Apparent effect of rabbit endogenous lentivirus type K acquisition on retrovirus restriction by lagomorph Trim5αs

To test the hypothesis that rabbit endogenous lentivirus type K (RELIK) could play a role in shaping the evolution of TRIM5α, the susceptibility of viruses containing the RELIK capsid (CA) to TRIM5 restriction was evaluated. RELIK CA-containing viruses were susceptible to the TRIM5αs from Old World monkeys but were unaffected by most ape or New World monkey factors. TRIM5αs from various lagomorph species were also isolated and tested for anti-retroviral activity. The TRIM5αs from both cottontail rabbit and pika restrict a range of retroviruses, including HIV-1, HIV-2, FIV, EIAV and N-MLV. TRIM5αs from the European and cottontail rabbit, which have previously been found to contain RELIK, also restricted RELIK CA-containing viruses, whereas a weaker restriction was observed with chimeric TRIM5α containing the B30.2 domain from the pika, which lacks RELIK. Taken together, these results could suggest that the pika had not been exposed to exogenous RELIK and that endogenized RELIK might exert a selective pressure on lagomorph TRIM5α.


Introduction
Retroviruses have a unique replication strategy that involves an obligatory integration step where the viral genome is inserted into that of the host. When this takes place in germ or embryonic cells, the virus can be transmitted vertically to the next generation and is termed an endogenous retrovirus (ERV). ERVs can be fixed in a population, where with time, most of them accumulate deletions, missense and non-sense mutations, leading to a loss of replication competency. While some of the ERVs are co-opted by the host for various functions ranging from development to intrinsic immunity, many are just relics left behind from an ancient infection [1,2]. Like the fossils in paleontology, ERVs provide a means to estimate the age of an infection [3], which could provide insights into the origins of related current exogenous viruses that are in circulation [4,5].
Lentiviruses belong to the genus of retrovirus with members that are a threat to human health, such as HIV-1 and HIV-2. They were thought to be a group of relatively modern viruses until the discovery of the rabbit endogenous lentivirus type K (RELIK) [6]. Compared with the primate lentiviruses, RELIK has a simpler genome, containing only three accessory genes (tat, rev and dUTPase). However, similar to the primate lentiviruses, their capsid (CA) protein binds cyclophilin A [7]. In addition, viruses containing the RELIK CA can also infect non-dividing cells, suggesting that these properties of lentiviruses were acquired a long time ago. Through PCR, RELIK sequences were identified in the genomes of several genera of the lagomorph order, including that of the hare, which placed the origins of RELIK at more than 12 Mya [8]. Analyses of the PCR products revealed a high degree of homology with the original sequences isolated from the European rabbit. This suggested that invasion of the germline occurred before the divergence of the Lepus genus, or that the exogenous virus was widespread among the different lagomorph genera that shared common geographical distributions. However, RELIK does not seem to be present in the pika family of lagomorphs [8,9].
Lentiviruses and other retroviruses are susceptible to host restriction factors [10]. One group of restriction factors targets the CA of the virus [1]. These include the murine factor Fv1 [11], as well as TRIM5a [12] which has been found in many species [13][14][15][16]. TRIM5a belongs to a large family of proteins containing the tripartite ring, B-box and coiled coil (RBCC) motif [17,18]. Some Trim proteins such as TRIM5a also contain the B30.2 domain at the C-terminus [19,20]. TRIM5a restricts a range of retroviruses, including lentiviruses [21,22], gammaretroviruses [23], spumaviruses [24] and betaretroviruses [25]. While its precise mechanism of action is still not known, it is clear that multimerization of the molecule that is mediated by the coiled coil is essential [26][27][28][29]. The B30.2 domain with its four variable regions [30] has been found to be the major determinant of restriction specificity [31]. It mediates binding to HIV-1 CA [26] and is under strong positive selection [32,33], indicating that it is the part of TRIM5a that recognizes the virus. Presumably, TRIM5a evolution is driven by exposure to retroviral CA. However, it remains unclear whether the selective pressure comes from endogenous or exogenous viruses [1].
TRIM5a orthologues have been isolated from lagomorphs such as the European rabbit and hare [15,34]. They restrict a range of retroviruses, including gammaretroviruses and lentiviruses. In addition, there is strong evidence of positive selection in residues in the B30.2 domain, suggesting that retroviruses could be involved in shaping the evolution of the lagomorph TRIM5as [35]. In the light of the prevalence of RELIK sequences in many lagomorph genera, we wondered whether this ancient lentivirus could play a role in the selection of the lagomorph TRIM5as. Hence, we set out to investigate the relationship between lagomorph TRIM5a restriction and viruses containing the RELIK CA. We report here that the TRIM5a from the cottontail rabbit and pika can restrict a wide range of retroviruses. In addition, viruses containing the RELIK CA were susceptible to restriction by TRIM5as from a number of primates and lagomorphs.

