TRIM5 alpha Drives SIVsmm Evolution in Rhesus Macaques

The antagonistic interaction with host restriction proteins is a major driver of evolutionary change for viruses. We previously reported that polymorphisms of the TRIM5α B30.2/SPRY domain impacted the level of SIVsmm viremia in rhesus macaques. Viremia in macaques homozygous for the non-restrictive TRIM5α allele TRIM5Q was significantly higher than in macaques expressing two restrictive TRIM5alpha alleles TRIM5TFP/TFP or TRIM5Cyp/TFP. Using this model, we observed that despite an early impact on viremia, SIVsmm overcame TRIM5α restriction at later stages of infection and that increasing viremia was associated with specific amino acid substitutions in capsid. Two amino acid substitutions (P37S and R98S) in the capsid region were associated with escape from TRIM5TFP restriction and substitutions in the CypA binding-loop (GPLPA87-91) in capsid were associated with escape from TRIM5Cyp. Introduction of these mutations into the original SIVsmE543 clone not only resulted in escape from TRIM5α restriction in vitro but the P37S and R98S substitutions improved virus fitness in macaques with homozygous restrictive TRIMTFP alleles in vivo. Similar substitutions were observed in other SIVsmm strains following transmission and passage in macaques, collectively providing direct evidence that TRIM5α exerts selective pressure on the cross-species transmission of SIV in primates.


Introduction
The epidemic of human immunodeficiency virus (HIV), including both HIV-1 and HIV-2, is a consequence of crossspecies transmission of lentiviruses from non-human primates (NHP) to humans [1,2]. HIV-1 is derived from cross-species infection of simian immunodeficiency virus in chimpanzees (SIVcpz) and HIV-2 from SIV in sooty mangabeys (SIVsmm) [3,4,5,6]. The cross-species transmissions of SIV were also observed between primates of different species in the wild [7,8,9]. However, not all cross transmissions will result in epidemic infection in the new species. For HIV-1, several cross-transmission events, which occurred independently, generated the different distinct lineages, termed groups M, N, O and P, but only group M resulted in the worldwide pandemic of acquired immune deficiency syndrome (AIDS) in humans. For HIV-2, at least eight distinct lineages, termed groups A-H, were generated by independent cross-transmission, and only groups A and B have spread in the human population [2]. The divergence of several host proteins, including apolipoprotein B mRNA-editing, enzymecatalytic, polypeptide-like 3G (APOBEC3G), Tetherin/BST-2, tripartite motif-containing protein 5a (TRIM5a) and SAM domain and HD domain-containing protein 1 (SAMHD1), constitute the specific restrictions preventing lentivirus crosstransmission among primates of different species [2,10]. Only the virus strains which escape these restrictions are able to establish epidemic infection in a new host. The evolution and selection by interaction between viruses and host restriction factors resulted in the appearance of species-specific lentiviral lineages infecting different primates. Studies on how HIV/SIV interacts with restriction factors and overcomes the species specific barrier will not only help us to trace the origin of HIV/SIV, but also help us to understand the pathogenesis of HIV-1 infection. Such knowledge provides useful information for the development of anti-HIV drugs and vaccines. In our study, we used the SIVsmminfected rhesus macaque model to study the relation between TRIM5a and SIV infection.
TRIM5a was first identified as a protein responsible for restriction of HIV-1 replication in macaque cell lines [11]. It is widely found and described as a retrovirus inhibitory protein in primates and several other mammals [11,12,13,14,15,16,17,18,19]. TRIM5a is a member of the tripartite motif or TRIM family of proteins which have RING finger, B-box, and coiled-coil structure domains. In addition to these three common domains shared by all TRIM family proteins, TRIM5a also has a B30.2/SPRY domain at its C terminus [20]. In some primate species, the TRIM5a B30.2/ SPRY domains are replaced by cyclophilin-A (CypA) due to alternative mRNA splicing and these TRIM-Cyp variants also have restrictive activity against some retroviruses [14,16,21,22,23,24,25]. TRIM5a blocks lentivirus replication at a post-entry stage before reverse transcription [11,26]. The detailed mechanism of TRIM5a restriction of lentivirus replication has not been well elucidated due to a lack of good methods for studying virus replication at this stage. However, binding to viral capsid protein is required for TRIM5a mediated restriction [26,27,28]. Several studies revealed that interaction between TRIM5a and lentiviral capsid protein resulted in dissociation of the viral core particle before reverse transcription [29,30,31,32,33], and triggering of the innate immune responses to restrict virus replication [34]. All the domains of TRIM5a are required for retroviral restriction activity and the B30.2/SPRY domain determines the specificity of capsid recognition and retroviral restriction [35,36,37,38,39]. Phylogenetic analysis of TRIM5a sequences from human and other species of NHPs indicated considerable interspecies variability and positive selection in the B30.2/SPRY domain, which revealed the important role of TRIM5a in fighting against virus infection during the evolutionary history of human and NHPs [18,40,41,42].
