Field evaluation of a gag/env heteroduplex mobility assay for genetic subtyping of small-ruminant lentiviruses

Small-ruminant lentiviruses (SRLVs) are a heterogeneous group comprising visna/maedi virus (VMV) and caprine arthritis encephalitis virus (CAEV), which infect sheep and goats. Like other lentiviruses, SRLVs are characterized by extensive genetic diversity. Nucleotide sequence databases of the gag, pol and env genes of viruses isolated throughout the world have allowed phylogenetic analyses to be performed (Rolland et al., 2002; Shah et al., 2004a; Zanoni, 1998). On the basis of these analyses, SRLV strains can be classified in two major phylogenetic groups, A (VMV-like viruses) and B (CAEV-like viruses), which can be further divided into different subtypes. To date, group A contains nine subtypes (A1–A9) and group B contains two subtypes (B1 and B2). A limited number of highly divergent SRLV strains recently identified in several European countries have been included in three additional SRLV groups, named C–E (Gjerset et al., 2006; Grego et al., 2007; Reina et al., 2006; Shah et al., 2004a). Whilst VMV and CAEV were originally considered to be specific to sheep and goats, respectively, most subtypes have been isolated from both host species (Gjerset et al., 2007; Leroux et al., 1997; Reina et al., 2006; Rolland et al., 2002; Shah et al., 2004a; Zanoni, 1998), suggesting that cross-species infection may occur under field conditions. This presumption has been confirmed by phylogenetic analyses of SRLV sequences isolated among sheep and goats from mixed flocks, providing direct evidence for natural interspecies transmission of subtype A4 in both directions (Shah et al., 2004b), as well as subtype B1 and group C from goats to sheep (Gjerset et al., 2007; Pisoni et al., 2005).

The development of an efficacious diagnosis of SRLV infection remains one of the biggest challenges in the control and eradication programmes that have been implemented in several European countries. Major obstacles to diagnosis development include delayed seroconversion and long-term-seronegative-infected animals, low viral loads impairing PCR tests and SRLV variability. As epidemiological studies have revealed the co-circulation of highly divergent genetic subtypes in restricted geographical regions (Germain & Valas, 2006; Gjerset et al., 2007; Grego et al., 2002, 2007; Leroux et al., 1997; Reina et al., 2006; Shah et al., 2004a), up-to-date information on the prevalent SRLV subtypes circulating in each country is of crucial importance to improve the efficiency of the diagnosis tests. In addition, the spread of multiple subtypes in a specific geographical region, combined with the ability of SRLV strains to cross the species barrier, may give rise to dual infection and recombination. However, only one case of dual infection and viral recombination has been reported recently in naturally infected goats (Pisoni et al., 2007), and whether interspecies transmission contributes to the evolution of genetic and biological properties of SRLV isolates is currently unknown.

Current tools for the characterization of SRLV diversity include sequencing of short genome segments, typically in the gag/pol and env regions, and, more rarely, full-genome sequencing. However, DNA sequencing methodologies are unsuitable for genetic analyses performed on large numbers of samples, limiting the detection of viral variants that are different from the local predominant subtypes. Large-scale, non-sequencing methods are thus required. Recently, the heteroduplex mobility assay (HMA) methodology has been adapted as a high-throughput genotyping tool for the study of SRLV genetic diversity (Germain & Valas, 2006). This method was originally designed for the V1–V2 region of the env gene to discriminate among SRLV groups. In the present study, the HMA was validated as an accurate approach to estimate the genetic divergence among intra-group SRLV env sequences. Moreover, the HMA was adapted for the SRLV gag region spanning the nucleocapsid-coding sequence. French SRLV field isolates among sheep and goats from mixed flocks were investigated using HMA analysis performed in parallel on both the env and gag genes. The results of gag/env HMA subtyping were confirmed by genetic and phylogenetic analyses.

Samples.
A total of 75 blood samples from SRLV naturally infected animals, 42 sheep and 33 goats, selected for their positive SRLV serological status using a commercially available ELISA (Chekit CAEV/MVV; Behring), was examined in this study. These animals came from five mixed flocks (A–E) located in three regions of southern France (Ariège, Corse and Lozère). Fifteen of the ovine lentiviruses have been partially described previously (Germain & Valas, 2006). Blood samples were collected into EDTA-coated tubes and genomic DNA was isolated using a DNeasy Tissue kit (Qiagen), according to the manufacturer's instructions.

