
been enforced by substrate-product order if terminase was activated
by binding to proheads; however, this is not the mechanism, rather
a trans-acting inhibitor prevents tern~inasefrom cutting in the
absence of proheads, and this dependence on proheads can be
relieved by eliminating the inhibitor (25).
Likewise, the dependency of late events in the cell cpcle 01) the
completion of early events may be due to either substrate-product
order or to a control mechanism (Fig. 1).If, for example, replicated
chromosomes are essential substrates for tiiitosis, then the depetldence is due to substrate-product order; alternatively, if the dependence is due to an inhibitor of mitosis produced in response to
unreplicated chromost~mes,then we would say the dependence is
due to control. Control might also be exerted by an activator; for
example, completion of DNA synthesis might produce an activator
of mitosis. Since it is difficult to distinguish control by activation
from substrate-product order by an empirical test, we will concelltrate our discussion of control mechanisms on those that act by
inhibition. We use the term control t o include regulatiot) at any
level: transcription, translation, or posttranslation.
How can one distinguish extrinsic control by inhibitiot) from
substrate-product order? The existence of a control tnechanism is
suggested when one finds chemicals, mutants, or other conditions
that relieve a dependent relationship; that is, conditions that permit
a late event t o occur even when an early, normally prerequisite event,
is prevented. We term such an observatiot) "relief of dependence."
This argument rests on the assumption that if dependence is
intrinsic to the mechanism of assetnbly as it is for the bacteriophage
T 4 tail, then one would not be likely to relieve this dependence by
mutational inactivation of a gene product. For example, one would
not likely find a mutation that would relieve the need for at) early
protein in the assetnbly sequence [however, see (26) 1. Similarly, if

Fig. 1. Models illustrating the dcpcndcncc o f

~nitosiso n replicated al~ct
x
d
undamaged DNA. (A) A
Damage
replicated, unda~nagcd
chro~nosomc
passcs
t h r o u ~ h chromosome
condensation and scgrceation. iR. <:., and l>'iA
y
chronloso~ne sustaining
a doublc-strand break is
illustrated; however, a
g,ap or a single-strand kcC Negative control
slon remaining after
DNA replication would
he cxpcctcd to have the
sarnc conscclucnccs. (6)
The broken chromoE~~~~
some is a n inadequate
D Relief of dependence
substrate for chromosome condensation (or
tor r ) m c later step) and
mitosis is blocked. (C)
The 1)NA break creates
a signal that inhibits chromosome condcnsation. (D) A mutation (like rod9)
climinatcs the negative inhibition so the damaged chromoso~ncpasscs
through mitosis and an error occurs.
A Normal mitosis

g-e

-) ,f

the dependence of nuclear division on DNA replication it) the cell

cycle is due to an intrinsic requirement of replicated chromosomes in
the nuclear division machinery, we would not expect to relieve this
dependence by mutation. We are aware of the fact that this empirical
criterion cannot be taken as rigorous evidence for the existence of a
control tiiechanistii. One can imagine rare mutations that will alter a
protein in such a way as to permit it to assenlble without its nor~l~al
substrate, but tilost mutations will eliminate a f~inction.We suggest
that a finding of "relief of dependence" provides prima facie
evidence fix control, especially whet) the mutation is shown to
elinlinate the function of a protein. By the relief of dependence
criterion, a number of control tnechanisms have recently been
identified in the cell cycle. We call these control mechanisms
checkpoints (27), because they appear to have the role of checking to
see that prerequisites (such as DNA replication in the case of
mitosis) have been properly satisfied.

Checkpoints in the Cell Cycle

We discuss it) this section a few cases where sufficient evidence
exists to suggest that the dependence of a late event in the cell cycle
on an earlp event is due to a checkpoint. We will describe first it)
some detail a control mechanism that we have worked on, the k 4 D 9
spste~llin yeast, which is responsible for tilaking nlitosis dependent
on the completion of DNA replication, and then discuss more
briefly a few other control systems.
Dcpewdc~nccoJtnitosis on DNA synthesis. In yeast, manlmaliat) tissue
culture cells, Aspeyqillus, and tnany other eukaq~oticorganisms,
arrest of DNA synthesis by specific inhibitors or by mutational
inactivation of replicatiot) enzymes prevents mitosis. In yeast, the
IZAL19 gene specifies a component of this control system (27).
Temperature-sensitive mutants defective in some D N A replication functions (DNA polymerase I, cdc17; DNA polymerase 111,
cdc2; and DNA ligase, cdc9) do not normally undergo mitosis at the
restrictive temperature. Dependency of mitosis on the completion of
DNA synthesis is relieved, however, L>y a complete deticiency of
RAIN; if these same cdc mutants have a rod9 gene defect, then the
cells continue through mitosis into the next cell cycle at the
restrictive temperature (28). A mutant tetiiperature-se~lsitive for
SCIENCE,

VOL.

