key: cord-0004902-2e79aqlc
authors: Cavanagh, D.; Brien, D. A.; Brinton, M.; Enjuanes, L.; Holmes, K. V.; Horzinek, M. C.; Lai, M. M. C.; Laude, H.; Plagemann, P. G. W.; Siddell, S.; Spaan, W. J. M.; Taguchi, F.; Talbot, P. J.
title: Revision of the taxonomy of theCoronavirus, Torovirus andArterivirus genera
date: 1994
journal: Arch Virol
DOI: 10.1007/bf01309782
sha: 28e3ef3bd2c392372d4babeae8ce6561dd3fe99b
doc_id: 4902
cord_uid: 2e79aqlc

nan

Introduction coronavirus species had increased by six, two from each of swine and rat and one from each of turkey and cattle. In addition, it had been observed that they replicated in the cytoplasm and matured by budding through endoplasmic reticulum and not at the cell surface. Currently recognized members of the genus are shown in Table 1 .

Subsequently it was demonstrated that these viruses had a single, high molecular weight, single-stranded RNA genome which had a 3' poly(A) tail and was infectious. It was associated with a single protein, the nucleocapsid protein (N). Virions had a density in sucrose generally in the range 1.15-1.18 g/ml [27, 38] . Although coronavirus virions contain few polypeptides, the characterization of these took some time because of the presence of contaminating host cell proteins, variation between species with regard to processing (cleavage, glycosylation) and differences in gene complements. It was eventually established that all coronaviruses had, in addition to N, a large, heavily glycosylated surface projection or spike glycopolypeptide (S) of about 200 kDa, which was cleaved into two glycopolypeptides in some species, and a small integral protein (M), about 25 kDa, associated with only a few glycans (Fig. 1) . A subset of the genus (Table 1) has an additional glycoprotein, the haemagglutinin-esterase (HE) protein, which is not essential for replication, at least not in some cell types (Table 1) . A virion-associated, small membrane (sM) protein has also been described; todate it has only been looked for and demonstrated in IBV and TGEV [13, 20] (Fig. 1) . Replication is within the cytoplasm and five or more subgenomic mRNAs are generated, these forming a 3' coterminal nested set (Fig. 2) . Table 1 ). The sM protein has not been identified in all coronaviruses and its exact association with the virion envelope is unclear A leader sequence of 60-70 or so nucleotides, corresponding to sequence at the 5' terminus of the genome, is present at the 5' end of all mRNas. Only the unique genes in the 5' unique of each mRNA are translated. The genes for the structural proteins are located in the 3' portion of the genome, in the order 5'-S-M-N-3' [29, 30, 36] (Fig. 2) . The era of cloning and sequencing revealed that coronavirus genomes comprised about 30.000 nucleotides, the 5'-most gene, number 1, comprising approximately 20.000 nucleotides and encoding the putative RNA-dependent RNA polymerase [2, 3, 14, 19] .

The broad picture painted above applies in general to the Torovirus genus, of which Berne virus is the most thoroughly studied member at the molecular level [31] (Table 1 ). In addition to many of the characteristics already described above for coronaviruses, gene 1 of both genera comprise two overlapping open reading frames (ORFs), la and tb (Fig. 2) , ribosomal frame-shifting being involved in the translation of the second ORF [4, 32] . The carboxy-terminal half of the S protein has a coiled-coil structure and the M protein has three membrane-spanning regions [31, 35] . In consequence, the Coronaviridae Study Group proposed to the ICTV Executive Committee that the genus Torovirus should join Coronavirus in the Coronaviridae. This proposal was accepted at the mid-term meeting of the Executive Committee in April 1992 [6] and was ratified at the Table  b HE pseudogene known for BEV No such protein described dsM currently identified only for IBV and TGEV

Congress of Virology, 1993. The nomenclature for genes, mRNAs and proteins of the coronaviruses [7] has been adopted for the toroviruses [31] . Major characteristics of these viruses are summarized in Table 2 and Figs. 1 and 2.