Nested PCR.
A 394 bp fragment spanning the V1–V2 variable region of the env gene and a 238 bp fragment spanning the nucleocapsid-coding region of the gag gene were amplified by nested PCR. The gag and env fragments corresponded to nt 1605–1842 and nt 6352–6745 in the Cork genome, respectively (Saltarelli et al., 1990). PCR amplification of the gag fragment was performed using the primer pairs P14/P15 in the first round and P40/P41 in the second round (Shah et al., 2004a). The PCR primer pairs specific for the env gene were A5/A31 in the first round and A51/B31 in the second round (Germain & Valas, 2006). First-round amplifications were performed using 500 ng genomic DNA in a 50 µl reaction containing 2 mM MgCl₂, 0.2 mM each dNTP, 300 nM each primer and 2.5 U Taq DNA polymerase (iTaq; Bio-Rad). The PCR cycling conditions were 3 min at 95 °C, followed by 35 cycles of 40 s at 94 °C, 60 s at 57 °C and 60 s at 72 °C, with a final extension of 4 min at 72 °C. Second-round PCRs were performed with 2 µl of the first-round reaction used as template. The cycling conditions were the same for amplification of the gag fragment, but the hybridization step for amplification of the env fragment was carried out for 50 s at 60 °C. All PCRs were carried out in a Bio-Rad Icycler thermocycler.

HMA.
Heteroduplex formation reactions were carried out as described previously (Germain & Valas, 2006). The relative mobilities of the heteroduplexes were calculated by dividing the migration distance of the heteroduplex bands (or the midpoint between two bands) by the migration distance of the corresponding homoduplex band. The same experimental conditions were carried out for gag and env HMA. A panel of 33 reference SRLV isolates was used in the env HMA analysis, comprising the prototype strains of subtypes A1 (Islandic strain K1514), B1 (American strain Cork) and B2 (Switzerland strain 5720), three French caprine lentiviruses (strains 680, 032 and 786) representative of subtype B1, and 27 French ovine lentiviruses belonging to either subtype B1 (six strains) or subtype B2 (21 strains), which have recently been characterized in our laboratory (Germain & Valas, 2006).

DNA sequence analysis.
Specific PCR products were purified from agarose gels using a QIAquick Gel Extraction kit (Qiagen). Sequencing of both strands using a standard ABI BigDye terminator reaction (MilleGen) was performed on uncloned PCR fragments or on PCR products cloned into the linear pDrive vector using a cloning kit (Qiagen). Nucleotide sequences were aligned for subsequent phylogenetic analysis using CLUSTAL W version 1.8 (Thompson et al., 1994). Manual rearrangements of the alignments, including gap exclusion and length adjustment, were applied to achieve optimal results. Pairwise genetic distances were calculated using MEGA version 3.1 with the Tamura–Nei substitution model. Phylogenetic construction was carried out using the neighbour-joining method (Saitou & Nei, 1987) implemented in MEGA with the Tamura–Nei gamma distance (Tamura & Nei, 1993). The statistical confidence of the topologies was assessed using 1000 bootstrap replicates (Felsenstein, 1985).

Subtype reference strains.
The GenBank accession numbers of the previously published SRLV sequences are as follows: Cork (M33677[GenBank] ), 680 (AJ400718[GenBank] ), 032 (AJ400720[GenBank] ), 786 (AJ400721[GenBank] ), 4668 (AY445885[GenBank] ), S93 (AF338226[GenBank] ), 85/34 (U64439), WLC1 (AY362038[GenBank] ), SA-OMVV (M31646[GenBank] ), K1514 (M60609[GenBank] ), EV-1 (S51392), 5560 (AY454175[GenBank] ), 5561 (AY454176[GenBank] ), 5720 (AY454218[GenBank] and AY842774[GenBank] ), A-S76 (AY842749[GenBank] ), A-S450 (AY842748[GenBank] ), A-S2013 (AY842750[GenBank] ), B-S6164 (AY842757[GenBank] ), B-3056 (DQ149845[GenBank] ), C-S192 (AY842753[GenBank] ), C-S269 (AY842754[GenBank] ), C-S3034 (AY842755[GenBank] ), D-S536 (AY842738[GenBank] ), D-S9103 (AY842740[GenBank] ), D-S553 (AY842739[GenBank] ), E-S46 (AY842763[GenBank] ), E-S26 (AY842761[GenBank] ) and E-S218 (AY842762[GenBank] ), 5692 (AY454208[GenBank] ) and CA1GA (AF322109[GenBank] ).

