Perspectives opinion

Redefining viruses: lessons from Mimivirus Didier Raoult and Patrick Forterre

Abstract | Viruses are the most abundant living entities and probably had a major role in the evolution of life, but are still defined using negative criteria. Here, we propose to divide biological entities into two groups of organisms: ribosome-encoding organisms, which include eukaryotic, archaeal and bacterial organisms, and capsidencoding organisms, which include viruses. Other replicons (for example, plasmids and viroids) can be termed ‘orphan replicons’. Based on this suggested classification system, we propose a new definition for a virus — a capsid-encoding organism that is composed of proteins and nucleic acids, self-assembles in a nucleocapsid and uses a ribosome-encoding organism for the completion of its life cycle. The Darwinian revolution created a new approach to classification by proposing a common origin for living organisms. Since then, scientists have grouped animals and plants phylogenetically, rather than by gross appearance. The genetic revolution and our ability to build trees based on genetic similarities provided support for this method of classification. Over the past 30 years, the development of more efficient sequencing strategies has led to the reclassification of organisms into a universal tree of life based on ribosomal RNA sequences1. Viruses, however, lack ribosomes and have not yet been incorporated into this universal tree of life. Until now, the genetic information that is encoded by viruses was not thought to contain sufficient information to allow their general phylogenetic classification, and consequently no clear definition of viruses is currently available. This is unfortunate, as viruses are the most abundant living entities on the planet2 and metagenomic studies from randomly sequenced environmental samples have revealed that viral genes constitute the largest part of the genosphere2,3. Recent research has revealed an important role for viruses in various evolutionary scenarios, including the origin of DNA and mammals4–7. Here, based on our knowledge of archaea, archaeal viruses8 and intracellular

bacteria9, and the recent discovery of the largest known virus, Mimivirus10–13 (FIG. 1), we propose a new definition for the virus life form. Of course, any attempt to redefine an entire field will be controversial; however, a debate of this issue, using all of the currently available data, is needed. We propose a definition of viruses (and cells) that is based on the hypothesis that viruses are more than just parasitic nucleic acids and that the presence of either capsids or ribosomes forms the basis of the principal classification system in the living world. Defining viruses — a history According to Karl Popper14, definitions are based on the data and tools that are available at a specific moment in time. In the nineteenth century, the word ‘microbes’ was coined by Sedillot15 to define cellular microorganisms that were only visible using a microscope. In the middle of the twentieth century, microorganisms were divided into two groups, eukaryotes and prokaryotes, based on cellular structural features16. Eukaryotic cells have a nucleus and a nuclear membrane, whereas prokaryotic cells do not (although Planctomycetes, such as Gemmata obscuriglobus, are bacteria that have a nucleus and a nuclear membrane)16. In the last part of the twentieth century, molecular-biology tools opened the way for a new classification system

nature reviews | microbiology

for all cellular organisms. Carl Woese17,18 discovered the existence of three different ribosomes in the living world, which replaced the old prokaryote–eukaryote dichotomy with a trinity — archaea, bacteria and eukarya. All cellular organisms could thus be placed together in a universal tree of life. Viruses, however, were missing from this picture. Unlike most other microorganisms, viruses are obligate intracellular parasites that cannot replicate independently. They can infect organisms from all three domains of life, and can even parasitize other viruses; for example, the delta agent (with the hepatitis B virus19) and satellite viruses (with an adenovirus or tobacco mosaic virus (TMV)20,21). Despite their ubiquity and enormous importance to human health, viruses have long been neglected by evolutionary biologists, and are thought to be derived from cells. Indeed, as a direct consequence of the cellular theory that was established in the nineteenth century, living organisms and cellular organisms are synonymous to most scientists. Viruses were initially thought to be infectious agents that are not visible under a microscope and can be filtered through 0.22 µm ultrafilters (hence the name ‘ultravirus’)22,23. During the twentieth century, researchers developed two theories about viruses. The bacteriologists Felix d’Herelle, who discovered bacteriophages, and Macfarlane Burnet, who received the Nobel Prize in medicine in 1960, believed that viruses were organisms24,25 (as did Louis Pasteur), whereas Wendell Stanley26, who crystallized TMV and received the Nobel Prize in chemistry in 1946, believed that viruses were biomolecules. Later, while promoting the eukaryote–prokaryote dichotomy, Andre Lwoff 22 defined viruses as small (one dimension smaller than 0.2 µm), infectious, but not autonomous, agents that cannot divide by binary fission, and consist of proteins and a single type of nucleic acid. Lwoff insisted that viruses are not organisms and maintained that the infectious element of the virus is the nucleic acid, unlike bacteria or other pathogens, in which the infectious agent is the organism itself (although this theory has been contradicted recently27). volume 6 | april 2008 | 315