246

DNA ligase illustrates the effect of the RAD9 checkpoint (Fig. 2).
The analysis exploits the fact that in S. cerevisiae different phases of
the cell cycle are accompanied by distinctive changes in daughter
bud morphology. During incubation at the restrictive temperature
for 3 hours, GI (unbudded) cells containing the cdc9 mutation
became blocked before mitosis (large budded cells) (Fig. 2, A and
B). Arrested cells had completed the bulk of DNA synthesis
[confirmed by analysis of DNA content by flow cytometry (27)],
but become arrested presumably because many unligated singlestrand lesions remain in the DNA (28). In contrast, cdc9 GI cells that
also have the rad9 defect typically proceed past mitosis and enter the
next cell di$sion at the restrictive temperature (Fig. 2, C and D).
Analysis of m'st nuclear morphology of growing cells shifted to the
restrictive temperature confirms that cdc9 cells are blocked before
chromosome separation in mitosis (Fig. 2E), whereas cdc9-rad9 cells
are not arrested, but display a distribution of cells in different phases
of the cell cycle.
The cdc9-rad9 double mutant illustrates a principle that we think
may apply to many checkpoints, that elimination of the checkpoint
may have catastrophicor subtle consequences depending on prevailing conditions. Temporary inactivation of DNA ligase activity is not
lethal for most cells, as shown by the ability of cdc9 mutant cells to
retain viability after a brief incubation at the restrictive temperature.
In the absence of RAD9, however, DNA ligasdeficient cells die
much more rapidly at the restrictive temperature (Fig. 3) (29).
Therefore, relief ox the dependence of mitosis on DNA synthesis is
lethal when completion of DNA synthesis is blocked. Alternatively,
if cells are not perturbed by an interruption of DNA replication (or
by extrinsic DNA damage, see below), then RAD9 is not an essential
function; cells have indistinguishablegrowth properties whether the
RAD9 gene is intact or defective (27). The only effect of complete
deficiency of the RAD9 checkpoint is that rad9 cells lose chromosomes spontaneously at a rate 21 times higher than that of wild-type
cells [rate of loss of one chromosome from a disome strain; 6.0 x
in a rad9 deletion and 2.9 x
in Rad' (27)l. We imagine
that these two phenotypes, cell death or chromosome loss, are
extremes of the same primary role of this checkpoint, to ensure that
chromosomes have been fully replicated and are intact before mitosis
is initiated. Thus the order of DNA synthesis and mitosis is
apparently established independently of the RAD9 checkpoint, but
the RAD9 product ensures the order if DNA synthesis is intermpted. An attractive hypothesis to explain how RAD9 inhibits mitosis is
to suggest that it negatively regulates a function essential for mitosis.
It will be of interest to determine whether the RAD9 gene product
interacts with MPF or other components known to play essential
roles in the initiation of mitosis.
The RAD9 control system was initially identified in a search for
mutations that permit cell division of cells with defective genomes
due to damage induced by x-irradiation (27). Mutations in the
RAD9 gene allow cells with DNA damage to proceed through cell
division, whereas irradiated wild-type cells arrest in Gz until the
damage is repaired. Mitosis in most other eukaryotic cells (30) is also
dependent on undamaged DNA, since x-irradiation and other DNA
damaging agents arrest cells before mitosis. The dependence of
mitosis on completion of DNA replication is relieved in mammalian
somatic cells by caffeine (31). Like rad9 defects in S. cerevisiae,
caffeine treatment of irradiated mammalian cells permits their entry
into mitosis (and decreases cell viability) and yet has no observable
effect on this transition in unirradiated cells (32).
Temperature-sensitive recessive mutations in mouse BHK cells
(tsBN2) and Aspergillus (bimE7) qualify as checkpoints since both
relieve dependency of mitosis on DNA synthesis (33). In both
mutants, cells blocked in DNA synthesis enter mitosis without
completing DNA synthesis when mutant cells are shifted to the
3 NOVEMBER 1989

restrictive temperature. Death of cells carrying these mutations is

probably not due solely to relief of dependence, as bimE7 mutants
shifted to the restrictive temperature are arrested in mitosis. The
cellular functions are unknown, but deficiency for either must
render inactive a checkpoint that prevents mitosis should DNA
synthesis be blocked. Both are suggested to be negative regulators of
additional functions essential for mitosis.
Dependence of anaphase on metaphase and delay as evidence of a
checkpoint. In the examples of dependence cited above, inhibition of

"3

- --=ar

. E
Cell

:'3

*;

Fig. 2. DNA ligase-defective cells are arrested

before mitosis only if the
R A D 9 gene is present.
Progression in the cell
cycle of individual cells
was determined by photomicroscopy. ( Aand B)
cdc9 and (C and D) cdc9rad9 cells were grown at
the permissive temperature then plated on thin
agar slabs. Photographs
of the same fields were

