Genomic organization, sequence divergence, and recombination of feline immunodeficiency virus from lions in the wild

Background Feline immunodeficiency virus (FIV) naturally infects multiple species of cat and is related to human immunodeficiency virus in humans. FIV infection causes AIDS-like disease and mortality in the domestic cat (Felis catus) and serves as a natural model for HIV infection in humans. In African lions (Panthera leo) and other exotic felid species, disease etiology introduced by FIV infection are less clear, but recent studies indicate that FIV causes moderate to severe CD4 depletion. Results In this study, comparative genomic methods are used to evaluate the full proviral genome of two geographically distinct FIV subtypes isolated from free-ranging lions. Genome organization of FIVPle subtype B (9891 bp) from lions in the Serengeti National Park in Tanzania and FIVPle subtype E (9899 bp) isolated from lions in the Okavango Delta in Botswana, both resemble FIV genome sequence from puma, Pallas cat and domestic cat across 5' LTR, gag, pol, vif, orfA, env, rev and 3'LTR regions. Comparative analyses of available full-length FIV consisting of subtypes A, B and C from FIVFca, Pallas cat FIVOma and two puma FIVPco subtypes A and B recapitulate the species-specific monophyly of FIV marked by high levels of genetic diversity both within and between species. Across all FIVPle gene regions except env, lion subtypes B and E are monophyletic, and marginally more similar to Pallas cat FIVOma than to other FIV. Sequence analyses indicate the SU and TM regions of env vary substantially between subtypes, with FIVPle subtype E more related to domestic cat FIVFca than to FIVPle subtype B and FIVOma likely reflecting recombination between strains in the wild. Conclusion This study demonstrates the necessity of whole-genome analysis to complement population/gene-based studies, which are of limited utility in uncovering complex events such as recombination that may lead to functional differences in virulence and pathogenicity. These full-length lion lentiviruses are integral to the advancement of comparative genomics of human pathogens, as well as emerging disease in wild populations of endangered species.

The effects of FIV infection and disease are well described in domestic cat (Felis catus) but less so in exotic felids. FIV Fca infection in domestic cat is analogous to HIV infection of humans causing early flu-like symptoms, followed by severe weight loss, chronic wasting disease, and increased susceptibility to rare cancers and opportunistic disease, neurologic disease and death [19,20]. Captive and wild populations of two species, the African lion (Panthera leo) infected with FIV Ple and the puma (Puma concolor), infected with FIV Pco exhibit less severe disease associations. However, infected lions show a dramatic decline in CD4+ subsets, a reduction of the CD4+/CD8+ ratio, reduction of CD8+β high cells, and expansion of the CD8+β low subset relative to uninfected lions [21][22][23]. Further, FIV Pco infected puma display a more generalized response of lymphopenia expressed as a significant decline in total lymphocytes, CD5+ T-cells, and CD5lymphocytes as well as a significant reduction in CD4+ Tcells [23]. Like lions, seropositive pumas have a significant decline in CD8+β high cells but differ by not showing compensatory expansion of CD8+β low cells relative to controls [23]. The results observed with FIV-infected lion and puma parallels human (HIV) and Asian monkey (SIV) CD4+ diminution, and suggests there may be an immunological cost of FIV infection in these two species of large cats.
Identification of genetic correlates of FIV virulence, infectivity, and pathogenicity in different cat species is limited due to a paucity of complete genome sequence. Only subtypes A, B and C from domestic cat FIV Fca [24][25][26], subtypes A and B from puma FIV Pco [14,27] and a single strain (FIV Oma ) from Pallas cat (Otocolobus manul) [16] have been sequenced in entirety. Here we present full-length provirus sequenced from FIV Ple subtype B isolated from lions in the Serengeti National Park in Tanzania and FIV Ple subtype E from lions dwelling in the Okavango Delta in Botswana. These two FIV Ple subtypes exhibit a range in sequence divergence throughout the genome, share motifs unique to this lion-specific lentivirus, yet also exhibit unusual and significant differences in the env gene.