Evaluation of the env HMA
To determine the genetic divergence of SRLV samples and to discriminate subtypes B1 and B2 from each other, HMA was first performed on heteroduplexes formed by pairwise combinations of V1–V2 env fragments from 30 French SRLV sequences belonging to subtypes B1 and B2 (Germain & Valas, 2006). Heteroduplexes were detected when the degree of divergence exceeded about 6 %, which corresponds to the upper range of the viral quasispecies diversity found within individual hosts (Herrmann et al., 2004; Leroux et al., 1997). A linear equation between the two variables was observed for distances ranging from 6 to 24 %, which covers the range of diversity of group B. Analysis of the relative mobilities (RMs) of heteroduplexes revealed that the intra-subtype heteroduplexes exhibited an RM>0.74, whereas the inter-subtype heteroduplexes showed an RM<0.75, indicating that this discrimination zone can be defined as a diagnostic window for the subtype identification of SRLV group B. Taken together, these results demonstrated the good predictability and validity of the env HMA as an SRLV genetic screening methodology.

Evaluation of the gag HMA
A 238 bp fragment spanning the coding region of the p14 nucleocapsid protein was selected for HMA analysis. The choice of this gag region was guided by (i) the availability of sequences representative of the different SRLV subtypes, (ii) the performance of the PCR primers allowing detection of all subtypes (Shah et al., 2004a) and (iii) the variation in size of the PCR products due to group- and subtype-specific deletions. Indeed, alignment of reference sequences of SRLV subtypes revealed that variable domains with small deletions and insertions surrounded the highly conserved central domain of the p14 protein containing the zinc finger motifs (Fig. 1a). Subtype reference sequences of CAEV-like viruses differed from those of VMV-like viruses by deletions in both terminal domains, except for subtype B1 lacking the deletion in the C-terminal domain. These genetic features between subtypes were confirmed by analysis of all available SRLV sequences retrieved from databases, showing that all CAEV-like viruses (35/35) contained a deletion in the N-terminal domain of the p14 protein, compared with only 1/20 VMV-like virus. Given that subtype B2 was represented by only one gag sequence, the p14-coding sequence of nine French ovine viruses that have previously been classified into subtype B2 on the basis of V1–V2 env sequences (Germain & Valas, 2006) was determined. As shown in Fig. 1(b), eight French isolates contained a deletion in the N-terminal domain, whilst the remaining strain, E-S218, showed no deletion. However, phylogenetic analysis revealed that strain E-S218 belonged to group A on the basis of the gag sequence (see Fig. 4b), indicating a possible recombinant origin between SRLV groups A and B. In addition, all viruses classified in subtype B2 (9/9) contained a deletion in the C-terminal domain of the p14 protein, whilst 88 % (29/33) of subtype B1 viruses showed no deletion within this domain.

(39K): Fig. 1. Alignment of amino acid sequences spanning the p14 nucleocapsid-coding region used in the gag HMA. (a) Alignment of reference sequences of SRLV subtypes of groups A, B and C. Reference sequences correspond to strains EV-1 (A1), 85/34 (A2), 5692 (A3), 4668 (A4), 5560 (A5), Cork (B1), 5720 (B2) and CA1GA (C). (b) The sequences of nine previously characterized French ovine viruses were aligned with the reference sequences of subtypes B1 and B2. Five residues in the N-terminal domain and 15 residues in the C-terminal domain are missing in the p14 nucleocapsid protein. Deletions are indicated by dashes. A consensus sequence (Cons) is shown below each alignment. Zinc finger motifs are underlined.