$-qM

were
made initially
at the time
shiftedcells
to
the restrictive temperature for DNA ligase (A
and C) and 3 hours later
Spi~
(B and D). Unbudded
cells have not replicated
DNA and require ligase activity in the first cell cycle. The fate of GI
(unbudded) cells was quantitated by analysis of several fields; 78% (49 out of
63) of GI cdc9 cells are arrested with one large bud, whereas only 21% (12
out of 57) of GI cdc9-rad9 cells are arrested with one large bud. Most cdc9-rad9
GI cells (79%,45 out of 57) generated at least a third bud indicating the
original GI cell had proceeded through mitosis and one progeny cell had
initiated bud formation (and DNA synthesis) in the next cell cycle. The fates
of cells that were budded at the time of ligase inactivation were also analyzed
[note three in (A) and two in (C)]. Most of the budded cells have completed
replication and do not require ligase activity in the first cell cycle but do
require ligase for the second cell cycle. Eighty percent of budded cdc9 cells
generate two cells each arrestingwith one large bud, whereas 45% (45 out of
99) of budded cdc9-rad9 cells generate two cells each with one large bud and
55% (54 out of 99) proceed to a subsequent cell cycle. Combining data from
unbudded and budded cells 79% of cdc9 cells and only 36% of cdc9-rad9 cells
are blocked as large budded cells after 3 hours. Cytological examination
(described below) shows that even the few cdc9-rad9 cells that are arrested
with one large bud have proceeded past mitosis, whereas most of the cdc9
cells have arrested at mitosis. About 40% of the cdc9-rad9 cells at the time of
plating were inviable and did not change their shape after 3 hours [note
unchanged large budded cell in the comer of (C) and (D)]; these cells were
excluded from quantitative analysis. (E) DNA ligase inactivation results in
the arrest of cells before mitosis. cdc9 cells grown at the permissive temperature (23C) were shifted to restrictive temperature (36C) for 3 hours, fixed,
and analyzed for cell (top), nuclear (middle), and microtubule (bottom)
morphology. Two cells are shown. M, mitochondria; N, nucleus; S, spindle;
C, cytoplasmic microtubules. Cell morphology was determined by differential-interference-contrast microscopy and nuclear and microtubule morphologies, by epiiluorescence from the DNA-binding dye DAPI (4',6-diamidino-2-phenylindoledihydrochloride) and from FITC-conjugated antibodies to antibodies to tubulin, as described ( 3 8 ) .The position of the nuclei and
the presence of a short spindle at the neck of each large budded cell has been
described (36). When cells were shifted from growth at 23C to the
restrictive temperature for cdc9, the percentage of cells with DNA at the neck
of a large budded cell increased from 11to 84% for cdc9 and only from 11to
32% for cdc9-rad9. As expected, the distribution of cells in other parts of the
cell cycle showed that most cdc9 cells were arrested before mitosis, whereas
most cdc9-rad9 cells were not arrested before mitosis (27). Strains: 598-3
M A T a , cdc9-8 ura3 ade2 ade3 leul can1 cyh2 sap3 trpl SCE:: U R A 3 (provided
by L. Kadyk), and 7851-3-2 M A T a cdc9-8 rad9:: T R P l hpl ura3 leu2 ade3 ade2
his3 leul. Both are congenic with A364a. The R A D 9 deletion we constructed
by deleting 93% of the internal coding sequence and inserting a DNA
fragment containing the T R P l gene, and will be described in detail
elsewhere.
DNA

ARTICLES 631

Fig. 3. Rapid loss of viability in cells

defective both for DNA ligasc and for
the RADY gene. cdcY ( 0 )and cdc9-vadY
(0)

cells growing at 23C were shifted to

the restrictive temperature and viability
was determined by plating for viable
colonies at the permissive temperature
(23C). The cell viability reported is
relative to viability at the time of ternperaturc shitt. Results were reproducible
in separate experiments and with other
co~lgc~lic
strains.

0.1

0 1 . 5 3

Time (hours) at 36"

an early event rcsulted in complete arrcst of a late event. It is clcar in

other cases that inhibition of an carly event merely delays a late
event. Unless careful studies are carried out thc delay might be
overlooked. In cases where delay occurs, a checkpoint may exist. The
delay in thc late cvcnt indicates its dependence on thc carly cvent;
rnoreovcr, the eventual occnrrencc of the late event in the absence of
the early event indicates that relicf of dependcncc eventually occurs
spontanconsly making it unlikely that dcpendencc is due to a
substrate-product relationship.
A control lnechanisin is snggestcd by the obsen~ationthat whcn a
chromosome lags in finding its way to the rnetaphase plate, anaphasc is delayed, often until the lagging chrornosolne arrives at the
inctaphasc plate (34). Perhaps a similar checkpoint is present in yeast
cells, sincc cell division is greatly delayed in thosc divisions at which
loss of ccntrorncrc-containing plasmids occurs (35); the dclay of
division occurs at a spccific stagc in the cell cycle probably corrcsponding to mitosis.
Dependence og.centrosome duplication on D N A synthesis. In yeast (36)
and mammalian cells (37),inhibition of DNA synthcsis by ternperature-scnsitivc mutations in replication enzymes or by aphidicolin,
respectively, arrest spindle pole body or centrosolne duplication at a
stage characteristic of inetaphase. A inutation of ycast, espl, blocks
DNA replication and nuclcar division but not spindle pole bocty
duplication, so that nuclei accumulate with as many as cight spindle
pole bodies (38); thus, this csscntial genc plays a role in preventing
spindlc pole body duplication in the absence of DNA synthcsis.
Dependence c!freplicon initiation on other replicons. In mammaliail cclls
large contiguous regions of DNA, corresponding roughly to the
cytogcnetic bands of chromosomes, replicate coordinately as a result
of the fact that large arrays containing as many as a hnndrcd
replicons initiate replication synchronously (39). The initiation of
DNA synthesis has been found to be exquisitely sensitive to singlestrand Icsions (40); onc single-strand brcak may inactivate initiation
in as inany as a hundred replicons, presulnably those actjacent to one
anothcr. This rcsult is surprising, since the chromosome is thought
to be organized into topological domains corresponding roughly to
single
replicons.
The coordinate inhibition of initiation in clustcrs of
rcplicons is likely due to a control mcchanisin, since cclls from
illdividuals with thc gcnctic discase, ataxia telangicctasia, a syndroine predisposing individuals to canccr, are resistant to this
inhibition (41). I11 thesc cells dependency of DNA synthesis on
intact template is relieved and might be the cause of radiation
sensitivity. In unirradiated wild-type cclls, this control may prevent
reinitiation of any unligated strancts remaining froin thc previous S
phase; replication of a chromatid containing a single-strand lesion
632