Material and methods (a) DNA constructs
Structures of chimeric gag constructs are illustrated in figure 1. The construction of pEIAV-RELIK(CA) has been described previously [7]. pFIV-RELIK(CA) was made by substituting the EcoRI/Tth111I fragment containing the FIV CA flanked by MA and NC sequences in pFP93 with a corresponding fragment containing the RELIK CA flanked by FIV sequences. The fragment was generated by overlapping PCR using primer pairs FIVF/FIVRE-LIKRev2, FIVRELIKF3/FIVRELIKRev4 and FIVRELIKF5/ FIVRev in reactions with templates pFP93 [36], pRELIKCA [7] and pFP93, respectively. The first two fragments were joined in a second reaction with primer pair FIVF/FIVRELIKRev4. The resulting product was joined to the third fragment from the first reaction using primer pair FIVF and FIVRev. In order to make EIAV-RELIK(MA-CA), the XhoI/AleI fragment from pEIAV-RELIK(CA) that contain the EIAV MA was replaced by a fragment containing MA from RELIK (synthesized by Genescript based on the sequence reported by Katzourakis et al. [6]). This fragment was generated using overlapping PCR by performing the first reactions with primer pairs PONYXhoIF/MARELIKRev2, MARELIKF3/ MARELIKRev4 and MARELIKF5/RELIKCARev using pEIAV-RELIK(CA) [7], pRELIKMA and pEIAV-RELIK(CA) as templates, respectively. The first two fragments were joined using primer pair PONYXhoIF/MaRELIKRev4, and the resulting product was    joined to the third fragment in the first reaction using primer pair PONYXhoIF/RELIKCARev. TRIM5as from the European (Oryctolagus cuniculus) and cottontail (Sylvilagus floridanus) rabbit were amplified from cDNAs prepared from SIRC and Sf1ep cells, respectively, using the primer pairs RabbitT5F and RabbitT5Rev. mRNAs from these cell lines were isolated using the total mRNA extraction kit from Qiagen and reverse transcribed to cDNA using the first strand synthesis kit from Roche. The TRIM5a PCR products were cloned into pENTR-D-TOPO (Invitrogen) and sequenced. They were then transferred into pLGatewayIEYFP and pLGate-waySN by an LR recombination reaction using LR Clonase to create the retroviral delivery vectors.
A draft genomic DNA sequence from American pika (Ochotona princeps) can be found at http://www.ensembl.org/Ochotona_ princeps/Info/Index/. From this sequence, primers targeting sequences in TRIM5a exon7 and exon8 were designed and used to amplify the B30.2 region from pika genomic DNA (a generous gift of Dennis Lanning, Loyola University). The product was cloned into TopoBlunt and sequenced. Nucleotide and predicted amino acid sequences of the B30.2 coding region were identical to that in the database. To test the restriction properties of pika TRIM5a, the B30.2 domain was fused to the RBCC of the TRIM5a from the European rabbit. This was achieved by overlapping PCR using the primer pair TopoRabbitT5F/SIRCPIKARev3 to amplify the sequences encoding the RBCC from the European rabbit TRIM5a and primer pair SIRCPIKAF3/PIKARev to amplify exon8 from the Pika TRIM5a encoding the B30.2 domain. The two PCR products were used in a second reaction with primer pair TopoRabbitT5F/PIKARev to generate the chimeric TRIM5a that was cloned into pENTR/D-Topo before transferring to pLgatewayIRESEYFP and pLgatewaySN by LR recombination. Sequences of these primers are shown in table 1.