Genomic Organization and Sequence Divergence of FIV Ple
Subtypes FIV Ple subtypes B (accession number EU117991) and E (accession number EU117992) share a similar genome organization with other FIV which consists of LTR, gag, pol, vif, orfA, env, and additional small ORFs that may represent accessory genes including rev ( Table 1). The total proviral genome size was conserved between FIV Ple subtype B (9899 bp) and subtype E (9891 bp) ( Table 1). FIV -Ple gag encodes three putative structural proteins of matrix, capsid and nucleocapsid. Pol is conserved and encodes key viral enzymes of protease, reverse transcriptase, RNAase, dUTPase and integrase. FIV Ple vif, an accessory protein essential for viral replication, resembles that of FIV Fca . OrfA in FIV Ple is similar to FIV Fca and likely corresponds to HIV tat, which targets transcription factors in the LTR. FIV -Ple env encodes the putative leader, surface (SU), and transmembrane (TM) regions of the envelope glycoprotein, essential components for viral binding to and entry into the host cell. FIV Ple rev is similar to HIV/FIV rev, and is thought to be critical in viral replication. FIV Ple rev appears to be encoded by splicing two exons: the first in the leader region of env, the second located near the 3' region adjacent to env (Table 1).
The LTR of FIV Ple contains transcription and regulatory elements common to other FIV. These include the direct 2 bp repeat (IR) defining the 5' and 3' termini of LTR, AP-4, Aml-1 (EPB20), AP-1, TATA box, Poly A, and the cap transcription initiation site ( Figure 1). FIV Ple subtypes have additional transcription factors characteristic of FIV, but placed in alternate locations within the LTR U3 including NF-AT and CREBP-1/c-Jun. These and other motifs were determined by homology search with a threshold value of 85% with the Motif Search database [28] [see Additional file 1]. Overall, lion LTRs are not identical between subtypes B and E, differing by 15% in nucleotide substitutions, comparable to that observed between FIV Fca subtypes A, B and C (Figure 1, Figure 2A).
Deep genetic divergence between FIV strains from different cat species made alignments problematic. For coding regions, we first translated each gene into amino acid residues, which are less divergent as changes occur at a lower rate of substitution, to serve as a "scaffold" for alignment of nucleotides using the program RevTrans [29]. Our results indicate that pol (3657 bp) is the most conserved gene across FIV, albeit exhibiting substantial average pairwise genetic distances of 60% and 54% for nucleotide and amino acid data, respectively ( Table 2). Gag sequences (1551 bp) differed by an average pairwise genetic distance of 65.8% for nucleotides, a 53.2% amino acids (Table 2). However, vif (870 bp), orfA (351 bp), and env (2958 bp) were highly divergent. For these genes, sufficient homology existed to both identify the gene, and to create a multiple sequence alignment across all FIV yet, phylogenetic models for patterns of substitution at variable sites were saturated resulting in an average genetic distance of 100% for both nucleotide and amino acid data ( Table 2). Such differences in rates of evolution between viral genes corroborate previous findings describing functional con-straints for gag and pol [7,8,17], while also demonstrating that vif, orfA, and env rapidly evolve in each host species.

Phylogenetic Analyses of FIV Ple Subtypes
The evolution of FIV Ple subtypes is defined by separate phylogenetic analyses of each viral gene as well as combined data of concatenated sequences representing the entire coding region of FIV. LTR, gag, pol, vif and orfA affirm the species-specificity of FIV both in individual gene analyses (Figure 2A-E) and in the combined concatenated data phylogeny excluding env ( Figure 2F). The three subtypes of FIV Fca from the domestic cat exhibit the least amount of genetic divergence within each viral gene phylogeny. Sharing a monophyletic lineage with distantly related FIV Oma , the FIV Ple subtypes B and E have intermediate levels of genetic distance with each viral gene exam-Phylogenetic reconstruction based on nucleotide sequence of LTR and coding genes from full-length FIV nucleotide sequences excluding env Figure 2 Phylogenetic reconstruction based on nucleotide sequence of LTR and coding genes from full-length FIV nucleotide sequences excluding env. (A-E) Shown are the maximum likelihood trees (ML) which are identical to tree topologies using maximum parisimony (MP) and minimum evolution (ME) for each gene region. See methods and Additional file 3 for specific parameters as implemented in PAUP ver 4.10b. (E) OrfA phylogeny does not include FIV Pco subtype A due to lack of sufficient homology for proper gene identification. (F) Phylogenetic tree of concatenated combined data of coding genes gag, pol vif, and orfA. All nodes supported by 100% bootstrap proportions in ME, MP and ML analyses except for relative positions of FIV Fca subtypes which were supported by bootstraps >50% but less than 100% within the FIV Fca clade.  [4,35]. Thus, the pattern of FIV Fca divergence may represent recent emergence combined with rapid viral diversification within the domestic cat world-wide. In contrast, the puma is one of the oldest species within Felidae, sharing an evolutionary lineage with the African cheetah (Acinonyx jubatus) and the New World jaguarundi (Puma yagouaroundi) and arose approximately 4.5 MYA [34]. The extreme divergence between subtypes A and B within the FIV Pco lineage suggests an ancient origin of FIV infection of puma, a result consistent with the published pol-RT phylogeny marked by high levels of intra-subtype divergence of FIV Pco subtypes from throughout the host species range [4,8,11]. Lastly, the African lion species arose approximately 2 MYA and spread throughout Africa, Asia and the Americas [34]. However, due to episodes of population reduction followed by expansion from East Africa and recolonization, genomic diversity in modern lion populations coalesces to approximately 325,000 years ago and is confined the African continent [36]. FIV Oma is found in wild populations of the Eurasian Pallas cat [4], a species that arose during the late Pleistocene [34]. The monophyletic lineage of Pallas cat FIV Oma and African lion FIV Ple observed here suggest more ancient inter-species transmission as the last time lions and Pallas cats were in geographic contact was during the Pleistocene when lion ranges spread throughout Asia, providing a possible opportunity for FIV transmission between these species [37].