would gcncrate a doublc-strand break leading to the production of

an acentric chrolnosolne fragrncnt.
Dependence o f D N A reinitiation on mitosis. Mammalian cells inhibited in mitosis by colchicine (42) or mutations (43) dclay for Inany
hours but eventually reconstitute an interphase nucleus and replicate
their DNA without completing chroinosorne segregation. 'Ilese
observations may idcntify a chcckpoint that makes reinitiation of
DNA replication dcpendent on mitosis. This dependcncc has been
reproduced in vitro in extracts of Xenopus eggs where added DNA is
asselnbled into a nuclcus and rcplicates once (44); nuclei can
rercplicate without completing initosis if they are treated with
suggested
agents that inakc thein permeable, an observation that h..~s
a spccific lnodel for this ctcpendencc (45).
Depentlence ofmitosis on cqv~nlth.The product of the WEE1 gene of
S . pombe is not csscntial, but its prcsence delays mitosis; dclction of
WEE1 leads to cells that are sinaller than norinal at initosis, and
increased dosage leads to proportionatcly larger cells at mitosis (46).
This obscrvation may identify a checkpoint that integrates growth
and division (11).
Checkpoints in bacteria. In Esclzevichia coli, cell division is dependent
on coinpletion of DNA replication (47) and on the prcscnce of
undamaged DNA (48) as wcll. The SOS regulatory system of E. coli,
likc the R A D 9 system of ycast, is responsible for thc arrest (depcndency) of cell division in responsc to an inhibition of DNA
replication and in rcsponsc to DNA damage (49). DNA damage is
recognized (directly or indirectly) by thc KecA protein, which in its
activated form stimulates proteolytic cleavage of the LexA protcin,
an inhibitor of S u l A gene transcription. SUM protein inhibits cell
c~visionpossibly by inhibiting the FtsZ protcin. Null mutations of
S u l A arc insensitive to division arrest by DNA damagc, thus
lneeting the criterion of a checkpoint. Deficicncy for S u l A generatcs
a higher frequency of anucleate cells when DNA synthesis is
inhibited than is the case for wild-type cells (50).
A checkpoint that couplcs F plasmid replication and segregation
to I:'. coli cell division has becn identified (51). A gene, CcdB, carricd
by stable mini-F plasmids inhibits 1:. coli cell division when
replicarion of the plaslnid is inhibited. Cells containing plaslnids
lacking this function producc plasmid-frce cells at mulch higher
frcquencics than thosc containing this fiinction.

Some Embryonic Cell Cycles Lack Some

Checkpoints
Many experimental rcsults snggcst that soine early embryonic cell
cycles are controlled diEcrently than somatic cell cycles (11); thc
differences may be attributable to thc cxistencc of fewer checkpoints
in some embryonic cell cycles.
Pcrhaps thc inost definitivc cvidence for a difference in the control
of the cell cycle between solnatic cells and soine embryonic systems
arises from the consequences of inhibiting DNA synthesis. As
discussed above, chromosome conetensation, claboration of the
mitotic spindle, and cytokincsis are all prevented in yeast and other
somatic cells when DNA synthcsis is inhibited. However, in Llvosophila one or morc aberrant nuclcar divisions may occur after
inhibition of DNA synthesis (52); centrosoines continuc dividing
for many divisions up to the time of ccllular blastodernl (53), and,
although cytolunesis is not completed, the egg surface undergoes
pcriodic budding. Thus, soine of the events of cell division are being
activated. I11 early Xo~ncyusembryos (54), ccll division continues at a
normal ratc after inhibition of DNA synthesis until the mid-blastnla
transition; anucleate cells arc thc products of thesc divisions.
Some embryonic cells ditfer from somatic cclls in their response to
broken DNA. Although solnatic cells arrcst in G2 in responsc to xSCIENCE, VOI,. 246