(c) Infectivity and restriction assays
The infectivities of the RELIK chimeric viruses were measured by endpoint titration. Ten-fold dilutions of the virus stocks were made and added to each well of a 12-well plate that had been seeded with 5 Â 10 4 cells per well the day before in the presence of 10 mg ml 21 of polybrene. The cells were then stained for b-galactosidase 2 days post-transduction, and the number of blue cells counted [39].
To study the restriction of primate TRIM5as on EIAV and EIAV-RELIK(CA), vectors carrying the primate TRIM5as as well as EYFP were used to deliver the genes into CrFK cells so that more than 80 per cent were transduced. Two days posttransduction, the cells were seeded into 12-well plates and challenged with different dilutions of the EIAV and EIAV-RELIK(CA) virus. The number of infected cells was determined by staining for b-galactosidase 2 days later.
The restriction phenotype of the lagomorph TRIM5as was tested by the two-colour FACS assay as described previously. Briefly, CrFK cells were transduced with delivery vectors carrying TRIM5a and the EYFP marker. These were challenged with a panel of retroviruses (HIV-1, HIV-2, SIVmac, FIV, EIAV, PFV, SFV, FFV, B-MLV and N-MLV) 2 days post-transduction. The percentages of cells with and without restriction factor that were infected were determined by FACS analyses.
Cat CrFK cells were found to be permissive for the EIAV-RELIK chimeric virus. Single cell clones expressing lagomorph TRIM5as were derived from CrFK cells by transduction with limiting dilutions of retroviral vectors carrying both TRIM5a as well as the G418 resistance marker. The transduced cells were selected on medium containing G418 (1 mg ml 21 ) for two weeks until visible colonies appeared. Well-separated colonies were picked, expanded and tested for restriction.

(d) Protein analyses
Viruses were pelleted through a 20% sucrose cushion at 100 000g for 3 h at 48C. The pellets were washed once with PBS and resuspended in SDS loading buffer before separation by SDS -PAGE. The protein bands were then stained with Imperial protein stain (Thermo Scientific).

(e) Nucleotide sequence accession numbers
The gene sequences determined in this study have been submitted to GenBank (accession nos KC425460 and KC425461).

(a) Isolation and sequence characterization of TRIM5as from lagomorphs
To examine the influence of RELIK on TRIM5a evolution, we set out to compare the sequences and functional properties of TRIM5a from lagomorphs with and without endogenized RELIK (figure 2a). We prepared functional TRIM5a constructs from European and cottontail rabbit that contain RELIK, and from pika which does not [6,8,34]. The TRIM5a of the European and cottontail rabbits was amplified and cloned by RT-PCR from the SIRC and Sf1ep cell lines, respectively. Two different versions of TRIM5a were cloned from cottontail rabbit, presumably corresponding to two different alleles of TRIM5. A source of pika mRNA was not available to us, so the pika construct was prepared by amplifying exon8, containing the primary specificity determinants for retroviral restriction by TRIM5a [31], from pika genomic DNA and fusing it to the RBCC domain of the European rabbit TRIM5a cDNA. These four clones were sequenced and compared with one another and other lagomorph TRIM5a sequences deposited in the NCBI and EBI databases. The sequence of the pika exon 8 was identical to that found in the draft pika genome sequence. An alignment of the European and the cottontail rabbit (determined here) with the hare [34], and the pika (inferred from the genomic DNA sequence) TRIM5a proteins is shown in figure 2b. The sequences show marked diversity, particularly in the regions that corresponded to the variable regions (V1-4) previously identified in the primate TRIM5as [30,31], with the greatest differences occurring in the V1 region. In this region, the European rabbit and hare sequences are two residues shorter than the pika, whereas the cottontail rabbit alleles lack an additional six residues. In addition, the V3 region of the pika is one residue longer than those of the other species. Differences in the lengths of these regions are frequently observed in primate TRIM5a [30,40]. A particular feature of genes involved in evolutionary conflicts is a high frequency of non-synonymous nucleotide substitutions. We therefore analysed the sequence data for the presence of positively or negatively selected sites using the DataMonkey webserver. Fourteen positively selected codons were identified (table 2); no negatively selected sites were seen. Thirteen of selected sites mapped within the B30.2 domain with 12 present in the different variable regions, consistent with previous observations on TRIM5a and confirming that lagomorph TRIM5a has been subjected to positive selection during the course of evolution [34,35].

