Mimivirus: the emerging paradox of quasi-autonomous viruses.

Abstract

What is a virus? Are viruses alive? Should they be classified among microorganisms? One would expect these simple questions to have been settled a century after the discovery of the first viral disease. For years, modern virology successfully unravelled the huge diversity of viruses in terms of genetic material, replication mechanism, pathogenicity, host infection, and more recently particle structure, planet-wide distribution and ecological significance. Yet, little progress was made in understanding their evolutionary origin(s), as well as the fundamental nature of their relationship with the cellular world. Thanks to the recent studies on Mimivirus and other large DNA viruses, we are now entering a new era where the most basic concepts about viruses are revisited, including their true nature, how fundamentally different they are from cellular microorganisms, and how essential they might have been in the major innovations that punctuated the evolution of life.

Full text

Page 1

Mimivirus: the emerging paradox of
quasi-autonomous viruses
Jean-Michel Claverie and Chantal Abergel
Structural and Genomic Information Laboratory, CNRS-UPR 2589, Aix-Marseille University, Mediterranean Institute of
Microbiology, Parc Scientifique de Luminy, Case 934, 13288 Marseille Cedex 9, France
What is a virus? Are viruses alive? Should they be classi-
fied among microorganisms? One would expect these
simple questions to have been settled a century after the
discovery of the first viral disease. For years, modern
virology successfully unravelled the huge diversity of
viruses in terms of genetic material, replication mech-
anism, pathogenicity, host infection, and more recently
particle structure, planet-wide distribution and ecologi-
cal significance. Yet, little progress was made in under-
standing their evolutionary origin(s), as well as the
fundamentalnatureoftheirrelationshipwiththecellular
world. Thanks to the recent studies on Mimivirus and
other large DNA viruses, we are now entering a new era
where the most basic concepts about viruses are
revisited,includingtheirtruenature,howfundamentally
different they are from cellular microorganisms, and
how essential they might have been in the major inno-
vations that punctuated the evolution of life.
Mimivirus: the dogma buster
Initially misinterpreted as Bradfordcoccus, a Gram-
positive bacteria infecting Acanthamoeba, Mimivirus
waited a decade for its true viral nature to be recognized
[1]. The record-breaking size of the particle (750 nm in
diameter) was the main emphasis of this very first report,
as the historical characteristic feature used to distinguish
viruses from other ‘‘microbes’’ was their ability to go
through a filter with an average pore size of 500 nm in
diameter. By becoming the first ‘‘non-filtering’’ virus,
Mimivirus was breaking a century-old criterion. By com-
parison, the largest viral particles knownat that time were
those of Chlorella viruses or Poxviruses with dimensions of
200 – 350 nm. The anomalous size of Mimivirus particles
probably precluded the discovery of its relatives in marine
(or aquatic) environments, until recent metagenomic stu-
dies suggested that they might be quite abundant [2,3].
Indeed, virologists studying aquatic viruses had been
traditionally isolating them by filtration through 0.3 mm
pore-size filters, a protocol convenient for bacteriophages,
but leaving Mimivirus, its relatives and many other undis-
covered giant viruses, in the bacterial fraction. Breaking
with a tradition more than a century old, Mimivirus also
became the first virus with individual particles clearly
visible under an ordinary light microscope.
ThesubsequentdeterminationoftheMimivirusgenome
sequence resulted in the breaking of two additional dogma
strongly rooted in modern microbiology beliefs if less
historical than the size barrier. First, with a dsDNA gen-
ome size of 1.2 Mb [4] predicted to encode more than 900
genes (recently increased to 981 [5]), Mimivirus was the
first virus found to exhibit a larger genetic complexity than
many intracellular parasitic bacteria. Second, Mimivirus
was also the first virus to possess several of the most
central components of the protein translation apparatus,
including four amino-acyl tRNA synthetases that were
later proven to be functional [6].
Other studies on Mimivirus led to the discovery of more
unique, albeit less fundamental, features in its particle
structure [7,8], genome delivery and packaging [9], and
transcriptional regulation [5,10,11]. Having lost the
straightforward criteria of size, genome simplicity, and
translation competence as absolute boundaries separating
cellular organisms from giant viruses, we are now facing
the problem of redefining what we are prepared to call a
‘‘virus‘‘. This new definition, or set of discriminating prop-
erties, will have to be flexible enough to allow the classi-
fication of the most unexpected microorganisms that will
with no doubt turn up from the many ongoing and future
microbiome explorations.
Is there a natural limit to the size and complexity of
large dsDNA viruses?
With the discovery of Mimivirus, the historical (and oper-
ational)criterionoffilterability(particlediameter<500 nm)
has been broken, together with the notion that a virus
genome has to be simpler than that of a cellular organism.
A legitimate question is then: can we replace these
traditional limits by new ones derived from biochemical
and biological first principles? In other words, are there
natural limits to a virus particle and/or genome size?
Answering this question is philosophical, and central to
guide our future search for more atypical viruses.
The general organisation of the capsids (or capsid
heads) of dsDNA viruses (including phages infecting
Eubacteria and Archaebacteria) varies from highly sym-
metricalobjectstoavarietyofirregular morphotypes(such
as the round-shaped poxviruses [12], or the bottle-shaped
archaebacterial phages [13]). There is no apparent
relationship between the size of the virus particle and
its morphotype. Large capsids, for instance, might exhibit
anicosaedral shape (i.e. Mimivirus, 500 nm in diameter) or
an irregular one (i.e. Vaccinia virus, 350 nm in diameter).
In addition, they might also differ by the presence of an
outerlipidmembrane(envelopedviruses),aninternallipid
Opinion
Corresponding author: Claverie, J.-M. (Jean-Michel.Claverie@univmed.fr)
0168-9525/$ – see front matter ? 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.tig.2010.07.003 Trends in Genetics 26 (2010) 431–437
431

