|Home | About | Journals | Submit | Contact Us | Français|
Although few simian rotaviruses (RVs) have been isolated, such strains have been important for basic research and vaccine development. To explore the origins of simian RVs, the complete genome sequences of strains PTRV (G8P), RRV (G3P), and TUCH (G3P) were determined. These data allowed the genotype constellations of each virus to be determined and the phylogenetic relationships of the simian strains with each other and with nonsimian RVs to be elucidated. The results indicate that PTRV was likely transmitted from a bovine or other ruminant into pig-tailed macaques (its host of origin), since its genes have genotypes and encode outer-capsid proteins similar to those of bovine RVs. In contrast, most of the genes of rhesus-macaque strains, RRV and TUCH, have genotypes more typical of canine-feline RVs. However, the sequences of the canine and/or feline (canine/feline)-like genes of RRV and TUCH are only distantly related to those of modern canine/feline RVs, indicating that any potential transmission of a progenitor of these viruses from a canine/feline host to a simian host was not recent. The remaining genes of RRV and TUCH appear to have originated through reassortment with bovine, human, or other RV strains. Finally, comparison of PTRV, RRV, and TUCH genes with those of the vervet-monkey RV SA11-H96 (G3P) indicates that SA11-H96 shares little genetic similarity to other simian strains and likely has evolved independently. Collectively, our data indicate that simian RVs are of diverse ancestry with genome constellations that originated largely by interspecies transmission and reassortment with nonhuman animal RVs.
Group A rotaviruses (RVs) are a major cause of acute dehydrating diarrhea in infants and children under the age of 5 years worldwide. These infections lead to approximately 527,000 deaths each year, the vast majority occurring in developing countries (33). RVs are also responsible for gastroenteritis in many other animal species, notably mammals and birds (16, 38). RVs are members of the family Reoviridae and possess a genome consisting of 11 segments of double-stranded RNA (dsRNA). The prototypic genome of a group A RV encodes six structural proteins (VP) and six nonstructural proteins (NSP) (5). The mature RV virion is a nonenveloped triple-layered icosahedral particle. The inner most protein layer is formed by the core lattice protein VP2. Attached to the interior surface of the VP2 layer near the fivefold axes are complexes of the viral RNA-dependent RNA polymerase VP1 and the RNA capping enzyme VP3. Collectively, VP1, VP2, VP3, and the dsRNA genome form the core of the virion (5, 11). The core is surrounded by VP6, the sole constituent of the intermediate protein layer of the virion. The antigenic properties of VP6 are used in classifying RV isolates into groups. The outer protein layer of the virion is composed of trimers of the VP7 glycoprotein penetrated by spikes of the VP4 attachment protein (50). The properties of VP7 and VP4 form the basis of a dual classification system defining RV G types (glycosylated) and P types (protease sensitive), respectively. At present, 23 G genotypes and 31 P genotypes have been recognized in the literature based on sequence analyses (17, 39, 42, 45, 47). Recently, a comprehensive sequence-based classification system was established for the RVs which, together with a uniform nomenclature, allows each genome segment of the virus to be assigned to a particular genotype. In the comprehensive classification system, the acronym Gx-P[x]-Ix-Rx-Cx-Mx-Ax-Nx-Tx-Ex-Hx defines the genotypes of VP7-VP4-VP6-VP1-VP2-VP3-NSP1-NSP2-NSP3-NSP4-NSP5 encoding genome segments (17, 18).
