key: cord-0006262-u5npu6ng
authors: Attoui, H.; Mohd Jaafar, F.; Biagini, P.; Cantaloube, J.-F.; de Micco, P.; Murphy, F. A.; de Lamballerie, X.
title: Genus Coltivirus (family Reoviridae): genomic and morphologic characterization of Old World and New World viruses
date: 2002
journal: Arch Virol
DOI: 10.1007/s007050200005
sha: 4b0028de12501d3ea790afecc6f5cf85277d829c
doc_id: 6262
cord_uid: u5npu6ng

We report a genomic and morphologic study of the European Eyach (EYA) virus (genus Coltivirus, family Reoviridae) and a comparative analysis with the American Colorado tick fever (CTF) virus (the type species of the genus). The previously established, but distant, antigenic relationship between these viruses was strengthened by genetic findings (presence of cognate genes, amino acid identity between 55 and 88%, similar conserved terminal motifs, suspected read-through phenomenon in segment 9 of both viruses) and by indistinguishable ultramicroscopic morphologies. Moreover, putative constitutive modifying enzyme activities were suspected to be carried out by homologous viral proteins (RNA-dependent RNA polymerase, methyl/guanylyl transferase, NTPase). These findings, together with the comparative analysis to genomes of south-east Asian isolates, support the recent classification of arboviruses with 12 segments of dsRNA within two distinct genera (genus Coltivirus and genus Seadornavirus) and raise interesting questions about the evolutionary origins of coltiviruses. The previously proposed hypothesis that EYA virus was derived from an ancestral virus introduced in Europe with the migration of lagomorphs from North-America, would imply a divergence date between American and European isolates of over 50 million years ago (MYA). This analysis allows for the first time to propose an evolutionary rate for virus dsRNA genomes which was found to be in the order of 10(−8) to 10(−9) mutations/nt/year, a rate similar to that of dsDNA genomes.

For many years after the discovery of yellow fever virus [36] and the confirmation of Carlos Finlay's earlier finding that the virus is transmitted by mosquitoes, arthropod-borne viruses (arboviruses) were considered to be a singular unique group. It was even supposed that the arboviruses formed a natural taxonomic group [1] . However, in the 1960s and 1970s it became evident that the various arboviruses fall into several taxonomic groups [45] . Among the last of the arboviruses to be characterized and classified were those that are now members of the family Reoviridae, that is the member viruses of the genera Orbivirus, Coltivirus and Seadornavirus. This was despite the very early discovery of some member viruses: e.g., the first African horse sickness virus was discovered by M'Fadyean in 1900 [27] and Colorado tick fever virus was discovered by Florio in 1946 [16] . It was not until 1968-1971 that the first biophysical, biochemical and morphological characterization of these viruses permitted their taxonomic placement [9, 33] .

Members of the family Reoviridae have dsRNA genomes divided into 10 to 12 segments [28] . The member viruses of the genus Orbivirus (type species Bluetongue virus) have 10 segments, the members of the genus Coltivirus (type species Colorado tick fever virus) have 12 segments, and the members of the recently identified genus Seadornavirus (type species Banna virus; previously coltivirus subgroup B) have 12 segments [2, 28] . Among the member viruses of these taxa, the coltiviruses stand out because of their role as human pathogens. The New World virus, Colorado tick fever virus, the etiologic agent of Colorado tick fever, is widespread in the Rocky Mountain region of North America. Its distribution matches that of its vector ticks: Dermacentor andersoni [15] , other Dermacentor species, Ixodes species, Haemaphylis leporispalustris and Otobius lagophilus [13] . The antigenically related Old World virus, Eyach virus, was isolated in Europe in 1976 [12, 37] and indirectly incriminated in human neurological disease shortly thereafter [26] . Its distribution matches that of its vector ticks: Ixodes ricinus and Ixodes ventalloi. Other members of the genus are the California hare coltivirus S6-14-03 from California and the Salmon River virus from Idaho.

The first basis for distinguishing the coltiviruses from other viruses and relating them to each other was serologic cross-reactivity. The viruses exhibit no cross-reactivity with other viruses, but Colorado tick fever virus and Eyach virus cross-react in complement-fixation tests. The next basis for distinguishing the coltiviruses was virion morphology and morphogenesis. At the time of the first electron microscopic study of Colorado tick fever virus in 1968, only a few other arbovirus had been observed to have similar morphology, namely African horse sickness virus and bluetongue virus, two of the founding members of the genus Orbivirus which was formalized in 1971 [9, 32, 35, 43] . After Eyach virus was isolated in 1976, its morphology was studied cursorily, but until now little has been formally reported.