© 2008 Nature Publishing Group

Perspectives

MV

N VF

2 µm

Figure 1 | Mimivirus infecting Acanthamoeba polyphaga. Transmission electron microscope image that shows A. polyphaga infected with Mimivirus. Note the giant virus factory. N, nucleus; VF, virus Nature Reviews | Microbiology factory; MV, Mimivirus virions.

Viruses were thus tacitly defined by most molecular biologists as molecular genetic parasites that use cellular systems for their own replication22. Indeed, with such a broad definition, many types of selfish genetic elements (such as plasmids, transposons, retroposons, viroids and virusoids) were determined to be viruses. For example, Koonin et al.28,29 recently grouped all infectious-material-containing nucleic acids as either selfish elements and/or viruses and used these terms synonymously. Each of the definitions for viruses has recently been challenged by the discovery of viruses that are larger than cellular organisms10,30. Indeed, both the particle and genome sizes of viruses now overlap significantly with those of bacteria, eukaryotes and archaea. Mimivirus, the largest known virus, is visible with an optical microscope, contains a 1.2 megabase chromosome that encodes nearly 1,000 putative genes and harbours both RNA and DNA11. Moreover, if exposed to a Gram-stain procedure, Mimivirus stains Gram-positive, and was thought for a period to be a ‘Legionella-like organism’. Interestingly, Mimivirus was identified as a virus only a few years ago, when its icosahedral capsid was observed using an electron microscope13. The size of Mimivirus challenges the definition of a virus and even the definition of a microorganism as a living entity.

Defining organisms and living entities The general consensus of what constitutes life can be sampled by consulting a global resource such as Wikipedia (the largest free online encyclopaedia; see Further information), which defines life as ‘‘a condition that distinguishes organisms from inorganic objects’’. However, there is no universal definition of life. The frequently used ‘reproduction’ criterion does not apply to sterile organisms. The distinction between parasites (replicators) and free-living organisms cannot be used to distinguish between organisms. There is now no clear-cut limit between mitochondria, small symbionts, intracellular bacteria (such as Rickettsia spp. and Candidatus Carsonella spp.) and free-living bacteria in size or phylogenetically31. Some recent definitions of life, such as another from Wikipedia — “life is a characteristic of self organizing, self recycling systems consisting of populations of replicators that are capable of mutation, around most of which homeostatic, metabolizing organisms evolve” — clearly include viruses. We can also paraphrase Engels32 and define life as “the mode of existence of living organisms”, which brings us to the problem of organism definition. The definition of an organism is a difficult problem in itself and is subject to controversy. An organism has been defined