Common amino acid motifs in the replicase genes of coronaviruses and toroviruses Significant amino acid similarities between the coronavirus and torovirus replicase proteins strengthen the inclusion of Torovirus within the Coronaviridae. As with other positivestranded RNA viruses there are conserved domains for putative RNA-dependent RNA polymerase and NTP-dependet helicase activities. These are within the second, lb ORF, of the putative replicase gene. The corresponding coronavirus domains resemble those of torovirus more than that of the other positive-stranded viruses [31] , with the exception of the arteriviruses, of which more below. Most of the torovirus ORF la has yet to be sequenced. An important aspect of the evolution of coronaviruses is recombination. Homologous recombination has been demonstrated experimentally with routine hepatitis coronavirus (MHV) [181] . Perhaps the most surprising aspect of recombination has been the discovery that some coronaviruses (Table 1) possess an additional gene, encoding an haemagglutininesterase protein (HE) which has some 30% amino acid identity with the amino-terminal subunit of the haemagglutinin-esterase-fusion (HEF) glycoprotein of influenza C virus (a negative-strand RNA virus), implying that non-homologous recombination has occurred during the evolution of coronaviruses [21] . Berne torovirus also has an ORF, number 4, situated between the N and M genes, which encodes a polypeptide that has about 30% identity with both the coronavirus ItE protein and the influenza C virus HEF protein [33] . Notwithstanding the fact that the torovirus ORF 4 only encodes 142 amino acids in contrast to the > 400 amino acids of the HE and HEF protein, this further demonstrates the evolutionary links between toroviruses and coronaviruses. It should not be inferred that the HE gene-containing coronaviruses are more closely related to toroviruses than are those that lack HE, or that there is a closer relationship with influenza C virus. However, in the context of virus taxonomy, the possession by both coronaviruses and toroviruses of a gene, HE, with the coding potential for a protein sharing significant amino acid identity is an additional reason for including these viruses within one family.

Another non-homologous recombination event is believed to account for the finding that the protein by the carboxy-terminus of ORF 1 a of Berne virus has some 30% identity with the 30 kDa protein encoded by the first ORF of gene 2 of murine hepatitis virus.

The evolutionary implications of these proposed non-homologous recombination events are discussed in more detail by Snijder and Horzinek [31] . Suffice it to say that these amino acid sequence similarities in non-structural genes support the inclusion of Coronavirus and Torovirus in the same family.

Although the S and M polypeptides of the two genera are of similar size and have some overall structural features in common, they have virtually no amino acid sequence identity. The coronavirus N polypeptide is two to three-fold bigger than its torovirus counterpart ( Table 2 ) and, with the genomic RNA, forms a virion which, in ultrathin sections, exhibits disc-, kidney or rod shapes which are not seen in coronaviruses [39] (Fig. 1) . While generation of several 3' co-terminal subgenomic mRNAs is a feature of both genera, as is the presence of conserved sequences upstream of each gene, the presence of leader sequences (60-70 nucleotides long) on the 5' termini of the mRNAs has been found for coronaviruses but not toroviruses. A number of ORFs which are present in coronavirus genomes are absent from toroviruses. One such ORF, 3c and 4 in the case of infectious bronchitits virus and transmissible gastroenteritis virus, respectively, has been shown for these two species to encode a small membrane protein (sM) which is associated with the virion envelope although detail of its disposition is not known [13, 20] (Fig. 1) . These observations, plus the possession by coronaviruses of ORFs not present in Berne virus, requires that these two virus groups should occupy different genera.