Since 1997, the use of serology and electron microscopy for taxonomic purposes has declined as knowledge of the molecular biology and genetics of the viruses has advanced; at first this advance followed upon analysis of the nucleotide sequence of the smallest viral genome segments, then upon that of the entire genome [3] . We present here the complete genomic sequence of the prototype strain of Eyach virus and a comparative analysis of the genomic sequences of Colorado tick fever virus, additional strains of Eyach virus, and California hare coltivirus S6-14-03. We also present here a morphologic and morphogenetic comparison of Colorado tick fever virus and Eyach virus as a way of extending molecular comparisons to at least on set of phenotypic characters. These analyses have allowed us to consider the close phylogenetic relationships among the member viruses of the genus Coltivirus and the more distant relationships among the various genera of the family Reoviridae.

Eyach virus (EYA virus) strain Fr577 and EYA virus strain Fr578 were kindly provided by C. Chastel, as lyophilized infected mouse brain. EYA virus strain Gr and California hare coltivirus S6-14-03 (CTFV-S6-14-03) were kindly provided by N. Karabatsos as frozen mouse brain. The Colorado tick fever virus (CTF virus) strain Florio was purchased at the American Type Culture Collection.

CTF virus strain Florio, EYA virus strain Fr577 and EYA virus strain Fr578 were propagated by intracerebral inoculation of suckling mice. The mice were killed on day 7 post-infection and brains were homogenized in PBS by vortexing with glass beads or prepared for electron microscopy (below). EYA virus strain Gr, obtained as frozen infected mouse brain, was used for genome extraction without further propagation. California hare coltivirus S6-14-03 was adapted to cell culture by 4 passages in BHK-21 cells. The virus was then propagated in BHK-21 cells at 37 • C under 5% CO 2 , using Eagle's minimum essential medium supplemented with 2% fetal bovine serum and penicillin G, kanamycin, streptomycin, at 100 IU/ml, 100 g/ml and 100 g/ml, respectively. dsRNA extraction and preparation CTF virus strain Florio and California hare coltivirus S6-14-03: dsRNAs were extracted from virus infected cell cultures using a guanidinium isothiocyanate procedure (RNA NOW, Biogentex), as described previously [2, 4] .

EYA virus strain Fr578: dsRNA was extracted from 5 infected mouse brains after homogenization in phosphate buffered saline using the same RNA NOW method.

EYA virus strains Gr and Fr577: dsRNA was extracted from individual mouse brains using the same method.

Because EYA virus cannot be grown in cell culture, the sequence of its dsRNA genome segments was obtained by a strategy based on specific stepwise elimination of dsRNA segments (designated the "sequential segment subtraction method", "3SM") [4] . This involves RNase H hydrolysis of specific RNA template segments, after their hybridization The extracted viral dsRNAs from California hare coltivirus S6-14-03, EYA virus strain Fr577 and EYA virus strain Gr were copied into cDNA in the presence of random hexanucleotides and MuMLV superscript reverse transcriptase (Gibco BRL) as described previously [5] . PCR primers were designed from the sequences of the first, sixth, seventh and twelfth genome segments of CTF virus and EYA virus (primer sequences are shown in Table 1 ). The protocols used for PCR amplification, cloning and sequencing were as described previously [5] .

The methods used for the cloning and sequencing of CTF virus strain Florio has been described previously [3] .

The VP1 sequences of CTF virus strain Florio and EYA virus strain Fr578 were compared with the sequences of putative RNA-dependent RNA polymerases of representative strains of viruses representing eight genera of the family Reoviridae. GenBank accession numbers are provided in Table 2 .

All sequence alignments were generated by the Clustal W program [44] and the BLAST program contained in the DNATools program package (version 5.01.661, S.W. Rasmussen). Phylogenetic analyses were performed with the MEGA program [21] using the p-distance determination algorithm. Sequence relatedness was reported as percentage identity or percentage genetic distance. Tree drawing was performed with the Treeview program [34] . Comparison of sequences obtained in this study with those in databases was performed using the NCBI BLAST program (http://www3.ncbi.nlm.gov/BLAST). The Pfam program (http://www.sanger.ac.uk/Pfam/search.shtml) was used to search for previously described protein family-domains. The Motif program (http://www.motif.genome.ad.jp) was used to analyse theoretical protein sequences for the presence of known functional amino acid motifs.

Thin section electron microscopy Brains of suckling mice infected with CTF virus strain Florio were randomly cut into 1 mm 3 blocks and processed for sectioning. Tissue was fixed for 60 min in 2.5% glutaraldehyde, postfixed for 30 min in 1% phosphate-buffered osmium tetroxide, dehydrated in a graded ethanol series, and embedded in an Araldite-Epon mixture [30] . Scraped cell monolayers were pelleted by centrifugation at 630 g for 5 min and thereafter processed as brain tissue except for a reduction of glutaraldehyde fixation time to 20 min. Sections were cut with glass knives and stained with lead citrate [37] . Calibration of magnifications was made from photographs of a 54,864 line per inch diffraction grating replica (Ladd Research Industries, Inc., Burlington, Vermont, USA) taken at each magnification step used.

