ITMC17 9/30/06 0:03 Page 271

17
The evolution of viruses

Like all living organisms, viruses are not static entities but are subject to evolutionary pressures
and undergo evolutionary change.

Chapter 17 Outline
V iral genomes accumulate mutations in the
same way as all other nucleic acids and,
where conditions enable a mutant to multiply
17.1 The potential for rapid evolution in RNA
at a rate faster than its fellows, that mutant virus
viruses: quasispecies and rapid evolution
will have an advantage and will succeed the
17.2 Rapid evolution: recombination
parental type. Where viruses have an evolu-
17.3 Evolution of measles virus
tionary advantage is that many polymerases
17.4 Evolution of myxoma virus
(notably in the RNA viruses) lack proofreading
17.5 Evolution of influenza virus
mechanisms, so that mistakes in replication
are not corrected. Thus, mutants accumulate
more rapidly than seen with DNA-based organisms where DNA replication has proofreading.
Viruses also have a shorter generation time relative to their host. In any discussion of evolu-
tion, it is pertinent to ask where viruses first came from (see Section 1.9). The absence of any
fossil records and the scarcity of other evidence have not, of course, prevented scientists from
speculating about their origins. The two prevailing opinions are that viruses have either arisen
(i) from degenerate cells that have lost the ability for independent life, and/or (ii) from escaped
fragments of cellular nucleic acid.
The history of viruses extends as far back as man’s historical written or pictorial records. Amongst
the earliest references is a 3500-year-old bas relief sculpture from Egypt depicting a man sup-
porting himself with a crutch. Careful examination shows that the calf muscle of his right leg is
withered in the characteristic aftermath of paralytic poliomyelitis. A slightly later excavation of
the same civilization found the mummified body of the pharaoh, Rameses V, whose face bears
the characteristic pock marks of smallpox. Such evidence tells us of the long association of these
viruses with mankind.
ITMC17 9/30/06 0:03 Page 272

272 PART III VIRUS INTERACTIONS WITH THE WHOLE ORGANISM

Whatever their origins, viruses have been a great biological success, for
no group of organisms has escaped their attentions as potential hosts.
Viruses are largely species-specific with respect to their host and usually
do not cross species boundaries. It seems likely that every animal species
has the same range of viruses as occurs in humans. In Chapters 13–16,
the various ways in which viruses interact with their hosts are discussed,
illustrating how viruses cause a variety of changes, ranging from death
to the imperceptible. Evolution of any successful parasite has to ensure
that the host also survives. The various possible virus–host interactions
can be thought of as different ways in which viruses have solved this
problem.

17.1 THE POTENTIAL FOR RAPID EVOLUTION IN RNA VIRUSES:

QUASISPECIES AND RAPID EVOLUTION

The lack of a proofreading function in the polymerases of RNA viruses

means that base substitutions (mutations) occur at the rate of between
10−3 and 10−5 per base per genome replication event (i.e. one mutation
for every 1000–100,000 newly synthesized bases). Not all of the result-
ing mutants will be viable but many are, resulting in an extremely het-
erogeneous population of viruses. This is called a quasispecies. This has
a number of implications. For example, it is not possible to define the
genome sequence of that virus population precisely, and any sequence
in a database will represent only one member of the quasispecies. The
quasispecies phenomenon also endows RNA viruses with the ability to
adapt rapidly to and to exploit any environment they occupy. This is seen
on both a micro and a macro level, and arises because one or more mem-
bers of the quasispecies will inevitably have a selective advantage over
others. The micro level is infection of the individual, and virus evolution
during the life time of the individual is seen, particularly in life-long infec-
tions with HIV-1 (see Chapter 19) and hepatitis C virus (see Section 20.8).
It is likely that in these infections, the immune system is a major select-
ive force, and that the quasispecies phenomenon allows these viruses to
persist in the face of that immune response. The macro level of evolu-
tion is seen in viruses with a worldwide distribution. This could impinge
upon people with different genetic backgrounds, especially those that
affect the MHC (HLA) haplotype that controls T cell immunity. This leads
to the establishment of different specific genetic sequences, known as
genotypes, or clades of a virus in different parts of the world, and a
classification that is based on sequence rather than serology (e.g. HIV-1,
Chapter 19).
Influenza A virus does not cause persistent infections but virus vari-
ants evolve under the selection pressure of antibody stimulated during
ITMC17 9/30/06 0:03 Page 273

CHAPTER 17 THE EVOLUTION OF VIRUSES 273

an acute infection. These variants are in turn passed on to susceptible

individuals in a country or a continent. Thus, the virus evolves with the
effect that, in approximately 4 years, it is not recognized by the immune
memory established by the original host, and he or she can be reinfected
(Section 17.5). This is called antigenic drift, and essentially results in the
formation of new serotypes of virus.
Not all RNA viruses exhibit such obvious variation, even though they
have the same potential for change as HIV-1 and influenza virus. For ex-
ample, the overall antigenicity of polioviruses types 1, 2, and 3, and measles
virus has not changed over their known history of approximately 50 years,
and the original vaccines still provide the same level of protection. Thus,
by comparison with HIV-1 and influenza virus, the polioviruses and measles
virus are stable. However, their genomes do accumulate mutations, and
minor changes in antigenicity (in individual epitopes) of the virion can
be detected by probing with monoclonal antibodies. In addition, viruses
from different global areas can be distinguished by their genotype. It is
not understood why, for example, measles virus is antigenically stable
and influenza virus is not, as in many ways these two infections are sim-
ilar (both viruses are highly infectious, spread by the respiratory route,
and cause acute infections). Presumably measles virus is under a less effect-
ive or more constrained evolutionary selection pressure than influenza
virus, but exactly what these pressures are is not known.