(b) Functional characterization of TRIM5a from lagomorphs
To examine the anti-retroviral activities of the lagomorph TRIM5as, CrFK cells were transduced with retroviral vectors encoding the different TRIM5a proteins, and then challenged with a panel of retroviruses, including gammaretroviruses, lentiviruses and foamy viruses ( figure 3). Individual TRIM5as displayed slightly different restriction phenotypes. The foamy viruses PFV and SFV were not restricted by any TRIM5a, whereas the lentiviruses HIV-1, HIV-2, EIAV, FIV and the gammaretrovirus N-MLV were restricted by all four factors. FFV was restricted only by the TRIM5as from the cottontail rabbit, whereas B-MLV and, to a lesser extent, Mo-MLV were restricted by one of the two cottontail alleles. SIVmac was susceptible to that from the pika and one of the cottontail alleles. Because the pika construct possessed an identical RBCC with the European rabbit, we can conclude that the altered specificity for SIVmac is due to a change in the B30.2 domain. However, changes in the overall profile of restriction activity appear relatively small despite significant changes in TRIM5a sequence.

(c) Chimeric RELIK lentiviral Gag constructs
We next wanted to test whether the sequence changes in TRIM5a were associated with an alteration in the ability to restrict RELIK. We had previously described the construction of a chimeric lentiviral Gag, EIAV-RELIK(CA), consisting of the RELIK capsid (CA) in a background of EIAV Gag [7] that produced infectious viruses when co-transfected with an EIAV vector and VSV-G into 293T cells. However, the viral titres were significantly lower than the parental EIAV and could only be measured accurately by endpoint dilution after concentration of the viral supernatant. Before embarking on restriction studies with this virus, we wished to investigate whether we could understand the reasons for this low titre. CA from EIAV-RELIK(CA) and EIAV-RELIK(MA-CA) was larger than that of EIAV, whereas the MA band for EIAV-RELIK(MA-CA) was larger than those from EIAV-RELIK(CA) and EIAV. In addition, the CA band for FIV-RELIK(CA) was larger than that of FIV and the same size as those from EIAV-RELIK(CA) and EIAV-RELIK(MA-CA). These results indicated that properly processed viruses were made from all chimeric Gag-pol constructs. Furthermore, the intensities of the bands from different viruses were comparable, suggesting that the viruses with chimeric Gags were assembled, released and processed with similar efficiencies to the parental Gags. The infectivities of the viruses with chimeric Gags were investigated by titration on a panel of cells of different origins, including human (TE671), canine (D17), feline (CrFK) and rabbit (SIRC) (figure 4b). In general, the titres were highest on feline CrFK cells and lowest in rabbit SIRC cells. This was consistent with previous reports of the existence of several blocks to lenti-and retroviral infection in rabbit cells [41]. Hence, CrFK cells were used for all subsequent studies. Introduction of the RELIK MA into the background of EIAV Gag