Page 2

membrane (e.g. phycodnaviruses), or a peptidoglycan-like
outer layer (Mimivirus). On the other side of the size
spectrum, the particle (50 nm in diameter) of a recently
described archaeal virus is a simple membrane vesicle
enclosing an ssDNA genome, without detectable nucleo-
or capsid-like proteins [14]. In our search for more atypical
viruses, we must thus be prepared to encounter virions
(nucleic acid-containing particles) hardly recognizable as
such. Moreover, following the finding that Mimivirus
particles are covered by a fibre layer of peptidoglycan
[4,8,15]andgiventhecapacityofvirusestoexchangeentire
metabolic pathways with their hosts [16], it is not biologi-
cally absurd to predict that virions might be found encased
in chitin (as found in the Chlorella cell-wall), calcite (as
found in coccolithes), or silicate (as already proposed for
virus-likeparticles associatedwithdiatoms(M. Heldal and
G. Bratbak, unpublished results).
Restricting the comparison to dsDNA viruses enclosed
in a proteinaceous capsid exhibiting an icosahedral sym-
metry (i.e. similar to a regular solid made of 20 triangular
faces), the known variation in size is truly amazing, ran-
ging from 45 nm in diameter for a typical polyomavirus
virus to 500 nm for the Mimivirus particle (not including
the outer peptidoglycan layer). This factor of 11 in capsid
diameter, converts to a factor of 1330 in terms of internal
volume.Thepolyomavirus(e.g.SV40)genomeisabout5 kb
in size and encodes seven proteins [17]. Using the DNA
packing density of the polyomaviridae as a reference, the
Mimivirus capsid could easily incorporate a 6.5 Mb DNA
molecule, encodingmore than6000 proteins!Thus, there is
no physical limitation for giant viruses with capsid dimen-
sions similar to that of Mimivirus to pack far more complex
genomes. Even more so, as the DNA packing density in
these viruses infecting eukaryotes is about ten times less
than that found in bacteriophages, where it reaches 0.56 b
per cubic nanometer [18]. At this density, the internal-
most spherical compartment (about 400 nm in diameter) of
the Mimivirus particle would allow 18 Mb of DNA to be
packed, i.e. a genome larger than that found in many free-
living unicellular eukaryotes (e.g. Micromonas, Ostreococ-
cus, Saccharomyces), and comfortably larger than the
reduced genomes of parasitic eukaryotes, such as 2.5 Mb
in Microsporidia [19].
In summary, in the absence of a clear physical prin-
ciple limiting the size of a viral genome that could be
packaged in a sufficiently large capsid, the only natural
limits one might think of are: (1) that a virus particle
should remain smaller than the cell it infects (i.e. in the
micrometre range); and (2) that a virus genome is
expected to be smaller than that of its host. However,
the example of the bacteriophage G (498 kb genome)
infecting Bacillus megaterium (?5.1 Mb genome) [20],
indicates that the difference between the virus and its
host might not be huge. We thus would not be utterly
surprised if a giant virus with many more genes than
Mimivirus and/or with a larger capsid was discovered in
the near future.
Virus= virion: a common misconception
Since their discovery as filtering microbes, viruses have
been defined by their virions, the viral particles produced
during infection [21]. The confusion between the virus
and the virion is constant in the media (where the frigh-