17.2 RAPID EVOLUTION: RECOMBINATION

Recombination is the other major force in virus evolution and takes

place in a cell that is simultaneously infected by two viruses. Usually, the
two genomes are highly related with regions of homology between their
genomes that permit the replicating enzyme to move from one strand
to another. Thus at a stroke, the daughter molecule has some of the
properties of both parents. However, both parts of the resulting genome
have to be sufficiently compatible for the progeny to be functional. Both
DNA and RNA genomes undergo recombination. Detailed monitoring
of the HIV-1 pandemic shows new recombinant viruses arising between
different clades, and such plasticity makes the prospect of control by
vaccine problematic (see Chapter 19). Viruses with segmented genomes
can also undergo recombination by acquiring an entire genome seg-
ment from another virus, and this occurs at higher frequency. This
form of recombination is known as reassortment. The effect on antigenic-
ity is enormous as a virus can acquire an entirely new coat protein in
a single step. The prime example is influenza A virus where this pro-
cess, called antigenic shift, is responsible for pandemic influenza in man
(Section 17.5). However, a viable recombinant requires compatibility
ITMC17 9/30/06 0:03 Page 274

274 PART III VIRUS INTERACTIONS WITH THE WHOLE ORGANISM

between all eight genome segments, and this puts some limitation on the
creation of new viruses.

17.3 EVOLUTION OF MEASLES VIRUS

Measles virus

• Member of the paramyxovirus family.

• Infects only humans.
• Unusual in that every infection causes disease.
• Infection results in lifelong immunity.
• Measles can be countered by use of a live vaccine.

F. L. Black studied the occurrence of the measles in island popula-

tions (Table 17.1), and found a good correlation between the size of the
population and the number of cases of measles recorded on the island
throughout the year. A population of at least 500,000 is required to pro-
vide sufficient susceptible individuals (i.e. new births) to maintain the virus
in the population. Below that level, the virus eventually dies out, until
it is reintroduced from an outside source.
On the geological time-scale, humans have appeared only recently and
have only existed in population groups of over 500,000 for a few thou-
sand years. In the days of very small population groups, measles virus
could not have existed in its present form. It may have had another strat-
egy of infection, such as persistence, which would have allowed it to infect
the occasional susceptible passer-by, but there is no evidence of this.

Table 17.1
Correlation of the occurrence of measles on islands with the size of the island population.
(From Black, F. L. 1966. Journal of Theoretical Biology 11, 207–211).

Island Population New births per year Months of the year in

group × 10−3)
(× × 10−3)
(× which measles occurred
(%)

Hawaii 550 16.7 100

Fiji 346 13.4 64
Solomon 110 4.1 32
Tonga 57 2.0 12
Cook 16 0.7 6
Nauru 3.5 0.17 5
Falkland 2.5 0.04 0
ITMC17 9/30/06 0:03 Page 275

CHAPTER 17 THE EVOLUTION OF VIRUSES 275

However, Black has speculated upon the antigenic similarity of measles,

canine distemper, and rinderpest viruses. The latter infect dogs and cattle
respectively, animals that have been commensal with humans since their
nomadic days. Black suggests that these three viruses have a common
ancestor which infected prehistoric dogs or cattle. The ancestral virus
evolved to the modern measles virus when changes in the social beha-
vior of humans gave rise to populations large enough to maintain the
infection. The first such population occurred 6000 years ago when the
river valley civilizations of the Tigris and Euphrates were established.

17.4 EVOLUTION OF MYXOMA VIRUS

Myxoma virus

• Member of the poxvirus family.

• Natural host is the South American rabbit in which it causes only minor skin outgrowths.
• Also infects the European rabbit causing myxomatosis, with lesions over the head and body
surface; infection is 99% lethal.
• Is spread by arthropod vectors that passively carry virus on their mouth parts: mosquitoes in
Australia and rabbit fleas in the UK.

Myxoma virus was released in England and Australia upon wholly sus-
ceptible host populations of the European rabbit in an attempt to con-
trol this serious agricultural pest. This experiment in nature was carefully
studied with respect to the changes occurring in the virus and the host
populations and provides an object lesson in the problems of biological
control.
In the first attempts to spread the disease in Australia, myxoma virus-
infected rabbits were released in the wild but, despite the virulence of
the virus and the presence of susceptible hosts, the virus died out. It was
then realized that this was due to the scarcity of mosquito vectors, whose
incidence is seasonal. When infected animals were released at the peak
of the mosquito season, an epidemic of myxomatosis followed. Over the
next 2 years, the virus spread 3000 miles across Australia and across the
sea to Tasmania. However, during this period it became apparent that
fewer rabbits were dying from the disease than at the start of the epi-
demic. The investigators then compared the virulence of the original virus
with virus newly isolated from wild rabbits by inoculating standard lab-
oratory rabbits. Two significant facts emerged: (i) rabbits infected with
new virus isolates took longer to die, and (ii) a greater number of rab-
bits recovered from infection. From this it was inferred that the virus had
evolved to a less virulent form (Box 17.1). The explanation was simple:
ITMC17 9/30/06 0:03 Page 276

276 PART III VIRUS INTERACTIONS WITH THE WHOLE ORGANISM

Box 17.1

Evidence for the evolution of myxoma virus to avirulence in the European

rabbit after introduction of virulent virus into Australia in 1950

After release into Australia viruses were isolated from wild rabbits and their virulence tested
by infecting laboratory rabbits. The percentage of survial in the laboratory rabbits and the
mean survival times were calculated.