316 | april 2008 | volume 6

as “An individual living system such as animal, plant, fungus or microorganism” by Wikipedia, “An individual animal, plant or single-celled life form” by the Oxford English Dictionary Online and “Any living structure capable of growth and reproduction” by Chambers Reference Online (see Further information). The definitions from Wikipedia and the Oxford English Dictionary Online exclude intracellular parasites, symbionts, organelles and viruses. The definition from Chambers Reference Online, however, includes viruses and nucleic acids, as it does not retain the word cell. Mimivirus changed the perception of viruses and could also change the definition of an organism. Mimivirus virions are assembled at the periphery of a large membrane-bound nucleus-like structure — the viral factory — within the host cell (FIG. 1). Although viral factories have been described for most eukaryotic viruses (both RNA and DNA)33, the viral factory of the Mimivirus is especially spectacular12 and, when first observed using an electron microscope, was initially thought to be the nucleus of its giant amoebae host. JeanMichel Claverie34 proposed that viruses are entities that are associated with an intracellular viral factory, and should not be confused with virions. Interestingly, from this view, a virus is similar to an intracellular organism, which therefore further blurs the boundary between cellular organisms and viruses. The virus definition can also be modified by the distinction between a virus and a virion. A virus can be generated from synthetic oligonucleotides by wholegenome assembly to produce infectious virions35. Therefore, we believe that a virus can be entirely defined by its coding capacity. As for bacteria, it was recently shown that genome transplantation from one species to another is possible, and that cells which were transplanted with the genome of Mycoplasma mycoides were phenotypically identical to M. mycoides27; here, the genome defined the whole bacterium. Experiments which showed that synthesized or purified nucleic acids from either viruses or bacteria can infect hosts and be replicated, show that there are no fundamental differences between these living entities. Based on these recent data, we believe that organisms and living entities can be defined by genome analysis. Finally, we retain the more liberal definition of organisms and living entities as it applies to viruses, and thus reclassify viruses according to their genome content. www.nature.com/reviews/micro

© 2008 Nature Publishing Group

Perspectives J K A L D

Mimivirus Nanoarchaeum equitans Candidatus Carsonella ruddii Encephalitozoon cuniculi

O M N

J K A L D O M N P T B U Z V C G E F H I Q R S

P

Functional gene category

Ribosome-encoding organisms To determine a natural classification for all living organisms, we suggest that it is necessary to redefine what is meant by a cellular organism. We can now compare the genetic information that is encoded by cellular organisms from the three domains of life as well as that of a virus of similar proportions30. As illustrated in FIG. 2, the only significant differences in the distribution of clusters of orthologous groups (COGs) of gene categories between the smallest cellular organisms — Candidatus Carsonella ruddii (a bacterium)36, Nanoarchaeum equitans (an archaeon), Encephalitozoon cuniculi (a eukaryote) and Mimivirus — are the number of genes that are involved in translation, which is much lower in Mimivirus owing to the lack of ribosomal proteins, and the lack of any COGs that are involved in energy production and conversion in the virus. Although the absence of these genes is a negative characteristic that cannot be used to group viruses together, it is also a positive feature that groups all cellular organisms. These genes might have been lost independently many times during a parasitic mode of life by convergent evolution. Such convergent evolution was noted for intracellular bacteria that had lost most of the genes that encode metabolic pathways31, and this is also the case for mitochondria, chloroplasts and symbionts. In particular, the genes that are involved in protein synthesis, specifically those that encode ribosomal proteins and ribosomal RNA, are among the few genes that are conserved in all cellular organisms, including the smallest intracellular parasites37. This is because the last universal common ancestor (LUCA) probably possessed a sophisticated ribosome that contained at least 34 ribosomal proteins that are shared by all archaeal, bacterial and eukaryotic organisms. The descendants of the LUCA (or some of its predecessors) have superseded all other cellular life forms that could have used other mechanisms to synthesize their proteins. Although some RNA viruses (for example, arenaviruses) do contain ribosomes within their capsids, these ribosomes are native to their hosts, and the absence of genes that encode ribosomal proteins is common to all viruses. Thus, we suggest that all cellular organisms can be adequately defined as ribosome-encoding organisms (REOs), as opposed to viruses. Interestingly, in contrast to the view that is advocated by Lwoff 22, mitochondria and chloroplasts would be classified as REOs based on this definition (instead of as cellular organelles) because they contain their own translation apparatus1. There is indeed no clear difference, either morphologically or