Negative contrast electron microscopy Carbon-coated grids prepared according to the technique of Simpson and Hauser [41] were rendered hydrophilic by exposure to ultraviolet irradiation. Very small quantities of infected mouse brain tissue were teased apart in large drops of 2% sodium silicotungstate stain, pH 7.0 [17] . Grids were floated on the drops and dried by touching to filter paper. Cell cultures were differentially centrifuged prior to negative staining. Low speed pellets (630 g, 5 min) and ultracentrifuge pellets (100,000 g, 3 h) were dispersed in drops of stain and treated in the same way as brain tissue. Magnification calibration was also performed at the kilovoltage and magnification steps used for photographing negative contrast specimens. Spraying and pseudoreplication techniques were employed but did not yield further resolution of virus structure.

The sequence of the 12 segments of the genome of CTF virus strain Florio has been determined: GenBank accession numbers are AF133428, AF139758-AF139764, AF000720, AF139765, U72694 and U53227, respectively for segments 1 to 12.

California hare coltivirus S6-14-03 genome sequence Partial sequences of segments 1, 6, 7, and 12 of California hare coltivirus S6-14-03, as determined in this study, has been deposited in GenBank: accession numbers are AF343051 (1477 bp), AF343054 (1241 bp), AF343057 (1446 bp) and AF343060 (614 bp).

Full-length sequences of segments 1-12 of EYA virus have been determined: GenBank accession numbers are AF282467 to AF282478. The full-length sequence of segment 12 of EYA virus strain Fr578 was found here to be exactly the same as the previously published partial sequence of the same segment [5] .

Partial sequences of segments 1, 6, 7, and 12 of EYA virus strain Fr577 and EYA virus strain Gr, as determined in this study, have been deposited in GenBank: accession numbers are AF343052 (1477 bp) and AF343053 (1477 bp), AF343055 (1491 bp) and AF343056 (1491 bp), AF343058 (1241 bp) and AF343059 (1241 bp), AF343061 (617 bp) and AF343062 (617 bp).

The size of the various segments of CTF and EYA viruses are shown in Table 3 . The largest open reading frame (ORF) was determined for each segment. This permitted us to deduce the sizes of the 5 -and 3 -non-coding regions (NCR), and those of the putative proteins encoded by each segment (see Table 3 ). The proteins encoded by the various dsRNA segments were designated VPx, where x refers to the number given to the RNA segment, based on its size.

Comparison of the genomic sequence of CTF virus, three strains of EYA virus, California hare coltivirus S6-14-03 and the Seadornaviruses Genomic sequences of CTF virus strain Florio, California hare coltivirus S6-14-03, EYA virus strain Gr, EYA virus strain Fr577 and EYA virus strain Fr578 were compared with those of viruses which are now recognized as members of the genus Seadornavirus. The latter included Banna virus strain BAV-In6423 strain BAV-In6969 strain BAV-In7043 and strain BAV-Ch, and Kadipiro virus strain KDV-Ja7075. These comparisons indicated a maximum of 15% amino acid identity between member viruses of the two genera. This low level of similarity is comparable to values found when comparing viruses in different genera of the family Reoviridae. As a frame of reference, the level of amino acid identity between homologous proteins of different viruses within various genera across the family Reoviridae is between 20 and 45% [28] .

Analysis of the NCRs of the various genomic dsRNA segments permitted the identification of conserved motifs located at each terminus. In the positive strand of each genome segment of CTF virus strain Florio, the motif [5 -SACUUUUGY-3 , where Y = C or U, S = G or C] was found in the 5 -NCR. The motif [5 -WUGCAGUS-3 , where W = A or U] was found in the 3 -NCR. The 5 -and 3 -terminal tri-nucleotides of all segments were found to be inverted complements. In all segments except segment 11 the 5 -NCR tri-nucleotide was GAC; in segment 11 it was CAC. In all segments except segment 11 the 3 -NCR tri-nucleotide was GUC; in segment 11 it was GUG. In the positive strand of each genome segment of EYA virus strain Fr578, the motif [5 -GACAWUUK-3 , where K represents G or U] was found in the 5 -NCR. The motif was found in the 3 -NCR. The 5 -and 3 -terminal tri-nucleotides of all segments were found to be inverted complements. In all segments the 5 -NCR tri-nucleotide was GAC. In all segments the 3 -NCR tri-nucleotide was GUC that is the same as in CTF virus.

Analysis showed the terminal motifs of the two viruses to be quite similar. The consensus sequence was found in the 5 -NCR and in the 3 -NCR. The conserved terminal sequences of CTF virus and EYA virus although similar, differ in the nucleotide at the fifth position of 5 -NCR of the positive strand (U in case of CTF virus and A or U in the case of EYA virus). A comparison of these sequences to those of member-viruses of different genera of family Reoviridae is presented in Table 4 . This comparison showed that the last nucleotide at the 3 termini is a cytosine residue for viruses belonging to the different genera of family Reoviridae except phytoreoviruses which have a uracil residue. The first nucleotide at the 5 termini is a guanosine residue for viruses belonging to the different genera of family Reoviridae except fijiviruses and cypoviruses which have an adenine residue.