Mean rabbit Mortality Year of isolation

survival time (days) (%)
1950–1951 1952–1953 1955–1956 1963–1964

<13 >99 100 4 0 0

14–16 95–99 13 3 0
17–28 70–95 74 55 59
29–50 50–70 9 25 31
>50 <50 0 17 9

mutation produced virus variants which did not kill the rabbit as quickly
as the parental virus. This meant that the rabbits infected with the
mutant virus were available to be bitten by mosquitoes for a longer period
of time than rabbits infected with the original virulent strain. Hence
the mutants could be transmitted to a greater number of rabbits. In other
words, there was a strong selection pressure in favor of less virulent mutants
which survived in the host in a transmissible form for as long as possible.
The second finding concerned the rabbits themselves, and the pos-
sibility that rabbits were evolving resistance to myxomatosis. To test this
hypothesis, a breeding program was set up in the laboratory. Rabbits were
infected and survivors were mated and bred with other survivors.
Offspring were then infected, the survivors mated and so on. Part of each
litter was tested for its ability to resist infection with a standard strain of
myxoma virus. The result confirmed that the survivors of each genera-
tion progressively increased in resistance. However the genetic and
immunological basis for this is not understood.
This work shows (i) how a virus which is avirulent and well adapted
to peaceful coexistence with its host can cause lethal infection in a
new host, and (ii) how evolutionary pressures rapidly set up a balance
between the virus and its new host which ensures that both continue to
flourish. The latter remains a stumbling-block for biological control of pests
that attack animals or plants. Today, rabbits are still a serious problem to
agriculture in the UK and Australia.
ITMC17 9/30/06 0:03 Page 277

CHAPTER 17 THE EVOLUTION OF VIRUSES 277

17.5 EVOLUTION OF INFLUENZA VIRUS

Influenza A viruses

• Members of the orthomyxovirus family.

• Comprise 144 viruses (subtypes) that naturally infect aquatic birds; cause subclinical gut infec-
tions and virus is spread by the oral–fecal route.
• Can evolve to infect man, via domesticated poultry; causes respiratory disease that is spread by
infected droplets; serious morbidity and mortality, although can be subclinical.
• Initial introduction of a new influenza virus in man is called an antigenic shift; virus then
undergoes gradual change or antigenic drift – Darwinian evolution in response to positive
pressure from virus-specific antibody. Shift and drift refer to changes in the two major surface
proteins of the virus particle.
• Influenza can be countered by use of a killed vaccine, and recently by a live vaccine, and by
antiviral drugs.

Background

There are three groups of influenza virus, types

A, B, and C, that all cause respiratory disease in
man and can be distinguished by the antigenicity
of their internal virion nucleoproteins. Type A
viruses cause the worldwide epidemics (pan-
demics) of influenza, and both type A and B
viruses cause epidemics. Type C viruses cause only
minor upper respiratory illness and will not be
discussed further. In terms of natural history, the
primary hosts of influenza A viruses are wild
aquatic birds (such as ducks, terns, and shore
birds). Influenza B viruses infect only humans.
In man A and B viruses cause disease only in the
winter, usually January and February in the
Northern hemisphere, and June and July in
Fig. 17.1 Animal species that are naturally
the Southern hemisphere. At the equator, virus
infected with influenza A viruses. Wild birds
is present at a low level throughout the year. of the sea and shore form the natural reservoir
The cause of this periodicity is not known. In (top). Known routes of transmission are
addition, a limited number of other species are indicated by continuous arrows and probable
naturally infected with type A influenza viruses. routes of transmission by broken arrows.
These and the directions of transmission are
shown in Fig. 17.1. More recently (2005), cats and
dogs were found for the first time to be infected.
Typical influenza in man is a lower respiratory infection with fever
and muscular aches and pains, but can range from the subclinical to
pneumonia (where the lungs fill with fluid). In the elderly and people
ITMC17 9/30/06 0:03 Page 278

278 PART III VIRUS INTERACTIONS WITH THE WHOLE ORGANISM

Box 17.2

Influenza viruses have a formal descriptive nomenclature

Influenza virus type A

Subtype (example) H3N2
Where H is the hemagglutinin and N is the neuraminidase. There may be any permutation
of H subtypes 1–16 and N subtypes 1–9.
Strain (example) A/HongKong/1/68(H3N2)
The strain designation indicates (in order): the type, where the strain was isolated, the iso-
late number, the year of isolation, the subtype.
The nomenclature of nonhuman strains also includes the host species, e.g. A/chicken/
Rostock/1/34(H7N1).

with underlying clinical problems of the heart, lungs, and kidneys, and in
diabetics and the immunosuppressed, influenza can be life-threatening.
Immunity to an influenza virus is effective at preventing reinfection by
that same strain. The viral antigens relevant to protective immunity are
the envelope glycoproteins, the hemagglutinin (HA) and neuraminidase
(NA) (see Figs 3.14 and 3.15), and immunity is mediated by virus-specific
antibody. Despite the induction of the immune response repeated infec-
tion with influenza virus is common. The repeated infections are possible
because the HAs and NAs of influenza A and B viruses evolve continu-
ously so that previously acquired immunity is rendered ineffective. The
processes involved in the evolution are discussed below.
Influenza A viruses are classified formally as types, subtypes, and strains
(Box 17.2). There are 16 subtypes of hemagglutinin and nine subtypes
of neuraminidase. All 144 permutations are found in nature in wild aquatic
birds. By comparison, influenza in man is a side show, with currently
only two subtypes, H1N1 and H3N2, in circulation.