In addition to interspecies variability, intra-species polymorphisms of TRIM5a were also reported in humans [43,44,45,46], rhesus macaques and sooty mangabeys [47,48]. In rhesus macaques, an insertion/deletion polymorphism at amino acid 339-341 of the B30.2/SPRY domain resulting in TFP/Q polymorphisms confers differential restriction against multiple lentiviruses when tested in a single-cycle infectivity assay [47,48]. Five distinct alleles of TRIM5 have been described in macaques (Mamu-1 through Mamu-5). Mamu-1, 2 and 3 encode a TFP polymorphism in the SPRY domain and are restrictive for SIVsmm-infection, whereas, Mamu-4 and 5 encode a Q at this position and are permissive for SIVsmm infection [48]. A TRIM-CypA splice variant is also observed at a fairly low frequency in the rhesus populations and is also restrictive for SIVsmm-infection. The restrictive genotypes, TRIM5 TFP/TFP , and TRIM5 TFP/Cyp and TRIM5 Cyp/Cyp represent 31, 10 and 2% respectively of a large cohort of rhesus macaques [unpublished data, WEJ, JSM] that have been genotyped for TRIM 5 (n = 1742). In our previous study, we found that these TFP/Q polymorphisms also affected SIVsmm replication in vivo. Viral loads in macaques with restrictive TRIM5 TFP/TFP and TRIM5 TFP/Cyp genotypes are approximately 100-to 1,000-fold lower than in macaques with permissive TRIM5a Q alleles [49]. Modest association of TRIM5a genotypes and infection was also reported in SIV-mac251-infected rhesus cohorts, though the magnitude was smaller than what we observed in SIVsmE543-3 infected cohorts (,1.3 log) [50]. SIVsmE543-3 has a history of only two passages in rhesus macaques after its initial isolation from sooty mangabeys [51], and thus it may not be completely adapted for replication in rhesus macaques. This contrasts with the repeated passage of SIVmac in rhesus macaques that has resulted in significant adaptation. Hence SIVsmE543-3 infected rhesus cohorts with different TRIM5a genotypes provide us a good model to study how SIV can overcome TRIM5a restriction.

SIVsmm overcame TRIM5a restriction at late stages of infection of macaques
To investigate whether SIV could overcome TRIM5a restriction, we monitored the plasma viremia and peripheral blood CD4+ T cell counts of four SIVsmE543-3-infected rhesus macaques that had different TRIM5a genotypes. All of these four macaques were intravenously inoculated with a high dose of SIVsmE543-3 as previously described [52]. Of these four macaques, one (Rh447) was homozygous for the restrictive TRIM5 TFP allele, two (Rh458 and Rh063) were heterozygous for the restrictive TRIM5 TFP and TRIM5 CypA alleles, and one (Rh444) was homozygous for the permissive TRIM5 Q allele. The macaques with the restrictive TRIM5 TFP/TFP (Rh447) and TRIM5 TFP/CypA (Rh458 and Rh063) genotypes had much lower plasma viral loads than the macaque with the permissive TRIM5 Q/Q genotype (Rh444) during the acute stage of infection (Fig. 1). The peak plasma viremia at the acute stage of infection varied from 10 3 to 10 5 copies per ml in these three macaques (Rh447, Rh458 and Rh063), at least two logs lower than in the permissive macaque Rh444. All three macaques with restrictive alleles maintained stable peripheral blood CD4+ T cell counts during acute infection (above 1000 CD4+ T cells per ml blood), while a significant loss of CD4+ T cells was observed in macaque Rh444. These results were consistent with a previous report, which revealed that TRIM5 TFP and TRIM5 CypA alleles restrict SIVsmE543-3 replication in rhesus macaques [49]. However, we also observed that at a later stage of infection, plasma viral loads in macaques Rh447, Rh458 and Rh063 increased to 10 6 to 10 7 copies per ml, which was comparable to that observed in the permissive macaque Rh444. Increasing viremia was accompanied by a significant loss of CD4+ T cells (less than 300 CD4+ T cells per ml blood). All the macaques progressed to AIDS with opportunistic infections within two to three years post infection despite the differences in their TRIM5a genotypes. These results suggested to us that SIV may have escaped TRIM5 TFP and TRIM5 CypA restriction at late stages of infection.
To confirm this hypothesis, we isolated SIV RNA from plasma samples of macaques Rh447 (at 78 w.p.i., weeks post infection) and Rh458 (at 110 w.p.i.), and obtained 8.5 kb RT-PCR products including the full gag-pol-env coding region. The PCR products were subcloned into the SIVsmE543-3 LTR backbone to construct full-length virus clones. Infectivity of these virus clones was measured on cell lines stably expressing different rhesus TRIM5a alleles by single-cycle infectivity assay as previously described [49]. As shown in Fig. 2, a representative virus clone, 447-1 from macaque Rh447 (TRIM5 TFP/TFP ) replicated in cell lines expressing TRIM5 TFP or TRIM5 Q alleles but was still restricted by the TRIM5 CypA allele. A virus clone, 458-1 from

Author Summary
Human immunodeficiency virus (HIV) resulted from the transmission of simian immunodeficiency viruses (SIV) from nonhuman primates followed by adaptation and expansion as a pandemic in humans. This required the virus to overcome a variety of intrinsic host restriction factors in humans in order to replicate efficiently. Similarly, SIV encounters restriction factors upon cross-species transmission between nonhuman primates, specifically from a natural host species such as sooty mangabey monkeys to rhesus macaques. Previously we observed significant differences in the levels of virus replication of SIV among rhesus macaques due to subtle differences in one of these restriction factors, TRIM5 among individual macaques. Although a restrictive version of TRIM5 resulted in lower viremia, we also observed that the virus spontaneously mutated in the viral capsid gene and that these mutations were associated with escape from TRIM5 restriction. In the present study, we found that introduction of these escape mutations into the parental virus confers resistance to TRIM5 both in tissue culture and in macaques. These studies provide direct evidence that TRIM5 is a critical factor influencing the cross-species transmission of SIV in primates.
