Understanding restriction factors and intrinsic immunity: insights and lessons from the primate lentiviruses

Abstract

Primate lentiviruses include the HIVs, HIV-1 and HIV-2; the SIVs, which are endemic to more than 40 species of nonhuman primates in Africa; and SIVmac, an AIDS-causing pathogen that emerged in US macaque colonies in the 1970s. Because of the worldwide spread of HIV and AIDS, primate lentiviruses have been intensively investigated for more than 30 years. Research on these viruses has played a leading role in the discovery and characterization of intrinsic immunity, and in particular the identification of several antiviral effectors (also known as restriction factors) including APOBEC3G, TRIM5α, BST-2/tetherin and SAMHD1. Comparative studies of the primate lentiviruses and their hosts have proven critical for understanding both the evolutionary significance and biological relevance of intrinsic immunity, and the role intrinsic immunity plays in governing viral host range and interspecies transmission of viruses in nature.

Primate lentiviruses

By the end of the 20th century, HIV had spread from central Africa to encompass the entire globe. Currently, it is estimated that approximately 0.8% of the human population between the ages of 15 and 49 years is infected [1]. Both HIV-1 and HIV-2 are derived from SIVs, which were transmitted from natural hosts (African primates) to humans, probably in the early-to-mid 20th century [2–4]. The four currently known subtypes of HIV-1 (M, the pandemic strain; N; O; and P) are thought to be the result of four single, independent transmissions of SIV from chimpanzees (SIVcpz) into humans. In the case of groups M and N, SIVcpz is probably the direct ancestor. The HIV-1 group O and P viruses cluster closest with SIV of gorillas (SIVgor), which itself is derived from SIVcpz [5,6], indicating that both chimps and gorillas are reservoirs for lentiviruses capable of infecting humans. Another reservoir species, the sooty mangabey, harbors viruses (SIVsmm) closely related to HIV-2 groups A through I, with each of the nine groups likely originating from an independent transmission of SIVsmm [7–12]. While HIV-1 M has spread globally, even the most frequently found clades of HIV-2, A and B, are mostly confined to western Africa. All other subtypes are rare, sometimes consisting of a single infected individual. But, in all cases, it is possible that the viruses underwent a significant period of adaptation to humans as hosts between the time of the original transmissions and the emergence of HIV/AIDS in the 1980s [13,14].

At present, over 40 naturally occurring types of SIV have been identified in primate populations in Africa [14,15]. There is serological evidence that humans are frequently exposed to retroviruses from multiple nonhuman primate reservoirs, including viruses that were already the sources of HIV-1 groups M, N (SIVcpz from chimpanzees), O and P (SIVgor from gorilla) and HIV-2 groups A through I (SIVsmm from sooty mangabeys) [2–4,12,16–17]. In fact, cross-species transmission appears to be a common theme in the natural history of these viruses. The host and virus phylogenetic trees have different topologies, indirect evidence that the distribution of extant viruses among modern African primates must reflect, at least in part, a history of interspecies jumps (Figure 1) [13–14,18].

Figure 1. Tanglegram of primate species and primate lentiviruses.

(Left) Primate species; (right) primate lentiviruses. Dotted lines connect the primate lentivirus(es) to their respective hosts. Subsets of the HIV-1 and HIV-2 types infecting humans are highlighted in blue (HIV-1) and green (HIV-2), respectively; viruses isolated from captive macaques are highlighted in red. The primate phylogeny is derived from [19]. The primate lentivirus phylogeny is based on pol sequences obtained from the Los Alamos National Library HIV sequence database. Neighbor-joining trees were generated using the Tree Builder algorithm in Geneious (Auckland, New Zealand), pruned with Mesquite [20] and rendered using FigTree [21].

Phylogenetic analyses also suggest a role for recombination in cross-species transmission and emergence of several SIV lineages, such that some species of SIV appear to be descended from two or more distinct founder viruses [14]. Recombination between retroviruses requires that coinfection of the same cell in a single individual with both parental viruses must have occurred [22]. In addition to HIV-1 and HIV-2, there is evidence that SIVcpz of chimpanzees (Pan troglodytes sp.), SIVgor of gorillas (Gorilla gorilla gorilla), SIVmus-1 and SIVmus-2 [23] of the mustached monkey (Cercopithecus cephus), SIVgsn [24] of the greater spot-nosed monkey (Cercopithecus nictitans), SIVagmSab of the sabaeus subspecies of African green monkey (Chlorocebus sabaeus) [25], SIVdrl of drills (Mandrillus leucophaeus) and SIVmnd-2 of mandrills (Mandrillus sphinx) are the results of natural cross-species transmissions. Phylogenetic comparisons suggest that emergence of at least five of these, SIVcpz [26], SIVagmSab [27], SIVrcm, SIVdrl and SIVmnd-2 [28], involved recombination between two or more parental SIV strains (reviewed in [29]). The relative importance of recombination for interspecies transmission and emergence of new viruses remains an open question.