Two mechanisms of evolution

Influenza A viruses undergo two types of change affecting their major

surface glycoproteins called antigenic shift and antigenic drift (Box 17.3).
Since the start of modern virology there have been four shifts that
occurred in 1889, 1918, 1957, and 1968. A shift results in a pandemic
(a worldwide epidemic), is always preceded by an abrupt change in hemag-
glutinin subtype and, apart from the 1968 virus, also by a change in neur-
aminidase subtype. Drift results in epidemics, and is caused by gradual
evolution under the positive selection pressure of neutralizing antibody.
A new shift virus immediately starts to undergo continuous antigenic drift.
ITMC17 9/30/06 0:03 Page 279

CHAPTER 17 THE EVOLUTION OF VIRUSES 279

Box 17.3

Evidence for antigenic drift and shift in influenza virus

• Antigenic drift causes epidemics of influenza A and B in man. It results from a gradual
change in antigenicity of the influenza virus particle hemagglutinin (HA) and neuraminidase
(NA) proteins, through amino acid changes that affect antibody epitopes (see Figs 3.14
and 3.15). It takes ≥4 amino acid substitutions in ≥2 antigenic sites, and approximately
4 years, to evolve a drift variant that can escape immunity previously acquired to the
parent virus and so cause significant disease.
• Antigenic shift causes influenza pandemics in man. Only influenza A viruses are
involved. Shift results from an influenza virus acquiring a new HA and usually new NA.
This can arise by genetic reassortment of an existing human virus with a bird virus that
provides the new HA and NA genes. It may also arise by a bird virus adapting/evolving
to grow in man, and may involve intermediary mammalian hosts. Once formed, a shift
virus causes a sudden pandemic that spreads around the world in 1–2 years, but it
takes an unknown time (years?) to evolve to this point. A new shift virus starts drifting
immediately.

Influenza B viruses only undergo drift and this is believed to be because

they infect only man and have no other animal reservoir.
The complex evolutionary processes that influenza viruses undergo can
be better understood against a background of their biology. These viruses
are highly infectious and cause an acutely cytopathogenic infection.
Thus, they are a victim of their own success which results in a near uni-
versal immunity among hosts who, given adequate conditions, can live
for around 80 years. Apparently, the production of immunologically naïve
individuals by new births is not enough to allow influenza A and B
viruses to survive. Thus, they have evolved strategies for changing their
antigens in order to increase the proportion of susceptible individuals in
the population.
The phenomena of antigenic shift and drift in influenza have
significant implications for human infection and they are monitored by
a worldwide network of laboratories, coordinated by the World Health
Organization (WHO), that isolate and classify currently circulating
influenza viruses. In this way new strains can be quickly spotted and the
vaccine changed appropriately.

Antigenic shift

The appearance of shift viruses in man is shown chronologically in

Fig. 17.2. Virus was first cultivated in the laboratory in 1933 by intranasal
inoculation of ferrets with some nasal secretion from a virologist (Wilson
ITMC17 9/30/06 0:03 Page 280

280 PART III VIRUS INTERACTIONS WITH THE WHOLE ORGANISM

10M Smith) who had influenza, and was then suc-

1889
? cessfully passaged in embryonated chickens eggs.
1890 1900
1880 However, retrospective information has also been
obtained by studying influenza virus-specific
?
H2N2
H3N8 1910 antibodies in human sera that had been kept
? stored in hospital freezers for other purposes. As
1918 40M
1920
can be seen, antigenic shift has occurred sporadic-
2000 H1N1
1997 H5N1 (bird)
ally from the first recorded shift of 1889 at inter-
1990
vals of 11, 18, 39, 11, and more than 32 years.
H1N1 (pig) 1930 A brief history of the terrible 1918 pandemic is
H1N1
1980
H3N2 shown in Box 17.4. Shift, by definition, introduces
H2N2
1977 1940 a virus into a population that has no pre-existing
1976 1970 immunity, so it is always associated with an
1950
1968 1960 explosive pandemic with high morbidity and
0.7M 1957 mortality, although these vary between different
1M shifts. Since 2001, a new H1N2 reassortant virus
Fig. 17.2 History of antigenic shifts of has been isolated in many countries in mainly
influenza A viruses in humans. The outer young people. However, this virus has not
circle denotes the year of emergence of a spread widely, presumably because it is sensitive
new subtype that is shown in the inner circle, to existing immunity to its parent human H1N1
the duration of the reign of that subtype, and human H3N2 viruses. Approximate mortal-
and when it is replaced by another subtype. ity figures are included in Fig. 17.2.
A time scale is shown in the middle circle. Until 1977, only one virus subtype was in
Approximate worldwide mortality figures circulation at any one time, and this virus was
for each shift are indicated in millions. The replaced completely when a new subtype was
1900 shift did not cause a serious pandemic.
introduced. What causes the original subtype to
Currently (2006) H3N2 and H1N1 subtypes
coexist. Some occasional infections of
disappear is not known. In 1977, during the reign
people with bird or pig viruses that are not of the H3N2 subtype, an H1N1 virus appeared
transmitted person to person are noted in that was identical to the H1N1 virus that had been
the inner circle (but see text). in circulation in 1950. Thus this was not strictly
speaking a shift, but a reintroduction. Where
the 1977 virus came from is not known. It had not been infecting the
human population during its “absence” as it would have undergone 27
years of antigenic drift (see below). It is as if it had been frozen, and some
say that this is literally what happened. However, this is conjecture. At
the time of writing, the drifted descendants of the 1968 H3N2 and the
1977 H1N1 cocirculate and continue to undergo antigenic drift. However
a new shift could occur at any time.