(31K): Fig. 4. Phylogenetic trees based on the SRLV env gene V1–V2 region (a) and gag gene p14 region (b). Nucleotide sequences of ovine and caprine French isolates were compared with reference strains of SRLV subtypes of groups A and B. Trees were constructed using the neighbour-joining method and statistically evaluated by bootstrap analysis. Bootstrap values greater than 90 % are included at all leading nodes. (a) French ovine isolates for which env sequences have been determined previously are indicated in italics. French strains used as reference standards in the env HMA are underlined. (b) Subtypes of group A strains are indicated in parentheses. To identify clearly the origin of French isolates, the identification of the flocks (A–E) and the prefixes S and G designating sheep and goat species, respectively, were added to the identification numbers of the SRLV strains. Bars, 0.05 nucleotide substitutions per site.

The gag HMA was evaluated by pairwise combinations of the amplicons obtained from the reference sequences of subtypes B1 and B2, and group A. As shown in Fig. 2(a), intra-subtype combinations shared similar electrophoretic patterns, consisting of the homoduplex band at the bottom of the gel and fast-migrating heteroduplex bands, except that the homoduplex bands of subtype B2 strains migrated much faster than those of subtype B1 strains due to the deletion in the C-terminal domain of the p14 protein. As expected, inter-subtype comparisons including one sequence of subtype B2 generated two distinct homoduplex bands and slow-migrating heteroduplex bands. Finally, the inter-group heteroduplexes exhibited very low mobilities, resulting from the combined effects of deletions and base substitutions, compared with those formed from intra-subtype combinations (Fig. 2b). These results indicated that the gag HMA was suitable to discriminate between SRLV groups as well as to discriminate subtypes B1 and B2 from each other.

(33K): Fig. 2. Analysis of gag HMA profiles. (a) Pairwise combinations of DNA fragments spanning the p14 nucleocapsid-coding region of the gag gene were generated between SRLV reference subtypes, as indicated. (b) Box plots of heteroduplex mobilities obtained with intra-subtype (left panel), inter-subtype (middle panel) or inter-group (right panel) comparisons of SRLV reference sequences. Values within the second and third quartiles are boxed and medians are indicated. Whiskers (vertical bars) indicate values up to 1.5 box lengths beyond the box. Values outside the box plots are indicated by open circles.

Validation of the gag/env HMA on field isolates
SRLV subtype screening using the gag/env HMA was validated on blood samples from naturally infected sheep and goats that were seropositive for CAEV antigens by ELISA. Five French flocks were investigated, four mixed flocks (A, B, C and E) in which sheep and goats were kept together with permanent contact between adult animals, and one mixed flock (D) in which sheep and goats were reared in juxtaposed pens of a large barn without permanent contact between the two animal species. The V1–V2 env fragments were amplified by PCR from blood samples of 42 sheep and 33 goats. Seropositive goats from flock E gave negative results by PCR and were no longer considered in this study. In order to assign the SRLV env subtype to the unknown samples by HMA analysis, PCR products were tested individually with analogous fragments amplified from a panel of plasmids containing the V1–V2 env fragment of previously characterized ovine viruses originating from these flocks. All representative electrophoretic patterns of the env HMA are shown in Fig. 3. All electrophoretic patterns produced heteroduplexes migrating below the single-stranded DNA band, indicating that the 75 env sequences examined in this study belonged to the SRLV group B. All sheep samples from flocks A and D showed either no heteroduplex or very fast-migrating heteroduplexes (RM>0.75) when tested with their reference sample, supporting a molecular homogeneity and their unambiguous classification in subtype B2. Sheep samples from flocks B and C exhibited heteroduplexes with either very fast (RM>0.75) or slow (RM<0.75) electrophoretic mobilities when tested with their reference sample, indicating the presence of both subtypes B1 and B2. Sheep samples from flock E produced three distinct electrophoretic patterns when tested with their reference sample, one pattern with no heteroduplexes and two patterns with heteroduplexes migrating either just below or above the subtype discrimination zone (RM=0.75). As the sample used as the reference standard belonged to the subtype B2, HMA analysis indicated the presence of two variants of subtype B2 and one variant of subtype B1 in sheep from flock E.