T B U Z V C G E F H I Q

Translation Transcription RNA processing and modification Replication, recombination and repair Cell-cycle control, mitosis and meiosis Post-translational modification, protein turnover and chaperones Cell-wall and membrane biogenesis Cell motility Inorganic ion transport and metabolism Signal-transduction mechanisms Chromatin structure and dynamics Intracellular trafficking and secretion Cytoskeleton Defence mechanisms Energy production and conversion Carbohydrate transport and metabolism Amino acid transport and metabolism Nucleotide transport and metabolism Coenzyme transport and metabolism Lipid transport and metabolism Secondary metabolite biosynthesis, transport and catabolism General function (prediction only) Function unknown

R S 0

20

40

60

80

100

120

140

160

Number of COGs

Figure 2 | Clusters of orthologous groups (COGs) in Mimivirus and traditional cellular organisms. Distribution by categories of cluster of orthologous groups’ homologues in Mimivirus, compared with cellular organisms from the three domains of life that have the smallest currently known Nature Reviews | Microbiology genomes: Nanoarchaeum equitans (archaea), Candidatus Carsonella ruddii (bacteria) and Encephalitozoon cuniculi (eukarya).

genetically, between mitochondria, symbionts and intracellular bacteria such as Candidatus Carsonella spp. and Rickettsia spp. Capsid-encoding organisms By analysing all infectious materials other than REOs, which range from a few hundred base pairs, such as the single-stranded RNA molecule that is carried by capsid borrowed from a helper virus (virusoid and satellite RNA)19, to the giant Mimivirus, it is clear that no single common protein exists in the virosphere. There is no genetic equivalent in this group to the ribosomal-RNA or universal proteins that are common to REOs. Furthermore, virus-specific proteins are only found in subsets of viral groups. Consequently, protein phylogenies have only been useful to tentatively establish a classification for selected virus groups. For example, according to Iyer et al.38, nucleo–cytoplasmic large DNA viruses (NCLDVs), such as

nature reviews | microbiology

Mimiviruses, could be classified based on a set of conserved proteins that are involved in viral DNA replication and transcription11,38. Although the viral-factory structure can be viewed as an ‘organismal’ form of the virus, it cannot be used to define viruses because, first, the presence of viral factories has not yet been demonstrated in archaeal and bacterial viruses (possibly for methodological reasons or because the whole cell is transformed into a viral factory) and, second, it is not known if all viral factories share a common feature, although they do disseminate their genetic information in the same way (through the linear or exponential multiplication of nucleic acids and massive production of virions). We propose that the expression of a capsid is the only positive determinant that can be considered to define viruses. The viral capsid is a necessary structure that is used by the viral factory to disseminate the virus outside of the REO host and infect new hosts. Indeed, volume 6 | april 2008 | 317

© 2008 Nature Publishing Group

Perspectives PBCV1 Vp54

PRD1 P3

I

DG

B

I C

HE F

C′ F′

B′ I′

G′ D′ H′E′

STIV

DG

B C

HE F B′

C terminus

C terminus N terminus

B

B′

F′ G′ C′E′ H′ I′ D′

G D I C B HE F N terminus

N terminus B

B′

G′ C′ H′E′ F′

B′ I′ D′

C terminus B

B′

Figure 3 | Capsid proteins from viruses that infect organisms fromNature all three domains of life. Reviews | Microbiology Comparison of the major capsid-protein structures from viruses that can infect the three domains of life: the major capsid protein of virus Paramecium bursaria Chlorella virus 1 (PBCV1) (Vp54; Protein Data Bank (PDB) code 1M4X), which infects Chlorella-like eukaryotes; the bacteriophage PRD1 coat protein P3 subunit (PDB code 1cjd), which infects the bacterium Escherichia coli; and the capsid protein of virus Sulfolobus turreted icosahedral virus (STIV) (PDB code 2BBD), which infects the hyperthermophilic archaeon Sulfolobus solfataricus P2. These three capsid proteins contain the typical double-jelly-roll fold (shown in red) that is absent from cellular proteins, which confirms that these structures originated from a common ancestral protein. Other features shared by these viruses suggest that this protein was already a capsid protein of an ancestral virus that was present at the time of the last universal common ancestor, or even earlier. C terminus, carboxyl terminus; N terminus, amino terminus.