The mechanism of antigenic shift

As indicated, the natural reservoir of influenza A viruses comprises wild

aquatic birds, such as ducks, terns, and shore birds. These avian viruses
are tropic for the gut (not the respiratory tract), cause a subclinical infec-
tion, and are evolutionarily stable. Many of these birds migrate enormous
ITMC17 9/30/06 0:03 Page 281

CHAPTER 17 THE EVOLUTION OF VIRUSES 281

Box 17.4

1918 influenza

The 1918 shift virus was unusual in several ways. It is estimated that in just 1 year it killed
around 40 million people worldwide. (The enormity of this can be appreciated when com-
pared to the 34 million that have died in 25 years of the AIDS pandemic.) The highest mor-
tality was in young adults and not the elderly and other high risk groups. The overall mortality
rate was 25 times greater than other pandemics (2.5% compared with 0.1%). Neither the
virulence of the 1918 virus nor its age tropism is fully understood, and it is feared another
high mortality virus might arise at any time.
The seemingly intractable problem of analyzing the virus from 1918 that had never been
cultivated was solved by extracting fragments of virus genome from the preserved tissue of
victims of that pandemic. These are either pieces of lung preserved in formalin in the US
army pathology archives or from the exhumed bodies of people who died and were buried
in the Arctic permafrost. Virus RNA has been analyzed by RT-PCR and and the complete
genome has now been sequenced. Phylogenetics allows hypothetical family trees of
influenza A viruses to be constructed, but the resulting relationship of the 1918 virus with
other viruses is perplexing. The analysis of the sequence suggests that the virus adapted rapidly
from birds to man and was not a reassortant.
Reverse genetics has allowed the functions of 1918 RNAs to be investigated by making
infectious virus which has one or more of its RNAs replaced by a 1918 virus RNA. The main
finding is that the 1918 HA converts an avirulent virus into one that is pathogenic for mice.
This reassortant virus stimulates macrophages to make cytokines and chemokines that attract
inflammatory cells into the lung and causes local hemorrhage – as did the virus in people
in 1918. Exactly how the 1918 HA differs in this regard from other HAs is not clear.
A complete infectious 1918 virus, A/Brevig Mission/1918, has now been recreated in the
USA using reverse genetics and viral RNA from a body preserved in the permafrost. This
has caused immense concern worldwide as nobody in the world has immunity to the 1918
virus, and the consequences of an escape would be horrific. However, it was considered
that the need to understand what made the virus so terrible made the risk worth bearing.
A raft of safety measures including a high level containment laboratory were used for this
experiment. Compared with an H3N2 virus, also derived by reverse genetics, the 1918 virus
needed no added protease to cleave the HA and form infectious virus, grew to higher titer
in mouse lung, caused more lung pathology, and killed chicken embryos. The H3N2 virus
with the 1918 HA had increased virulence, but was still less virulent than the complete 1918
virus, showing that other 1918 genes also contribute to virulence. In short, the reconstructed
1918 virus is highly virulent. However, what makes it virulent is still not understood.

distances (e.g. from Siberia to Australia) and spread virus as they go via
infected feces. In essence, antigenic shift in the circulation of human
influenza virus occurs when virus from a wild bird infects a human being
(Fig. 17.3). However, like most viruses, avian influenza virus is species
ITMC17 9/30/06 0:03 Page 282

282 PART III VIRUS INTERACTIONS WITH THE WHOLE ORGANISM

Box 17.5

Some major evolutionary groupings of vertebrates

Class Reptilia: reptiles

Class Aves: birds
Class Mammalia:monotremes (e.g. platypus)
marsupials (e.g. kangaroo)
placentals (e.g. man)

specific and does not readily in-

fect other bird species let alone
Adaptation mammals which are members of
Migration, through
another major taxonomic group
fecal passage (Box 17.5). Thus, the virus has
transmission
to adapt by progressive muta-
Pandemic tion before the jump to humans
can occur.
Gut infection Respiratory It is thought that the first
and subclinical infection
and disease link in the chain is the infection
Recombination of freerange domestic poultry
with existing human (mainly chickens, ducks, turkeys,
virus
Stable virus Evolving virus Shift virus and quail) by migrating wild
birds. This is an avirulent respir-
Fig. 17.3 Summary atory infection, although it is not understood how the switch from a
of the events leading gut to respiratory infection comes about. The infection passes from bird
up to an antigenic to bird until mutations take place in the protease cleavage site of the
shift of human
hemagglutinin precursor polypeptide. Cleavage is necessary for virus
influenza virus. The
infectivity. Normally, the HA can only be cleaved by a protease that is
evolutionary time
scale is not known present solely in the respiratory tract, but after an increase in the num-
but probably takes ber of basic amino acid residues in the HA cleavage site it can be cleaved
several years. by another protease that is widely distributed throughout tissues, and
this allows the virus to grow throughout the body and cause serious dis-
ease. At this stage the virus is able to jump from Class Aves to Class
Mammalia, and multiply (but not yet spread) in humans (Fig. 17.3). One
of the major gaps in modern virology is the understanding of animal virus
transmission in general, and what gene or genes are responsible for its
control. Another variation in the evolutionary progression of virus from
wild birds to humans may involve domestic pigs.
In rural areas poultry and pigs are often kept together, giving ample
opportunity for the crucial bird-to-mammal adaptation. At this point
the virus has two evolutionary options. It can continue to accumulate
ITMC17 9/30/06 0:03 Page 283

CHAPTER 17 THE EVOLUTION OF VIRUSES 283

mutations and become better adapted to humans, or it can H2N2 H3Nx

recombine with a human influenza virus strain and so NA
acquire genes that are already fully adapted. In fact, it may
do both. One essential adaptation is a change to the viral
attachment site on the HA protein so that the virus can use HA
as a receptor an N-acetyl neuraminic acid that has an α2,6
linkage to the carbohydrate moiety on a glycoprotein or gly-
colipid (as is found in cells of the human respiratory tract),
rather than the α2,3 linkage that is most common in birds.
The advantage to the virus of achieving this by recom-
bination is that at its simplest, there need only be a recom- Pool of RNA
bination of the RNA segment that encodes the new segments
hemagglutinin with the seven existing segments from the in a cell