tening pictures of prickly particles always illustrate
articles on each new flu pandemic), but is also rampant
in the scientific literature, such as when debating the
position of viruses in the living world [22–24]. The crystal-
lization of the tobacco mosaic virus (i.e. of its inert
particles) contributed to the view that viruses should
not be considered alive, despite their capacity to evolve
and self-reproduce (two hallmarks of living entities) [21].
Indeed, viruses (as particles) do not exhibit any metabolic
activity, a situation a priori incompatible with life. How-
ever, the detailed study of Mimivirus’s replication cycle
helped changing this viewpoint. The delivery of the Mimi-
virus’s particle inner core within the host cytoplasm is
followed by an eclipse phase after which a spectacular
intracytoplasmic virion factory appears (Figure 1) [25–
28]. As was already documented for poxviruses [29], these
factories become the site of translation and transcription
as well as replication of the viral genome [28], using the
host’s pool of metabolic precursors. At this stage, the
functional resemblance between Mimivirus and an intra-
cellular parasitic bacterium is striking, and the numerous
cellular and biochemical functions involved in rebuilding
from scratch this transient microorganism upon each
infection cycle is what justifies the complexity of the
Mimivirus genome. We thus proposed that the intracellu-
lar factory corresponds to the real virus, whereas the
virion should be reappraised as the mere vehicle used
to spread its genome from cell to cell [23,28,30]. Within
this new conceptual framework, equating virus and virion
is like confusing a seed with a tree, or an egg cell with a
human being, even though both of them exhibit the same
genome (Figure 2). This new way of looking at large DNA
viruses infecting eukaryotes is making some progress in
the community, where some authors extended it to
viruses infecting Eubacteria or Archaebacteria with the
‘‘virocell’’ concept, whereby the infected host cell becomes
the true virus [24,31].
The notion that the Mimivirus virion factory should be
considered a transient microorganism received additional
support with the discovery that they could be ‘‘infected’’ by
theirownvirus,Sputnik.Atoddswithpreviouslydescribed
satellite viruses, Sputnik does not infect Acanthamoeba
host cells in the absence of its helper Mimivirus, and the
replication and transcription of its genome appear to be
performed and regulated exclusively by the Mimivirus-
encoded machinery [5,28]. Ending with the production of
icosahedral particles, the whole Sputnik replication cycle
takes place within Mimivirus intracytoplasmic factories
[32–35], interfering with the production of Mimivirus
particles. These properties led the authors to propose
the (still controversial) concept of virophage to distinguish
this bona fide ‘‘virus of virus’’ from satellite viruses co-
infecting the same cellular host with a helper virus. This
new kind of parasitism might be a common feature among
large DNA viruses infecting eukaryotes, as another vir-
ophage (Mavirus) was recently discovered associated with
CroV, a virus infecting Cafetaria roenbergensis, a single-
cell flagellate from marine environments (C. Suttle and M.
Fisher, unpublished results).
Opinion
Trends in Genetics Vol.26 No.10
432