the viral capsid has been called the ‘virus self ’ by Dennis Bamford and colleagues39, who first identified clear homologous traits between capsid proteins and the capsid architecture of viruses that were infecting bacteria (enterobacteria phage PRD1) and eukarya (an adenovirus)40. Later, it was shown that the double-jelly-roll fold (which has not been found in any cellular protein) is also present in the capsid proteins of Paramecium bursaria Chlorella virus 1 (PBCV1), an NCLDV that infects eukaryotic algae41, and Sulfolobus turreted icosahedral virus (STIV), an archaeal virus that was isolated from a Yellowstone hot spring42 (FIG. 3). Modelling experiments have shown that the capsid protein of Mimivirus contains the same fold, which suggests that it is present in all NCLDVs43. All these viruses are double-stranded DNA viruses and have an internal lipid layer (with the exception of adenoviruses). These observations favour the hypothesis that an ancient form of virus that had this type of capsid predates, or was a contemporary of, the LUCA43. Interestingly, the double-jelly-roll fold, which is common to double-stranded DNA viruses that have an internal lipid layer, is also present in the capsid proteins of some single-stranded RNA viruses, and single-jelly-roll folds are observed in the capsid proteins of many other DNA and RNA viruses. It will be important to determine if, as suggested by Rossmann and co-workers41, all these jelly-roll folds are evolutionarily related, which would suggest a common and ancient origin for some DNA and RNA viruses. Capsids probably

have multiple origins, as different unique folds are present in the capsid proteins of otherwise apparently unrelated viruses (and again are absent from any cellular proteins) which links head-and-tailed bacterial viruses (Caudovirales) and eukaryotic herpesviruses44. Future work should compare the coat proteins of viruses that have non-icosahedral morphologies (filamentous, rod-shaped or pleomorphic) to identify additional folds that are unique to viral capsids. Capsids might originally have been storage devices that were designed to protect nucleic acids that were accidentally released from lysed cells. These structures could have been selected within an intracellular parasite because of their ability to disseminate multiple replication copies by uncoupling replication of the parasite genome from that of the host genome. Capsids that possess the associated mechanisms that are used to exit from one host cell and enter a new host are specific and complex structures that could have appeared independently several times; however, we argue that this event would not have occurred frequently over the course of evolutionary time. In any case, the appearance of capsids was a crucial event in the early evolution of life that resulted in divergence among all subsequent organisms. We therefore propose to define viruses as capsid encoding organisms (CEOs). The presence of a capsid defines a group of living entities that contain nucleic acid and a capsid, and overlaps one of the trivial virus definitions.

318 | april 2008 | volume 6

In addition to capsids, analyses of the genomes of viruses and related elements, such as plasmids, have revealed the existence of replication proteins, such as the superfamily III helicases, protein-primed DNA polymerases and rolling-circle initiator proteins, that have no cellular homologues, but are present in viruses that infect organisms in different domains. This suggests that these proteins were never encoded by cellular genomes or that they originated in ancient cellular lineages that were wiped out by the descendants of the LUCA38. The diversity of viral RNA- and DNAreplication mechanisms and their associated proteins indicates that various types of replicons originated in an ancient virosphere and, possibly, even predate the LUCA4,29. Consequently, the origin of viruses could stem from an association between cassettes of capsid-encoding genes and particular replicons (including an origin of replication and genes that encoded replication-machinery proteins which could have used this origin for self-replication). Interestingly, most genes that are encoded by viruses which infect organisms in all three domains of life have no cellular homologues, which is in contrast to the traditional view that viruses are derived from genetic elements that escaped from cells and became infectious. The persistence of two different names for viruses — those that are associated