human strain that control person-to-person spread. This

scenario would account for the 1968 shift where an H3N2
virus replaced the H2N2 virus.
H3N2
Recombination in influenza viruses is very efficient as the
genome is segmented – it comprises eight single-stranded,
Fig. 17.4 Recombination
negative-sense RNAs. The hemagglutinin and neuramin-
(reassortment) between an
idase proteins are encoded by distinct RNA segments. When existing human influenza A virus
a cell is infected simultaneously with more than one strain (H2N2) and a new virus from
of virus, newly synthesized RNA segments reassort at random the wild bird reservoir (H3Nx,
to the progeny (Fig. 17.4). Many, though not necessarily all, where x represents an unknown
of the 28 (256) possible genetic permutations that can be formed neuraminidase subtype; see text)
between two viruses are genetically stable. Reassortment that gives rise to antigenic shift.
occurs readily in cell culture, in experimental animals, and The two viruses simultaneously
in natural human infections between all type A influenza infect a cell in the respiratory
viruses, but not between A and B viruses. Even after a reas- tract, and the eight genome
sortment event, more mutational adaptation may be needed segments from each parent assort
independently to progeny virions.
before the virus is able to cause a recognized pandemic. Thus
The example shows a novel
it may be that a shift virus is present in the human popula- progeny virion (H3N2) that
tion, undergoing mutational adaptations for a few years comprises the RNA segment
before it causes a pandemic. encoding the H3 avian
hemagglutinin and the seven
Tracking influenza viruses in domesticated animals remaining segments from the
existing human virus.
Two instances of abortive antigenic shifts took place in the
eastern USA in 1976 and in Hong Kong in 1997. These viruses were noted
and notable because they caused death in a small number of young
people, and it was feared that they signalled the start of a new lethal
pandemic. Both had their origins in nonhuman animals. The 1976 virus
was also isolated from local domestic pigs, and the 1997 H5N1 virus from
chickens that had been brought as live birds to the local poultry market.
Of 18 people infected with H5N1 virus, six (33%) died, but these repres-
ent deaths among hospitalized, seriously ill cases, and it is not known
ITMC17 9/30/06 0:03 Page 284

284 PART III VIRUS INTERACTIONS WITH THE WHOLE ORGANISM

Percentage relatedness to the how many mild cases there were. However,
100
antigens of the original neither virus had the ability to spread from
person to person, and no more infections
virus isolate

were seen. Since that time, surveillance has

50 increased, particularly in southeast Asia, of
domestic poultry and infected people in order
to get an early warning of any impending
pandemic. The H5N1 virus was isolated only
Years from poultry in 2001 and 2002, but then
caused some cases of fatal disease in humans
Fig. 17.5 Diagram showing antigenic drift of type in 2003, 2004, and 2005. It has been isolated
A influenza virus in humans. This could represent from countries in southeast Asia including
either the HA or NA. Each point is a virus strain
Indonesia, Thailand, Vietnam, Japan, South
isolated in a different year.
Korea, Cambodia, Laos, and China. Human
infection rates are low but mortality has
(1933) 1957 1968 1977 2005
been put as high as 67%. Usually there is
Relatedness to the antigens
of the original shift virus

H2
H1 H5N1 infection of people’s own chickens,
N2 H3
N1 and presumptive bird-to-human transmis-
sion. There may have been very occasional
human-to-human transmission but, in the
N1 H1 absence of a pandemic, this is clearly (and
H1N1 H2N2 H3N2 H3N2 + H1N1 fortunately) still an inefficient process.
Outbreaks in poultry cause alarm and have
Fig. 17.6 Diagram showing the course of provoked culls of 20 million or so birds by
antigenic shift and drift of influenza A viruses in the authorities. Other economic costs are
man. The first virus, isolated in 1933, was H1N1. also huge as trading partners ban imports of
This arose by antigenic drift from the 1918 virus. poultry products. The known species range
Other shift viruses appeared in 1957 (H2N2) and of influenza A viruses was extended in 2003
1968 (H3N2). A 1950 H1N1 virus reappeared in when H5N1 virus spread with some fatalities
1977. Drift is shown schematically. The 1957 N2 to tigers and leopards in a Thai zoo, as a result
was acquired by the H3N2 shift virus, and has of animals being inadvertently fed infected
drifted from 1957 to the present day. chicken carcases, and domestic cats have
been infected experimentally. In the last
few years various subtypes have been isolated from poultry and humans,
including H9N2 virus which causes only mild respiratory disease in man
in China and Hong Kong, H7N7 virus in Holland and Belgium (with one
human fatality), H7N1 in Italy, H7N2 virus in the USA, and H7N3 in
Canada; viruses isolated from pigs in Europe and North America include
H1N1, H1N2, and H3N2 subtypes. In 2005, H3N8 virus was found for
the first time in dogs in the USA. This overall increase in virus isolation
may partly be the result of better surveillance rather than any rise in virus
dissemination. To avoid the possibility of serious infection, workers
handling infected birds wear safety suits and are offered prophylactic
anti-influenza drugs.
ITMC17 9/30/06 0:03 Page 285

CHAPTER 17 THE EVOLUTION OF VIRUSES 285

Antigenic drift

Influenza A viruses have been isolated every year from humans around
the world since their discovery in 1933. Each new isolate is tested sero-
logically with antibody to all other influenza strains. It soon became appar-
ent that the hemagglutinin and neuraminidase of new isolates were
antigenically slightly different from those of the previous year, and that
the difference increased incrementally year by year. This difference is
reflected in one or more amino acid substitutions in one or more of the
antigenic sites on the HA or NA. This phenomenon is explained by the
assumption that influenza virus mutants carrying modified antigenic
determinants have an evolutionary advantage over the parent virus in
the face of the existing immune response. However, although both the
HA and NA drift, the HA is the major neutralization antigen and there-
fore appears to be evolutionarily the more significant. Thus year-by-year
new amino acid substitutions – and new epitopes – appear. Epitopes
Type A
can also disappear if a mutation creates an attachment site for a carbo-
hydrate moiety that then sterically prevents antibodies from reacting
with those epitopes. The name “antigenic drift” is very apt (Fig. 17.5).
The nature of the selective force which drives this process is discussed
below. In practice a drift variant that can cause significant disease arises
about every 4 years, and this has on average four or more amino acid
substitutions in two or more antigenic sites. At the same time the “old”
strain is completely replaced by the “new” strain. Influenza B viruses also
undergo antigenic drift, but this is a slower process and also several dif-
ferent strains cocirculate. The reason for this difference is not known. Type B