Comparisons of nucleotide and amino acid sequences showed that all of the genome segments of EYA virus have cognate segments in the CTF virus genome (Fig. 1) . According to the standard designation of segments in which numbers are assigned in a decreasing order of size (S1 being the largest and S12 the smallest), there is one change in homologous EYA virus and CTF virus segments: segment 6 of EYA virus strain Fr578 is homologous to segment 7 of CTF virus strain Florio and vice versa.

Identification of RNA-dependent RNA polymerase motifs Motifs found in the polymerase sequences of all members of the family Reoviridae, were identified in the VP1 of EYA virus and CTF virus. These are motifs SG (positions 754-755) and GDD (positions 816-818 for CTF virus and 815-817 for EYA virus). The search for sequence homologies using the local BLAST program revealed matches between VP1 of CTF virus and EYA virus (amino acid 500 to 930) and the RNA-dependent RNA polymerases of Nilaparvata lugens reovirus (accession number D49693; identity: 24%, similarity: 39%) and rice ragged stunt virus (accession number U66714; identity: 24%, similarity: 38%). These findings are consistent with the hypothesis that the first genome segments of CTF virus and EYA virus encode the viral RNA-dependent RNA polymerase (VP1(Pol)).

Analyses of amino acid sequences using NCBI's BLAST program, showed that amino acid 75-105 of the VP2 of CTF virus and amino acid 75-110 of VP2 of EYA virus matched (identity: 46%, similarity: 53%) a sequence of the DNA adenine methyltransferase of Chlorella virus SC-1A (family Phycodnaviridae, accession number U65738). The analysis of this region of the Chlorella virus DNA methyltransferase showed that it is conserved among a number of DNA methyltransferases.

Analyses of VP5 showed the presence of a potential phosphamide linkage site (amino acid 342-347 of CTF virus: LNYDKY and for EYA virus: LNYIKH) comparable to that found in the protein encoded by segment 2 of members of the genus Orthoreovirus (position 1166-1171: ANPDKF, 40). This protein sequence is thought to be involved in the fixation of guanosine forming the cap structure found on the positive sense strand of orthoreovirus dsRNA.

The NCBI's BLAST program revealed that the region between amino acid 380 and 550 of VP6 of CTF virus resembled a number of nucleotidebinding proteins, such as myosin and kinesin and two distinct purine nucleotide phosphohydrolases (NTPases). These NTPases are the Archaeoglobus fulgidus purine NTPase (43% similarity, amino acid 195-356, accession number AE001032) and the Sulfolobus acidocaldarius purine NTPase (41% similarity, amino acid 236-408, accession number Y10687). In the homologous EYA virus protein (VP7), the sequence between amino acid 330 and 550 also resembled nucleotide-binding proteins, such as myosin, kinesin and ATPases.

The VP10 sequence of CTF virus and EYA virus contains the pattern [hhhhGx4GKSxnhhhhDD], where h indicates a bulky hydrophobic residue (amino acid 385 to 428) corresponding to a nucleotide-binding site in a number of protein modification enzymes [20, 25] . The local BLAST analysis showed that the region of the protein between amino acid 358 and 589 is homologous to proteins encoded by segment 7 of rice ragged stunt virus (genus Oryzavirus, family Reoviridae) and Nilaparvarta lugens reovirus (genus Fijivirus, family Reoviridae). In these latter viruses, segment 7 respectively encodes the non-structural protein NS7, and a core protein which has nucleotide-binding activity. Moreover, in the same region (amino acid 379 to 511), the Pfam analysis also revealed the presence of a kinase domain.

The RGD sequence (arginine-glycine-aspartic acid) is characteristic of a number of cell binding proteins and in the case of viruses this motif is thought to mediate the binding of viral structures to cellular integrins [6, 39] . Three RGD sequences were identified in the VP1 (amino acid 891-893), VP2 (amino acid 942-944) and VP4 (amino acid 172-174) of CTF virus. All of them are conserved in the homologous EYA virus proteins. RGD motifs in VP1 and VP2 are present at the peaks of a single hydrophilic domain, as shown using a Kyte and Doolittle hydropathy plot [22] . These RGD sequences may therefore be functional. An additional RGD motif is present in a hydrophilic region of CTF virus VP1 (amino acid 1071-1073) but this site is not conserved in EYA virus VP1. Other significant sequence features -read-through phenomenon Analysis of the nucleotide sequence of segment 9 of both EYA virus and CTF virus, revealed the presence of an opal stop codon belonging to the class of so-called "leaky" stop codons. This stop codon is flanked at its 3 -end by a cytosine residue, a configuration that has been shown in retroviruses [14] and alphaviruses [42] to allow "read-through." This phenomenon results from the incorporation of arginine, cysteine or tryptophan in the usual stop codon [14] . Accordingly, two proteins might be synthesized, the first encoded by the ORF between nucleotides 41 and 1054 (ORF1 encoding the protein VP9), the second encoded by the ORF between nucleotides 41 and 1846 (ORF2 encoding VP9 ). Comparison of the nucleotide sequences of segment 9 of CTF virus and EYA virus shows that the sequence variation is mainly a function of the third position of the codon. The ORF1 variations at the third position of the codons were calculated using the number of differences estimation method and found to represent 66.4% of the overall variation in this ORF, while the remaining 33.6% were shared by positions 1 and 2. An identical magnitude of variation was found in the part of ORF2 beyond the leaky stop codon. Clearly, this nucleotide variation was not randomly distributed (66.5% at the third position versus 33.4% at positions 1 . This result indicates that this region of segment 9 is submitted to the same kind of genetic selective pressure as regions encoding viral proteins. It is therefore very likely that this region of ORF2 is translated during viral replication. In addition, using the Kyte and Doolittle hydropathy plotting system (with an 11 amino acid sliding window) on the deduced proteins (beyond the suspected leaky stop codon) showed that the putative proteins of CTF and EYA viruses are very similar (Fig. 2) . This indicates that the function of the encoded proteins of the two viruses is highly conserved.