Intrinsic immunity: early insights from animal models of aids

Beginning in the 1970s, and prior to the emergence of AIDS in humans or the discovery of HIV-1, AIDS-like symptoms began to affect macaques housed in National Primate Research Centers in the USA (Figure 2). Shortly after the detection and isolation of HIV-1, the cause of AIDS in macaque colonies was identified as an HIV-like virus, now known as SIVmac [30]. This was followed fairly quickly by the first description of HIV-2 [31], and the discovery of naturally occurring SIVs in many African primates. In fact, SIVmac and HIV-2 are both descended from SIVsmm [14,32]. Experimental infection of macaques quickly became the predominant experimental model for preclinical AIDS research, and the discovery of several intrinsic immune effectors can be traced from early work with the SIV/macaque model.

Figure 2. Timeline of events and discoveries in the field of primate lentivirology from the 1960s to present.

Red: viral transmissions, bold: discoveries of viruses, blue: discoveries of host restriction factors. Special interest references are indicated below the timeline.

SHIVs

Despite the obvious relatedness of HIV-1 to the newly discovered SIVs, researchers discovered that HIV-1 could not replicate in monkey cells [33]. This prompted the creation of recombinant SIV/HIV chimeric viruses (SHIV), in which subsets of viral genes from one parent virus were replaced with the homologous regions of the other [34–36]. While env and some of the accessory gene loci of SIV could be substituted with HIV-1 sequence and still form viruses that replicated in monkey cells, replacement of the SIV gag gene with HIV-1 gag resulted in a severe restriction of replication in monkey cells. The capsid (CA) domain within gag turned out to be the determinant of the species-specific patterns of SIV and HIV-1 infectivity, and subsequent work determined that the CA-dependent block manifests after viral entry, but prior to integration [37].

Hofmann and colleagues conducted a broad survey of primate cell lines, comparing single-cycle infectivity of HIV-1 and SIVmac [38]. They tested cell lines of human, Old World monkey, New World monkey, prosimian, rodent, lagomorph, carnivore and bovid origin, using viruses pseudotyped with a heterologous envelope protein to circumvent cell-specific or species-specific differences in receptor expression. They found that the early block to HIV-1 in rhesus macaque cells was part of a larger pattern in which almost all Old World monkey cells were refractory to HIV-1 infection. By contrast, New World monkey cells were generally susceptible to single-cycle HIV-1 infection, but were not readily infected with SIVmac (with the exception of cell lines from owl monkeys). Similar blocks were seen in cells from rabbits and cows, and prosimian cells were resistant to both virus types [38].

Ultimately, a screen for rhesus macaque cDNAs that could block HIV-1 infection of human cells led to the identification of TRIM5α as the responsible factor [39]. TRIM5α is a cytoplasmic protein with retrovirus-specific antiviral activity, and work from a number of laboratories has shown TRIM5α-mediated restriction of a range of retroviruses, including gammaretroviruses, betaretroviruses, lentiviruses and spumaviruses [39–44]. TRIM5α belongs to the large family of tripartite motif (TRIM) proteins. TRIM proteins are involved in many cellular functions, including antiviral defense [45–48]. TRIM5α can restrict diverse lentiviruses; it binds the viral capsid core through directed interactions mediated by its C-terminal SPRY domain, forms a multi-meric lattice and causes the premature disassembly and proteasomal degradation of the capsid structure [49–54].

Sequencing of the TRIM5 coding sequence from multiple species and cell culture experiments confirmed that the species-specific patterns of restriction could be explained by divergence in the amino acid sequences of the SPRY domain of TRIM5α [44,55–58]. The restrictive properties of TRIM5 were also discovered by two groups working with owl monkey cells (Aotus sp.) – surprisingly, they found that the TRIM5 locus of owl monkeys does not encode the α-isoform, which contains a C-terminal SPRY domain [49,59]. As the result of an ancient, LINE-mediated retrotranspositional insertion, owl monkeys instead express a chimeric protein in which the C-terminal SPRY domain has been replaced with a Cyclophilin-A (CypA) domain. Because CypA can bind to the capsid proteins of some lentiviruses, owl monkey TRIM-CypA1 can act as a restriction factor [49,59]. A strikingly similar event occurred in the Old World macaques, also resulting in a TRIM5–CypA chimeric protein (discussed in a later section) [60–64].