Antigenic drift is at least as important in causing human influenza as

antigenic shift. In the UK there are 4000–14,000 deaths associated with Fig. 17.7 Model of
antigenic drift of
epidemics of influenza every year, and this extrapolates to 400,000 to
influenza type A
1,400,000 deaths per year worldwide. Thus, in the twentieth century, drift
and type B viruses.
viruses were responsible for approximately 40 to 140 million deaths world- Points on the same
wide. (In the UK an influenza epidemic is formally defined as 400 cases level represent drift
of influenza per 100,000 people.) variants that arise
in the same year.
The mechanism of antigenic drift The branch length
indicates the
As soon as a new shift virus appears and infects people, it begins to drift relative change
(Fig. 17.6). Drift of influenza A viruses is linear due to the dominating in antigenicity
effects of favorable mutations (Fig. 17.7). Influenza B viruses undergo less from virus in the
drift than type A viruses, and multiple virus lineages coexist. Drift hap- preceding year. Drift
is shown for an
pens on a global scale. It can only be theorized how drift takes place as
arbitrary 7-year
it is an assumption that drift variants arise from virus circulating in the period. See text for
previous year. It is generally believed that variants are selected by neu- further discussion.
tralizing antibody (Box 17.6).
ITMC17 9/30/06 0:03 Page 286

286 PART III VIRUS INTERACTIONS WITH THE WHOLE ORGANISM

Box 17.6

Testing antigenic drift in the laboratory

• Antigenic drift has been modeled in the laboratory using a neutralizing monoclonal anti-
body (Mab) specific for the HA. Virus and MAb are mixed together and inoculated into
an embryonated chicken’s egg. This is a surprisingly efficient process and, after just one
passage, the progeny virus is no longer neutralized by the selecting MAb. This “drift” virus
is also known as an escape mutant, and represents the growth to population dominance of
an antigenic variant virus that already existed in the inoculum. Sequencing shows there
is a single amino acid substitution in the expected antigenic site. If the drift virus is sub-
jected to a MAb to a different epitope, then the process repeats and the progeny virus
now has two amino acid changes compared with the original (Fig. 17.8). However if two
or more MAbs are mixed together to resemble an antiserum, no progeny virus – mutant
or wild-type – is produced at all. Thus drift can only take place if an antiserum effectively
contains antibody to only a single epitope.
• Influenza virus HA molecules have five antigenic sites (although sometimes one is hid-
den by a carbohydrate group). In H3 viruses these are labeled A–E. Each site comprises
around 10 epitopes, thus there are 50 epitopes in all, and in theory during infection the
immune system should make antibodies to all of these. However when tested by immu-
nizing rabbits with virus and measuring the amount of antibody made to individual epi-
topes, it was found that one epitope in site B was dominant, and there were only traces
of antibody to two other epitopes (Fig. 17.9). The amounts varied between animals. In
mice the results were similar but the dominant epitope was in site A. Thus for unknown
reasons the immune system responds selectively to foreign epitopes.
• While some antisera were completely neutralizing, others were shown to select escape
mutants like a MAb. Thus it would seem that drift operates because certain individuals
have an antibody response that is highly biased to one epitope. The derivation of the ≥4
amino acid substitutions in ≥2 sites, referred to above, could therefore be achieved as shown
in Fig. 17.10. The biased antibody response may be genetically controlled, and drift vari-
ants may arise in a subpopulation with the relevant antibody response. Further, since adults
suffer repeated influenza infections, they are likely to make a complex antibody response
that cannot give rise to drift variants. The simplest antibody response is likely to occur in
children after their first infection, and this may be responsible for the selection of drift
variants.

Unsolved problems relating to influenza virus

Despite considerable knowledge of influenza virus (the complete genome

sequence, the atomic structure of the HA protein, infectious nucleic acid)
there are major areas of ignorance. Most of these concern viral epidemiology
ITMC17 9/30/06 0:03 Page 287

CHAPTER 17 THE EVOLUTION OF VIRUSES 287

Selection no. 1

Incubate and
Virus "Drift"
inject
+ progeny virus
MAb1 resistant
to MAb1

+ MAb2
Embryonated egg

"Drift" Incubate and

progeny virus inject
resistant to both
MAb1 and MAb2

Embryonated egg

Selection no. 2

Fig. 17.8 “Antigenic drift” in the laboratory: a single neutralizing monoclonal antibody can select
a population of influenza virus escape mutants that is no longer neutralized by the selecting MAb.
Another round of selection with a second MAb produces virus that now carries two amino
acid substitutions, but selection with the two MAbs simultaneously is completely neutralizing
(not shown).

which is immensely difficult to study. Space precludes dis-

cussion but the following problems still require an answer
about an important though common virus:
Anti-HA titre

• Why is influenza a winter disease?