Caspid-encoding organisms Viruses of Bacteria

Viruses of Eukarya

Viruses of Archaea

Bacteria Archaea

Eukarya

Ribosome-encoding organisms

Figure 4 | Redefining viruses. Representation of viruses with their capsids, and the three Nature Reviews | Microbiology domains of life that have evolved from the last universal common ancestor. The three domains have ribosomes, but lack a capsid. The newly defined viruses have a capsid, but no ribosome. Other infectious elements are not shown.

www.nature.com/reviews/micro © 2008 Nature Publishing Group

Perspectives with bacteria (bacteriophages) and those that are associated with eukaryotic cells (viruses) — is, therefore, confusing. In our proposal, viruses that infect cellular organisms from all three domains of cellular life should be unified under a common name, CEOs, and classified according to their capsids, as previously suggested by Bamford39. For practical reasons, it will be also useful, in some cases, to discriminate viruses by the domain of their host: as bacterioviruses (those that infect bacteria; previously known as bacteriophages), archaeoviruses (those that infect archaea) or eukaryoviruses (those that infect eukaryotes). To summarize, we propose to redefine viruses as CEOs that are composed of proteins and nucleic acids that self-assemble in a nucleocapsid, do not multiply by binary fission and use an REO for the synthesis of their proteins and production of the energy and precursor molecules that are required for their life cycle (FIG. 4).

worlds have evolved in parallel. One form of life expresses ribosomes and comprises three domains: archaea, bacteria and eukarya. The other form of life expresses capsids that produce virions which infect REOs from each of these three domains. Didier Raoult is at the Unité des Rickettsies, IRD-CNRS UMR 6236, IFR‑48, Faculté de Médecine, 27 Bd Jean Moulin, 13385 Marseille, France. Patrick Forterre is at the Institut Pasteur, 25 rue du Docteur Roux, 75015 Paris1, France, and the University Paris Sud, Institut de Génétique et Microbiologie, CNRS, UMR 8626 IRF‑115, Centre d’Orsay, 91405 Orsay, France. Correspondence to D.R. e‑mail: [email protected] doi:10.1038/nrmicro1858 Published online 3 March 2008 1. 2. 3. 4.

Orphan replicons How can we classify infectious genetic elements that do not encode either capsids or ribosomes (including viroids, virusoids, RNA satellites, transposons19 and plasmids)? These replicons could have originated either from CEOs that had lost their capsid-encoding genes or from ancient RNA (or DNA) replicons that were unable to obtain capsid proteins from a CEO for their propagation. Interestingly, these replicons can parasitize both REOs and CEOs, and some use capsids from helper viruses. We suggest that these elements be grouped together under the term ‘orphan replicons’. The question then arises as to whether we should consider these elements as organisms. Because the term organism implies at least a minimal level of integration (the association of several organs into a functional unit), we propose to reserve the term organism for biological entities which encode both genes that are involved in their replication (a replicon cassette) and genes that encode either ribosomes or capsids. Conclusions Human beings like dichotomies. In biology, the animal–plant dichotomy was eventually replaced by the prokaryote–eukaryote dichotomy. Indeed, this attraction to dichotomies could partly explain why the prokaryote– eukaryote division persists, despite the vast amount of molecular evidence that indicates the existence of three domains of ribosomeencoding cells1. Here, we propose to reinstall a primary dichotomy in the classification of the living world between REOs and CEOs. We conclude that two connected natural

5. 6. 7.

8. 9. 10. 11. 12.

13.

14. 15. 16. 17. 18.

19.

20. 21. 22. 23.