Accessory genes & live-attenuated SIV

All primate lentiviruses share the three canonical gag, pol and env retroviral genes, which encode the basic structural and enzymatic proteins required for infection and replication. Additionally, they possess a variable repertoire of accessory genes (Table 1). nef, vif and vpr are shared among all primate lentiviruses, while vpu is only found in certain viruses, such as SIVgsn, SIVgor, SIVcpz and HIV-1. Similarly, a different accessory gene called vpx is only found in some viruses, such as SIVrcm, SIVsm, SIVmac and HIV-2. While the accessory proteins are not required for basic replication of the virus in most cell lines, their importance in vivo has been proven by both experimental and naturally occurring accessory gene deletion mutants. For example, SIVmac239 strains deleted in one or more accessory genes are attenuated to various degrees in vivo, ranging from mild (Δnef/Δvpr), to moderate (any combination involving Δvpx), to severe (any combination involving Δvif) [65–67].

Table 1.

Accessory gene repertoire of representative primate lentiviruses.

Accessory genome composition	Examples
vif, vpr, nef	SIVagm, SIVmnd-1, SIVcol
vif, vpr, vpx, nef	HIV-2, SIVmac, SIVsmm, SIVmnd-2, SIVdrl, SIVrcm
vif, vpr, vpu, nef	HIV-1, SIVcpz, SIVgor, SIVgsn, SIVmus-1

Similar effects were documented in cases of human HIV-1 infection wherein the virus had inactivating mutations in nef. For example, the Sydney Blood Bank Cohort comprises a small group of patients unintentionally infected with a naturally occurring nef-deleted strain of HIV-1 originating from a single blood donor. All members of the cohort qualified as long-term nonprogressors, initially, and even after 30 or more years of infection with this attenuated HIV-1 strain, a subset of individuals maintain their nonprogressor status [68,69]. In an unrelated longitudinal study, only nef-deleted HIV-1 could be recovered from a long-term nonprogressor patient [70]. In both cases, long-term attenuation of HIV was not mediated by adaptive immunity, but rather caused by abnormalities of the viral accessory gene repertoire.

Research on accessory gene function has led directly to the identification of at least three restriction factors, APOBEC3G (and related enzymes of the APOBEC3 cluster), BST-2/tetherin and SAMHD1 [71–75]. In all three cases, the antiviral function of these factors is actively thwarted by the activity of one or more lentiviral accessory proteins. The first restriction factor to be identified was APOBEC3G (A3G), which is targeted for degradation by viral infectivity factor proteins (Vif) [71]. A3G is a cytidine deaminase that incorporates into assembling virions and causes G-to-A hypermutation of the proviral DNA during reverse transcription in the subsequent target cell [76,77]. Additionally, there is evidence that A3G may also inhibit the reverse transcription complex in a deaminase-independent manner [78–80]. Vif recruits a multiprotein E3 ubiquitin ligase complex to mediate proteasomal degradation of A3G, thereby preventing incorporation of A3G into budding virions [81–84]. Exclusion of A3G from virions prevents both inhibition of the reverse transcription complex and hypermutation during reverse transcription.

HIV-1 mutants with an inactivated vpu gene display a striking phenotype in cell culture: electron microscopy images reveal mature virions built up in thick layers around cells, unable to separate from the cell surface [72,75]. mRNA microarray experiments identified BST-2, also called tetherin, to mediate this phenotype. BST-2 localizes to the plasma membrane, where it prevents freshly budded virions from moving away from the cell. Vpu counteracts BST-2 through direct binding and endocytosis-mediated removal of BST-2 from the plasma membrane. Not all primate lentiviruses have a vpu gene (Table 1), and in such viruses, the anti-BST-2 activity is usually provided by Nef [85,86].

SAMHD1 was identified in proteomic screens as a major Vpx-interacting protein [73,74]. SAMHD1 is a myeloid cell-specific phosphohydrolase, which depletes the intracellular concentration of dNTPs, thus severely inhibiting synthesis of viral DNA by the reverse transcription complex. This phosphorylation-dependent activity is restricted to resting cell populations, and missing in activated cells [87]. Depletion of the dNTP pool would be expected to inhibit other virus types as well, and a diverse panel of retroviruses as well as the DNA virus HSV-1 have indeed been found to be sensitive to SAMHD1-mediated restriction [88,89]. Primate lentiviruses use Vpx (and in some cases, Vpr [90]) to target SAMHD1 for proteasomal degradation.