Page 3

Giant viruses (Giruses) versus parasitic cellular
organisms: where is the frontier?
OnceweaccepttheequalityoftheMimivirusintracellular
cytoplasmic virion factory with a transient microorgan-
ism, the paradox of the exceptional size of the Mimivirus
genomevanishes,asitbearsnorelationshiptothesimpler
taskofbuildingacapsid,thatweknowcanbeencodedbya
handful of genes [17]. A more meaningful comparison can
now be made with the genome size and gene count of
obligate intracellular parasitic bacteria such as Chlamy-
Figure 1. Life cycle of Mimivirus in Acanthamoeba castellanii. (a) Schematic representation (center) and transmission electron microscopy images illustrating the key steps
of the infectious cycle. Mimivirus is represented as red hexagons, the virion factory as a blue oval, the cell nucleus as a grey circle, and the vacuole as a white circle. (b)
Detail of an Acanthamoeba cell exhibiting two early phases of Mimivirus infection. One particle (VP, upper right) is still intact within a phagocytic vacuole while a second
one (S, bottom left) is now in the cytoplasm, stripped of its outer layers to become the ‘‘seed’’ of the future virion factory.
Opinion
Trends in GeneticsVol.26 No.10
433

Page 4

dia, Rickettsia, or the lesser known Trophyrema, all of
them with genome size (?1 Mb) and gene count (<1000)
similar to those of Mimivirus [15]. As seen in all cellular
parasites, these bacteria exhibit a clear tendency toward
genomic reduction, with the loss of many biosynthetic
capabilitiesthattheycompensatebyobtainingtheneeded
metabolites (such as amino acids, nucleotides or ATP)
directly from the cytoplasm of the host cell. Interestingly,
some of these (normally) obligate parasitic bacteria
remain at the limit of the free-living life style, such as
Trophyrema whipplei, which is capable of growing in a
specially designed axenic medium [36,37]. More extreme
cases of genome reduction are exhibited by the vertically
transmitted sap-eating insect symbionts such as the
Gamma proteobacteria Buchnera (with genome sizes as
smallas0.4Mb)[38]andCandidatus(?0.16Mb,codingfor
less than 200 proteins)[39]. In the latter case, the genome
symbiont has lost all genes involved in cell envelope
biogenesis and metabolism of nucleotides and lipids
[39]. Is it still possible to draw a line between the virus
and the cellular world given these bacterial cells display-
ing amazingly defective genomes and Giruses exhibiting
an increasing diversity of biosynthetic pathways [4,16]?
Both Giruses and (until now) bacterial symbionts encode
theirownDNAreplicationandtranscriptionmachineries.
Giruses thus cannot be distinguished from minimal cells
on the basis of the corresponding genes. The next possible
frontier becomes the presence or the absence of a trans-
lationapparatus,thusdividingbiologicalentitiesbetween
ribosome-encoding
archaeal and bacterial organisms) and capsid-encoding
organisms [40,41]. However, this criterion might not be as
absolute as once thought [4,15]. On one hand, no basic
principle precludes virus genomes from being packaged in
a non-proteinaceous capsid, such as a simple lipid vesicle
[14]. On the other hand, the dogma that viruses entirely
delegate the translation of their proteins to their host is
alreadybroken.Wefoundninegenescentraltotheprotein
translation process in the Mimivirus genome: four ami-
noacyl-tRNAsynthetases,fourtranslationfactorsandone
peptide chainrelease factor. All thesegenes are expressed
during the infectious cycle [5], and the function of two of
themhasbeendirectlyvalidated[6].tRNAsarealsofound
in many large DNA viruses, such as in Mimivirus, where
their expression has been validated [11]. As for the genes
encoding the ribosomes themselves, it is true that they