Several years ago, Nakagomi et al. provided evidence by RNA-RNA hybridization assays that RVs originating from different animal species could be resolved into genogroups based upon the existence of unique species-specific genome constellations (29-31). More recently, the concept that RVs preferentially retain certain species-related genome constellations has been further supported by whole-genome sequencing (8, 24). For human RVs, two major genogroups (Wa-like genogroup 1 and DS-1-like genogroup 2) and one minor genogroup (AU-1-like genogroup 3) have been described (8, 17, 30). Although these genogroups are generally species specific, it is believed that the human AU-1 genogroup is of feline origin (31) and that the human Wa and DS-1 genogroups share common ancestor with porcine and bovine RVs, respectively (17). Another recent study based on full genome sequence data has indicated that the rarely seen human G3P RVs are of feline or canine origin (46). Two additional sequence-based studies have indicated that human RVs with P specificity may have originated after interspecies transmission from rabbit RVs and RVs from hosts belonging to the order Artiodactyla (i.e., hoofed mammals with even toes, including ruminants and pigs) (19, 20). These examples indicate that interspecies transmission of entire RV gene constellations from one host species to another may contribute significantly to viral evolution. In addition to interspecies transmission, complete genome sequencing of RVs have revealed multiple examples of naturally occurring inter- and intragenogroup reassortment (17, 19, 21-23, 37, 41).
The simian RV strains, notably RRV and the SA11 derivatives (e.g., SA11-Cl3 and SA11-4F), have been used extensively as models in the study of all aspects of RV biology, including characterizing genome replication and virion assembly, delineating high-resolution structures of viral proteins and the virion capsid, and describing the functions of viral proteins. Moreover, the RRV strain was used to create a set of human-simian reassortant viruses that formed the basis of the first commercially licensed RV vaccine (Rotashield; Wyeth Laboratories) (10). Serological analyses have indicated that simian RVs are probably endemic in wild nonhuman primate (NHP) species in Africa (32). However, whether or not unique genogroups or preferred genome constellation exist for the simian RVs has not been determined, because of the lack of comprehensive genetic data. Most simian RVs isolated to date (e.g., rhesus macaque viruses RRV  and TUCH  and the pig-tailed macaque virus PTRV ) have been recovered from monkeys kept in captivity in the United States. An important exception is the SA11 isolate, which was recovered from a vervet monkey in South Africa (15). Simian RV infections occur mostly in young monkeys, similar to human RV infections in children (32, 40).
To gain further insight into the origins and properties of simian RVs, we sequenced and contrasted the genomes of PTRV, RRV, and TUCH with other RVs, including SA11-H96 (G3P), the only previously fully sequenced simian RV (41). Our results reveal that these four simian RVs are of divergent ancestry and have evolved by combinations of interspecies transmission and reassortment with RVs naturally occurring in other animal species. Thus, the simian RVs do not possess a common genome constellation nor define a unique genogroup. Although frequently used as disease models, the simian RVs show limited genetic similarity with the human RVs (genogroups 1 and 2) responsible for most human disease.
Monkey kidney (MA104) cells were propagated in 199 medium containing 10% fetal bovine serum. The cell culture-adapted RV strains PTRV (G8P), RRV (MMU18006; G3P), and TUCH (G3P) were activated by treatment with 10 μg of trypsin (T0303-1G; Sigma-Aldrich)/ml and grown in MA104 cells infected with ≤1 PFU of virus. Infected cells were maintained in serum-free 199 medium until the cytopathic effects were complete.
Genomic dsRNA was extracted from PTRV using TRIzol (Invitrogen). The dsRNA of PTRV was converted into cDNA using the FLAC (full-length amplification of cDNA) method (14). Briefly, anchor primers were ligated onto the ends of the dsRNA using T4 RNA ligase. Afterward, the dsRNA segments were resolved by electrophoresis on a 1% agarose gel, and each was recovered by using a QIAquick gel extraction kit (Qiagen). cDNAs were synthesized for each dsRNA segment using AMV reverse transcriptase (New England Biolabs) and then PCR amplified with ExTaq DNA polymerase (Takara). The cDNAs were cloned into the pGEM-T Easy vector (Promega).