The Arterivirus genus contains several known species (Table 1) . It might seem unlikely that the genus Arterivirus, for many years assigned to the family Togaviridae, on account of its morphology (Fig. 1) and genome size (13-15 kb), would be removed from this family and be considered more closely related to the Coronaviridae. This, however, is the case, the ICTV Executive Committee accepting the recommendation that the arteriviruses should be removed from the Togaviridae in April 1992 [6] . The basis for this move was that key features of the arterivirus genome organization, transcription and translation were atypical of the togaviruses but bore a strong resemblance to those of the Coronaviridae (Fig. 2 and Table 2 ). The Y-most gene, number 1, comprises two large ORFs, la and lb, translation of lb involving ribosomal frame-shifting [10] . Within ORF la are two domains, one belonging to the protease superfamity which comprises the chymotrypsin-like and picornavirus 3C-like proteolytic enzymes, the other domain belonging to the papain-like superfamily, wich are also present in the coronavirus ORF 1 a (this region of the torovirus genome has not been sequenced) [19, 34] . Moreover, the polymerase and helicase domains identified in ORF lb of coronaviruses and toroviruses are not only present in ORF lb of arteriviruses, but also the arterivirus domains have greater amino acid similarity to those of the Coronaviridae than to other positive-strand RNA viruses [6, 12] . The genes encoding the surface glycoproteins, the M and N proteins occur in the genome in that order, as with the corona-and toroviruses. Six subgenomic RNAs are produced in a 3' co-terminal, nested-set arrangment and each message has a leader RNA derived from the 5' terminus of the genome [8, 16] (Fig. 2) . Although smaller than its counterpart in the coronaviruses and toroviruses, the arterivirus M protein is assumed to have a triple membrane-spanning domain ( Table 2) .

These similarities have led to the arteriviruses being considered as members of a coronavirus 'superfamily', along with coronaviruses and toroviruses, although the term 'superfamily' has no formal place in viral taxonomy [10] . One way of reflecting these similarities would have been to simply place the genus Arterivirus within the Coronaviridae. However, this suggestion has not won overall approval because it over-emphasizes the points of genomic and replication similarity at the expense of reflecting major structural differences between arteriviruses on the one hand and the other two genera on the other. Arterivirus virions are only half the diameter of the other two genera and have an isometric nucleocapsid (Fig. lc) . Unlike the surface projections of the corona-and toroviruses, the arterivirus surface glycoproteins, of which there are at least two (GL and Gs), are not prominent and do not have a coiled-coil structure ( Table 2 ). The genome reflects the small size of the virions, being only about 13-15 kb, and the nucleocapsid is formed with a very small N protein (Table 1) .

Two alternative taxonomic proposals have been debated, neither of which has found general agreement. One involves the creation of a new family, provisionally called Arteriviridae for convenience, the relationship between members of this family and the Coronaviridae being recognized by placing them together in a higher order taxon, an order. The other, not necessarily less contentious, suggestion is that the three genera should be placed in one family divided into two subfamilies.

According to this scheme the arteriviruses would be in a subfamily, provisionally called Arterivirinae, the other two genera being within another subfamily e.g. Coronavirinae, within the Coronaviridae. The subfamily taxon has been utilised in a number of virus families. For example, all members of the Herpesviridae have the same overall morphology and virion structure, possess a large double-stranded DNA genome and have similar replicative features. The family is divided into three subfamilies, the divisions being largely based on biological criteria e.g. host cell range, length of replication cycle [28] . The results of comparison of gene complements, genome organization and amino acid similarity of proteins supports the division of this family into three subfamilies, with one exception, namely Marek's disease virus. This is currently within the Gammaherpesvirinae, based on its tropism for tymphocytes, but it more closely resembles members of the Alphaherpesvirinae on molecular criteria. The Baculoviridae, which comprises two subfamilies, contains viruses of invertebrates which are all rod-shaped ('baculo' from the Latin baculum = stick) but otherwise show substantial variation in size, one species also having a long tail-like projection [40] .

Members of the Eubaculovirinae subfamily form within occlusion bodies while those in the second subfamily, Nudibaculovirinae, do not. There is substantial variation in genome size among the family members. Further delineation of the members into groups is anticipated as data increases [40] .