Restriction factors & intrinsic immunity in primates

Comparative approaches have proven useful in the study of intrinsic immunity and the restriction factors, particularly in the case of primate hosts (including humans) and their lentiviruses (HIV-1, HIV-2 and the SIVs). Molecular evolutionary analyses, including phylogenetic comparisons and statistical tests of positive selection (e.g., d(N)/d(S) estimates), have proven especially informative when combined with cell culture experiments. Such approaches consistently indicate that the restriction factor genes in primates, including those encoding APOBEC3G, TRIM5, BST-2/tetherin and SAMHD1, have been subject to strong, recurring bouts of positive selection throughout much of primate evolution [91]. In several cases, evolutionary analyses have guided experiments to map interacting domains and key residues involved in virus–host interactions [53,90,92–95]. Dating of signatures of positive selection in host intrinsic immunity genes and cell culture experiments strongly suggest that viruses related to the extant primate lentiviruses were already present several million years ago and infected ancestors of the Old World primates [96], a possibility that is indirectly supported by the presence of endogenous SIV-like sequences in the genomes of some prosimians [97].

Evolutionary studies, genetic surveys, laboratory experiments and animal models of lentiviral infection together suggest that intrinsic immunity has a significant impact on viral host range in nature, and may serve as an important barrier to cross-species transmission of viruses and emergence of new viral diseases. The following sections focus on four restriction factors as examples of how comparative approaches are shedding light on the evolutionary significance and biological relevance of intrinsic immunity.

TRIM5

TRIM5 belongs to a very old and large gene family, and TRIM family members can be found widely among metazoans [98,99]. The mammalian TRIM5 gene has been most extensively studied in primates, and comparison of the TRIM5 locus across primate lineages reveals a very high degree of protein sequence diversification, particularly in the C-terminal SPRY domain [55,56]. On at least two separate occasions, the SPRY domain has been replaced with a CypA domain, giving rise to the TRIM-CypA fusions in South American owl monkeys (Aotus sp.) and Asian macaques (Macaca sp.) [49,60]. The presence of degraded CypA sequences downstream of the TRIM5 locus in the genomes of some primates suggests that additional TRIM-CypA fusions may have arisen more than once during the evolution of Old World primates [100].

In primates, the TRIM5 SPRY domain has evolved under strong positive selection for at least 23 million years, and substitutions in positively selected sites affect virus-target specificity in cell-culture experiments, consistent with a protein engaged in long-standing or repeating cycles of genetic conflict with viral pathogens [55]. Positively selected residues largely fall into variable ‘patches’ in the SPRY domain coding sequence, and in some lineages these same variable regions have also undergone changes in length due to in-frame duplications of short stretches of sequence [55,56]. For example, the V1 region of African green monkey TRIM5α and related species has a duplication resulting in 60 additional nucleotides (Figure 3) [55,56]. It is therefore likely that positive selection has acted upon both kinds of variation – amino-acid substitutions and length differences in variable regions – in shaping the interactions of primate TRIM5 with retroviral pathogens.

Figure 3. Some key events in the evolution of primate restriction factors, overlayed on a tree depicting evolution of New world and Old world primates.

Branches are not to scale.

Among Old World primates, the TRIM-CypA allele has only been found in Asian macaques, suggesting an origin subsequent to the migration of macaque ancestors out of Africa. Although the CypA domain of macaque TRIM-CypA is much younger than the SPRY domain, there is also evidence for functional diversification of the CypA domain in the macaque lineage [101]. Some residues in the TRIM5 coiled-coil (CC) domain also appear to have evolved under positive selection [55,58]. Selection on residues in the SPRY domain can be attributed to direct interaction with the capsid proteins of retroviral pathogens; selection on the CC domain is harder to explain, but it is worth noting that the CC domain of another TRIM protein, TRIM25, is bound by a viral antagonist – the NS1 protein of influenza A virus [102]. As with TRIM25, perhaps the CC domain of primate TRIM5 is (or was) also the target of an unknown virus-encoded antagonist that drove diversification of these residues.

Altogether, rhesus macaques possess at least three classes of TRIM5 alleles, encoding three distinct TRIM5 proteins that differ in target recognition properties, referred to as rhTRIM5^Q, rhTRIM5^TFP and rhTRIM5-CypA [60–64]. The rhTRIM5^Q and rhTRIM5^TFP alleles contain C-terminal SPRY domains, and differences in specificity map to a complex Q/TFP polymorphism at position 339–341. The third, rhTRIM5-CypA, expresses a fusion protein in which the SPRY domain is replaced with a CypA domain – TRIMCypA fusions inhibit lentiviruses by interacting with a specific loop in lentiviral capsids (the 4–5 loop, also known as the CypA-binding loop). At least four TRIM5 alleles have been reported in crab-eating macaques (Macaca fascicularis), mfTRIM5^TFP, mfTRIM5^Q, mfTRIM5-CypA1 and mfTRIM5-CypA2 [101]. In captive pigtailed macaques (Macaca nemestrina) TRIM5α-encoding allele(s) have not been found, and it is possible that the TRIM5-CypA allele (mnTRIM5-CypA) became fixed in this species [62–63,101].