Genomic dsRNA was recovered from infected cell lysates using TRIzol-LS (Invitrogen) and denatured by incubation with 50% dimethyl sulfoxide. The RNA was used as a template for the preparation of viral cDNAs using a OneStep RT-PCR kit (Invitrogen) and appropriate segment-specific primers (available upon request). Reverse transcription-PCR products were gel purified and sequenced with an ABI Prism BigDye v3.1 terminator cycle sequencing kit and detected with an Applied Biosystems 3730 DNA analyzer. In a few cases, the terminal segment-specific primers used in preparing RRV cDNAs extended into the open reading frame, preventing the de novo identification of some N- and C-terminal residues of proteins. The numbers of masked terminal residues are as follows: VP1 (two N-terminal residues: two C-terminal residues [2:2]), VP2 (4:0), VP4 (7:1), VP6 (2:0), NSP3 (2:0), and NSP5 (3:0).
The cDNAs of PTRV, RRV, and TUCH were sequenced with an ABI Prism BigDye v3.1 terminator cycle sequencing kit (Applied Biosystems Group). The dye terminator was removed by using Performa DTR cartridges (Edge Biosystems), and sequences were determined with an Applied Biosystems 3730 DNA analyzer. Sequence files were analyzed and assembled by using Sequencher 4.5 software (Genes Code Corp).
Phylogenetic and molecular evolutionary analyses were conducted by using the MEGA software (version 4) (44). Genetic distances were calculated by using the Kimura-2 correction parameter at the nucleotide level, and phylogenetic trees were constructed by using the neighbor-joining method with 500 bootstrap replicates. VP7 and VP8* images were generated by using the UCSF Chimera-Molecular Modeling System (35).
The genotypes of the genome segments for PTRV, RRV, and TUCH were assigned based on the recommendations of the Rotavirus Classification Working Group (RCWG) (18), using the online RotaC genotyping tool (http://rotac.regatools.be/).
The nucleotide sequences of the VP1, VP2, VP3, VP4, NSP1, VP6, NSP3, NSP2, VP7, NSP4, and NSP5 genes of RV strains RRV, TUCH, and PTRV are available in GenBank under the accession numbers EU636924 to EU636934, EF583010 to EF583013 and FJ816611 to FJ816617, and FJ422131 to FJ422141, respectively.
PTRV was recovered from a stool sample of a young female pig-tailed macaque with diarrhea housed at the Medical Lake breeding facility of the National Primate Center, University of Washington, Seattle, in 1990. Previous analysis established that PTRV (G8P) has a “long” electropherotype, a subgroup I VP6, and a genotype A NSP4 (9), all characteristics typical of bovine RVs. Likewise, RNA-RNA hybridization assays have indicated that PTVR and bovine RVs are closely related, since eight to nine of their RNA segments formed heteroduplexes with electrophoretic migration patterns similar to those of homoduplexes (9).
Complete genome sequencing identified the following genotype constellation for PTRV: G8-P-I2-R2-C2-M2-A3-N2-T6-E2-H3. This constellation is typical of P bovine RVs, suggesting that PTRV may have originated by interspecies transmission from a bovine to a simian host (Table (Table1).1). Phylogenetic analyses confirmed that all 11 gene segments of PTRV cluster with those of bovine RVs (Fig. (Fig.1).1). Interestingly, many of the PTRV gene segments (VP1, VP2, VP3, VP6, VP7, NSP1, NSP3, and NSP5) also cluster with several human P and nonbovine animal RV strains: 111/27-05, B10925-97, Hun5, MG6, PA169 (human), RC/18-08 (sable antelope), Chubut (lama guanaco), and OVR762 (ovine) (Fig. (Fig.1).1). This clustering pattern is not surprising, since it was recently shown that these P human and animal strains share a largely conserved consensus genotype constellation with P bovine RVs (Table (Table1)1) (19). Thus, although PTRV may have originated from interspecies transmission of a bovine RV to a pig-tailed macaque, it is possible that RVs infecting other members of the Bovidae family or the Artiodactyla order may have been the source of PTRV.