A major characteristic of members of the Poxviridae is their morphology (brick-shaped or ovoid virions) and large size. The members of the Chordopoxvirinae subfamily, which contains eight genera, infect vertebrates and a second subfamily, Entomopoxvirinae, comprises viruses of insects [11] .

The use of subfamilies has recently been introduced for the Paramyxoviridae in order to reflect that the genera Paramyxovirus and Morbillivirus have more in common with each other than with the third genus of the family, Pneumovirus [24] . The Pneumovirinae have more but smaller genes and little amino acid identity with the other two Paramyxoviridae genera but all three genera have very similar virion and nucleocapsid morphology, genome size and broadly similar genome organization.

The proposal to create two subfamilies within the Coronaviridae has the merit of drawing attention to both the similarities and differences between the three genera but with the emphasis being on the similarities. Implicit in this proposal is that the common characteristics should be weighted more than the differences and that the similarities can be accommodated taxonomically without the introduction of a taxon higher than family. One consequence of this scheme is that the taxon order would remain available for future use in a context that might include other virus families. The name Coronaviridae should not, of itself, be an impediment to this proposal. It can be argued that the name Coronaviridae has been in use for 25 years, long enough for it to evoke in the minds of virologists those many characteristics, described above, which define its two existing genera -and of which the existence of a 'corona' (and other morphological features?) is perhaps now the least important. However, the possibility of a new name for a family embracing all three genera, within two subfamilies, could be given further consideration.

Morphology was a major criterion by which the above families were first defined. Clearly, if morphology and fundamental aspects of architecture such as nucleocapsid structure are to continue to be of pre-eminent importance in the defining of virus families, then Arterivirus cannot be in the Coronaviridae but should be assigned to a new family. However, it remains highly desirable to reflect replicase homologies and the similarities in genome organization and replication strategy by a recognized taxon. The next higher taxon is that of order. Recently the ICTV Executive Committee set a precedent by recognizing the first virus order, Mononegavirales. This comprises the families Filoviridae, Parau~vxoviridae and Rhabdoviridae [25] . Members of these three families are all enveloped but have readily distinguishable virion morphology and substantial differences in genome size, the genomes being approximately 11 to 16 kb. Common features include a non-segmented, negativesense genomic RNA, a helical nucleocapsid, the initiation of primary transcription by a virion-associated RNA-dependent RNA polymerase, similar gene order and a single 3'

promotor. Arguably, an order containing the Coronaviridae and a yet-to-be-created family for the arteriviruses would be based on similar principles, namely major differences in morphology and genome size but common genome type and organization, conserved replicase domains, generation of several mRNAs in a nested-set configuration, ribosomal frame-shifting within gene 1, replication within the cytoplasm and virion formation by budding through intracellular membranes.

At its 1992 mid-term meeting the ICTV Executive Committee recognized a second order, Caudovirales, comprising the three bacterial virus families Myoviridae, Siphoviridae and Podoviridae [26] . The common feature which has led to their inclusion in an order is the possession by these bacteriophage of a tail and the use of this structure for attachment to host cells, and as a conduit through which the viral DNA passes into host cells. Members of the three families differ with respect to the nature of the tail (contractile, non-contractile/ long or non-contractile/short) and in other respects e.g. 4-to 5-fold range in genome size, number of proteins. Thus the order Caudovirales has been defined on criteria ( a particular morphological, structural and associated functional feature) quite different from those applied to the Mononegavirates (where nature of the genome, genome organization and overall replicate strategy are major uniting factors). It would appear, therefore, that there are no hard-and-fast rules governing the inclusion of families into an order. In this context, creation of an order for the arteriviruses, coronaviruses and toroviruses would seem to be a logical way of highlighting the similarities and differences between these viruses.

Whether this solution is preferable to the 'one family, two subfamily' scheme remains a matter of debate. Further discussion is required within ICTV as to whether, as in the past, gross morphological differences require that viruses be placed in different families or whether genome organization, replication strategy and sequence similarities, unavailable until relatively recently, should be considered as uniting factors which outweigh morphological criteria for the definition of a family. The ICTV Executive Committee has initiated the formation of a provisional Arterivirus Study Group (Chair: Dr. M. Brinton) to further evaluate the taxonomic status of the arteriviruses concurrent with the continuing debate within the Coronaviridae Stud)' Group.