TRIM5a Drives SIV Evolution macaque Rh458 (TRIM5 TFP/CypA ) was able to replicate in cell lines expressing any of the alleles. Several other clones from Rh447 and Rh458 were tested and similar results were obtained (data not shown). In contrast, SIVsmE543-3 was restricted by the TRIM5 TFP and TRIM5 CypA alleles while SIVmac239 was resistant to all the alleles in the assay, which was consistent with the previous report [49]. These results confirmed that virus from late stage plasma had escaped TRIM5a restriction. Furthermore, the escape was specific to the restrictive alleles expressed by the macaques from which they were derived. Thus clones from Rh447 (TRIM5 TFP/TFP ) only escaped TRIM5 TFP restriction while clones from Rh458 (TRIM5 TFP/CypA ) escaped both TRIM5 TFP and TRIM5 CypA restriction, suggesting that the appearance of escape virus clones is due to the specific selection of TRIM5 restriction in vivo.
Amino acid substitutions in capsid associated with SIVsmm escape from TRIM5a restriction To further investigate how SIVsmm overcame TRIM5 restriction in these animals, we isolated SIV RNA from plasma samples sequentially collected from Rh444, Rh447, Rh458 and Rh063 at different time points during the course of infection ( Fig. 1), cloned and sequenced the full gag coding regions. The capsid sequences were aligned and compared to the capsid sequences from the full length clones described above (447-1 and 458-1) and SIVsmE543-3 as shown in Fig. 3. In the permissive macaque Rh444 (TRIM5 Q/Q ), 4 of 10 clones had random amino acid substitutions and the other clones were identical to SIVsmE543-3 in capsid sequence. In contrast, common amino acid substitutions, ''P37S'', ''R98S'' and ''L135V'' were found in capsid sequences of 13 clones from macaque Rh447 (TRIM5 TFP/ TFP ). ''P37S'' and ''R98S'' were also found in all the clones from macaques Rh458 and Rh063 (TRIM5 TFP/CypA ). Clones from Rh458 and Rh063 had amino acid substitutions in the ''GPLPA87-91'' region, which is the site of CypA binding. Most of the clones from Rh458 and Rh063 also had a ''P159S'' substitution which is located at the C-terminal domain of the capsid protein. The capsid sequences were also compared to SIVmac239, which is not restricted by any of the rhesus TRIM5a alleles. SIVmac239 has mutations and a deletion, with ''LPA'' substituted by ''QQ'' in the CypA binding loop and a ''R97S'' (equivalent to R98S in SIVsmE543-3) substitution, consistent with important roles of these mutations in escaping from TRIM5a restriction, but does not have the P37S substitution. To confirm whether these substitutions conferred virus escape from TRIM5 restriction, we introduced the observed spontaneous mutations that appeared to be associated with escape from TRIM TFP (P37S and R98S) into the SIVsmE543-3 capsid and constructed a series of mutants carrying single or a combination of these substitutions ( Fig. 4 A-D). Their infectivity was measured on cell lines stably expressing a representative rhesus TRIM5 TFP allele (Mamu-2), a representative TRIM5 Q allele (Mamu-4) and the one TRIM5 CypA allele, as shown and listed in Fig. 4. While SIVsmE543-3 WT was restricted by both the TFP and CypA alleles, introduction of the P37S mutation resulted in some improvement in infectivity (Fig. 4B) while introduction of the R98S substitution had no effect on infectivity ( Fig. 4C; p.0.05, Fig.  S1). Introduction of both in combination (SIVsmE543-3 S 37 S 98 ( Fig. 4D) significantly improved infectivity relative to wild type (p,0.001, Fig. S1) to levels similar to SIVmac239. These results indicated that both the ''P37S'' and ''R98S'' substitutions are required for escape from TRIM5 TFP restriction in the context of the SIVsmE543-3 capsid. The requirement for the two substitutions was unexpected since SIVmac239 which is permissive for TRIM TFP only has the equivalent of the R98S substitution. As predicted from our sequence analysis of escape variants in macaques, the P37S and R98S substitutions had no effect on TRIM Cyp restriction.