remain, for now, the privilege of cellular organisms. Even
the most reduced bacterial symbionts appear to have
retained all the genes necessary to build their own ribo-
somes (ribosomal proteins, 16S, 23S and 5S ribosomal
RNAs,andtRNAgenesforall20aminoacids)[39].However,
this requirement might not be absolute in the world of
parasitic eukaryotes, within which mechanisms for the
import of ribosomal proteins seem to exist [42]. A complete
setofribosomalgenesmightthusnotberequiredinthemost
reduced parasitic or symbiotic cellular organisms. By con-
trast, nothing would preclude a giant virus from encoding
someofitsownribosomalcomponents;forinstance,tomore
selectively translate its own mRNAs. This could actually be
the role of the two tRNA-modifying enzymes and of the
mRNA CAP-binding protein encoded by Mimivirus. For
boot-strapping their infectious cycle independently of the
host cell nucleus (Figure 1b), Poxvirus (or Mimivirus)
particles incorporate multiple copies of ready-to-go tran-
scription complexes [5,43]. For the same purpose, a Mimi-
virus particle could easily fit a few ribosomes (as already
documented for Arenaviruses packaging host ribosomes)
that measure only 30 nm at their largest dimensions. Thus,
there is no basic principle that would preclude virion fac-
tories from using ribosomesencoded (at least partially) bya
Girus genome. The apparent incompatibility between the
presence of encoded ribosomes and the world of viruses is
thusawaiting anexplanation. Thispuzzling dichotomy,ifit
remainsabsoluteinthefuturedespitethediscoveryofmore
giant viruses, should be taken into account in any scenario
we might propose for their evolutionary origin.
In summary, the convergence between some cellular
organisms increasingly host-dependent and harbouring
less than a minimal genome [39] and quasi-autonomous
Girus encoding increasingly cell-like virion factories and
simply using the cytoplasm as a rich medium, suggests
that the sole analysis of a gene content might not be
sufficient, in the most extreme cases, to distinguish be-
tween them. In terms of genomic complexity, a continuum
appears to exist between the world of viruses and parasites
or symbionts derived from cellular organisms.
organisms(includingeukaryotic,
The evolutionary hypotheses on the origin of Giruses
Despite its many unique features [28], Mimivirus (and its
close relative [32,44]) appears to belong to a class of large
Figure 2. The true nature of large DNA viruses. The relationship between the virion
factory (or the virocell [24]) to the virus particle, is the same as that between
humans and their egg cells. Both propagate their cognate organism by packing
their genome into a specialized ‘‘container’’ that initiates a developmental program
leading to new organisms. The gene content is not correlated to the egg or virion
stage, but to the complexity of the resulting organisms. We see the intracellular
virion factories (the dark circle within the infected Amoeba, bottom left) as the true
viral ‘‘microorganism’’, which is the result of a developmental program initiated
upon infection by a specialized genome dissemination device, the virion.
Mimivirus was crucial in this change of perspective due both to the spectacular
dimension of its virion factory (up to 6 mm in diameter) and its exceptional gene
content (981 genes), an overkill for the simple task of encoding an icosahedral
capsid. Both the egg and the virion are generic devices used to disseminate the
genomes of vastly different organisms covering a large spectrum of phylogenetic
distances. Consequently, their overall structure and morphology do not provide
reliable clues to the evolutionary relationship of the cognate organisms.
Opinion
Trends in Genetics Vol.26 No.10
434