Pace, N. R. Time for a change. Nature 441, 289 (2006). Bergh, D., Borsheim, K. Y., Bratbak, G. & Heldal, M. High abundance of viruses in aquatic environments. Nature 340, 467–468 (1989). Edwards, R. A. & Rohwer, F. Viral metagenomics. Nature Rev. Microbiol. 3, 504–510 (2005). Forterre, P. The origin of viruses and their possible roles in major evolutionary transitions. Virus Res. 117, 5–16 (2006). Villarreal, L. P. Viruses and the Evolution of Life 1–426 (ASM, Washington, 2005). Prudhomme, S., Bonnaud, B. & Mallet, F. Endogenous retroviruses and animal reproduction. Cytogenet. Genome Res. 110, 353–364 (2005). Forterre, P. Three RNA cells for ribosomal lineages and three DNA viruses to replicate their genomes: a hypothesis for the origin of cellular domain. Proc. Natl Acad. Sci USA 103, 3669–3674 (2006). Prangishvili, D., Forterre, P. & Garrett, R. A. Viruses of the Archaea: a unifying view. Nature Rev. Microbiol. 4, 837–848 (2006). Blanc, G. et al. Reductive genome evolution from the mother of Rickettsia. PLoS Genet. 3, e14 (2007). La Scola, B. et al. A giant virus in amoebae. Science 299, 2033 (2003). Raoult, D. et al. The 1.2-megabase genome sequence of Mimivirus. Science 306, 1344–1350 (2004). Suzan-Monti, M., Scola, B. L., Barrassi, L., Espinosa, L. & Raoult, D. Ultrastructural characterization of the giant volcano-like virus factory of Acanthamoeba polyphaga Mimivirus. PLoS ONE 2, e328 (2007). Raoult, D., La Scola, B. & Birtles, R. The discovery and characterization of Mimivirus, the largest known virus and putative pneumonia agent. Clin. Infect. Dis. 45, 95–102 (2007). Popper, K. The Logic of Scientific Discovery 1–544 (Taylor & Francis Group, London, 2002). Pasteur, L. La théorie des germes et ses applications en médecine et en chirurgie. C. R. Acad. Sci. 86, 1037–1043 (1878). Fuerst, J. A. Intracellular compartmentation in planctomycetes. Annu. Rev. Microbiol. 59, 299–328 (2005). Woese, C. R. There must be a prokaryote somewhere: microbiology’s search for itself. Microbiol. Rev. 58, 1–9 (1994). Woese, C. R., Kandler, O. & Wheelis, M. L. Towards a natural system of organisms: proposal for the domains Archaea, Bacteria, and Eukarya. Proc. Natl Acad. Sci. USA 87, 4576–4579 (1990). Fauquet, C. M., Mayo, M. A., Maniloff, J., Desselberger, U. & Ball, L. A. Virus Taxonomy: Classification and Nomenclature of Viruses. Eighth Report of the International Committee on Taxonomy of Viruses 1–1259 (Academic, San Diego, 2005). Dodds, J. A. Satellite tobacco mosaic virus. Curr. Top. Microbiol. Immunol. 239, 145–157 (1999). Atchison, R. W., Casto, B. C. & Hammon, W. M. Adenovirus-associated defective virus particles. Science 149, 754–756 (1965). Lwoff, A. The concept of virus. J. Gen. Microbiol. 17, 239–253 (1957). Raoult, D. The journey from Rickettsia to Mimivirus. ASM News 71, 278–284 (2005).