Discordant env phylogeny between FIV Ple subtypes reveals ancestral FIV recombination events in the wild
The patterns of phylogenetic divergence between FIV strains from different cat species are concordant between all viral gene regions with one notable exception, the env  Figure 3A and 3B). A closer examination of the env gene shows only two shared regions of homology between FIV Ple subtypes. The first spans the sites 1-519 of env, containing exon 1 of rev (Table 1), within the leader region exhibiting 80% nucleotide and 68% amino acid homology between FIV Ple subtypes. The second region occurs at the terminal 3' region of env (sites 2506-2958) with 87% and 71% genetic identity for nucleotides and amino acid, respectively. Based on comparison with FIV Fca , this region of FIV Ple may be the rev responsive element (RRE), which is critical for targeting rev to the nucleolus of the cell [38]. As rev is conserved between lion subtypes, it is likely that RRE must remain conserved as well.
By contrast, the SU and TM regions of env differ substantially between FIV Ple subtypes ( Figure 4). A contiguous region of env, from amino acid sites 181 through 931 (green in Figure 4), shows that FIV Ple subtype E is more similar to FIV Fca than to FIV Ple subtype B. Further, env of FIV Ple subtype B, concordant with results from other gene trees (Figure 2A-E), shares more homology with FIV Oma (blue in Figure 4). Moreover, the lack of monophyly between FIV Ple subtype B and FIV Oma (Figure 3) is a consequence of the recombinant env of FIV Ple subtype E, as exclusion of this subtype from the analyses (data not shown) recovered the monophyletic relationship observed with other genome regions (Figure 2A-E).
The predicted env protein from both FIV Ple strains were compared to other published FIV strains with respect to inferred structural elements, with particular focus on regions known to be important for receptor binding. Con-Phylogenetic reconstruction based on nucleotide sequence of fulllength proviral FIV including env and separate analysis of env Figure 3 Phylogenetic reconstruction based on nucleotide sequence of fulllength proviral FIV including env and separate analysis of env.
A. Phylogenetic tree of concatenated combined data of coding genes gag, pol vif, orfA and env. B. Phylogenetic tree of env sequences only. Shown is the maximum likelihood tree (ML) identical to tree topology using maximum parisimony (MP) and minimum evolution (ME) for each gene region. See methods and Additional file 3 for specific parameters as implemented in PAUP ver 4.10b. All nodes supported by 100% bootstrap proportions in ME, MP and ML analyses except for relative positions of FIV Fca subtypes which were supported by bootstraps >50% but less than 100% within the FIV Fca clade. Figure 4) and variable regions (red in Figure 4) and epitope binding sites (orange in Figure 4) were identified based on their locations in the domestic cat FIV sequences [39]. The V3-V5 regions shared least homology between the two strains. In FIV Fca , this region has been shown to contain the CXCR4 binding site [40], neutralizing antibody binding sites [41][42][43] and several epitopes important for cell tropism and cell line adaptation [44][45][46]. Within the V3-V5 region, several biochemical differences have been noted between domestic and non-    (8)(9)(10) and the other non-domestic FIVs (13)(14). A similar trend of lower charge and more cysteine residues in B than E was noted in V3, the region implicated as receptor binding domain for FIV [44,46,47]. In contrast, the more conserved regions flanking V3-V5 were more positively charged in FIV Ple subtype B than in FIV Ple subtype E, but contained similar numbers of cysteine residues and putative N-glycosylation sites. Such differences suggest that substantial divergence may occur in secondary and tertiary structures at the receptor-binding region of these two lion lentiviruses.