irradiation, some einbryos seem to be insensitive to brokcn DNA crrors in thc scgrcgation of other organellcs. A likcly period to look
sincc Rana pipiens oocytes fertilized with heavily irradiated sperm for thc deliberate elimination of defective cells is at the time when
divide to proctucc haploid organisms (55) and carly cleavage stage cell divisions slow down, transcription begins, groups of cclls
Dvosophila embryos continuc nuclear division, DNA rcplication, and become asynchronous, and gastrulation commences (thc mid-blasccntrosoine duplication for sevcral cycles with little or no initotic tula transition in Xenopus and division fourteen of Dvosoplzila). It
delay aftcr x-irradiation (56). We spcculate that the Rana pipiens and may be pertincnt that nuclei of Drosophila cmbryos that fail to dividc
Dvosoplzila analogs of thc KAD9 checkpoint are inactive in these or that merge with other nuclei fall into the intcrior of thc cgg at this
time in einbryogenesis and do not contribute to the larval cclls (65).
embryos.
Thegnu mutation of Drosophila has a phenotype that supports the Thc Drosophila embryo has the capacity to replace lost nuclei even
idea that many cell cycle cvcnts are indcpcndcnt of one atlothcr in when the number lost is quite large; when nuclei are inactivatcd by
the embryo (57); gnu embryos replicate DNA and centrosomcs ultraviolet irradiation before the fourtecnth division, compensatory
without undergoing nuclear division. In contrast, yeast mutations divisions occur that approximately restore the appropriatc numbcr
that block nuclcar division also prcvcnt the next round of DNA of nuclei before ccllularization (66). The existence of a monitoring
system for ancuploicty cdoes not seem too difficult to imagine in vicw
replication (58) and spindle pole bocty duplication (36).
Quite dramatic results have becn obtained in thc study of of the phenomena of dosage compensation (67) and X chronlosome
enucleatcd einbryos where centrosome duplication and cortical inactivation (68), situations in which differences in chromosome
contractions characteristic of cytokinesis continue for many cycles, ratios are dctccted.
A variety of abnormal cells arising from infidelity of the mitotic
often with normal kinetics (515, 59, 60). However, thcse results arc
not necessarily informative with respect to our search for thc proccss havc been detected in humans including aneuploidy, genc
presencc of checkpoints, sincc thcse control mechanisms probably amplification, and lnultipolar nlitoses. Sporadic cases of mitotic
require signals gcnerated by the failurc of onc evcnt that are received infidelity may not inerit special attention since inany causes are
by and serve to inhibit soine other event; with thc nucleus or possible. However, when initotic infidelity is rampant and rcproanother organelle rcmoved it is unavailable to send signals. For ct~icibleas it is in many typcs of human tcilnors it may be fruitful to
cxample, sca urchin embryos havc becn rcported to continue to consider pcrturbations of the checkpoints that norinally ensure
prelnaturcly condensc thc chromosomes of fertilizing spcrin aftcr initotic fidelity as potential causcs.
cnuclcation but not after inhibition of DNA synthesis with aphidicolin (62).
REFERENCES A N D NOTES

Fidelity in the Embryonic Cell Cycle

The function of chcckpoints in the cell cycle is to cnsure the
completion of early events before late cvents begin. When checkpoints are eliminated by nlutation or other means, ccll dcath,
infidelity of chromosome transmission, or increased susceptibility to
environmental perturbations (like DNA dainaging agents) result.
Since some early embryos (Dvosophila, Xenopus) lack the checkpoint that makes mitosis dependent on DNA replication, the
considerations discussed above would prcdict that thc cmbryonic
cell divisions in these orgatlislns might occur with less fictelity than
the somatic cell divisions of thc same organism. T o our knowlectge,
therc is no data available at present on thc fidclity of chromosoinc or
other organclle distribution during embryonic cell division in
comparison to somatic cell divisions of the samc organism; we will,
nevcrthclcss, pursue the iinplications of this thought.
If this prediction of a lower fidclity in soine embryonic divisions is
verified, we will need to consider why cinbryonic dcvclopment has
sacrificed fidelity. We suggest that chcckpoints would dclay cell
division in those cclls whcre stochastic problems require correction
and would lead to asynchrony in a cell population. It appears that
those embryonic systems that have eliminated the chcckpoint,
making mitosis depenctent on DNA replication, arc the systems
whcre rapid and synchronous division is evident. Perhaps checkpoints have been eliminated becausc synchrony and speed wcre
important for cmbryonic developn~ent.
What is the valuc of fidelity to inetazoans and what price would
the embryo pay in sacrificing fidclity? Dcvclopincntal abnorn~alities
(63) and catlccr (64) secln to bc much greatcr risks in organisms with
chrolnosolnal atleuploicty. It seclns unlikcly that embryonic systeins
would incur thcse risks. We suspect thcrcforc that thesc cmbryonic
systeins havc evolvcd some compensating systein to allay increascd
risk of cancer and developlnental aberrations. We suggest that
embryonic systems utilizing rapid synchronous divisions will ctetect
and eliminate atleuploid cells and possibly cells that have incurrcd
3 NOVEMBER 1989

1.
2.
3.
4.