Page 5

dsDNA viruses referred to collectively as nucleo-cyto-
plasmic large DNA viruses (NCLDV) [45] and infecting a
variety of eukaryotic hosts. NCLDVs include Poxviruses,
IridovirusesandPhycodnavirusesasmajorrepresentatives.
A maximum-likelihood reconstruction of NCLDV evolution
yielded a set of 47 conserved genes [45]. The existence of
these ‘‘core’’ genes suggests strongly that they were present
in the genome of their last common ancestor, a virus encod-
ing a complex machinery of replication, transcription,
morphogenesis, and metabolic pathways making it rela-
tively independent from host cell functions [45]. However,
the existence of vastly different genome sizes (from 1.2 Mb
for Mimivirus down to 102 kb for the smallest Iridovirus)
amongNCLDVs,theerraticdistributionofthesecoregenes,
and the lackofa bona fide‘‘coregenome’’asinitially defined
[46](withonlyfivegenesconservedinallanalyzedgenomes)
suggestsascenariomorecomplexthanverticaldescentfrom
acommonancestorcoupledwithlineage-specificgenelosses
[45].Incontrast,vastlydifferentvirusesinfectinghostsfrom
differentdomains(Eubacteria,ArchaeaandEukarya)share
homologous capsid proteins and/or ATPases for genome
packaging, suggesting that they evolved from a common
ancestral virus, eventually predating the last universal
common ancestor [47]. In the wake of previous hypotheses
linking DNA viruses to the emergence of the eukaryotic
nucleus and the shift from RNA to DNA cellular genomes
[48–51], we propose an alternative hypothesis combining a
vertical descent scenario with the trapping of an ancestral
DNAvirusandthepossibilityforproto-nucleitoreverttoan
infectious viral-like state, as still observed today in some
parasitic red algae (Figure 3) [52]. This new scenario
accounts for the wide variation in genome size and gene
content among NCLDVs, together with the sharing of hom-
ologous capsid components by NCLDVs and many other
unrelated viruses. In this framework, the viral particle (i.e.
thecapsidproteinandthepackagingATPase),althoughthe
most characteristic feature of the virus world, becomes a
mere vehicle successfully selected to package parasitic gen-
omes with vastly different evolutionary histories, providing
a natural explanation for the lack of correlation between
particle structure and viral genome complexity (Figure 2).
Therefore, the highly complex Mimiviridae would have
originated late, near the completion of this iterative evol-
utionary process, packaging a large complement of ances-
tral genes (eventually predating the Cambrian explosion
leading to modern organisms [53]). Reductive evolution, a
well documented process common to all intracellular para-
sites [54,55], would then be responsible for the progressive
streamlining of the genome of the ancestral Mimiviridae to
its present state [4,28].
Some authors remain strongly opposed to the hypoth-
esis of any highly complex ancestor virus [24], and insist on
explaining the anomalous size of Mimivirus genome by an
exceptional propensity to acquire genes from its host and
other viruses by lateral transfer [22,56,57]. This conserva-
tive model suffers three main shortcomings: (1) the lack of
similarity of most Mimivirus genes (> 60%) with extant
cellular (or viral) genes [4,28]; (2) the small proportion
(?10%) of Mimivirus genes due to recent horizontal trans-
fer [58]; and (3) the violation of the rule that intracellular
parasites are invariably submitted to genome reduction as
observed in the many Acanthamoeba-infecting bacteria
that have been studied [59]. In this context, the oversized
capsid used by Mimivirus to pack its genome is consistent
with the notion that it was inherited from a Mimiviridae
ancestor possessing a much larger genome.
Concluding remarks
Since the initial discovery of Mimivirus, many phyloge-
neticanalyseshavebeenpublishedtosupportorinvalidate
Figure 3. An iterative model for the origin of the eukaryote nucleus and the simultaneous emergence of various NCLDV families. (a) A primitive DNA virus (a bacteriophage
ancestor) gets trapped within an RNA cell initiating a proto-nucleus. (b) Cellular genes are progressively recruited to the proto-nucleus, pushed by the selective advantages
of DNA genomes. (c) This situation remains unstable for a while with some of the proto-nuclei reverting to a viral state (generalizing the use of the original capsid structure
as a genome vehicle). (d) These viruses infected other cells at various stages of their rapid ongoing evolution. (e) The emergence of nucleated cells with biochemically more
stable DNA genomes might have coincided with the end of the pre-darwinian era [60] where a collective mode of evolution (characterized by unstable proto-cellular
organisms overwhelmed by too frequent horizontal gene transfers) switched to a cellular mode of evolution (with the dominance of gene inheritage by vertical descent,
both in the cellular and viral world). This hypothetical scheme provides a mechanism for the emergence of various overlapping virus lineages, all using the bacteriophage
inherited capsid structure, predating the emergence of the eukaryotes and exhibiting various reassortments of viral and ancestral cellular genes. (The figure is adapted from
Ref. [30].)
Opinion
Trends in GeneticsVol.26 No.10
435