To determine whether PTRV exhibits similarities to bovine RVs at the protein level, we contrasted the surface-exposed residues of their outer capsid glycoprotein VP7 and spike protein VP4. Such analysis was made possible by the recent description of the atomic structure for the VP7 trimer (1) and the earlier description of the atomic structure for the VP8* globular head of the VP4 spike protein (4). The location of structure-based antigenic domains on the VP7 trimer (7-1a, 7-1b, and 7-2) (Fig. (Fig.22 and and3)3) and the VP8* monomer (8-1, 8-2, 8-3, and 8-4) (Fig. (Fig.33 and and4)4) were predicted based on the characterization of antibody-escape mutants and the identification of serotype-specific amino acid changes. The VP7 antigenic domains 7-1a, 7-1b, and 7-2 include the previously described antigenic regions A (residues 87 to 101) and D (residue 291); C (residues 208 to 221), E (residues 189 to 190) and F (residues 233 to 242); and B (residues 142 to 152), respectively (36). Comparison of PTRV VP7 with those of the G8 bovine strains Tokushima 9503 (P) and O-agent (P), and the ovine strain OVR762 (P) identified a single difference (A87V) of a surface-exposed residue (Fig. (Fig.3C).3C). PTRV VP7 differs from that of the G8 human strain 69M (P) by one additional residue (T171A). Only the A87V change maps definitely within an antigenic domain (7-1a). The location and limited number of changes, combined with their conservative nature, suggests that the antigenicity of PTRV VP7 is identical to, or nearly so, with that of several G8 bovine RVs. This hypothesis is supported by the results of plaque-reduction assays which have shown that antiserum raised against PTRV (G8P) displays a high level of neutralizing activity to 69M (G8P) (9). As might be anticipated from these data, the amino acid identity of PTRV VP7 with the VP7 proteins of Tokushima 9503, O-agent, OVR762, and 69M is extremely high (>98%) (Table (Table22).
Analysis of PTRV VP8* showed that it included only three surface-exposed residues (D111N, P157S, and H202Y) differing from VP8* of the P bovine O-agent strain (G8) (Fig. (Fig.5C).5C). Likewise, these were the only differences noted between PRTV VP8* and VP8* of SA11-5N, a reassortant virus with a bovine VP4 gene. In addition to the D111N, P157S, and H202Y replacements, VP8* of the P bovine strain NCDV (G6) contains one additional difference (N185T). Only the D111N change is located within a defined antigenic domain (8-3); thus the antigenicity of the PTRV VP8* protein can be expected to closely mirror that of several P bovine RVs. This prediction is in agreement with the results of plaque reduction assays which have shown that antiserum raised against PTRV (G8P) exhibits a high level of neutralizing activity to NCDV (G6P) (9). The results are also consistent with analyses showing that the amino acid identity of PTRV VP8* is greater than 96% with the VP8* proteins of O-agent, SA11-5N, and NCDV (Table (Table33).
Along with antigenic domains, the VP8* proteins of most animal RVs and a few human strains contain a sialic acid-binding site that facilitates virus attachment to the host cell (4). Amino acid residues that form the sialic-acid binding site of PTRV VP8* are identical to those of many P bovine RVs but differ from those of other simian RVs (Table (Table4).4). Since PTRV grows to high titers in macaques and spreads efficiently within macaque colonies, the sialic acid binding site characteristic of P bovine viruses can be presumed to successfully recognize carbohydrates displayed on macaque enterocytes (13). Collectively, analysis of the PTRV genome sequence and structural antigenic domains suggests that PTRV originated by interspecies transmission of a G8P bovine virus into a macaque.