C. S. Ncwlon, Miimhiol. Rev. 52, 568 (1988).

A. Kornbcrg, D N A Hcpliiatton (Freeman, San Francisco, 1988).
S. M. Gasser and U. K. Lacmmh, Trends Go1t.t 3, 16 (1987).
1). Maz~a,Iirt. Rt.11. (:ytol. 100, 49 (1987); J. R. McIntosh, h f d . Ccll Uiol. 2, 115
(1983).
~ ~ yMerhanhinr, V. L. Dcllarco, P. E.
5. J. R. McIntosh, in Ancriploidy: E t i ~ i l ~ iand
Voytek, A. Hollacndcr, Eds. (Plenum, New York, 1985), pp. 197-229.
6. T. J. Mitchison, A n n u . l<c~>.
Cell R ~ o l 4. , 527 (1988).
7. 1. W. New~mrtand 13. I. Forbes. A n n u . Rt.11 i3locht.r~.56. 535 119871.
8. k. W. irk;, Int. Rev. k y t o l . 15,'49 (1983).
9. J. M. Mitchison and P. Nurse, 1. Ct.11 Sci. 75, 357 (1985).
10. We use the tcrm early cmbryonic cell cycle to refer to the early divisions of
organisms like Xeilopus and Dr<rcophila in which the divisions arc rapid, synchronous, occur without growth, and are iulder the control of the maternal gcnomcthat is, thosc befixc cyclc fourteen of Dmcophila and those before the mid-blastula
transition of Xtnopus. We use the tcrm somatic cell cyclc to rcfcr to the divisions of
cukaryotic microorganisms as well as most metauxu1 cclls, excluding only thosc
defined as carly cmbryonic.
11. A. W. Murray and M. W. Kirshner, Scit.ilr 246, 614 (1989).
12. P. N. Rao, in I'rrmaturr Chroinoroinc Conde~tcarioil:Application iil Uasrr, Clitliral atld
Mritatioil Rt.st.arrh, P. N. Rao, R. T. Johnson, K. Spcrling, Eds. (Academic Press,
New York, 1982), pp. 1-42.
13. J. R. Pringlc and L. H. Hamvell, in M<~lerular
Biolopy ofthe Y r a ~ Sarihaminyccr:
t
Id!/+
Cyclc and Iilht.rirailie, J. N. Strathem, E. Joncs, J . Broach, Eds. (Cold Spring Harbor
Laboratory, Cold Spring Harbor, NY, 1981), vol. 1, pp. 97-142.
14. T. Hirano and M. Yanagida, in M~~lccular
and Cell R i o l ( ~ f yoJYeast~,E. F. Walton,
Ed. (Rlackie, Glasglow, Scotland, 1988), pp. 223-245.
15. P. Baum, C. Furlong, R. Hycrs, I'ror. .Vat/. Acad. Sii. L1.S.A. 83, 5512 (1986); M.
Rose and G. Fink, Ct.11 48, 1047 (1987).
16. T. C. Huhkcr, J. H . Thomas, 13. Hotstein, J . Cell Uiol. 106, 1997 (1988).
17. M. Rudd and J. L. Campbell, Pr~ii.Narl. Acad. Sci. L1.S.A 84, 2838 (1987); K. C.
Sitney, M. E. Rudd, J. L. Campbell, Ccll 56, 599 (1989).
18. 13. G. Harkcr, A. L. Johnson, L. H. Johnston, M G G ( h f o l . Gt.11. Co1t.t.) 200, 458
(1985).
19. M. Han, M. Chang, U. J. Kim, M. Gn~nstcin,Ccll 48, 589 (1987).
20. J. H. Thomas and 13. Rotstcin, ihid. 44, 65 (1986).
21. C. Holm, T. Goto, J . C. Wang, 13. Rotstc~n,ihid. 41, 553 (1985).
22. 13. Koshland and L. H. Hartwell, Scicnre 238, 1713 (1987).
23. P. H. Bcrgct, in V i r u ~Structure and Arceinbly, S. Casjens, Ed. (Jones and Hartlctt,
Boston, l985), pp. 149-168.
24. P. H. Rcrgct and J. King, in Racteriophafe T 4 , C. K. Mathews, E. M. Kuttcr, G.
Mos~g,P. B. Berget, Eds. (American Society fix Microbiology, Washington, DC,
1983), pp. 246-258.
25. H . Murialdo and W. L. Fife, Golctia 115, 3 (1987).
26. In some mutants defective in genes needed for T4 head morphogcncsis, the major
head subunit assembles into aberrant stnictures; most of thcsc stn~cturcsarc easily
distinguished from normal heads. [See U. K. Laemmli, E. Molbert, M. Show, E.
Kellcnbcrgcr, J , Mol. Uiol. 49, 99 (1970).
27. T. A. Weinert and L. H . Hamell, Sciolre 241, 317 (1988); unpuMishcd
observations.
ARTICLES