Of the two allele types with SPRY domains, the Q at position 339 (rhTRIM5^Q alleles) is ancestral, whereas the TRIM5^TFP allelic lineage arose sometime in the past 10 million years (Figure 3) [51]. Mutations that sensitize the macaque virus SIVmac239 to restriction by TRIM5^TFP are closely clustered in one region of CA (see below) [51].

The first in vivo evidence that TRIM5 has an impact on lentivirus host switching came from retrospective analysis of two experimental cohorts of SIVsmm-infected rhesus macaques [103]. Genotyping of over 40 SIVsmE543-3-infected rhesus macaques in the first cohort revealed a significant correlation between plasma viral loads and TRIM5 genotype. Specifically, animals expressing alleles that were restrictive in cell culture experiments (genotypes rhTRIM5^TFP/TFP, rhTRIM5^TFP/CypA and rhTRIM-5^CypA/CypA) had the lowest viral levels in plasma, whereas animals homozygous for the nonrestrictive rhTRIM5^Q allele had the highest levels of viremia. Heterozygous animals (animals with one restrictive allele and one permissive allele) had intermediate viral loads. Notably, over time, viral loads increased in a small number of animals with restrictive TRIM5 genotypes. Virus sequences recovered at late time points from these monkeys shared an R97S substitution at the base of the helix 4–5 surface loop of the viral CA protein. This was confirmed as a rhTRIM5^TFP escape mutation, and cell culture experiments confirmed that this residue functions as an important determinant of TRIM5 sensitivity [51,103]. Arginine is also found at the homologous position in HIV-1 CA, and comparison of the HIV-1 and SIVmac CA crystal structures indicate that the R-to-S change results in a rearrangement of hydrogen bond networks at or near the base of the CypA binding loop [51].

A second cohort, consisting of four rhesus macaques infected with a primary SIVsmm isolate (SIVsmE041) was also analyzed [103]. All four animals possessed at least one restrictive TRIM5 allele. While three of the four monkeys have controlled SIVsmE041 replication for 7 years postinfection [103] [Kirmaier A, Hall LR, Kaur A, Johnson WE, Unpublished Data], virus rebounded in the fourth animal around 4 months postinfection. Viral sequences recovered from this animal’s plasma revealed the same R97S escape mutation in CA [103].

The importance of the R97S mutation was confirmed by two related studies [51,104]. Efforts to map the binding site of rhTRIM5α on CA demonstrated that viruses with either a serine or arginine at CA position 97 scored the highest impact on differential recognition by rhTRIM5α [51]. During experimental infection of rhesus macaques with SIVsmm two changes in capsid emerged: P37S and R98S (equivalent to R97S in SIVmac) [104]. When the two mutations were introduced into the SIVsmE543-3 clone, the virus displayed partial resistance to rhTRIM5^TFP in vitro, and ex vivo it replicated significantly better in rhesus macaque primary blood mononucleated cells than the parental SIVsm clone [104]. Thus, multiple independent lines of evidence show that the emergence of SIVsmm as pathogenic SIVmac in rhesus macaques in the 1970s involved adaptation of CA to evade recognition by rhesus TRIM5 proteins. The outcome of experimental cross-species transmission of SIV in rhesus macaques is therefore largely dependent on the TRIM5 genotype of the recipient monkey, and provides direct proof that adaptation in CA facilitates escape from TRIM5-mediated restriction in vivo.

APOBEC3G

Comparison of the complete A3G coding sequences from 11 primate species, including New World monkeys, Old World monkeys and hominids, suggests that A3G has been evolving under positive selection for at least 33 million years [92]. Since APOBEC3 is encoded in all placental mammals [105,106], the theoretical upper limit of APOBEC3-mediated host–virus genetic conflict could be up to 100 million years before present [19]. A3G has also provided evidence for the existence of ancient primate lentiviruses. For example, a multiresidue insertion event in a subset of Old World monkeys (Colobinae), which allows the evasion of Vif-mediated degradation, suggests that SIV-like viruses may have circulated among the ancestors of modern primates as far back as 12 million years ago [96].

