Common variants of rhesus TRIM5
can be grouped into three allelic classes: TRIM5CypA
, and TRIM5Q
. We established the existence of all six possible genotypes in a large colony of captive rhesus macaques, using archived genomic DNA samples from the Genetics Core of the New England Primate Research Center. In this colony, we observed frequencies of 46% (TRIM5TFP/TFP
), 36% (TRIM5TFP/Q
), 5% (TRIM5TFP/CypA
), 10% (TRIM5Q/Q
), 1% (TRIM5Q/CypA
), and 2% (TRIM5CypA/CypA
). These values indicate allele frequencies in one particular colony of rhesus macaques; because of differences in animal husbandry practices and potential founder effects, these values do not necessarily reflect the distribution of genotype frequencies within other captive colonies or in wild rhesus monkey populations. Nonetheless, the presence of allelic variation in the rhesus TRIM5
gene can be exploited to study the impact of TRIM5
expression in vivo.
The rhesus macaque TRIM5 coding sequence is highly polymorphic.
To ask whether TRIM5-mediated restriction plays a role in cross-species transmission and emergence of primate lentiviruses, we tested six representative alleles of rhesus macaque TRIM5
for restriction activity against four closely related viruses of old-world monkeys, SIVmac239, SIVsmE543, SIVsmE041, and SIVstm/37.16 (). SIVmac239 is a molecular clone of a highly adapted, emergent virus of rhesus macaques 
, generated in the 1980s by experimental passage of SIV-positive plasma through a series of five monkeys 
. In all likelihood, SIVmac239 is descended from a cross-species transmission event that took place in the 1960s in captive colonies of rhesus macaques 
, probably as the unintended consequence of experiments involving transfer of biological material from SIVsm-positive sooty mangabeys to rhesus macaques 
. Regardless of origin, as a result of this long association with macaques, experimental infection with SIVmac239 reproducibly results in high levels of persistent viral replication 
. In contrast, SIVsmE543-3 is a molecular clone derived by intentional inoculation of a rhesus macaque with plasma from an SIVsm-infected sooty mangabey, followed by passage through one additional rhesus macaque 
; thus, opportunity for SIVsmE543-3 to adapt to macaques was limited to only two animals. As a result, SIVsmE543-3 replication in macaques is highly variable, with acute viral loads ranging from 103
viral RNA copies/ml plasma, and set-point values from <100 to 108
. Variation in SIVsmE543-3 infected animals is consistent with an influence of genetic variation in a host gene or genes 
. SIVsmE041 is a biological isolate cultured directly from an SIV-positive sooty mangabey 
and has therefore not experienced any prior adaption to rhesus macaques. SIVstm/37.16 is an SIV isolate from a different species, the stump-tailed macaque (M. arctoides
), and represents an independent cross-species transmission event involving transmission of SIVsm directly to M. arctoides
. The relevant properties of these four viruses are summarized in .
Overview of SIV strains used in this study.
Infectivity of all four viruses was measured on cell lines stably expressing six common alleles of rhesus macaque TRIM5, including three TRIM5TFP alleles, two TRIM5Q alleles, and the TRIM5CypA allele (). SIVmac239 was resistant to all six. SIVsmE041, SIVsmE543, and SIVstm were also resistant to TRIM5Q but unlike SIVmac239 were sensitive to both TRIM5CypA alleles and TRIM5TFP alleles (). Of the four SIV strains tested, only SIVmac239 is the product of decades of replication and spread in rhesus macaques. Thus, the comparison suggests that sensitivity to TRIM5TFP and TRIM5CypA alleles represents the ancestral phenotype and that emergence of SIVsm (as SIVmac) in rhesus macaques required acquisition of adaptive changes to overcome those particular types of alleles. In contrast, the SIVsm variants that first invaded rhesus macaques were probably inherently resistant to TRIM5Q alleles.
Differential restriction of SIVmac and SIVsm strains by multiple alleles of rhesus macaque TRIM5.
To analyze the impact of TRIM5
variation on cross-species transmission directly, we acquired samples from two independent SIVsm/macaque cohorts. The smaller cohort consisted of four sooty mangabeys and four rhesus macaques that had been experimentally inoculated with SIVsmE041 
. Infection of sooty mangabeys (natural hosts of SIVsm) resulted in ready take of virus and persistent infection (unpublished data), whereas infection of macaques resulted in a transient infection followed by a decrease in viral replication to a point near or below the detection limit (). Using archived DNA, we determined that the four macaques included three TRIM5TFP/TFP
homozygotes and one TRIM5TFP/CypA
heterozygote. Thus, the alleles present in these four animals were all of the same types that restricted SIVsmE041 in tissue culture (). In one animal, replication resurged during the first year, reaching ~104
RNA copies/ml plasma (). Capsid sequences recovered from this animal revealed the appearance of two fixed changes, R97S and V108A (SIVmac239 numbering) at the late time point (). In contrast, amplification and sequencing from acute infection, from the SIVsmE041 inoculum, and from an infected sooty mangabey, revealed only the ancestral states at both positions (R97 and V108) ().