We then focused on substitutions in the CypA binding loop, initially introducing the ''LPA89QQ'' substitution observed in SIVmac239 since SIVmac239was not restricted by TRIM5 CypA [49]. Introduction of ''LPA89QQ'' into SIVsmE543-3 greatly improved its infectivity on the cell line expressing the TRIM5 CypA allele (Fig. 4E, p,0.001, Figure S1) but had no effect on infectivity in the TRIM5 TFP cell line. The mutant SIVsmE543-3 QQ 98 replicated on the TRIM5 CypA cell line as well as on the control cell line that lacked TRIM5a expression. We also examined the LPA89QQ substitution in combination with the two substitutions we had identified as critical for escape from TRIM TFP (Fig. 4F, G and H). All the mutants carrying other substitutions combined with ''LPA89QQ'', including E543-3 S 37 QQ 89 , E543-3QQ 89 S 98 , and E543-3 S 37 QQ 89 S 98 , also replicated well in the TRIM5 CypA cell line. These results indicated that the ''LPA89QQ'' substitution is sufficient to confer virus escape from TRIM5 CypA restriction. As expected both the P37S and R98S substitutions were required to confer escape to TRIM TFP in the context of the CypA binding loop changes (Fig. 4H). Introduction of the spontaneous substitution in the CypA binding loop that we observed in our TRIM TFP/CypA macaque, H458 (G87Q) also resulted in a significant improvement in infectivity in the TRIM CypA cell line (data not shown) indicating that there are multiple avenues to TRIM5a Drives SIV Evolution PLOS Pathogens | www.plospathogens.org achieve resistance to TRIM CypA . As expected, the mutant E543-3 S 37 QQ 89 S 98 , which carried the combination of substitutions ''P37S'', ''R98S'' and ''LPA89QQ'', was able to replicate on cell lines expressing either TRIM5 TFP or TRIM5 CypA alleles. All of these mutants replicated well on cell lines expressing the permissive TRIM5 Q alleles. These results are summarized in Table 1 and confirmed that the amino acid substitutions carried by late-stage clones from macaques with TRIM5 TFP or TRIM5 CypA alleles helped the virus overcome TRIM5a restriction.

Escape from TRIM5a restriction improved virus fitness in rhesus macaques with restrictive TRIM5a alleles
We further investigated whether the escape from TRIM5 restriction improved virus fitness in rhesus macaques that expressed restrictive TRIM5 alleles. Due to the low frequency of the TRIM5 CypA allele in rhesus macaque populations, we did not have sufficient macaques that were homozygous for the TRIM5 CypA allele for in vivo studies. Hence we only compared the replication of SIVsmE543-3 and SIVsmE543-3 S 37 S 98 , the mutant which carries both the ''P37S'' and ''R98S'' substitutions, in TRIM5 TFPhomozygous macaques. Before inoculation into macaques, we investigated virus replication in cultured PBMC. After activation with phytohemagglutinin (PHA), PBMC from 18 macaques with a TRIM5 TFP/TFP genotype were infected with SIVsmE543-3 and SIVsmE543-3 S 37 S 98 at a multiplicity of infection (M.O.I.) of 0.001. PBMC from five macaques with a TRIM5 Q/Q genotype served as positive controls. Virus production was measured by monitoring reverse transcriptase (RT) activity of supernatant collected at 3-day intervals and quantified with the phosphor imaging. Fig. 5 A and B show the replication kinetics of wild type and the variant SIV in PBMC of a representative TRIM5 TFP/TFP and TRIM5 Q/Q macaque, respectively. In PBMC from macaque RhDCWW (TRIM5 Q/Q ), SIVsmE543-3 replicated as well as mutant SIVsmE543-3 S 37 S 98 . Meanwhile in PBMC from macaque RhDBF7 (TRIM5 TFP/TFP ), the mutant replicated better than wild type SIVsmE543-3. To make a quantitative comparison, the area under the replication curve (AUC), which is indicative of total amount of virus production during the culture, was calculated and compared pairwise between these two viruses. As shown in Fig. 5 C and D, the mutant SIVsmE543-3 S 37 S 98 replicated better than wild type in 18 TRIM5 TFP/TFP macaques and the difference was significant (Paired T test, P,0.01). In contrast, the replication of the two viruses in 5 TRIM5 Q/Q macaque PBMC was not significantly different (Paired T test, P.0.05).
Then we compared the replication of these wild type SIVsmE543-3 and the putative TRIM TFP -resistant variant (SIVsmE543-3S 37 S 98 ) in vivo. Twelve rhesus macaques with TRIM5 TFP/TFP genotypes were divided into two groups; relative  Table 2. Before inoculation, viruses were expanded on PBMC from the permissive TRIM5 Q/Q macaque, RhDCCW. Gag sequences of the two virus stocks were evaluated after expansion on PBMC and no additional mutations were found in either of the two virus stocks (data not shown). Each macaque was inoculated intrarectally (I.R.) with 1000 TCID 50 (5610 5 RNA copies of virus) and the infection was evaluated by monitoring plasma viral RNA load. Four weeks later, any of the macaques that remained uninfected were inoculated intrarectally on a weekly schedule with the same amount of virus until they became infected. There was no correlation between expression of a known restrictive MHC-I genotype and either acquisition or viral load in acute phase of infection. We compared the acquisition of infection of the group receiving SIVsmE543-3 and SIVsmE543-3S 37 S 98 by Kaplan-Meier curves as shown in Fig. 6A. In the group inoculated with the variant, SIVsmE543-3 S 37 S 98 , four of six macaques were infected after the first inoculation, while in the group inoculated with the wild type SIVsmE543-3 only two macaques were infected after the first inoculation. Macaques inoculated with SIVsmE543-3 S 37 S 98 required significantly less exposure to get infected than macaques inoculated with SIVsmE543-3 (log-rank test, P = 0.0452). It required 3.5 inoculations to infect half of the macaques with SIVsmE543-3, while it only required one inoculation by SIVsmE543-3 S 37 S 98 . We also compared the plasma viral loads of these two groups after infection. The plasma viral loads for each macaque and the median plasma viral load for each group are shown in Fig. 6 B and C. Macaques infected with SIVsmE543-3 S 37 S 98 had significantly higher plasma viremia compared with macaques infected with SIVsmE543-3, with mean differences of 105-fold at peak, and 25fold at 8 w.p.i. As a measure of cumulative virus replication, we also compared the area under the curve (AUC) during the acute phase of infection (1-8 weeks) and observed a 70-fold higher level

TRIM5a exerted selective pressure on SIVsmm transmission into macaques
We were interested in determining whether the changes we observed in SIVsmE543 could be generalized to other SIVsmm/ mac strains so we investigated the capsid sequences of SIV isolates commonly used in NHP models for evidence of TRIM5 selection.