The RRV (MMU18006) strain was isolated in 1975 from a juvenile rhesus macaque with diarrhea that was born in the California National Primate Research Center (43). The TUCH strain was isolated in 2002 from an asymptomatic juvenile rhesus macaque housed at the Tulane National Primate Research Center in Louisiana (25). Previous studies have established that RRV (G3P) and TUCH (G3P) have “long” electropherotypes, subgroup I VP6s, and genotype C NSP4s (6, 12). Complete genome sequencing identified the genotype constellations for RRV (G3-P-I2-R2-C3-M3-A9-N2-T3-E3-H6) and for TUCH (G3-P-I9-R3-C3-M3-A9-N1-T3-E3-H6). Thus, seven genes of RRV and TUCH share common genotypes (Table (Table1).1). Comparison of the RRV and TUCH genotype constellations with other completely sequenced RV genomes revealed an unexpected finding. As shown in Table Table1,1, RRV and TUCH share eight and seven genotypes, respectively, with several G3P strains of canine (A79-10, CU-1, and K9), feline (Cat97), and human canine/feline-like (HCR3A, RO1845) RVs. In addition, two other human strains, which are believed to be of feline origin, AU-1 (G3P) and T152 (G12P), also share five to six genotypes with RRV and TUCH (Table (Table1).1). These data suggest that RRV and TUCH are related to one another and have a shared ancestry with canine/feline RVs. Moreover, these data reveal that overall RRV and TUCH are more closely related to the human AU-l/feline-like RVs than to human Wa- or DS-1-like RVs.
To probe the origins of RRV and TUCH in greater detail, phylogenetic dendrograms were constructed for each of the 11 viral gene segments (Fig. 1A to C). The analysis showed that the VP3 (M3), VP7 (G3), NSP1 (A9), NSP3 (T3), NSP4 (E3), and NSP5 (H6) genes of RRV and TUCH clustered, albeit only distantly, within the same genotype as canine, feline, and canine/feline human RVs. The dendrograms also showed that the RRV VP4 (P) and TUCH VP1 (R3) genes clustered, again distantly, within the same genotype as the canine/feline/human RVs. For the VP2 gene, both RRV and TUCH, together with AU-1 and T152, clustered within the C3 genotype, whereas the canine/feline/human RVs were contained within the M2 genotype, suggesting the possible reassortment of an RRV/TUCH ancestor with a human DS-1-like (genogroup 2) RV (Fig. (Fig.1A).1A). Other genes of RRV and TUCH that are not of the canine/feline/human genotype may have also originated by reassortment. These include the VP4 (P), VP6 (I9), and NSP2 (N1) genes of TUCH, which have genotypes unrelated to canine/feline/human RVs (Fig. (Fig.1).1). In particular, the NSP2 gene of TUCH is most closely related to the Wa-like strain IAL28 (92% nucleotide identity), suggesting reassortment with a human Wa-like (genogroup 1) virus. On the other hand, the only representatives of the P and I9 genotypes are the VP4 and VP6 genes, respectively, of the TUCH virus, rendering the origins of these genes obscure. RRV was found to possess VP1 (R2), VP6 (I2), and NSP2 (N2) genes with genotypes characteristic of RVs infecting Bovidae or related hosts. Thus, some genes of RRV may have originated by reassortment between an ancestor and a bovine or other ruminant RV.
The RV strain SA11 was initially isolated in South Africa in 1958 from a healthy vervet monkey and, after adaptation to cell culture, this and derivative reassortant strains have been used extensively as prototypes in the study of RV biology. SA11, like RRV and TUCH, was determined by serotyping to represent a G3 virus (25). A previous comparison of the complete genomes of several SA11 laboratory isolates showed that some represented reassortants formed in vitro by the introduction of a bovine VP4 (SA11-4F) or NSP2 (SA11-both) gene into the SA11 genetic background (41). SA11-H96, the strain believed to most closely resemble the original SA11 isolate, has a genotype constellation of G3-P-I2-R2-C5-M5-A5-N5-T5-E2-H5. Phylogenetic analyses indicated that for several genes [VP2 (C5), VP3 (M5), VP4 (P), NSP1 (A5), NSP2 (N5), NSP3 (T5), and NSP5 (H5)], SA11-H96 is the only representative of its genotype (Fig. (Fig.11 and Table Table1).1). For the remaining genes (VP1, VP6, VP7, and NSP4), SA11-H96 forms a distinct subcluster inside the R2, I2, G3, and E2 genotypes, respectively (Fig. (Fig.1).1). These findings indicate that SA11-H96 is not closely related genetically to any other known RV, including those that are of simian or human origin.