Attenuated replication of the sooty mangabey virus SIVsmE041 upon experimental cross-species transmission to rhesus macaques.
The second and larger cohort consisted of historical samples from 44 SIVsmE543-3-infected rhesus macaques. Genotype frequencies in this cohort were 30% TRIM5TFP/TFP
, 23% TRIM5TFP/Q
, 26% TRIM5TFP/CypA
, 12% TRIM5Q/Q
, 9% TRIM5Q/CypA
, and 0% TRIM5CypA/CypA
. Animals with two restrictive alleles (TRIM5TFP/TFP
) had dramatically diminished viral replication compared to TRIM5Q/Q
homozygotes, with mean (geometric) differences of 830-fold and 1,728-fold, respectively, by 8 wk post-infection (). Animals with one restrictive allele (TRIM5TFP/Q
heterozygotes) displayed intermediate levels of viral replication. Taken in conjunction with the clear differences in restriction of SIVsmE543-3 by TRIM5TFP
, and TRIM5Q
in vitro (), these results are consistent with allelic variation in TRIM5
having a significant impact on SIVsmE543 replication kinetics in rhesus macaques. We also obtained archived DNA from animal #E543, the source of the original SIVsmE543-3 clone 
, and determined that this animal had been a TRIM5Q/Q
homozygote. The fact that E543 bore a non-restrictive genotype may well have facilitated isolation of the original SIVsmE543-3 clone.
TRIM5 genotype and replication of SIVsmE543-3 in rhesus macaques.
Several reports describe a correlation between specific alleles of class I MHC
and enhanced control of SIVmac239/SIVmac251 infection in rhesus macaques 
. Specifically, MHC
alleles have been associated with lower levels of chronic phase viral replication in SIVmac239-infected animals 
. Several observations argue against the dramatic differences in viral replication of SIVsmE543-3 in rhesus macaques being due to Mhc class I rather than TRIM5. First, the TRIM5
and class I MHC
loci are on different chromosomes, reducing the probability of a chance association between suppressive alleles of TRIM5
and a specific allele or alleles of class I MHC
with activity against SIVsmE543-3. More importantly, the effects of class I Mhc on SIVmac239 replication do not manifest during the acute stage of infection 
, whereas the correlation with TRIM5
genotype is already apparent during acute infection (). Finally, Goldstein et al. demonstrated that variation in susceptibility of cells taken from naïve animals (prior to infection) and tested ex vivo correlated with viral replication levels in vivo, when the same animals were subsequently infected with SIVsmE543; such results argue in favor of an inherent, genetic cause of variation in susceptibility and against an effect of virus-specific adaptive immune responses induced by infection 
. Finally, we typed all 43 animals in the SIVsm543-3 cohort for the presence of the Mamu-B*08
alleles and found no association between these alleles and the observed differences in . Most importantly, none of the infected animals with TRIM5Q/Q
genotypes were Mamu-B*08
positive, so the statistically significant differences between those two groups cannot be attributed to Mamu-B*08
associated control of SIVsmE543. Thus, the observed correlation between TRIM5
genotype and SIVsmE543-3 replication levels is not due to a spurious association between suppressive alleles of TRIM5
and class I MHC
alleles previously associated with control of SIVmac239. It is important to note, however, that this result does not rule out a general influence of class I Mhc on viral replication levels in rhesus macaques, only that MHC
genotype does not explain the correlation depicted in . Among other things, allelic variation in MHC
class I may well contribute to the significant variation observed within groups ().
Three macaques in the SIVsmE543-3 cohort also had patterns of resurgent viral replication consistent with escape from suppression (). All three animals had two restrictive alleles and included one TRIM5TFP/TFP homozygote and two TRIM5TFP/CypA heterozygotes. We amplified gag sequences encoding the NTD of CA from all three animals, and from a TRIM5Q/Q animal, and compared these to the original SIVsmE543-3 clone (). Strikingly, an R97S change was present in every clone from all three animals, typically due to an AGA->AGC substitution, although 3/15 clones in one animal were AGA->AGT. No changes were found in the corresponding region of virus from the non-restrictive TRIM5Q/Q animal. Thus, an identical R97S change appeared independently in four animals with suppressive TRIM5 genotypes, including three SIVsmE543-infected animals () and one SIVsmE041-infected animal ().
Emergence of Trim5-resistant SIV in vivo.