Like TRIM5, APOBEC3G of any given primate species usually does not restrict the primate lentivirus naturally infecting that species, indicating Vif-specific adaptation and resistance to A3G of the native host [107–109]. Within primates, interspecies variation at amino acids 128, 129 and 130 affect sensitivity to different Vif proteins, suggesting that these are important determinants of A3G antilentiviral activity [76,108–114] and that different Vif proteins may bind the same or overlapping elements of A3G proteins [112–113,115].

Evidence for the in vivo significance of the A3G–Vif interaction for host range determination comes from the study of SIVagm in African green monkeys (Chlorocebus sp.). The four sub-species of African green monkey are each infected with a different subtype of SIVagm. A3G in these subspecies has been subject to recent diversifying selection in wild populations, with nonsynonymous SNPs clustering at surfaces targeted by Vif (A3G residues 128 and 130). These mutations in A3G were selected to allow evasion of SIVagm Vif proteins, implicating Vif as the selective pressure. A3G-driven evolution of Vif is a naturally occurring phenomenon, and the specificity of Vif proteins from the four SIVagm subtypes reflects adaptations to subspecies-specific A3G variants in African green monkey populations exposed to SIV in the wild rather than in the laboratory, demonstrating a case of contemporary selection and adaptation between primate lentiviruses and their natural primate hosts [116].

Further evidence for the importance of A3G antagonism comes from the finding that the Vif protein in some primate lentivirus lineages has evolved alternative or modified A3G interaction sites; for example, A3G of the mantled colobus monkey (Colobus guereza) features a multiresidue insertion at amino acid 66 and binding of SIVcol Vif is affected by this A3G domain rather than the 128/129/130 site [112]. This pattern could have occurred either because emergence of SIVcol (after cross-species transmission of a virus from another host) acquired the ability to antagonize this unique form of A3G, or alternatively, this novel A3G variant arose in the host population and rendered a previous allele of SIVcol Vif nonfunctional.

Gorilla A3G resists degradation by most lentiviral Vifs and should therefore effectively serve as a barrier to cross-species transmission of SIV [113]. Yet gorillas did acquire SIVcpzPtt once [5,6], implying that this virus was able to overcome gorilla A3G. Two SIVcpz Vifs indeed exhibit some activity in the presence of gorilla A3G [113], suggesting that partial resistance to gorilla A3G might have lowered the genetic barrier SIVcpzPtt had to overcome during its emergence as SIVgor.

SIVsmm is the ancestor of both SIVmac in macaques and HIV-2 in humans. Compton et al. reported that the primary clone SIVsmE041 is resistant to rhesus and human A3G, a property that could have aided in its cross-species transmission and emergence [112]. However, a survey of the rhesus macaque A3G locus revealed the existence of three A3G alleles, distinguished by unique mutations at amino acid position 59/60 (A3G^Y, A3G^LL and A3G^LR). SIVsmE041 Vif is not able to degrade A3G^LR. By contrast, SIVmac, the virus that emerged after transmission of SIVsmm into captive macaque colonies, efficiently degrades all three rhesus macaque A3G alleles. A G17E substitution in the Vif protein of all SIVmac isolates is responsible, and represents an adaptation that must have occurred during emergence of SIVmac [117]. These results establish that A3G can serve as barrier to cross-species transmission and emergence of primate lentiviruses.

BST-2/tetherin

The discovery of BST-2’s antiviral activity and its antagonist HIV-1 Vpu immediately raised the question of whether all primate lentiviruses are affected by BST-2, and in particular those that do not encode Vpu. Vpu is found in the Cercopithecine lineage of SIVs but is not encoded in the SIVsm/SIVmac/HIV-2 lineage. As it turned out, in the majority of primate lentiviruses, including many Cercopithecine viruses, Nef counteracts BST-2 instead [85,86]. So why did HIV-1 evolve to use Vpu? Human BST-2 has a five amino acid deletion in its cytoplasmic domain encompassing the Nef binding site, rendering it resistant to most lentiviral Nefs. The ancestors of HIV-1 and HIV-2 (SIVcpz and SIVsm) consequently had to re-establish BST-2 antagonism, which was accomplished by Vpu in HIV-1 and by the envelope protein in HIV-2 [118,119]. Independent acquisition of BST-2 antagonism in three different lentiviral accessory genes emphasizes the significance of BST-2 as a component of host antiviral defense. The five amino acid deletion in human BST-2 is also found in Neanderthal and Denisovan genomes, suggesting the deletion occurred >800,000 years ago; it is interesting to speculate that deletion of the Nef-binding site in the Homo sp. lineage reflects prior selection by an ancient virus with Nef or Nef-like activity [120].