Genotyping and phylogenetic analysis indicate that the RRV and TUCH strains are related to one another and have originated independently of the SA11-H96 strain. Thus, it was surprising to find that the exterior surface-exposed residues of the RRV VP7 show greater similarity to the corresponding residues of SA11-H96 VP7 than TUCH VP7. Specifically, only three surface-exposed residues (D123N, T212A, and A213T) differ between the RRV and SA11-H96 VP7 proteins, while seven residues differ between the RRV and TUCH VP7 proteins (Fig. (Fig.3).3). This parallels data showing that the amino acid identity of RRV VP7 with SA11-H96 VP7 is greater than that with TUCH VP7 (96 and 91%, respectively) (Table (Table2).2). Analyses of surface-exposed residues and amino acid identities also show that RRV VP7 is more similar to the VP7 proteins of the G3P canine/feline/human RVs (e.g., Cat97, K9, and RO1845) than to TUCH VP7 (Fig. (Fig.3C3C and Table Table2).2). Notably, residues that differ between RRV VP7 and the VP7 proteins of the G3P canine/feline/human RVs exclude the 7-1a domain and only one or two residues at the periphery of the 7-1a domains differ between the RRV and SA11-H96 VP7 proteins (D123N) and the RRV and TUCH VP7 proteins (D123N and S126T) (Fig. (Fig.3C).3C). The conserved nature of the 7-1a domain suggests that it is chiefly responsible for the antigenic properties that define the G3 serotype. Indeed, the 7-1A domain contains the antigenic epitope for the VP7-specific monoclonal antibody (MAb 159) that is used as a reference standard in identifying G3 RVs. The limited number of differences between the RRV and SA11-H96 VP7 proteins suggests that their antigenicity should be quite similar, which has been demonstrated previously (43). Whether the increased number of differences between the G3 VP7 proteins of RRV and TUCH measurably decreases the production of cross-reacting neutralizing antibodies in infected animals is unclear.
Few P and P RVs have been identified, ruling out a comparative analysis of their VP4 proteins. On the other hand, several P RVs have been described, most of which, like the RRV strain, are G3 viruses that have been isolated from feline, canine, or human hosts (Table (Table1).1). Sequence alignments reveal that the RRV P VP4 protein is most closely related to the P VP4 protein of the goat RV isolate GRV (Fig. (Fig.44 and data not shown). However, these proteins are somewhat evolutionarily distant, as illustrated by the fact that their VP8* components differ in 13 surface-exposed residues (Fig. (Fig.5C)5C) and have an amino acid identity of 92% (Table (Table3).3). The number of surface-exposed residues that differ increases to more than 26 when RRV VP8* is compared to the VP8* proteins to some other G3P viruses (e.g., K9, Cat97, HCR3A, A79-10, CU-1, and RO1845) (Table (Table1,1, Fig. Fig.5C,5C, and data not shown). Parallel with this increase, the amino acid identity of RRV VP8* with VP8* proteins of other G3P viruses falls to below 87% (Table (Table3).3). Thus, although RRV and several other viruses (e.g., GRV, K9, Cat97, HCR3A, A79-10, CU-1, and RO1845) have been assigned P genotypes based on their nucleotide sequences, the VP8* proteins of these viruses show significant variation in amino acid sequence. Since some of the variation maps to the antigenic domains of VP8*, it is possible that P viruses may display antigenic differences that influence the specificities of neutralization antibodies formed in infected animals (Fig. (Fig.44 and and5C5C).