Four lineages of SIV strains, including SIVsmm, SIVmac, SIVstm and SIVmne, were isolated from infected macaques during several independent SIV transmission events in the 1970s to 1980s [4,51,53,54,55,56,57,58,59,60,61,62,63,64]. The macaque passage history of commonly used SIV isolates is briefly summarized in Table 3 and the sequences of their capsid N-terminal domains are aligned and showed in Fig. 7. Phylogenetic analysis indicated that all of these SIV isolates originate from the cross-transmission of SIVsmm from sooty mangabeys, followed by unknown or intended experimental passages in different species including rhesus, pigtail and stump-tailed macaques [65,66,67]. Capsid sequences of primary clones of SIV directly isolated from sooty mangabey monkeys [66] were highly conserved. All SIVsmm TRIM5a Drives SIV Evolution clones from sooty mangabeys encoded a ''P37'' and ''R98'' in their capsid, similar to SIVsmE543-3. Some SIVsmm clones encoded an ''I91L'' variation in the CypA binding loop which does not affect sensitivity to TRIMCypA restriction [49]. The lack of variability in the sequence alignment of primary SIVsmm capsid proteins suggested that there was no TRIM5a restriction and selection during the passage of SIVsmm among sooty mangabeys. The limited passage of SIVsmm from the Tulane Primate Center in rhesus macaques resulted in the isolation of several SIVsmm clones including SIVsmH4, SIVsmE543-3, and SIVsmE660 clones FL6 and FL14 [4,54,68]. All of these clones, except for SIVsmE660-FL14, carried the''P37'', ''R98'' and ''LPA89'' observed in the capsids of primary SIVsmm clones. SIVsmE660-FL14 had a ''A91P'' substitution, which allowed escape from TRIM5 CypA restriction (data not shown), suggesting that the passage history of SIVsmE660 may have involved exposure to TRIM5 CypA selection. The long-term passage of SIVsmm in rhesus macaques resulted in the isolation of SIVmac [55,69]. SIVmac251 clones and SIV-mac239 had ''LPA89QQ'' and ''R97S'' (equivalent to R98S in SIVsmm) substitutions in their capsids. SIVmac142, which is independent of the SIVmac251/239 passage chain, had the ''P37S'' substitution in addition to the ''LPA89QQ'' and ''R97S''substitutions. SIVstm underwent an unknown history of passage in stumptailed macaques (Macaca arctoides) after the original SIVsmm crossspecies transmission. Only two SIVstm clones are available. SIVstm37.16, which was directly cloned from an infected stumptailed macaque [56], only has the ''R98S'' substitution in capsid. The other, SIVstm22579, isolated after two passages in rhesus macaques [57] has both the ''P37S'' and ''R98S'' in capsid identified as critical for escape from TRIM TFP restriction but no changes in the GPLPA motif. Finally, SIVsmPBj which has a history of passage in pigtail macaques has a substitution in the CypA binding loop of capsid, ''A91P'' conferring virus escape from TRIM5 CypA restriction, which is consistent with pigtail macaques exclusively carrying the TIRM5 CypA allele [22]. The variance of these mutations in different SIV clones was associated with their different passage history in macaques, suggestive of TRIM5 selection during their cross-transmission and adaptation. However a common pathway of escape mutations similar to what we observed spontaneously in SIVsmE543 was observed in all of these adapted viruses.

Discussion
The antagonistic interaction with host restriction proteins is considered a major driver of evolutionary change for viruses. We previously reported that the polymorphism of TRIM5a affected SIVsmm replication in rhesus macaques, which suggested that TRIM5a restriction also plays an important role in HIV/SIV evolution [49]. Here we showed that selection of TRIM5 resistant SIVsmm capsid mutants at late stages of infection in rhesus macaques expressing restrictive TRIM5 alleles. The resistance to TRIM5 restriction was associated with three common amino acid substitutions in the N-terminal domain of the capsid protein. By site-direct mutagenesis, we confirmed that these substitutions conferred TRIM5 restriction and improved the virus fitness in rhesus macaques with restrictive TRIM5 alleles. These results provide both in vitro and in vivo evidence to support the hypothesis that TRIM5 exerts selective pressure on SIV evolution.