The amino acid identity between homologous proteins of CTF and EYA viruses ranged from 55 to 88%, with similarity ranging between 72 and 93% (Fig. 2) . The most variable proteins are those encoded by segments 6, 7 and 12 with amino acid identities of 55, 57 and 63%, respectively. When sequences of segments 6, 7 and 12 of California hare coltivirus S6-14-03, EYA virus strain Fr577 and EYA virus strain Gr were included in analyses, the same magnitude of variability was observed, as reported in Table 5 . In the case of the protein encoded by segment 1 (the viral RNA-dependent RNA polymerase) high identity values were found (86-99%), as seen in Table 5 .

Analyses of the amino acid sequences of the viral polymerase proteins (Pol) of member viruses of particular genera in the family Reoviridae indicated relationships that could be used to develop a sense of the overall phylogenetic relationships within the family. There was 21-97% amino acid identity between the various rotaviruses, 97-99% between the various phytoreoviruses, 56-98% between the various orbiviruses, and 92-99% between the various orthoreoviruses. However, Fig. 3 .

Thin section electron microscopy

The same progression of events in the maturation of CTF virus and EYA virus was observed in mouse brain and in cell culture. The initial discernible change in cells was the appearance in the cytoplasm of granular areas (identified as viral inclusion bodies, VIB) of moderate electron density. These VIBs had definite yet unbounded margins and some involved large portions of the cytoplasm. Electron dense particles, 45 nm in diameter, were found scattered throughout these granular matrices and especially at their edges; these particles were indistinguishable from cores of progeny virus particles (Figs. 4, 5) . The observation of VIBs reveals possible internal structures that could correspond to early stages of morphogenesis, reinforcing the hypothesis that these granular areas correspond to the site of virus assembly. As infection progressed increasing numbers of virions appeared at the periphery of the granular matrices. Virions consisted of a homogeneous dense core, 45 nm in diameter, surrounded by an electron lucid layer, considered to be the capsid itself. The total diameter of virions was 75-80 nm (Figs. 6, 7 ). An additional envelope layer was found surrounding some virions located within cisternae of the endoplasmic reticulum. This envelopment appeared to be a consequence of passage of virions through endoplasmic reticulum membranes (Figs. 7, 8) . Enveloped particles had a diameter of 90-93 nm. No difference in any morphologic or morphogenetic character was discernable between CTF virus and EYA virus. With both viruses large numbers of multiple filaments appeared in complex arrays in association with virions and granular matrices (Fig. 9) .Arrays of these filaments had cross-striations. These filaments might be equivalent to the "tubules" consisting of the NS1 protein [28] , observed in orbivirus infected cells. Within  Fig. 8 . Cytoplasm of a neuron of a suckling mouse infected 7 days earlier with EYA virus. Individual virions were found within envelopes derived from cytoplasmic membranes. Some viruses, notably coronaviruses, arteriviruses and flaviviruses, migrate in this way to the cell surface and are released by exocytosis -however, we were not able to prove this mechanism of shedding here. Thin section stained with lead citrate. Magnification × 75,000. EYA virions are indicated by arrows neurons in the brains of suckling mice infected with CTF virus arrays of these filaments occurred within nuclei as well as in cytoplasm. With both viruses, in both suckling mice and cell cultures, relatively few virions were found free in extracellular spaces, even late in infection. Instead, virions accumulated within cells at various stages of degeneration. Virus release appeared to occur via cell lysis.

Negative contrast electron microscopy CTF virus and EYA virions were observed in negatively stained preparations; they were identical. Virions were round or polygonal in outline with regularly spaced polygonal surface features, which seemed to lie flat on a rather smooth virion surface. Virion mean diameter was 80 nm (range 73-89 nm) (Fig. 10A) . Quite prominent in all preparations were delicate round particles 50 nm in diameter (Fig. 10B) . These were made up of regular subunits. From observations of partially degraded virus particles and virus particles variably penetrated by stain, and by analogy with the morphological details of orbiviruses and reoviruses, it was concluded that the 50 nm particles represent the virion inner capsid and the 80 nm particles complete virions. As has been observed in the past, the detailed surface structure of CTF virus and EYA virus capsids seemed quite discernable, seemed quite available for analysis of symmetry. However, so few particles have been photographed in a way suitable for such analysis that this has not yet been done.