(d) Restriction of RELIK by TRIM5a from various primates
To examine whether RELIK was susceptible to TRIM5a restriction, we challenged a panel of permissive CrFK cells transduced with the TRIM5a from different primates with the EIAV-RELIK(CA) virus and EIAV. The results are shown in figure 5. As observed previously, EIAV-RELIK(CA) but not EIAV was restricted by the TRIM5CypA from the owl monkey and rhesus macaque [7]. Unlike EIAV, which is restricted more than 10-fold by the four ape TRIM5as (orangutan, gorilla, chimpanzee and human), EIAV-RELIK(CA) was restricted only by the orangutan TRIM5a. Both EIAV and EIAV-RELIK(CA) were restricted by the three TRIM5as from Old World monkeys (sooty mangabey, vervet and rhesus macaque), but not by those from the New World monkeys capuchin, squirrel monkey and common marmoset. EIAV-RELIK(CA) is also resistant to the TRIM5a from the cotton top tamarin (New World monkey), whereas EIAV is modestly restricted (about 10-fold). These results indicated that RELIK can be a target for CA-dependent restriction factors such as TRIM5a. They also show that the RELIK CA is sufficiently different from that of EIAV to enable it to escape from restriction by the TRIM5as from apes and the cotton top tamarin.

(e) Restriction of RELIK by lagomorph TRIM5as
The results from the sequence studies provided evidence for the presence of selection forces acting on lagomorph TRIM5as, most likely from retroviruses (figure 3 and [15,34]). Because the lagomorph TRIM5as were active against various lentiviruses, it was tempting to speculate that RELIK could be involved in driving the evolution of TRIM5a in lagomorphs. To test this hypothesis, single cell clones were derived from CrFK cells transduced with the lagomorph TRIM5a expression constructs by endpoint titration of retroviral vectors carrying the TRIM5a genes on CrFK cells, followed by G418-selection so that single colonies, which were well separated, could be isolated. This was to ensure that the clones contained a single copy of the lagomorph