Interestingly, the amino acid composition of the VP8* sialic acid binding sites varies markedly among G3P RVs, with those of the caprine GRV and human CMH22 strains exhibiting greatest similarity to the RRV sialic acid binding site (Table (Table4).4). Unexpectedly, the GRV and CMH22 sialic acid binding sites were found to be identical to that of the TUCH P VP8* and thus are even more closely related than the sialic acid binding sites of GRV and CMH222 are with the TUCH site. The GRV and CMH222 sialic acid binding sites also show notable similarity with that of SA11-H96. The significance of the amino acid variation observed for the sialic acid binding is not clear but may be a determinant affecting host range and cell tropism. However, our analysis establishes that based on amino acid composition, there is no single type of sialic acid binding site for simian RVs.
A principal goal in the isolation and characterization of NHPs RVs has been the desire to identify surrogates of human RVs that in large animal models will mimic the growth, pathology, and immune responses that are characteristic of human infections. To date, only five simian RVs have been described (PTRV, RRV, SA11-H96, TUCH, and YK-1); all have been used to one extent or another to probe the biology of the virus. Of these viruses, only the SA11-H96 isolate has been fully sequenced (41). To further understand the genetic origin and relatedness of the simian RVs to each other and to human and other animal RVs, the complete genomes of three additional strains—PTRV, RRV, and TUCH—were determined and phylogenetically analyzed.
Analyses of PTRV, isolated from a colony of pig-tailed macaques at the University of Washington, revealed that its genome was entirely of bovine/artiodactyl origin. Serological analyses showed that this virus was endemic in the colony from at least 1987 until 1994 (9). This implies that an ancestor of PTRV crossed the host species barrier successfully and was able to spread and persist in the new NHP host without undergoing reassortment. The successful introduction of the PTRV ancestor into the simian host illustrates once again that the species barrier for RVs is far from absolute. Notably, an earlier analysis of human RVs belonging genogroup 2 (DS-1-like) suggests that these viruses have a common origin with Bovidae RVs. In addition, the unusual human G6P strains are suspected of originating from a virus infecting the Bovidae or other member of the Artiodactyla (2, 17, 19).
RRV was one of the first animal RVs used in the development of human vaccine candidates (48). By generating reassortants with human RVs strains, a tetravalent RRV-based (serotype G1, G2, G3, and G4) vaccine (Rotashield; Wyeth Laboratories) was produced that was licensed and commercially used in infants and young children in the United States in 1998 to 1999 (7). Unfortunately, the vaccine was withdrawn from the market due to an association with an increased risk of intussusception in vaccinees (3, 27, 28). Our genomic analyses suggest that RRV may originate from the reassortment of a feline/canine ancestral RV with a Bovidae RV. The genes of RRV encoding VP1 (genotype R2), VP6 (genotype I2), and NSP2 (genotype N2) are closely related to those of Bovidae RVs (Fig. 1A and C); thus, these genes were probably introduced into the RRV genome by reassortment relatively shortly before the isolation of the virus in 1975. However, the genes of RRV and the feline/canine RVs that share the same genotypes—VP3 (M3), VP4 (P), VP7 (G3), NSP1 (A9), NSP3 (T3), NSP4 (E3), and NSP5 (H6)—are phylogenetically rather distantly related. Thus, the interspecies transmission of RRV from a feline or canine host to a NHP was most likely not a recent event.