Colorado tick fever virus was identified more than fifty years ago by Florio and colleagues as the etiologic agent of Colorado tick fever, one of the most important tick-borne viral diseases in North America [13] . The virus is the type species of genus Coltivirus. In its natural habitat the virus is transmitted between small mammals and the tick Dermacentor andersoni and other ixodid ticks [13, 15, 16] . Antigenic variants of the virus have been described; these have been isolated from ticks and vertebrates [8, 19] . Two related but distinct viruses have been identified in North America: one is California hare coltivirus S6-14-03, isolated as the name implies in California from a black-tailed jackrabbit Lepus californicus [23] ; the other is Salmon River virus, isolated in Idaho from a human patient with a CTFlike illness [31] . A virus related to CTF virus, identified as Eyach virus, has been isolated on several occasions in Europe [12, 37] . This virus exhibits a one-way cross-reaction with CTF virus in complement-fixation tests, but each virus has a distinct neutralization profile. More recently, comparison of the sequence of genome segment 12 of CTF virus and EYA virus confirmed this serological relationship [5] .

For some years the genus Coltivirus was divided into two subgroups, A and B. Subgroup A included CTF, EYA and the other North American and European viruses. Subgroup B comprised viruses isolated from mosquitoes and humans in Southeast Asia. However, recently dsRNA hybridisation analyses, serological analyses and comparison of the genomic sequences of the various viruses led to dropping the A and B subgroup designation and construction of a distinct genus, the genus Seadornavirus, for the Asian mosquito-borne viruses [18, 45] .

Genetic characterization of the coltiviruses was for some years limited to hybridisation analyses of genomes segments of field isolates of CTF virus; this work only confirmed the narrow range of CTF virus genetic diversity and suggested possible reassortment in nature of segments 4 and 6 [8] . Recent progress in viral genomics, such as the analyses of the nucleotide sequence of full genomes of the two major coltiviruses, as reported here, have permitted a more systematic approach and many new insights into the viruses and their evolutionary relationship.

These sequence analyses have shown that the organization of the genomes of CTF and EYA viruses is similar: (i) the genomes of the viruses are of comparable size (∼29,000 bp); (ii) each of the 12 segments of genomic dsRNA of the viruses are of comparable size; (iii) each genome segment is flanked by similar 5 -and 3conserved non-coding sequences; and (iv) there is strong evidence that in segment 9 of each virus there is a read-through phenomenon.

These findings are reinforced by the identification, in the cognate genes of the viruses, of sequences encoding similar motifs (functional domains) in nonstructural viral proteins as found in certain other viruses and organisms. However, the paucity of available data on coltivirus proteins did not permit clear resolution of many structure-function relationships. An important exception is the unambiguous identification of the viral RNA-dependent RNA polymerase encoded in segment 1 of the viruses. The specific SG and GDD motifs found match polymerase motifs found in the genomes of other member viruses of the family Reoviridae. Motifs for protein modifying enzymes including nucleotide phosphohydrolases (NTPases), guanylyltransferases and methyltransferases I and II were also identified. Such constitutive enzymes are known to be involved in capping positive strands of virion RNA segments in other member viruses of the family Reoviridae.

Classification criteria within the family Reoviridae, based on the quantification of genetic relatedness, have been formally accepted by the International Committee on Taxonomy of Viruses in its Seventh Report [28] . In particular, the degree of relatedness of VP1(pol) amino acid sequences has been accepted as a means of delineating genera and assigning viruses to particular genera. A VP1(pol) amino acid identity of less than 20% has been taken as a criterion for placing viruses in different genera [2] . Accordingly, the comparison of the CTF virus VP1(pol) sequence with that of EYA virus (amino acid identity 86%) unambiguously places the two viruses in the same genus (genus Coltivirus). Moreover, on the same basis this genus may be considered distinct from all other genera -the amino acid identity between coltivirus VP1(pol) and all other member viruses of the family Reoviridae is less than 20%. In particular, theVP1(pol) amino acid identity between the coltiviruses and the member viruses of the genus Seadornavirus (which are also arthropod-borne and also have 12 segment dsRNA genomes) is less than 10% [2] .

A number of other molecular and biological characters support the separation of the coltiviruses and seadornaviruses into distinct genera: (i) The genomes of CTF virus and EYA virus are significantly larger than those of Banna virus and Kadipiro virus (∼29,000 bp vs. ∼20,000 bp); (ii) The G + C content of CTF and EYA virus genomes is 47%, which is significantly higher than the 39% calculated from the full-length genome sequences of Genetic analyses have indicated that the most important differences between the member viruses of the genus Coltivirus are found in the genes encoded in segments 6, 7 and 12 (these segment designations are the same for CTF and EYA viruses). Amino acid identities here are 55%, 56% and 62%, respectively, lower than those seen for other segments of the two viruses. Together with their geographic and ecological isolation and their distant serological relationship, there seems to be a consistent basis for defining the two viruses as distinct species within the genus. The same basis is likely to be useful in the future for the demarcation of other species within the genus.