B. Env (2958 bp) A. Combined All Genes (9391 bp)
Recombination in lentiviruses is not uncommon. In the ongoing global HIV pandemic, at least 34 circulating recombinant forms from HIV-1 subtypes have been so far described in patients world-wide [48]. SIV full genome sequence comparisons increasingly depict extant primate lentiviruses with mosaic structures indicative of multiple recombination events over time [49][50][51][52][53][54]. In FIV Fca , recombination in the V3-V5 region of env was detected between subtypes A and B in feral cats [7], and different recombination frequencies occur between large regions of FIV Pco subtype B in domestic cat experimentally infected with FIV Pco B [31]. Whereas the frequency of FIV Ple recombination is not yet known, our studies show that over 40% of Serengeti lions in Tanzania are multiply infected with FIV -Ple subtypes A, B and C, which circulate freely within this large population [6] and thus offer opportunities for recombination.
The recombination of env in FIV infected lions has interesting evolutionary significance because the divergence in this region is extensive between the two subtypes. Therefore, subtype E recombination may represent an ancient event of recombination followed by a long period of divergence, or a more recent recombination with a highly divergent but as yet unsequenced strain either from lions or another African felid species. Although FIV Ple subtype E env is more similar to FIV Fca than to any other known FIV the extent of genetic divergence is still quite substantial, i.e. 64.4% nucleotide relative to FIV Fca subtype C ( Table  2), suggesting that if recombination has occurred recently, it is likely to have been with strain that has not yet been sequenced for the env gene. This recombination event may also have functional implications, as FIV Ple subtype E env has structural features more similar to pathogenic FIV Fca .
Further investigation into complete genome analyses of FIV Ple subtypes A, C, D and F as well as FIV from other seropositive African felids, will likely provide new insights into the role of recombination in env in the wild. Clinical studies will help to clarify the significance of these recombination events.

Conclusion
Ongoing efforts to sequence full genome FIV from all seropositive exotic cat species will be essential to understanding the evolutionary trajectory of these viruses including the origin and frequency of recombination within FIV. This study demonstrates the necessity of whole-genome analysis to compliment population/genebased studies, which are of limited utility in uncovering complex events such as recombination that may lead to functional differences in virulence and pathogenicity. The changes observed in the env gene as a consequence of recombination in FIV Ple will provide important clues to the natural history of these viruses and their hosts, and may lead to insights into genetic determinants of pathogenicity and virulence differences between domestic cat and lion FIV; findings with important implications for HIV pathogenesis in humans and virus attenuation in wild populations of endangered species.
Media was collected biweekly and subjected to microtiter reverse transcriptase assay as follows. and reversed transcribed to cDNA with Superscript II (Invitrogen) according to manufacturer's instructions. PCR was then performed to amplify a diagnostic region of pol as previously described [4]. Amplicons were sequenced to confirm the presence of a Botswana strain of FIV (FIV Ple subtype E). One ml aliquots of supernatant were frozen at -70°C. Aliquots were then thawed and used to inoculate 3 × 10 6 Mya-1 cells, which were grown 14 days to achieve positive RT values as above. Cells were supplemented with fresh media weekly and grown to 1 × 10 7 cells at which point cells were harvested by centrifugation and cell pellets were frozen at -70°C.