Our results also demonstrate the value of using SIV infected rhesus macaques to study the influence of restriction proteins during HIV/SIV cross-transmission. Several reports have revealed that interactions between the host restriction proteins APO-BEC3G and Tetherin/BST-2 with HIV/SIV resulted in the acquisition and/or evolution of the Vif, Vpu and Nef proteins [70,71,72,73]. Since it is impossible to trace back the cross transmission events which occurred hundreds or thousands of years ago, most of these studies were based on phylogenetic analysis of host restriction genes and SIV sequences in different primate species followed by in vitro infection inhibition assays. The question remaining however is in what way the selection and adaptation occur. In this paper, macaques expressing restrictive TRIM5 alleles were infected by two different routes. Three macaques (Rh447, Rh458 and Rh063) were intravenously inoculated with a high dose of wild type SIVsmE543-3. For these macaques, the restrictive alleles did not prevent infection but the virus maintained a low level of replication until the appearance of mutations associated with TRIM5 resistance. The other twelve macaques were inoculated by a more biologically relevant mucosal method using repetitive lower dose intrarectal inoculation with the wild type SIVsmE543-3 or the TRIM5-resistant SIVsmE543-3S 37 S 98 mutant. This allowed us to observe that the TRIM5resistant mutant had enhanced acquisition of infection in addition to improved viral replication when compared to wild type virus.
To determine whether this was a phenomenon that could be generalized to all SIVsmm strains, we attempted to trace the influence of TRIM5 selection on the cross-species transmission of SIVsmm into captive macaques in the 1970s. The primary SIVsmm clones derived directly from sooty mangabeys showed a high degree of conservation in the capsid protein despite being isolated from separate mangabeys in different U.S. national primate centers or from wild caught animals in West Africa [66]. None of the SIVsmm clones had any of the amino acid substitutions that have been associated with escape from TRIM5a restriction. Since all evidence suggests that SIVsmm has circulated among sooty mangabeys for quite a long time, it may be completely adapted to this host and not restricted by sooty TRIM5a. Polymorphism of TRIM5a has also reported in sooty mangabeys and four different alleles have been identified [47]. However, none of these sooty TRIM5a alleles restricted SIVsmE543-3 replication in single-cycle infectivity assays (data not shown) consistent with the long-term co-evolution of this virus in this species. It also suggests that TRIM5 restriction and  capsid. The mutations varied depending on their passage history in macaques. Although there is no record to trace exactly when and how SIVsmm was introduced into captive macaques, the sequence alignment of SIV capsid of these clones revealed that the TRIM5 genotypes of the macaques during virus passage probably also exerted selective pressure on SIV evolution in macaques. SIVsmE543 had only been passaged through two rhesus macaques and at least one of these rhesus was homozygous permissive for TRIM5 [49], which is consistent with the lack of escape substitutions in the capsid of this virus. Other viruses exhibited a range of escape mutations, with viruses passaged the most extensively in rhesus macaques, such as SIVmac239, exhibiting escape from both restrictive alleles and others such as those passaged in pigtail macaques (and thus only subjected to TRIM CypA selection), only having changes in the Cyclophylin A binding loop of capsid. Such a process of restriction and selection may also affect SIV infection in primates of other species, as is indicated by the observed variation of SIV capsid sequences from  primates of different species [74]. However the amino acid sites associated with resistance to TRIM5 may be not the same as what we observed in SIVsmm-infected macaques and will require further study for identification. Regarding the significance of TRIM5 polymorphisms in modulating HIV-infection of humans, TRIM5a polymorphism has also been reported in human populations. Several nonsynonymous SNPs were found in screening of more than 1000 samples from human populations. Two nonsynonymous SNPs, resulting in amino acid polymorphisms H43Y in the RING domain and R136Q in the coiled-coiled domain, were reported to affect the in vitro TRIM5a restriction of HIV-1 and were differently distributed in HIV-1 positive and negative cohorts [43,44,45,46,75]. However, whether these SNPs actually affect virus acquisition result is controversial since some groups suggested that the effect of TRIM5a polymorphisms on HIV-1 disease progression was minimal [46]. Since HIV-1 had already circulated in a small population in Africa for several decades before causing a worldwide pandemic in the 1980s [76], the virus may have already adapted and escaped from human TRIM5a restriction during early passage in humans. Furthermore, the TRIM5a sequences within human beings are much less divergent than within rhesus macaques and only few polymorphisms in human TRIM5a were located in the SPRY domain, which may be due to the lack of positive selection by other retroviruses during the evolution of humans [43]. The low variability of human TRIM5a may also explain the lack of restriction of human TRIM5a on HIV-1 replication.
Our studies also provide a model to study the mechanism of TRIM5a restriction. The conserved capsid protein is a potential target for anti-HIV drug development because of its important role in virus replication. Studies on interaction between TRIM5a and the HIV/SIV capsid will provide useful information for anti-HIV drug development. However, the mechanism of TRIM5a restriction on HIV/SIV replication remains unclear. A recent report by Pertel et al. indicated that interaction of TRIM5a with the retrovirus capsid lattice promotes innate immune signaling [34]. Another report by Battivelli et al. showed that the CTL escape mutants isolated from HIV-infected patients are more sensitive to human TRIM5a restriction than laboratory-adapted HIV-1 strains [77]. These reports revealed that TRIM5a restriction of retroviruses is not only mediated through capsid lattice dissociation and degradation brought on by interaction between TRIM5a and the capsid protein, but are also linked to innate and adaptive immune responses. Macaques infected with SIVsmE543-3 and its TRIM5-resistant variants will be a valuable animal model to study the mechanism of TRIM5 restriction in vivo.