(72K): Fig. 3. Genetic subtyping of French SRLV field isolates by env HMA. PCR-amplified fragments spanning the V1–V2 region of the env gene of ovine and caprine SRLV field strains from five flocks (A–E) were mixed with the corresponding fragments of reference strains, as indicated. Each reference strain was tested with the prototype CAEV (Cork) and VMV (K1514) strains to produce HMA patterns representative of intra- and inter-group comparisons, respectively. The subtypes of reference strains used in the HMA are indicated in parentheses. For each panel, the dotted line represents the electrophoretic migration of heteroduplexes with a relative mobility of 75 %. For each flock, the number (n) of samples showing a similar electrophoretic pattern is indicated below each representative HMA profile.

HMA analysis of goat samples from flock A revealed two distinct electrophoretic patterns with either very fast- or slow-migrating heteroduplexes, indicating the presence of both subtypes B1 and B2. In contrast, goat samples from flocks B–D produced heteroduplexes with comparable electrophoretic mobilities and were classified into subtype B1, according to the subtype assignment of their respective sheep reference sample. The genetic homogeneity of the virus population in goats from these three flocks was confirmed by pairwise comparisons between goat samples, showing HMA patterns with fast-migrating heteroduplexes (data not shown). In flocks A and B, some goat samples appeared to be genetically highly related to the sheep reference sample, as the HMA pattern produced no heteroduplex, strongly suggesting the occurrence of inter-species transmission of subtypes B2 and B1 in flocks A and B, respectively. One goat sample from flock A and nine goat samples from flock B produced no heteroduplex when tested with their reference sheep sample. Similarly, one sheep sample and seven goat samples from flock C produced heteroduplexes of similar electrophoretic mobilities when tested with the reference sheep sample, whereas no heteroduplex was formed with pairwise combinations of these unknown samples (data not shown). As HMA patterns generated by the combination of genetically highly related sequences produced no heteroduplex, these results strongly indicated the occurrence of inter-species transmission of subtypes B2 (flock A) and B1 (flocks B and C).

A positive gag HMA PCR product was obtained for 73/75 (97.3 %) of the analysed blood samples. PCR products were tested individually by HMA using as the reference standard the Cork CAEV strain representative of subtype B1. As described above, the samples producing HMA patterns with one homoduplex band and two subsets of heteroduplexes with fast or moderate mobilities were assigned to subtype B1, whilst those forming HMA patterns with one homoduplex band and slow-migrating heteroduplexes were assigned to group A. The samples producing HMA patterns with two homoduplex bands were classified into subtype B2. The SRLV gag subtyping results are presented in Table 1. Complete correlation was observed between the gag and env HMA subtyping for 67/73 samples (91.8 %). Five goat samples, four in flock B and one in flock D, gave discordant results by the gag/env HMA. These samples were classified into subtypes B1 and B2 by the env and gag HMA, respectively. Of 13 sheep samples analysed in flock E, 12 were classified into subtype B2 by the gag/env HMA, whereas the remaining sample was assigned to subtype B2 by the env HMA and group A by the gag HMA.