Of course, two phenotypic characters representing the expression of a large part of the genome of CTF virus and EYA virus are the structure of the virion itself and the complex morphogenetic steps leading to the construction of the virion. In this regard, the authors enjoyed reviewing studies of CTF virus done more than 30 years ago and comparing those observations with unpublished studies of EYA virus. The viruses proved indistinguishable, but there is still more to be done here: the elegant structural studies that have been done in recent years on reoviruses, rotaviruses and orbiviruses (bluetongue virus) using computer analysis of cryoelectron microscopy images and X-ray crystallography should be done on the coltiviruses. The overall similarity in virion strucural design of the member viruses of the family Reoviridae belies the differences that have evolved as the member viruses of the various genera have occupied such different niches. From the need to withstand a wide pH range and digestive enzymes in the vertebrate intestinal tract and transmission by the fecal-oral route (reoviruses, rotaviruses) to the need to withstand the systemic environment in diverse arthropod vectors and vertebrate hosts and transmission by arthropod bite (orbiviruses), to the need to withstand the unique physiology of plants and plant cells (fijiviruses, phytoreoviruses), very different characters, especially of virion surface structure, are called for. Aspects of the functional qualities of these structures should be addressed by comparative structural studies.

The identification of genetic, serologic and morphologic/morphogenetic relationships between CTF virus and EYA virus raises interesting questions about their evolutionary origin. Several hypotheses have been proposed: As a frame of reference, the molecular evolutionary rate of the polymerase gene of tick-borne flaviviruses (single stranded RNA viruses) has been estimated as 5×10 −5 mutations/nt/year [46] . This rate is based on host divergence of the various viruses between 4,000 and 6,000 years ago.

Even though there is little information on the molecular evolutionary rate of dsRNA viruses, it seems extremely unlikely that the coltiviruses could be evolving at a rate consistent with that in the first hypothesis. On the other hand, several lines of thought favor the second hypothesis: (i) It seems reasonable to expect that the stabilizing effect of double-strandedness, as seen with dsDNA genomes, should apply to dsRNA genomes and therefore the molecular evolutionary rate of viruses with such genomes might be expected to approach that of dsDNA genomes (perhaps less the extra stabilizing effect of DNA proofreading enzymes); (ii) It has been proposed that the alternating life cycles of arboviruses in arthropods and vertebrates provides a stabilizing effect -mutations, especially in polymerase genes, that might provide a survival advantage in one host might represent a disadvantage in the other host -so the replication and transmission of the parental type might be favoured overall; (iii) Tick-borne viruses may evolve slower than viruses transmitted by other arthropod vectors or directly from vertebrate host to host -tick vectors remain dormant for long periods during which virus replication cycles must be reduced, and tick vectors remain localized thereby limiting virus radiation and dispersal; (iv) The powerful evolutionary effect of reassortment, so common with dsRNA viruses, may be abrogated by the econiche isolation and habits of the tick vectors of the coltiviruses -that is, the level of reassortment seen with the other genus of dsRNA viruses, the orbiviruses, might reflect habits and habitats of their Culicoides spp. and mosquito vectors rather than the viruses themselves; (v) We know nothing about other coltiviruses or similar viruses, viruses that might "fill in the blanks" between CTF virus and EYA virus and provide support for one or another evolutionary hypothesis. For example, it is not inconceivable that another virus might be discovered that would indicate that CTF virus was derived from an EYA-like virus rather than the opposite.

Perhaps the most intriguing aspects of this examination of the various hypotheses for the molecular evolution of the coltiviruses pertain to: (i) the overall molecular evolutionary rate of viral dsRNA genomes in general, and (ii) the more ancient evolutionary history of the divergence of the ancestors of the member viruses of all the genera of the family Reoviridae and the other dsRNA viruses.

Another intriguing aspect of this examination of the genomes of the coltiviruses has been the discovery that part of the EYA virus VP7 is clearly distinct from the homologous region of CTF virus VP6 (amino acid identity < 24%), whereas the rest of the proteins are quite similar (amino acid identity > 60%). The distinct region of EYA virus VP7 is similar (∼ 50% similarity) to a sarcolemmalassociated protein of the host rabbit in Europe, Oryctolagus cuniculus. This raises the question of a possible recombination event between the genome of an ancestor of EYA virus and an mRNA of its lagomorph host. Recombination in dsRNA viruses (distinct from reassortment) has been reported (e.g., in rotaviruses), but only between viral genomes. Recombination between viral and cellular RNAs has been seen in member viruses of the genus Pestivirus (family Flaviviridae; positive-stranded ssRNA viruses) [29] . Clearly, the significance of the similarity between the EYA virus protein and that of its host warrants further investigation.