Our study also has quite practical applications to improve the models of SIV-infected macaques for HIV-1 research. The present study focused on escape mutations for one of the most common restrictive TRIM5 alleles to determine whether the spontaneous substitutions were associated with biological escape from TRIM5 restriction in vivo. Obviously a virus that has escaped both restrictive alleles is still required as a challenge virus for vaccine trials. SIVsmE543-3, although being pathogenic, has not been widely used as a challenge strain and thus may not be the ideal target for these studies. SIVsmE660, which was isolated from a SIVsmE543 infected macaque, is widely used as a challenge stock in the rhesus macaque model for vaccine evaluation. Several groups reported that the replication of SIVsmE660 was also affected by TRIM5 alleles: these studies suggest that TRIM restriction of this virus can affect acquisition of infection in repetitive ''low'' dose intra-rectal challenge models [78,79]. This effect can confound vaccine evaluation unless study groups are balanced for TRIM5 alleles or animals with restrictive alleles are excluded from the study. In our previous report, we described several infectious molecular clones from a SIVsmE660 virus stock and found that all of the clones were restricted by TRIM5 TFP , and one SIVsmE660 clone had TRIM5 CypA escape mutations due to substitutions in the CypA binding loop of capsid [54]. The results in this paper suggest that the introduction of the ''P37S'' ''R98S'' substitutions into SIVsmE660 clones may confer escape from TRIM5 TFP restriction, and provide virus clones not restricted by TRIM5 for use as challenge viruses in vaccine trials. The replication and pathogenesis of SIVsmE660 clones with these mutations is under evaluation in rhesus macaques.

Ethics
This study was carried out in strict accordance with the recommendations described in the Guide for the Care and Use of Laboratory Animals of the National Institute of Health, the Office of Animal Welfare and the United States Department of Agriculture. All animal work was approved by the NIAID Division of Intramural Research Animal Care and Use Committees (IACUC), in Bethesda, MD (protocol # LMM-6. The animal facility is accredited by the American Association for Accreditation of Laboratory Animal Care. All procedures were carried out under Ketamine anesthesia by trained personnel under the supervision of veterinary staff and all efforts were made to ameliorate the welfare and to minimize animal suffering in accordance with the ''Weatherall report for the use of nonhuman primates'' recommendations. Animals were housed in adjoining individual primate cages allowing social interactions, under controlled conditions of humidity, temperature and light (12-hour light/12-hour dark cycles). Food and water were available ad libitum. Animals were monitored twice daily (preand post-challenge) and fed commercial monkey chow, treats and fruit twice daily by trained personnel. Early endpoint criteria, as specified by the IACUC approved score parameters, were used to determine when animals should be humanely euthanized.

Animals
Colony-bred rhesus macaques of Indian origin (Macaca mulatta) were housed in a BSL2 facility using BSL3 practices. The TRIM5 genotype of rhesus macaques were determined as previously described [49], and MHC I genotypes were determined by the Rhesus Macaque MHC Typing Core facility at the University of Wisconsin. Four rhesus macaques (Rh444, Rh447, Rh458 and Rh063) were intravenously infected with the SIVsmE543-3 molecular clone as previously described [52]. Blood and plasma were collected sequentially. The viral RNA levels in plasma were determined by quantitative reverse transcriptase PCR (RT-PCR) and blood CD4+ T cell subsets were quantified by flow cytometric analysis as previously described [54].

RT-PCR
Virus RNA was isolated from plasma of macaque Rh447 and Rh458 with the QiaAmp Viral RNA kit (QIAGEN, Germany). Reverse transcription of viral RNA to single-stranded cDNA was performed using the SuperScript III first-strand synthesis system (Invitrogen, Carlsbad, CA) with primer R-R (59-TGC TTA CTT CTA AAT GGC AGC TTT-3) according to the manufacturer's instructions. Full-length gag-pol-env cassettes (including the entire gag, pol, vif, vpr, tat, rev, env genes and parts of the nef gene) were amplified by nested PCR using Platinum Taq Hi Fidelity (Invitrogen) polymerase. First-round PCR was performed by using 1 ml of bulk cDNA with primers Nar-F (59-GGTTGGCGC CCG AAC AGG GAC TT-39) and R-R under the following cycling conditions: 94uC for 2 min, followed by 35 cycles of 94uC for 30 s, 60uC for 30 s, and 68uC for 9 min, with a final extension of 68uC for 10 min. Second-round PCR was performed by using 1 ml of the first-round PCR product with primers Nar-F and Bgl-R (59-GGC CAA GAT CTG CTG CCA CCT CTG TC-39) under the same conditions used for the first-round PCR. PCR products were subcloned into SIVsmE543-3 by cutting with restriction enzymes NarI and BglII to construct chimeric virus clones.