• How does the same virus appear simultaneously in dif-
ferent places around the globe at the start of an epi-
demic (i.e. it does not appear to follow any apparent
transmission chain)?
• Why is the seasonal restriction of influenza apparently
unaffected by the year-round introduction of virus by Antibody to site A Total antibody
air travellers incubating influenza? Does this virus not Antibody to site B Antibody to site D
infect people in summer or does it infect them with-
Fig. 17.9 An epitope-biased serum
out causing disease?
antibody response in a rabbit injected
• How does a clinically significant drift variant travel from
with influenza A virus. Nearly all the
northern to southern latitudes in 6 months (and did HA-specific antibody is accounted for
so prior to frequent air travel)? by the response to a single epitope in
• How does a shift virus (other than the reoccurring H1N1 antigenic site B. (From Lambkin, R.
of 1977) replace the existing virus? and Dimmock, N. J. 1995. Journal of
• How does a drift virus replace the existing virus? General Virology 76, 889–897.)
• The hypothesis of antigenic drift outlined above needs
verifying.
• Why did the 1918 shift virus have a higher mortality
than other shift viruses?
ITMC17 9/30/06 0:03 Page 288

288 PART III VIRUS INTERACTIONS WITH THE WHOLE ORGANISM

Person 1 Person 2 Person 3 Person 4

Infected with Infected with Infected with Infected with

current virus the site A epitope the site A+B the site A+B+C
drift virus epitope drift virus epitope drift virus

An epitope-biased An epitope-biased An epitope-biased An epitope-biased

anti-HA response anti-HA response anti-HA response anti-HA response
e.g. to an epitope e.g. to an epitope e.g. to an epitope e.g. to an epitope
in site A in site B in site C in site D

Site A epitope Site A+B Site A+B+C Site A+B+C+D

drift virus epitope epitope epitope
drift virus drift virus drift virus

The human population as a whole is no

longer immune to progeny virus from person 4

New epidemic

New vaccine needed

Fig. 17.10 How antigenic drift may occur in nature, bearing in mind that clinically significant drift
viruses (that cause epidemics) have four or more amino acid residues changed in two or more
antigenic sites present on the hemagglutinin protein. Note that people with biased antibody
responses may be uncommon and that there may be any number of nonselective infections of
other individuals occurring between the four persons shown here, as indicated by the broken
arrow of transmission. A similar process could cause drift of the NA protein.

KEY POINTS

• Viruses are evolutionarily dynamic and respond to changes in the macroenvironment and
microenvironment (the body and immune system).
• Every living organism, plant or animal, amoeba or elephant, has its own extensive range of
viruses, and new infections of man can arise by nonhuman viruses crossing species, e.g. HIV,
influenza A virus. Cross-over is accelerated by loss of habitat and increased proximity of man
and animals.
• All established human viruses undergo Darwinian evolution as a result of evolutionary pres-
sure acting on random variants arising by mutation, although the scale of change ranges from,
say, measles virus (low) to influenza A virus (high).
• Many new infections in a host species result in high morbidity and mortality, but evolution
selects hosts that are genetically more resistant to infection and less virulent progeny virus
(e.g. myxoma virus in the European rabbit).
• A new shift variant of influenza A virus could arise at any time and because of its efficient
transmission such virus is a major threat to human health. Better antiviral measures than are
currently available are needed to combat a virus with the potential of the 1918 virus.
ITMC17 9/30/06 0:03 Page 289

CHAPTER 17 THE EVOLUTION OF VIRUSES 289

QUESTIONS

• Describe the processes of antigenic drift and antigenic shift in influenza virus and indicate,
with reasons, which is likely to generate a more pathogenic virus.
• Using examples, discuss the proposition that viruses with RNA genomes evolve more rapidly
than those with DNA genomes.

FURTHER READING

Black, F. L., 1966. Measles endemicity in insular Lamb, R. A., Krug, R. M. 1996. Orthomyxoviridae:
populations: critical community size and its evolu- the viruses and their replication. In Fields Viro-
tionary implications. Journal of Theoretical Biology 11, logy (B. N. Fields, D. M. Knipe, and P. M. Howley,
207–211. eds), Vol. 1, 3rd edn, pp. 1353–1395. Lippincott-
Domingo, E., Holland, J. J., 1997. RNA virus muta- Raven, Philadelphia.
tions and fitness for survival. Annual Review of Lipatov, A. S., Govorkova, E. A., Webby, R. J. et al.
Microbiology 51, 151–178. 2004. Influenza: emergence and control. Journal
Domingo, E., Webster, R. G., Holland, J. J., 1999. of Virology 78, 8951–8959.
Origin and Evolution of Viruses. Academic Press, Nicholson, K. G., Wood, J. N., Zambon, M. 2003.
New York. Influenza. Lancet 362, 1733–1745.
Fenner, F., Ratcliffe, F. N., 1965. Myxomatosis. Cam- Oxford, J. S. 2000. Influenza A pandemics of the 20th
bridge University Press, London. century with special reference to 1918: virology,
Griffiths, P. D., 2004. Should vaccines and antiviral pathology and epidemiology. Reviews in Medical
therapy for influenza now be deployed strategically? Virology 10, 119–133.
Reviews in Medical Virology 14, 137–139. Reid, A. H., Taubenberger, J. K., Fanning, T. G.,
Guan, Y., Poon, L. L. M., Cheung, C. Y. et al. 2004. 2004. Evidence of an absence: the genetic origins
H5N1 influenza: a protean threat. Proceedings of the of the 1918 pandemic influenza virus. Nature
National Academy of Sciences of the United States of Reviews Microbiology 2, 909–914.
America 101, 8156–8161. Webby, R. J., Webster, R. G., 2003. Are we ready
Kobasa, D., Takada, A. E. A., Shinya, K. et al. 2004. for pandemic influenza? Science 302, 1519–1522.
Enhanced virulence of influenza A viruses with the Also check Appendix 7 for references specific to each
haemagglutinin of the 1918 pandemic virus. Nature family of viruses.
(London) 431, 703–707.
ITMC17 9/30/06 0:03 Page 290