Completion of the sequence of the genome of the coronavirus avian infectious bronchitis virus

The primary structure and expression of the second open reading frame of the polymerase gene of the coronavirus MHV-A59

An efficient ribosomal frame-shifting signal in the polymerase-encoding region of the coronavirus IBV

Coronaviruses and their diseases

Genus Torovirus assigned to the Coronaviridae

Recommendations of the coronavirus study group for the nomenclature of the structural proteins, mRNAs and genes of coronaviruses

All subgenomic mRNAs of equine arteritis virus contain a common leader sequence

Rottier PJM (i992) Structural proteins of equine arteritis virus

Equine arteritis virus is not a togavirus but belongs to the coronavims-like superfamily

Classification and nomenclature of viruses. Fifth Report of the International Committee on Taxonomy of Viruses

Complete genomic sequence and phylogenetic analysis of the lactate dehydrogenase-elevating virus

TGEV coronavirus ORF4 encodes a membrane protein that is incorporated into virions

Nucleotide sequence of the human coronavirus 229E RNA polymerase locus

Toroviruses of animals and humans: a review

Lactate dehydrogenase-elevating virus (LDV): subgenomic mRNAs, mRNA leader and comparison of 3'-terminal sequences of two LDV isolates

Coronavirus -organization, replication and expression of genome

RNA recombination in animal and plant viruses

The complete sequence (22 kilobase) of murine coronavirus gene-1 encoding the putative proteases and RNA polymerase

Association of the infectious bronchitis virus 3c protein with the virion envelope

Sequence of mouse hepatitis virus A 59 mRNA2: indications for RNA recombination between coronavirus and influenza C virus

Lelystad virus, the causative agent of porcine epidemic abortion and respiratory syndrome (PEARS), is related to LDV aud EAV

Lactate dehydrogenase-elevating virus, equine arteritis virus, and simian hemorrhagic fever virus: a new group of positive-strand RNA viruses

Francki RIB, Fauq~et CM, Knuds~n DL, Br~wn F (eds) C~assi~-cation and nomenclature of viruses. Fifth Report of the International Committee on Taxonomy of Viruses

Classification and nomenclature of viruses. Fifth Report of the International Committee on Taxonomy of Viruses

The ICTVdb project and other initiatives

~ 979) C~r~aviridae

The family Herpesviridae: an update

The structure and replication of coronaviruses

The biology of coronaviruses

Toroviruses: replication, evolution and comparison with other members of the coronavirus-like superfamily

The carboxyterminal part of the putative Berne virus polymerase is expressed by ribosomal frame-shifting and contains sequence motifs which indicate that toro-and coronaviruses are evolutionary related

Comparison of the genome organization of toro-and coronaviruses: both divergence from a common ancestor and RNA recombination have played a role in Berne virus evolution

The 5' end of the equine arteritis virus genome encodes a papainlike cysteine protease

Coronaviruses: structure and genome expression

The molecular biology of coronavirnses

The proposed family Toroviridae: agents of enteric infections

Classification and nomenclature of viruses. Fifth Report of the International Committee on Taxonomy of Viruses

Authors' addreses: The Coronaviridae Study Group of the International Committee on Taxonomy of Viruses

ARFC Institute for Animal Health, Division of Molecular Biology

Uniformed Services University of the Health Sciences

Laboratoire de Virologie et Immunologie Moleculaires, Centre de Recherches de Jouey-en-Josas, 78352 Jouey-en-Josas

Afdeling Virologie

We are very grateful to C. Denny and D. Hawkins for producing the diagrams of coronavirus, torovirus and arterivirus virions and to Dr. E. Snijder for Fig. 2 and for critical reading of the manuscript.