To sequence the gag gene, virus RNA was isolated from plasma of macaque Rh444, Rh447, Rh458 and Rh063. Reverse transcription was performed as described above. 1.8 kb PCR products covering the full gag region were amplified from 1 ml of bulk cDNA with primers Nar-F (59-GGTTGGCGC CCG AAC AGG GAC TT-39) and Gag-R (59-GCT GAT GAT TCA ATT GTA ACA GG-39) under following cycling conditions: 94uC for 2 min followed by 35 cycles of 94uC for 30 s, 60uC for 30 s, and 68uC for 2 min, with a final extension of 68uC for 5 min. PCR products were cloned into PCR4-Topo vector with TOPO-TA cloning kit (Invitrogen) and sequenced. All sequences were deposited in GenBank under the following accession numbers (KC904980-KC905000). Gag sequences were aligned using ClustalX2 and compared with the parental SIVsmE543-3 clone (GenBank accession U72748) and SIVmac239 (GenBank accession M33262). Single-cycle infectivity assay Retroviral vector V1EGFP-SIV and Crandell-Rees Feline Kidney (CRFK) cell lines stably expressing six common rhesus TRIM5 alleles, including three TRIM5 TFP alleles, two TRIM5 Q alleles, and the TRIM5 CypA allele were described in a previous report [49]. The entire gag-pol-vif segment of SIVsmE543-3 mutants or virus clones from macaque Rh447 and Rh458 were subcloned into V1EGFP-SIV by cutting with restriction enzymes NarI and BstBI. Single-cycle SIV viruses were produced in 293T cells by co-transfection of V1EGFP-SIV and pVSV-G, titered and infectivity measured on CRFK cell lines stably expressing TRIM5 alleles as previous described [49].

Preparation of virus stocks
293T cells were maintained in Gibco GlutaMAx DMEM media plus 10% FCS, 100 U/ml penicillin, and 100 mg/ml streptomycin and transfected with 10 mg SIVsmE543-3 mutants plasmid using FuGENE 6 transfection reagent (Roche Diagnostics, Indianapolis, IN). Virus stocks were collected from the supernatant of transfected cells after 48 hours and filtered through a 0.22 um filter. The tissue culture infectivity dose (TCID 50 ) of virus stocks was tested on TZM-bl cells and calculated by the Reed & Muench method [80].
Viruses were expanded on rhesus PBMCs before inoculation into rhesus macaques. PBMCs were collected from a permissive TRIM5 Q/Q macaque, RhDCCW and cultured in complete RPMI 1640 media containing 10% interleukin-2 (IL-2) After stimulation with 2 mg/ml phytohemagglutinin (PHA) for 72 hours, activated PBMC were infected with SIVsmE543-3 and SIVsmE543-3 S 37 S 98 , respectively, at an M.O.I. of 0.001. Infected PBMCs were washed and cultured in complete RPMI 1640 media containing 10% IL-2. Virus stocks were collected at day six and filtered, followed by titration on TZM-bl cells and quantitation by realtime RT-PCR. The gag regions were amplified from virus PBMC stocks by RT-PCR and PCR products were sequenced as described above.

PBMC infection
PBMCs were collected from 18 macaques with TRIM5 TFP/TFP genotype and five with TRIM5 Q/Q genotype and activated with PHA as described above. 10 6 activated PBMCs were infected with 10 3 TCID 50 293T transfection virus stocks at 37uC for 60 minutes. The infected PBMCs were washed and cultured in complete RPMI 1640 media containing 10% IL-2. Virus production was monitored by reverse transcriptase (RT) activity of supernatant collected at 3-day intervals. RT values were quantified with Phosphor Imaging (FujiFilm, Japan). Replication curves were plotted based on RT values.

Animal inoculations
Twelve STLV, SRV and SIV seronegative rhesus macaques with TRIM5 TFP/TFP genotypes were divided into two groups. The distribution of major histocompatibility complex (MHC) class I alleles Mamu-A*01, Mamu-B*08, and Mamu-B*17 were balanced between the two groups. Each macaque was inoculated intrarectally (I.R.) with a 1:50 dilution of SIVsmE543-3 or SIVsmE543-3 S 37 S 98 PBMC virus stocks (1000 TCID 50 , 5610 5 RNA copies). After inoculation, viral RNA levels in plasma were determined by quantitative RT-PCR. Four weeks later, any of the macaques that remained uninfected were inoculated intrarectally on a weekly schedule with same amount of virus until viral RNAs became detectable in plasma. After infection, blood and plasma were collected and plasma viral RNA levels were determined.

Statistical analyses
All statistical analyses and graphic analyses were performed using GraphPad Prism5 (GraphPad Prism Software, La Jolla, CA). In vitro PBMC replication was assessed by Area under replication curves (AUC) calculation and pairwise compared by paired t test. Infectivity of SIV variants in the single cycle assay was calculated as a percentage of infectivity of the vector control in each cell line and compared statistically to the wild type using One Way ANOVA with Dunnett's Multiple Comparison Test. Kaplan Meier curves were plotted based on the inoculation number before infection and log-rank test was used to compare the acquisition of infection. Non-parametric Mann-Whitney-test was used for the comparison of viral loads between the two groups. Figure S1 Statistical comparison of SIVsmE543-3 and mutant replication on cell lines expressing different TRIM5 alleles (raw data shown in Fig. 4). The replication of SIVsmE543-3 and its mutants on cell lines expressing TRIM5 TFP (A), TRIM5 Q (B) and TRIM5 TFP (C) alleles are showed as the percentage of the replication on cell line expressing vector control. Differences of replication between mutants and wild type SIVsmE543-3 were compared by one-way analysis of variance (ANOVA) with Dunnett's post-test. Pairs of groups that differed significantly are indicated (**,p,0.01, ***,p,0.001). (TIF)