The genetic analyses reported here have allowed us to resolve several questions pertaining to the taxonomy of the coltiviruses, but they have raised even more questions. Extending these genetic analyses to a better understanding of the natural history of the coltiviruses will take much more work. One key will be developing a better understanding of the nature and extent of coltivirus variation in nature and a better understanding of the evolution of the various viruses in their complex and geographically and ecologically isolated niches. In this regard, we can hope that more and more coltiviruses will be discovered in nature and characterized in reference laboratories. Another key will be developing a better understanding of the variation in structure-function relationships of the various viral proteins, the architectural organization of the virions, and the nature of variances in viral phenotypes, and pathotypes. Taken together, the morphologic, serologic, biologic, and genetic characteristics that we have studied so far begin to paint the picture of a coherent set of viruses that have solved survival demands in unique fashion. The viruses are entrenched in their econiches and represent in some instances important threats to human health. The viruses are interesting adversaries, worthy of further investigation.

Fifty years with viruses

Complete sequence determination and genetic analysis of Banna virus and Kadipiro virus: Proposal for assignment to a new genus (Seadornavirus) within the family Reoviridae

Sequence determination and analysis of the full-length genome of Colorado tick fever virus, the type species of genus Coltivirus (family Reoviridae)

Strategies for the sequence determination of viral dsRNA genomes

Comparative sequence analysis of American, European and Asian isolates of viruses in the genus Coltivirus

Crystal structure of the top domain of African horse sickness virus VP7: comparison with bluetongue virus VP7

Les ectoparasites du lapin de Garenne, Oryctolagus cuniculus: Apports a son histoire

Genetic relatedness of Colorado tick fever virus isolates by RNA-RNA blot hybridization

Physiochemical and morphological relationships of some arthropod-borne viruses to bluetongue virus -a new taxonomic group. I. Physiochemical studies

Single-stranded DNA binding proteins required for DNA replication

Erve and Eyach: two viruses isolated in France, neuropathogenic for man and widely distributed in western Europe

Isolation of Eyach virus (Reoviridae, Colorado Tick Fever group) from Ixodes ricinus and I. ventalloi ticks in France

Ecology of Colorado tick fever

Identification of amino acids inserted during suppression of UAA and UGA termination codons at the gag-pol junction of Moloney murine leukemia virus

Colorado tick fever. Isolation of the virus from Dermacentor andersoni in nature and laboratory study of the transmission of the virus in the tick

The etiology of Colorado tick fever

The internal structure of an insect virus

Proceedings of midterm meeting

Antigenic variants of Colorado tick fever virus

Evolution of thymidine and thymidylate kinases: the possibility of independent capture of TK genes by different groups of viruses

Molecular evolutionary genetics analysis

A simple method for displaying the hydropathic character of a protein

Survey for evidence of Colorado tick fever virus outside of the known endemic area in California

Lapin de Garenne, Oryctolagus cuniculus L. et arbovirus dans le sud-est de la France résultats de deux enquêtes sérologiques

Novel families of putative protein kinases in bacteria and archaea: evolution of the "eukaryotic" protein kinase superfamily

Antibodies against some arboviruses in persons with various neuropathies

African horsesickness

Reoviridae

Molecular characterization of pestiviruses

Plastic embedding mixtures for use in electron microscopy

Orbiviruses and Coltiviruses

Physiochemical and Morphological relationships of some arthropod-borne viruses to bluetongue virus -a new taxonomic group. II. Electron microscopic studies

Colorado tick fever virus: an electron microscopic study

TreeView: an application to display phylogenetic trees on personal computers

Electron microscopy of neurotropic African horse sickness virus

Surgeons US Army (1900) The etiology of yellow fever: a preliminary note

Eyach, an arthropod-borne virus related to Colorado Tick Fever Virus in the Federal Republic of Germany

The use of lead citrate at high pH as an electron opaque stain in electron microscopy

RGDdependent entry of coxsackievirus A9 into host cells and its bypass after cleavage of VP1 protein by intestinal proteases

Complete nucleotide sequence of reovirus L2 gene and deduced amino acid sequence of viral mRNA guanylyltransferase

Structural components of vesicular stomatitis virus

The Alphaviruses: Gene expression, replication, and evolution

Bluetongue virus structure

CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice

) Virus Taxonomy. Seventh Report of the International Committee on Taxonomy of Viruses

Population Dynamics of flaviviruses revealed by molecular phylogenies

The authors wish to thank C. Chastel and N. Karabatsos for providing EYA virus strain Fr577, EYA virus strain Fr578, EYA virus strain Gr and the California hare coltivirus S6-14-03. We also thank Dr C. Goldsmith for preparation of some of the photomicrographs. This study was supported by EU Grant 'Reo ID' number QLK2-2000-00143. The "Unité des Virus Emergents" is an associated research unit of the Institut de Recherche pour le Développement (IRD). This study was supported in part by the IRD.