Rapid acquisition of anti-BST-2 activity by nef-deleted SIVmac after passage in three rhesus macaques further highlights the severe selective pressure exerted on viruses by BST-2. Viruses recovered from these animals had both acquired changes in the viral Env glycoprotein that specifically counteract rhesus BST-2 [121,122]. The changes in Env conferring anti-BST-2 activity were not in the same sites that HIV-2 Env uses to interact with human BST-2, a case of convergent evolution in the same gene of two viruses to gain resistance to BST-2-mediated inhibition.

Analyses of the primate BST-2 locus spanning 33 millions years of evolution revealed that the strongest signatures of selection fall into the Nef-interacting domain of BST-2 [123]. Among the many enveloped viruses that could have influenced the evolution of primate BST-2, viruses encoding Nef or a Nef-like protein may have played a significant role in shaping this locus for tens of millions of years.

SAMHD1

Some (but not all) primate lentiviruses use Vpx to counteract the effects of SAMHD1 [73–74,124]. The vpx and vpr accessory genes arose by gene duplication/recombination sometime during evolution of the primate lentiviruses [125]. So far, at least two studies have examined the SAMHD1 gene of primates for evidence of positive selection [90,93]. Both studies identified clusters of positive selection, and site-directed mutagenesis confirmed that selected residues are determinants of sensitivity/resistance to Vpx proteins. Curiously, HIV-1 (and related SIVs such as SIVcpz) does not encode Vpx or a Vpx-like activity, and HIV-1 replication in terminally differentiated myeloid cells can be significantly enhanced by SIV Vpx in trans [124] – it remains an open question as to why the HIV-1 lineage has not adapted to overcome SAMHD1, and whether there may be a selective advantage to not having a SAMHD1 antagonist and/or to limiting replication in specific myeloid cell types.

Future perspective

With respect to intrinsic immunity and the restriction factors, most research efforts have focused primarily on the molecular and biochemical details of mechanism, and to a lesser extent, regulation. Far less is known about the biological significance of intrinsic immunity in nature, particularly at the organismal and population levels, and with respect to emerging viral diseases and the jumping of viruses between hosts.

Molecular evolutionary studies and primate models together suggest that intrinsic/innate effectors such as TRIM5, APOBEC3, BST-2 and SAMHD1 may have their greatest impact as determinants of viral host range and the ability of viruses to transmit, spread and adapt to new host populations [91]. Although experimental infection of rhesus macaques has provided some insight, the degree to which restriction impacts cross-species transmission in outbred host populations in nature remains unknown and will be difficult to extrapolate from cell culture studies and animal models. Is sensitivity to any one restriction factor enough to ensure that a particular cross-species exposure is unlikely to result in spread and emergence of a new viral disease, and what is the effect of multiple restrictions functioning in tandem? An intensified focus on the genetics of interspecies transmission may help researchers determine if there are fundamental rules that can be used to help predict the likelihood of specific zoonoses and new emerging viral diseases. Understanding the genetics of restriction and intrinsic immunity also has practical benefits for animal models of AIDS, most of which are currently based on genetically outbred primates – for example, as some of the blocks to HIV-1 infection have been identified, several groups have been able to engineer simiantropic HIV-1 strains that circumvent restriction and replicate in macaques [126–128]. Moreover, because genetic variation in restriction factor genes is likely to be a confounding element in small, preclinical vaccine and pathogenesis studies in macaques, identifying and accounting for the contributions of different loci can be used to improve the statistical evaluation of outcomes.

Five restriction factors with known activity against lentiviruses (TRIM5, APOBEC3G, BST-2, SAMHD1 and, most recently, Mx2) can be used to illustrate the complexity of the problem. First, all five factors work by distinct mechanisms, attacking different weaknesses in the retroviral replication cycle. Unlike cell culture studies, where individual restrictions are studied in isolation, in nature two or more factors may come into play at once. When SIVsmm emerged as SIVmac in macaque colonies, the virus had to adapt to overcome blocks imposed by rhesus orthologs of TRIM5 and APOBEC3G [103,117]. Similarly, it is likely that emergence of HIV-1 in humans required, at a minimum, adaptation of Vpu as an antagonist of human BST-2, and it is likely that other adaptive changes were required [14].