Phylogenetic analyses are consistent with a minimum of two historical transmissions of SIVsm into macaques, one into stump-tailed macaques (SIVstm), and the other into rhesus macaques (SIVmac); both transmissions are thought to have occurred in captive macaques sometime prior to the 1970s 
. If TRIM5-mediated restriction influences cross-species transmission of primate lentiviruses, we predict the existence of adaptive changes in SIV isolates corresponding to such events. Indeed, alignment of the NTD of several lentiviruses in the SIVsm/SIVmac/HIV-2 lineage revealed multiple potential adaptations in SIVmac (). Most striking is an inferred R97S change, identical to the change that appeared in the SIVsmE041 and SIVsmE543 experimental cohorts described above ( and ). Based on phylogeny of the SIVsm/SIVmac/HIV-2 lineage 
, the R97S change was probably selected twice, once coinciding with emergence of SIVmac in rhesus macaques and once coinciding with emergence of SIVstm in stump-tailed macaques. Consistent with this interpretation, the underlying nucleotide substitutions are different in the two viruses (AGA->AGC in SIVstm, and AGA->TCA in SIVmac). However, S97 is also found in a small percentage of HIV-2 and SIVsm isolates; thus, it is possible that one or both historical transmissions were initiated by an SIVsm with serine at position 97, rather than de novo mutation as seen in the experimental cohorts. In either case, these combined observations strongly suggest that S97 is selectively advantageous in macaques.
Adaptations in the CA of SIVmac strains confer resistance to a subset of rhesus TRIM5 alleles.
The second putative adaptation is a highly unusual LPA/QQ substitution at the tip of the CA 4–5 loop (QQ89,90
in SIVmac239). How this change was generated is unclear but must have involved multiple point mutations and a net loss of three nucleotides. While QQ89,90
is very common among SIVmac isolates, P90
(the proline in LPA89–91
) is extremely well conserved in SIVsm and HIV-2 isolates (). The highly unusual nature of the LPA->QQ substitution, together with its location in a stretch of residues known to affect TRIM5-mediated restriction 
, mark it as a potential adaptation that arose to circumvent restriction by rhesus macaque TRIM5 proteins.
To ask whether the R97S and LPA/QQ changes in SIVmac strains arose as adaptations to overcome rhesus TRIM5, we reverted these sites to the ancestral sequence in the context of the macaque-adapted strain SIVmac239. We then tested SIVmac239S97R
for gain-of-sensitivity to different rhesus TRIM5
alleles. Indeed, the S97R reversion resulted in increased sensitivity to all three TRIM5TFP
alleles but had no effect on resistance to the TRIM5Q
alleles. In contrast, the QQ89,90
reversion resulted in sensitivity to both TRIM5TFP
alleles. This closely resembles the pattern displayed by SIVsm isolates (see ), confirming that these represent bona fide adaptations to overcome restriction by rhesus TRIM5
alleles. Because QQ89,90
is in the 4–5 loop of capsid, the adaptation probably functions by altering the TRIM5CypA binding site. The structural basis for the interaction between the TRIM5 B30.2/SPRY domain and viral CA is not yet well defined, and it is not clear why QQ89,90
also affects resistance to TRIM5TFP
alleles. However, this result is consistent with studies showing that site-directed mutations in the HIV-1 CypA-binding domain influence binding and restriction by TRIM5α proteins 
. A better understanding of the influence of CA positions 89–91 on restriction awaits more detailed, structural understanding of the interaction between CA and TRIM5α.
Neither reversion affected resistance of SIVmac239 to TRIM5Q, consistent with our observation that SIVsmE041 and SIVsmE543, which retain ancestral states at these sites, were also resistant to TRIM5Q (). We conclude that TRIM5Q alleles did not significantly hamper SIVsm colonization of rhesus macaques or its emergence as SIVmac in the 1970s. In fact, TRIM5Q/Q animals may have facilitated initial transmission of SIVsm among macaques, permitting higher levels of replication and increasing the probability of adaptation and spread.
Unlike the R97S change, we did not see the LPA89–91/QQ89,90 change appear in either experimental cohort ( and ), possibly because the multiple mutations involved make it a low probability occurrence. Once virus with the QQ89,90 motif in CA appeared, however, it probably contributed to emergence of SIVmac by facilitating spread among animals bearing restrictive TRIM5CypA alleles. Consistent with this hypothesis, experimental introduction of the QQ89,90 sequence at the homologous positions in the sooty mangabey strain SIVsmE041 by site-directed mutagenesis rendered the virus resistant to TRIM5CypA (). The mutation did not affect sensitivity to TRIM5TFP alleles or resistance to TRIM5Q alleles. Thus, reversion to the ancestral state in SIVmac239 (changing QQ89,90 to LPA89–91) resulted in sensitivity to restriction by TRIM5CypA, and the reciprocal substitution to recreate the evolutionarily derived state in SIVsmE041 (IPA to QQ) resulted in resistance to TRIM5CypA. These results are consistent with the hypothesis that the QQ89,90 sequence is an adaptation to overcome rhesus TRIM5CypA that arose during emergence of the SIVmac lineage. In the case of SIVmac239QQ/LPA, the reversion also affected resistance to TRIM5TFP alleles, suggesting that the effects of the adaptation may be influenced by additional differences between the SIVsm and SIVmac capsids. Finally, note that two mutants, SIVsmE041IPA->QQ and SIVmac239S97R, have similar patterns of restriction (compare ), consistent with the fact that these two viruses are identical at the two sites in question (i.e., both viruses are QQ89,90 and R97).