Table 1. HMA gag/env genetic subtyping of French SRLV field isolates

Phylogenetic analysis of SRLV env sequences
In order to confirm the subtype classification inferred by the gag/env HMA, 40 PCR products (17 env sequences and 23 gag sequences) of selected sheep and goat samples representative of the different HMA patterns were sequenced and phylogenetically analysed. As shown in Fig. 4(a), the 17 env sequences segregated together with the 14 previously reported env sequences originating from the same flocks into two large clades corresponding to subtypes B1 and B2. All samples showing heteroduplexes with RM>0.75 clustered with the subtype used as the reference standard. On the other hand, the samples from flocks A–D showing slow-migrating heteroduplexes (RM<0.75) clustered with a different subtype from that used as the reference standard. In contrast, sample S26 of flock E producing heteroduplexes with RM<0.75 was classified into subtype B2 together with sample S46 used as the reference standard in the env HMA, although both sequences clustered into very distant branches. Thus, overall, 30/31 samples (96.7 %) analysed gave concordant subtype results by env HMA and phylogenetic analysis. Moreover, the phylogenetic segregation of isolates within each subtype correlated strongly with the electrophoretic mobility of heteroduplexes. The samples showing no heteroduplexes clustered together with the reference strain used in HMA, whilst those showing fast-migrating heteroduplexes formed a distinct branch within the same cluster. In contrast, the samples showing heteroduplexes migrating just below the subtype discrimination zone formed a separate cluster. Phylogenetic analysis performed on the p14 gag fragment showed that the 23 French gag sequences segregated into three large clades, which were well supported by bootstrap analysis (Fig. 4b). These clades referred to group A and subtypes B1 and B2. The clustering of all samples correlated perfectly with the genetic subtyping inferred by the gag HMA, including the four samples (B-G6077, B-G4120, D-G602 and E-S218) chosen for sequencing and showing discordant results between the gag and env HMA. In the past few years, increasing data from phylogenetic analyses of viral sequences from many countries throughout the world have indicated that interspecies transmission may contribute to the genetic diversity and evolution of SRLV. However, the heterogeneity of SRLV has been studied mostly by sequencing of partial regions derived from only one gene, precluding the characterization of a large number of isolates and the identification of unusual viruses. The HMA technique has recently been developed as a large-scale, alternative subtyping method to discriminate SRLV groups on the basis of env sequences (Germain & Valas, 2006). Here, the env HMA was evaluated to predict the genetic divergence of SRLV strains and to discriminate between subtypes B1 and B2. A linear relationship between the electrophoretic mobility of heteroduplexes and genetic divergence was observed for distances ranging from 6 to 24 %, which covers the range of diversity of group B. Moreover, a discrimination zone was identified in the heteroduplex relative mobility pattern, which clearly distinguished between subtypes B1 and B2. These results demonstrated that the env HMA is a reliable tool for both the identification of SRLV group B strains and the accurate prediction of their subtype classification. Furthermore, the sensitivity of this method (6 % of divergence) precluded the detection of the viral quasispecies under our experimental conditions, thus allowing direct identification of cross-species infection.

The env HMA has two main limitations for the genetic subtyping of SRLV. Firstly, the env gene exhibits the highest variability within the lentiviral genome, impeding the PCR amplification of some SRLV strains. Secondly, it cannot differentiate alone between pure subtypes and recombinant forms. In order to circumvent these problems, we developed a gag HMA based on a 238 bp fragment spanning the p14 nucleocapsid-coding region. The PCR primers used for the amplification of this gag fragment have been used previously for successful amplification of SRLV sequences representative of all subtypes (Shah et al., 2004a). Moreover, sequence alignment revealed both group- and subtype-specific features in the p14-coding region that could be used as markers in the HMA. Indeed, all of the CAEV-like strains showed a short amino acid deletion in the N-terminal domain of the protein compared with all of the VMV-like strains, except for the prototype VMV K1514 strain. In addition, most of the reference sequences of subtype B1 differed from those of subtype B2, which contained an additional amino acid deletion in the C-terminal domain of the protein. This distinct feature between subtypes B1 and B2 was confirmed by sequence determination of the corresponding fragments of eight additional pure subtype B2 strains. These deletions, combined with base substitutions, allowed the gag HMA to assign subtypes B1 and B2 correctly.

The specificity of the gag/env HMA was verified subsequently using SRLV field isolates originating from seropositive sheep and goats from mixed flocks in France. The gag HMA subtype results correlated fully with sequence and phylogenetic analyses. Concordant results between the env HMA subtyping and phylogenetic analysis was observed for 30/31 (96.7 %) samples, one sample being classified into subtype B1 by the env HMA and into subtype B2 by sequencing. Overall, the HMA subtyping results revealed a predominance of subtypes B1 and B2 in the French mixed flocks investigated in this study. However, the relative distribution of these subtypes differs significantly in sheep and goats. Subtype B1 was found predominantly in goats, whereas subtype B2 was prevalent in sheep. Similar observations have been described for SRLV epidemiology in other European countries, including Italy and Switzerland (Grego et al., 2002; Shah et al., 2004a), suggesting that some SRLV subtypes may be better adapted to spread in goats and others in sheep. The HMA analysis also provided direct evidence for interspecies transmission of subtypes B1 and B2 in three out of four flocks for which PCR products were obtained from both sheep and goat samples, indicating that natural interspecies transmissions occur regularly. This was confirmed by DNA analysis showing that sheep and goat samples producing no heteroduplex when tested pairwise by HMA were phylogenetically closely related and displayed extremely low V1–V2 sequence divergences, ranging from 1.2 to 1.8 %. For comparison, these genetic distances are of the same extent as the viral quasispecies diversity within individual hosts (Herrmann et al., 2004; Leroux et al., 1997). To our knowledge, this is the first report of env sequences representative of subtype B2 in goats, confirming that no SRLV subtype is particularly prone to crossing the species barrier.