A survey of several different organisms reveals that the known restriction factors are often distributed across multiple chromosomes (Table 2). In organisms with multiple functionally distinct alleles at such loci, random reassortment will contribute to phenotypic diversity at the population level (Table 3). In rhesus macaques, for example, multiple alleles have been described for TRIM5 (three alleles), BST-2 (two alleles) and APOBEC3G (three alleles). Because the genes are located on three different chromosomes, there can be up to 108 (6 × 3 × 6) composite genotypes involving just these three restrictions in rhesus macaques. As other restriction factor genes are discovered and studied, this number is likely to increase. Other hypothetical examples are given in Table 3. Thus, this sort of intrinsic immune complexity may be what many viruses encounter in nature when they invade new, naive host populations.

Table 2.

Chromosomal locations of restriction factors in five mammalian species.

Locus	Homo sapiens	Pan troglodytes	Macaca mulatta	Callithrix jacchus	Rattus norvegicus
APOBEC3G	chr. 22	chr. 22	chr. 10	chr. 1	chr. 7†
BST-2	chr. 19	chr. 19	chr. 19	chr. 22	chr. 16
Mx2/MxB	chr. 21	chr. 21	chr. 3	chr. 21	chr. 11
SAMHD-1	chr. 20	chr. 20	chr. 10	chr. 5	chr. 3
TRIM5	chr. 11	chr. 11	chr. 14	chr. 11	chr. 1

^†The common rat encodes only one allele of APOBEC3.

chr.: Chromosome.

Table 3.

Hypothetical numbers of restriction factor genotypes based on numbers of functionally distinct alleles at five independently assorting loci.

Loci		Number of alleles
APOBEC3G	1	3	2	3
BST-2	1	1	2	3
Mx2	1	1	2	3
SAMHD1	1	1	2	3
TRIM5	1	3	2	3
Composite genotypes	1	36	243	7776

The existence of multiple independent restrictions resembles HAART, in which HIV-infected individuals are treated with a combination of inhibitors that target two or more distinct steps in the viral replication cycle. Just as resistance to HAART results from selection of multiple, separate adaptations on the part of the virus (often at a cost to viral fitness), the ability of a virus to become fully adapted to a new host is likely to require combinations of multiple, independent adaptations to circumvent intrinsic immune effectors acting on different parts of the viral replication cycle. We know that for APOBEC3G, BST-2 and TRIM5 one or two point mutations in viral genes suffice to abrogate restriction [94,103,117,129]. The presence of at least four restriction factors (APOBEC3G, BST-2, SAMHD1 and TRIM5; the recently discovered MxB [130–132] may be a fifth, and there are likely to be other undiscovered blocks to lentiviral infection) likely exerts a combined pressure on a newly transmitted virus, analogous to how HAART would in an HIV patient. The situation in outbred populations raises several important questions: given the number of restriction factors and the complexity of the genetics of restriction, is interspecies transmission of lentiviruses more likely between closely related hosts? In this regard, it is worth noting that HIV-1, which emerged from ape reservoirs, has spread much more efficiently in humans than HIV-2, which emerged from a more distant primate (sooty mangabeys). When viruses do successfully cross species lines, does emergence require simultaneous or near-simultaneous acquisition of multiple resistance mutations, and to what degree does recombination enhance this likelihood? Such questions are relevant to understanding interspecies jumping, adaptation and emergence of all kinds of viruses [133] – perhaps further investigation of the primates and their lentiviruses will help provide the answers.

EXECUTIVE SUMMARY.

Primate lentiviruses

• The primate lentiviruses include HIV-1 and HIV-2 in humans, SIVmac and related viruses found in captive macaques, and approximately 40 species of SIV endemic to African primates.

• The distribution of lentiviruses among modern primates is consistent with a natural history of interspecies transmission, adaptation and emergence of new viral lineages.

• Recombination may have played an important role in cross-species transmission of several primate lentiviruses.

Restriction factors (intrinsic immune effectors)

• The restriction factors include various intrinsic antiviral effectors, such as APOBEC3G, TRIM5α, BST-2/tetherin and SAMHD1.

• Restriction factors come in many different forms, act by different mechanisms and exploit weaknesses at different steps in the viral replication cycle.

• As a result of positive selection, restriction factor coding sequences in primates often exhibit significant interspecies divergence and dN/dS values >1.

• Primate restriction factors often exhibit significant nonsynonymous polymorphisms.

• Many of the known restriction factor genes map to different chromosomes, allowing for independent reassortment and formation of complex genotypes.

Viral accessory proteins

• Primate lentiviruses encode various accessory proteins (e.g., Vif, Vpu, Vpx, Vpr and Nef) that function as antagonists of host-encoded restriction factors.

• When primate lentiviruses jump between host species, viral accessory proteins can adapt to accommodate divergent restriction factor protein sequences found in the new host.

• Accessory proteins can also acquire new functions (neofunctionalization), at times facilitating adaptation to a new host.