Discussion
Consideration of the interaction between the various TRIM5as and their targets reveals three riddles that need answering to claim a full understanding of restriction specificity. First, how can Trim5a from one species recognize and interact with the CA protein of retroviruses from different genera if they share little or no sequence identity? Second, why will Trim5a from multiple species recognize a given virus even though the interacting residues, thought to lie in the variable regions of the Trim5a [30,31], differ significantly? Third, given the wide range of possible interactions reflected by questions one and two, why can a single amino acid change, for example the R332P alteration in human TRIM5a which allows restriction  of HIV-1 [42] or the R110E change in CA that controls N/B tropism in MLV [43], have such a major effect on restriction?
The properties of the lagomorph Trim5as we have examined provide further illustrations of these facets of TRIM5a restriction. They restrict a variety of viruses including lentiviruses, gammaviruses and foamy viruses (figure 3) with very different CA sequences. Pika and European rabbit recognize many of the same viruses despite having very divergent V1, V2 and V3 regions (figure 2). We know that single amino acid changes in V1 can alter the restriction properties of human TRIM5a [42,44,45] but a comparison of the Pika (32 amino acids) and European rabbit (30 amino acids) V1 region shows only three amino acid identities, yet both can restrict HIV-1, HIV2, FIV, EIAV and N-MLV but not PFV. SFV or B-MLV. Interestingly, the two alleles of cottontail rabbit, that are quite closely related (6/64 amino acid differences in V1 plus V2 plus V3), show significant differences in their restriction profiles. It particular, allele-2 was capable of restricting B-MLV, a property of naturally occurring Trim5 alleles not previously observed [31]. We note that multiple alleles of TRIM5a have previously been reported in European brown hare, Iberian hare and European rabbit [35]. It is therefore very tempting to suggest that this represents a further example of balancing selection among TRIM5a alleles [46].
It seems likely that the common fold of the CA protein from the Orthoretrovirinae [47] and similar properties underlying assembly in mature cores [29,[47][48][49], coupled with relatively non-specific interactions between the surfaces of CA and Trim5 [50] will prove important elements in addressing these questions. However, the case of the foamy viruses, which are susceptible to Trim5a restriction (figure 3 and [24]) but seem to form a rather different structure [51], provides an important complicating factor. Determining the structures of co-crystals between TRIM5a and a viral target would go a long way to resolving these issues. TRIM5a sequences, and therefore their target specificities, have been moulded in large part by a series of positive selection events [32,33,52]. However, the nature of the viruses driving these changes has been the subject of some debate. It is not clear whether the primary driving force for selection is the need to control ERVs, in which case viruses might be present in the species in which selection took place, or driven by exogenous infection in which case the virus might no longer exist or still be present in the species acting as a source for infection. One case in which a specific group of endogenous viruses has been implicated is the PtERV1 family that is present in chimpanzees but not in humans [53]. It was argued that the human TRIM5 changed in response to PtERV exposure protecting humans but leaving them susceptible to a later HIV-1 infection. However, others have subsequently questioned this suggestion [54]. The observations that (i) RELIK is present in some lagomorphs but not others [8,34], (ii) positive selection of TRIM5a was taking place in these animals [34,35], and (iii) our ability to generate an infectious viral vector containing RELIK CA [7] suggested that it should be possible to examine whether the acquisition of RELIK was associated with specific changes in TRIM5a resulting in RELIK restriction. Before embarking on these experiments, we wanted to see whether it was possible to generate a virus with higher titre than our original construct. We found that EIAV tolerates substitution with the RELIK CA better than FIV, despite RELIK sharing the same late domain with FIV, which is different from that in EIAV [55]. This could reflect the closer relatedness between EIAV and RELIK as observed from phylogenetic analyses [6,56]. The substitution seemed to compromise infectivity during early events, because processed Gag in the chimeric viruses was produced at levels similar to parental ones ( figure 4). In addition, inclusion of the RELIK MA together with CA in the context of the EIAV Gag further reduced the infectivity without affecting viral production. These results suggest that the different viral components continue to have an impact on each other even after virion maturation. In the absence of any evidence of direct interactions between the cleaved viral products, a possible explanation is that they recruit different cellular components that interact with the other viral products. These cellular factors will be more compatible to homologous viral proteins than those from a less related virus. Examination of these questions might be expected to shed further light on possible roles for Gag products in early, post-entry, events during infection. Comparison of replication of viral vectors containing the RELIK CA in the presence of TRIM5a from various lagomorphs revealed a gradient of restriction activity increasing from fourfold for pika, through four-and ninefold for the two cottontail alleles and 30-fold for the European rabbit (figure 6b). The various TRIM5a proteins show very similar effects on HIV-1 (figures 2 and 6a) arguing against any possible effects of TRIM5a expression level in the different clones. Such an increase in restriction of RELIK is consistent with a process driven by RELIK exposure. However, the increase is relatively small and these data by themselves do not address the question of whether selection is mediated by endogenous or exogenous viruses.
In the light of the absence of RELIK in the pika genome, one explanation for the low level of restriction of RELIK by the pika TRIM5a could be that the ancestors of the pika had not been exposed to the virus and, hence, the pika TRIM5a was not selected for providing resistance to this virus. Alternatively, the low level of restriction could be vestige of a successful protection against RELIK: the pika TRIM5a could have evolved under selection pressure from other retroviruses following the extinction of RELIK.
We also note that restriction is strongest by the TRIM5a from the European rabbit, which is the source of the RELIK sequence used in this study, possibly suggesting that the endogenized virus is most susceptible to the TRIM5a from its host. One target of TRIM5a restriction in HIV-1 has been mapped to the cyclophilin A-binding loop between alpha-helices 4 and 5 of the CA [57,58], though other regions may be important [59]. An exchange of the helix 4/5 region between SIV mac and HIV-2 resulted in altered susceptibilities to restriction by the Rhesus TRIM5a [60]. Although the RELIK sequences seem to be well conserved between the different genera in Leporidae with more than 90% homology, comparison of the cyclophilin A-binding loop between the RELIK CA from the European rabbit and the one sequence of the orthologue found in hares (accession no. FJ493032) revealed four substitutions and two deletions. Because small changes in this region are sufficient to influence restriction, it is tempting to speculate that the RELIK from hares will have different susceptibilities to lagomorph TRIM5as. To understand more about the relationship between endogenized lentiviruses and TRIM5a, it will be necessary to investigate the restriction of RELIK orthologues found in other genera of the Leporidae by the range of lagomorph TRIM5as.