Cell Culture of FIV Ple Subtype B: Serengeti lion Ple-458
Isolation and culture methods for FIV Ple Subtype B are similar to the methods described for Subtype E (above) with the following exceptions. FIV Ple Subtype B was isolated from PBMCs from a wild, sero-positive lion (Ple-458, Serengeti National Park), separated from heparinized whole blood by sucrose gradient centrifugation using Histopaque (Sigma). Cells were mixed with 10% DMSO with 90% fetal calf serum and viably frozen in nitrogen vapor in aliquots of ~10 6 cells per ml. Postfreezing, thawed PBMCs (10 6 cells) from the wild lion were co-cultivated with an equal number of lion donor cells (Ple-73, captive, National Zoological Park, Wash., D.C.; this lion was sero-positive but had repeatedly tested negative for virus isolation). All PBMCs were mitogen stimulated with concanavalin A (5 ug/ml) for 72 hrs. Cocultures were propagated in RPMI 1640 with 10% bovine serum and 10% human interleukin-2 (Gibco-BRL). Fresh media was added every 72 hours and new donor cells (10 6 cells) were added every 14 days. Replicating virus was confirmed in the supernatant by demonstrating both positive Mg 2 + -dependent reverse transcriptase (RT) and the presence of typical lentiviral particles seen by electron microscopy [10]. Virus rich supernatants were clarified by slow speed centrifugation and stored in liquid nitrogen freezers.
In order to expand the culture sufficiently to harvest viral supernatant for Western blot assays and to conduct the genetic analysis, 1 ml RT positive supernatant (LLV-2, SV lab) was used to inoculate 3201 cells (5 ml at 2 × 10 6 /ml), FeLV negative lymphosarcoma cells [56]. Cells were maintained in equal parts Leibovitz's L-15 media and RPMI 1640 with 20% fetal calf serum with glutamine (2x) and penicillin/streptomycin (1x). Initially, this culture was difficult to maintain in 3201 cells because it caused rapid cell death thus, in order to keep the culture alive, fresh media and naive 3201 cells had to be added every 3-4 days. After 21 days post infection (dPI), fresh media continued to be added to the culture every 3-4 days, but the addition of naïve 3201 cells was stopped and the % viability was allowed to decline (in the hope that a cell adapted virus could emerge that would enhance our ability to grow up viral stocks for use in Western blot assays). From dPI 28 to 49 the culture viability hovered between 18-24%, but after dPI 52 it was clear that both the viability and cell numbers began to improve (viability from 46 to 86%). By dPI 71 the cell viability was holding at >90% and the culture was growing at 40-50% per day. Infected cells for DNA extraction and genetic analysis of subtype B virus were harvested on dPI 88, centrifuged, and the pellets frozen at -70C.

DNA extraction, Cloning and Sequencing of FIV Ple Subtypes B and E
DNA was extracted and purified from frozen cell culture pellets following the manufacturer's protocols established for blood products (Quiagen). Following extraction, DNA quality was checked by gel electrophoresis, and quantified by spectrophotometer (NanoDrop).  [15,25,58]. Open reading frames were determined and regions of homology between FIV Ple with other FIV strains using pair-wise comparisons implemented by BLAST of two sequences [59]. The boundaries of both the 5'LTR and 3'LTR regions were identified by the conserved polypurine tract (PPT) shared by all FIV [60] and the primer binding site (PBS) which mark the boundary between the 3'LTR and the 5'LTR, respectively.

FIV
The genome of lion FIV Ple was compared with existing fulllength FIV by multiple sequence alignments of each viral gene. LTR regions were aligned using Clustal X [61] and verified and edited by eye using Se-Al ver 2.0 [62]. Due to large genetic divergence between FIV from different species, alignment for coding regions of FIV used the program REVTRANS ver 1.4 [29] which takes a multiple sequence file, translates that file into amino acid residues, aligns the amino acids, and uses this alignment as the scaffold for nucleotide alignment. Aligned multiple sequence files were imported into Modeltest ver 3.7 [63] and the optimal model of nucleotide substitution was selected using the AIC criterion (see Additional file 3). Viral genes were analyzed separately, as well as combined, for genome comparison and phylogenetic reconstruction. Phylogenetic trees based on nucleotide data were obtained using a heuristic search with three different optimality criteria of maximum likelihood (ML), minimum evolution (ME) and maximum parsimony (MP) as implemented in PAUP* ver 4.0b10 [64]. Conditions for the ML analysis included starting trees obtained by stepwise addition, and branch swapping using the tree-bisection-reconnection (TBR) algorithm. Specific conditions for the ME search included starting trees obtained by neighbor -joining, TBR branch-swapping algorithm, and no collapsing of zero-length branches. The MP analyses coded gaps as "missing", with step-wise addition of taxa and TBR branch swapping. Support for nodes within the phylogeny used bootstrap analysis with identical settings established for each method of phylogenetic reconstruction and retention of node bootstrap values greater than 50%. The number of bootstrap iterations consisted of 1000 for ME and MP methods and 100 for ML. Additional analyses were conducted on FIV coding sequences after translation into amino acids. Genetic distances between strains were derived using the Pam-Dayhoff model of amino acid substitution as implemented in MEGA verson 3.1 [65] with gamma-correction (alpha = 2.5) and pairwise deletion of missing data. data, and contributed to writing the manuscript. JLT assisted in experimental design and helped write the manuscript. SVW provided cell culture expertise, reagents, and helped write the manuscript. MR collected blood samples from animals in the wild, conducted cell culture of subtype B and helped write the manuscript. KS conducted cell culture of subtype E and helped in writing the manuscript. CW and HW contributed expertise and essential logistic support in obtaining lion blood samples. SJ O'B contributed expertise, reagents, and helped write the manuscript. All authors have read and approved the final version of the manuscript.