nature reviews | microbiology

24. Burnet, M. General discussion of virus nomenclature. Ann. NY Acad. Sci. 56, 627–630 (1953). 25. Burnet, M. Virus classification and nomenclature. Ann. NY Acad. Sci. 56, 383–390 (1953). 26. Stanley, W. M. Virus as a chemical agent. Rev. Med. (Mex.) 32, 209–212 (1952). 27. Lartigue, C. et al. Genome transplantation in bacteria: changing one species to another. Science 317, 632–638 (2007). 28. Koonin, E. V. & Martin, W. On the origin of genomes and cells within inorganic compartments. Trends Genet. 21, 647–654 (2005). 29. Koonin, E. V., Senkevich, T. G. & Dolja, V. V. The ancient virus world and evolution of cells. Biol. Direct 19, 1–29 (2006). 30. Koonin, E. V. Virology: Gulliver among the Lilliputians. Curr. Biol. 15, R167–R169 (2005). 31. Audic, S. et al. Genome analysis of Minibacterium massiliensis highlights the convergent evolution of water-living bacteria. PLoS Genet. 3, e138 (2007). 32. Engels, F. Dialectics of Nature 1–410 (Wellred Publications, London, 2006). 33. Novoa, R. R. et al. Virus factories: associations of cell organelles for viral replication and morphogenesis. Biol. Cell 97, 147–172 (2005). 34. Claverie, J. M. Virus takes center stage in cellular evolution. Genome Biol. 7, 1–10 (2006). 35. Smith, H. O., Hutchison, C. A., Pfannkoch, C. & Venter, J. C. Generating a synthetic genome by whole genome assembly: phiX174 bacteriophage from synthetic oligonucleotides. Proc. Natl Acad. Sci. USA 100, 15440–15445 (2003). 36. Nakabachi, A. et al. The 160-kilobase genome of the bacterial endosymbiont Carsonella. Science 314, 267 (2006). 37. Koonin, E. V. Comparative genomics, minimal genesets and the last universal common ancestor. Nature Rev. Microbiol. 1, 127–136 (2003). 38. Iyer, L. M., Aravind, L. & Koonin, E. V. Common origin of four diverse families of large eukaryotic DNA viruses. J. Virol. 75, 11720–11734 (2001). 39. Benson, S. D., Bamford, J. K., Bamford, D. H. & Burnett, R. M. Viral evolution revealed by bacteriophage PRD1 and human adenovirus coat protein structures. Cell 98, 825–833 (1999). 40. Bamford, D. H., Grimes, J. M. & Stuart, D. I. What does structure tell us about virus evolution? Curr. Opin. Struct. Biol. 15, 655–663 (2005). 41. Nandhagopal, N. et al. The structure and evolution of the major capsid protein of a large, lipidcontaining DNA virus. Proc. Natl Acad. Sci. USA 99, 14758–14763 (2002). 42. Rice, G. et al. The structure of a thermophilic archaeal virus shows a double-stranded DNA viral capsid type that spans all domains of life. Proc. Natl Acad. Sci. USA 101, 7716–7720 (2004). 43. Benson, S. D., Bamford, J. K., Bamford, D. H. & Burnett, R. M. Does common architecture reveal a viral lineage spanning all three domains of life? Mol. Cell 16, 673–685 (2004). 44. Baker, M. L., Jiang, W., Rixon, F. J. & Chiu, W. Common ancestry of herpesviruses and tailed DNA bacteriophages. J. Virol. 79, 14967–14970 (2005).

Acknowledgements

The authors thank A. Hecker and P.E. Fournier for help with the figures.

DATABASES Entrez Genome: http://www.ncbi.nlm.nih.gov/entrez/query. fcgi?db=genome PBCV1 | PRD1 | STIV | TMV Entrez Genome Project: http://www.ncbi.nlm.nih.gov/ entrez/query.fcgi?db=genomeprj Candidatus Carsonella ruddii | Encephalitozoon cuniculi | Escherichia coli | Gemmata obscuriglobus | Mycoplasma mycoides | Nanoarchaeum equitans | Sulfolobus solfataricus P2

FURTHER INFORMATION Didier Raoult’s homepage: http://ifr48.timone.univ-mrs.fr/ portail2/index.php?option=com_content&task=view&id=78 Chambers Reference Online: http://www.chambersharrap. co.uk/chambers/features/chref/chref.py/main NCBI COG s database: http://www.ncbi.nlm.nih.gov/COG / Oxford English Dictionary Online: http://www.oed.com/ Protein Data Bank: http://www.rcsb.org/pdb/home/home.do Wikipedia: http://www.wikipedia.org/ All links are active in the online pdf

volume 6 | april 2008 | 319 © 2008 Nature Publishing Group