Comparison of gag and env HMA subtyping results revealed a different classification of viral sequences from the same sample in five goats from flock B and one sheep from flock E. The goat samples were classified as subtype B1 by the env HMA and subtype B2 by the gag HMA. The sheep sample was classified as subtype B2 by env HMA and group A by gag HMA. Phylogenetic analysis performed on both gag and env sequences from the sheep sample and three out of five goat samples confirmed the subtype assignment inferred by HMA. This is the first time, to our knowledge, that sequences from different genes and classified into different SRLV subtypes have been found within the same host in sheep and goats. Thus, these data indicate the relevance of subtyping two different gene fragments. The different phylogenetic classification of gag and env sequences obtained from the same sample strongly suggests the occurrence of inter-subtype recombination. Dual infection with different subtypes rather than infection with recombinant forms is unlikely, as both gag and env sequences were concomitantly amplified from all samples from these two flocks, excluding the possibility of PCR amplification failure due to genetic variability of viruses. In addition, all samples gave no heteroduplex in the gag/env HMA when tested with themselves, indicating the presence of only one major variant in each sample (data not shown). Overall, these results suggest that dual infection rarely occurs, although such events can lead to inter-subtype recombination in sheep and goats. This is consistent with a recent study showing that only two goats from a commercial flock of about 300 goats that were seropositive for SRLV reacted strongly to both group A- and group B-specific antigens, one of them being infected with a recombinant virus carrying a mosaic A/B env sequence (Pisoni et al., 2007). The situation is quite different for other lentiviruses, particularly human and feline immunodeficiency viruses, for which dual infections and recombinant forms have been detected at an appreciable frequency (Bachmann et al., 1997; Quinones-Mateu & Arts, 1999; Robertson et al., 1995; Troyer et al., 2004). However, the frequency of dual infection in sheep and goats may be underestimated, as the viral load in peripheral blood is usually low and is subjected to temporal fluctuations (Klevjer-Anderson et al., 1984), limiting the simultaneous detection of divergent SRLV strains by PCR from blood samples.

In addition to the high incidence of genetic diversity and evolution of SRLV, recombination between SRLV groups A and B may yield viruses with distinct biological properties. Indeed, prototypic VMV and CAEV strains differ in their cytopathic phenotype and have been classified as lytic and non-lytic viruses, respectively (Querat et al., 1984). These SRLV strains also differ in their in vitro species tropism. In particular, the Icelandic VMV strain K1514 can use receptors for entry and/or cell-to-cell fusion in human cells (Bruett & Clements, 2001; Hötzel & Cheevers, 2001), whilst the lack of functional receptors is the only barrier that prevents CAEV replication in these cells (Mselli-Lakhal et al., 2000), suggesting that recombinant virus between highly divergent SRLV subtypes may achieve the complete virus life cycle in cells other than those of small ruminants. Importantly, SRLV field isolates belonging to subtypes including the prototypic CAEV and VMV strains are still circulating in European countries (Barros et al., 2004; Grego et al., 2002; Leroux et al., 1997; Rolland et al., 2002; Shah et al., 2004a). Up-to-date information on circulating SRLV isolates may thus be of crucial importance. In this regard, HMAs may offer an important contribution to improve the molecular biology-based surveillance of SRLV, and the accurate subtype prediction for both the gag and env region using this assay may be extremely helpful to estimate the frequency of inter-subtype genetic recombination.

This work was supported by grants from AFSSA and from the Etablissements Publics Régionaux de Poitou-Charentes and the Conseil Général des Deux-Sèvres. We are most grateful to Christian Baudry for helpful discussion and Thierry Vidard for technical assistance.