633

28. L. H . Johnston and K. A. Nasmyth, .Vature 274, 891 (1978).

29. K. 11. Schlcstl 1.1 a1 , h f ~ l (.,ell.
.
l j i o l . 9, 1882 (1989).
30. C;. Olivicri and A. Michcli, Muiar. Ro. 122, 65 (1983); V. W. Bums, Railiat Rcs.
4,41 1 (1956);C. Luckc-Iiuhlc, E. A. Rldkely, P. Y. Ch-ung, C. A. Tobias, ibrd. 79,
97 (1979); B. S. Baker, A. T. C. Carpenter, M. Gatti, Pnlf. ' l o p (:yti;qeirei, 7, 273
(1987); J. G. Cdrlson, (;i,rrrrics 23, 596 (1938).
31. ( 1 1 . ~ .u1d
~ 1 A. l'.lrdcc, l'ror. .\:a11 Acaii. S r i . 1 i . S . A . 79, 2942 (1982); 1'. M. Bussc,
S. K. liohc, I<. W. Joncs, L. J . Tolmach, R a i l i a ~ Iln 76, 292 (1978).
32, Ilo~vcvcr,a s~gniticantdistinction benvccn the ctfccts of caffeine ,und those of
K A I 19 deficiency 1s that cclls blocked in the middle of I)NA svnthcsis can enter
snitos~safter catitlne tre,ltmcnt, whcrc,ls rad9-dcficlcnt cells blocked in c,lrl\r S phase
rclnain arrested; only cclls blockcd late in S enter mitosis [T.A. Wclncrt and L. H.
H,lrtircll, utlpublishcd obscnrations; R. Schlegel ,uid A. B. Pardcc, Scic,ircc 232,
1264 (1986)l.
33. S. A. Ohmani, 1). R. tnglc, J. H. IDoonan; N. R. Morris, C c l l 52, 241 (1988); 'I.
N i s h i n ~ ~ t oR., Ishidd, K. Ajil-O,S. Yamamoto, T. Takahashi, J . C c l l P i i y ~ i c ~109,
l.
299 (1981).
34. A. Rajcr .lnd J . Molc-13ajrr, ( ~ l i r n i n ~ ~ s ~7,
~ r i558
r a (1956); (: I.. Kicdcr and S. 1'.
~iiiy
~ / O u y i w M.
, A. Rcsnick Ed. (I.iss, New York,
Alea.itldcr, In A r ~ c i ~ / ~ i ~:Llciliarri~irir
1989); R. I;. Lirklc, Ilaiiiiit. Rcc. 41, 516 (1970).
35. A LV. Murr.1) and J. W. Szostak, personal a)tnnlilnlccltlon.
36. 13. Iiycrs a ~ t dL. C;octhcll, (:old S/~rii;q11(1ul1orSyinp. Quarri Riol 38, 123 (1974).
37. J. R . Kdttncr and S G. I'hilllps, J (:ell Uiol 57, 359 (1973); R. Kuriyama and G.
G I{ortsy, ibid 91, 841 (1981).
38. 1'. li'1~11n.(:. Ylp, L C;octsch, 13. I<ycrs,,Lloi (;ell Uiol 8, 5386 (1988).
39. R. H.lrld, (>I1 l(ic11 2, 389 (1979); Y. F Lau and F. E. Arrighi, (:lirnri~ocorria 83,
721 (1981),
40. I. W.~td~~dl)c,
I<alirai l(1.3 58, 541 (1974); 13. I'ainter and B. K. Young, Rioiliirii.
lliiyliy,. A l r o 418, 146 (1976); L. F. I'ov~rk, J . M o l Biol 114, 141 (1977).
41. R. li l'atntcr and li. K. Young, 1'101. .\:(ril. .,lied. S l i . 1J.S A . 77, 7315 (1980); Y.
Shiloh, I:. 'I'abor, Y. Bcckcr, K K O C l~oirrid.Scr. 19, 111 (1985), K . B I'd~ntcr,
11~1,l, kl. 89.
42. 13, N liarl>cr and I I. C;. C;<III.I~I,
l>ro<.l l . ,So(. Loir~Ioii,Si,r. R 131, 258 (1942).
43. R. J. Wang ~ n c L
i Y I I ~I.:\,~J (.'c/I 11~5.101, 331 (1976).
4 4 J. J. I<loa. atid K. A. Laskc!, "Grirruc 332, 546 (1988).
45. R . A. I..lskry, M. 1'. F'~lrman,J. J. lilow, Siicnn. 246, 609 (1989).

46.
47.
48.
49.
50.
51.
52.
53.
54.