The TUCH strain has been characterized in an attempt to develop a reliable NHP model for studies on RV protection. Pathogen-free macaques, which were challenged with TUCH, remained clinically normal, but shed large quantities of RV antigen in their feces, which resolved by the end of a 14-day observation (25). This model has been used to study extraintestinal spread of RVs, and the CD4+ and CD8+ lymphocyte responses of macaques after RV challenge (40, 51). Genomics analysis surprisingly revealed that the majority of the genes (7 of 11) of the TUCH strain are shared with feline/canine RVs and RRV (Table (Table11 and Fig. Fig.1).1). These data suggest that TUCH is also a descendant of the hypothetical ancestor from which the feline/canine and RRV strains derive. Most likely, one or more reassortment events have taken place starting from the hypothetical ancestor, replacing the NSP2, VP4, and VP6 gene segments, yielding the TUCH strain. The NSP2 gene segment of TUCH (genotype N1) may have originated from a Wa-like human RV, whereas the origin of its VP4 (P) and VP6 (I9) remains unknown. As is the case for RRV, the phylogenetic relationships between TUCH on one hand, and RRV and the feline/canine RRV strains on the other hand, is rather distant, suggesting that the interspecies transmission from a feline/canine host to a simian host must have happened a long time ago. Indeed, this long time period would explain why so many amino acid differences are noted between the VP7 proteins of RRV and TUCH. Interestingly, RRV can cause disease in macaques, whereas TUCH cannot, raising the possibility that one of the unique genes (e.g., the P VP4) of RRV represents a virulence determinant. Although TUCH does not cause disease in macaques, it does grow to high titers and is capable of spread. Thus, the determinants of the virus that dictate productive growth in the animal are not necessarily the same determinants affecting disease characteristics.
No common genotype constellation has been established for simian RVs, which up to this point, can be blamed on the lack of comprehensive sequence data. Our analysis indicates that the PTRV strain is unlikely to represent a typical simian genotype constellation, since it most likely originated after the interspecies transmission of a Bovidae or artiodactyl RV to a colony of pig-tailed macaques prior to 1987, where it was able to spread, evolve, and sustain itself. The genetic constellations of the RRV and TUCH are also unlikely to represent typical simian RVs genogroups, since these strains appear to have undergone reassortment with bovine, human, and/or unknown RV strains relatively recently. However, based on the observed genetic divergence, the proposed interspecies transmission of RRV and TUCH ancestors from a canine/feline host to a simian host, or vice versa, must have occurred quite a while ago. In its new host, the virus must not only have survived but also evolved. If macaques were the host species in which this ancestor virus established itself and spread, then the genomes of RRV and TUCH could be representatives of a simian genogroup, in which a few gene segments were more recently replaced by reassortment. This hypothesis is supported by the observation that RRV and TUCH (with their related genotype constellations) were independently isolated from different states in the United States (California and Louisiana, respectively). Additional support for this hypothesis comes from the RV strain YK-1, isolated from an immunodeficient pig-tailed macaque in the Yerkes National Primate Research Center, Emory University in Atlanta, which possesses identical genotypes as RRV for VP7, VP4, and NSP4 (49). However, based on current information, SA11-H96 is the most likely representative of a typical simian RV genogroup, since none of the gene segments of SA11 are closely related to other known RV strains.
In summary, interspecies transmission and reassortment has contributed significantly to the emergence of the simian RV strains PTRV, RRV, and TUCH. Epidemiological and sequence data of simian RVs collected from animals living in the wild will be crucial to further elucidate the ancestry of these viruses. None of the simian RVs characterized thus far are closely related to the strains of human RVs (Wa-like genogroup 1 and DS-1-like genogroup 2) that are responsible for the vast majority of cases of rotaviral diarrheal disease occurring in infants and young children. The attractiveness of simian RVs as model agents of human disease would be greatly improved if future searches were to identify simian genogroup 1 or 2 viruses that emulated human disease in juvenile animals.
We are grateful to Sarah McDonald for critical revisions of the manuscript.
J.M. was supported by a postdoctoral fellowship from the Fund for Scientific Research (FWO) Flanders. J.T.P., Z.F.T., D.C., and H.Y. were supported by the Intramural Research Program of the National Institutes of Health (NIH), National Institute of Allergy and Infectious Diseases. The sequencing of PTRV was partially supported by startup funds to L.Y. from Virginia Polytechnic Institute and State University. K.S. was supported by NIH R21RR024871 grant awarded to the National Center for Research Resources.
Published ahead of print on 25 November 2009.