P. Russell and 1'. Nurse, (:ell 49, 559 (1987).

W. E. Howe ,und L). W. Mount, J Barteric~l. 124, 1113 (1975)
J . W. Little 'lsld I). W. Mount, C c l l 29, 11 (1982).
G. C. Walker, Miirohiol. Rcv. 48, 60 (1984).
A. Jaffc, R. ID'Ari, V. Norris, J. Uartcrkll. 165, 66 (1986).
'1'. Obmm and S. His-lgd, I'voi. ~ V a t l Acad.
.
S i i . l/.S..,t. 80, 4784 (1983).
B. A. Edgar and G. Schubigcr, C c l l 44, 871 (1987).
J. W. liaif and ID. M. Glover, J . C c l l Uiol. 107, 2009 (1988).
1). Kimelman, M. Kirschner, T. Schcrson, Cell 48, 399 (1987); M. 1)asso and J.
Ncwport, pcrsondl co~ilmunic,lt~on.
55. R. Rriggs, E. U. Grccn, 'I'. J . King, J . ESJI. Z o o / . 116, 455 (1951).
56. A. Schncidcr-Minder, 11ri J . Kadiai. U i o l . 11, 1 (1966).
57. M. Freeman, C. Nusslcin-Volhard, I). M. Glover, Cell 46, 457 (1986).
58. J. Culoni dtid L. H . Hartwell, Esp. Cell R e . 67, 389 (1971).
59. K. Ham, 1'. 'I'ydcman, M. Kirschner, P m l . .Y(rfl. .,tiail. S l i 1 i . S . A . 77,462 (1980).
60. E. Schicrcnbcrg and W. R. RTood,D c v Riol. 107, 337 (1985).
61. C;. Klllian, C:. Bland, J. Kuzara, 1). Nishioka, Esp. Ccll Kcr. 158, 519 (1985).
62. G. Sluder, F J . Miller, C:. I.. Rider, 1. C c l l Rrol. 103, I873 (1986).

63 C . J . F,psteisl, Y711' ( . ~ o r r ~ i ~ ~ j i i i1:)

~ r r( l~i ~I ri r ~ ~ i i r i ~liiiI1~111r1r~i~.
r ~ ~ r r r ~ ~I)iririipIi,>, : \ 1 ~ ~ ~ 1 r ~ r r r(irrd

1~111~,
a i d Cell B~ologyScrlcs (C;.~~ilbridge
Illli\.. I'rcss,
:\lodi,i\, \,ol. 18 of l)cvcloprnc1it~1l
NC\\.\i<)lk,1986).
64. J. J. Yilnis, .Siil'irii, 221, 227 (1983); J. M . Bisllop, ihr~l.235, 305 (1987); 1:
Mitclniati, .\'1111i17, 310, 325 (1984); E. M . K~lllliJ I ~ Lb.
~ 'I'I1cr111~1t1.( :oiri(,i (;i.riti
(.'y/i~q~,iii.~
22, 1 (1986); R. Holliday, 'Ticri~b(;orci. 5, 42 (1989).
65. W. Sullivan, J. S. Mindcn, B. M. Alberts, J . G.11 Riol., in press; J. S. Mindcn, 1). A.
Agard, J. W. Scdat, li. M. Alberts, ihiii. 109, 505 (1989).
66. G. Ydsuda and G . Schubigcr, persoslal communication.
67. E. Jdffc and C. l a r d , 'Tuciiiic Gciicr. 2, 316 (1986).
68. S. G. Gmnt and V. M. Chapman, Annir Rcll. Gcrrci. 22, 199 (1988).
69. Wc thank H. Hrcwcr, B. Bycrs, F. (:ross, W. Fatlg~llati,N IIolli~lgs\vortll,C;.
Mann, (:. Masloil, and G. Schuhigcr for criticlstll of'the rnanuscrtpt; M. Dasso, L.
Edgar, A. Murray, J. Nc\\,port, (:. Kiedcr, G. Schuh~gcr,J. S/ost,~k,13. S u l l ~ v ~ ~ n ,
and G. Yasuda fbr unpuhl~shcdinfbrtllation. O u r work was supported hy grants
from the NIII (GM17709)Institute of General Medical Sctcnccs and the Amcrlcan
Hustness Foundation fix (hnccr Kescarcli. 'I.A.W. was supported by a f'cllo\\~shi~~
fi-om the Jane (;oftin Chllds Foundation.

AAAS-Newcornb Cleveland Prize

To Be Awarded for an Article or a Report Published in Science

The AAAS--Newcomb Cleveland Prize is awarded to the
author of an outstanding paper published in Science. The value of
the prize is $5000; the winner also receives a bronze medal. The
currcnt competition period began with the 2 June 1989 issue and
ends wid1 the issue of 25 May 1990.
Keports and Articles that include original research data, theories, or syntheses alci are hmdamental contribLltions to basic
conse.
lulowlcdge or teclulical achievements of
qucnce arc eligible for consideration of the prize. The paper must
bc a first-time publication of the author's own work. Reference
to perrlIlcnt earlier work by the aLldlor may be included to give

63 4

Throughou~tthe competition period, readers are invited to

nominate papers appearing in the Reports or Articles sections.
Nominations must be typed, and the following information
provided: the title of the paper, issue in which it was published,
author's name, and a brief statement of justification for nomination. Nominations should be submitted to the AAAS-Newcomb
Prix>
924, 1333
Street, NW,
Washhlgton, I)C 20005, and must be received on or before 30
June 1990. Final selection will rest wid1 a panel of distinguished
scientists
Science.
The award
be presented at
1991 AAAS
meeting. In cases of multiple authorship, the prize will be divided
eqi~allybetween or anlong the authors.

SCIENCE, VOI.. 246