|Home | About | Journals | Submit | Contact Us | Français|
Tetherin/BST-2 is a host-encoded protein that restricts a wide diversity of viruses at the stage of virion release. However, viruses have evolved antagonists of Tetherin, including the Vpu and Nef proteins of primate lentiviruses. Like other host genes subject to viral antagonism, primate Tetherin genes have evolved under positive selection. We show here that viral antagonists acting at three independent sites of selection have driven the evolution of Tetherin, with the strongest selective pressure on the cytoplasmic tail domain. Human Tetherin is unique among the Tetherins of simian primates in that it has a 5-amino-acid deletion that results in the loss of the residue under the strongest positive selection. We show that this residue at amino acid 17 is the site of the functional interaction of Tetherin with Nef, since single amino acid substitutions at this single position can determine the susceptibility of Tetherin to Nef antagonism. While the simian immunodeficiency viruses SIVcpz and SIVgor are able to antagonize their hosts' Tetherin with Nef, human immunodeficiency virus type 1 (HIV-1) Vpu has evolved to counteract Tetherin in humans. We mapped the adaptations in the N-terminal transmembrane domain of Vpu that allow it to counteract human Tetherin. Our combined evolutionary and functional studies have allowed us to reconstruct the host-pathogen interactions that have shaped Tetherin and its lentivirus-encoded antagonists.
Humans and other primates encode a wide repertoire of proteins that intrinsically inhibit retroviral infections (53). Tetherin, also known as BST-2 or CD317, is an example of such an intrinsic antiviral protein that inhibits virus release by anchoring in the envelope of budding virions and directly tethering virions to the plasma membrane (35). This relatively nonspecific antiviral mechanism allows Tetherin to potently restrict a wide array of viruses, including human immunodeficiency virus (HIV) and other primate lentiviruses (16, 17, 24, 31, 39, 51).
A characteristic of host antiviral factors is that they often result in viruses evolving antagonists to counteract restriction. Indeed, viruses have evolved multiple independent antagonists to counteract Tetherin (reviewed in reference 50). For example, HIV type 1 (HIV-1) encodes a Vpu protein that potently antagonizes human Tetherin (31, 51) through interactions with the transmembrane domain of Tetherin, leading to its degradation via β-TrCP (9, 13, 14, 26, 28, 29, 38). However, Vpu is exclusive to HIV-1 and a specific lineage of primate lentiviruses including the simian immunodeficiency virus SIVcpz, the precursor of HIV-1, and the closely related SIVgor (6, 25, 47, 52). Primate lentiviruses that do not encode Vpu, such as SIVmac and SIVsm, instead use Nef to antagonize Tetherin (15, 56). HIV-2, which does not encode Vpu, encodes an antagonist of Tetherin in its envelope (4, 23). Viruses other than primate lentiviruses have also evolved antagonists of Tetherin. These include Ebola virus, which antagonizes Tetherin through its glycoprotein (GP) (17), and Kaposi's sarcoma-associated herpesvirus (KSHV), which is able to counteract Tetherin with its K5 protein (27).
The evolutionary “arms race” between host antiviral genes and the virally encoded antagonists of these antiviral genes can be inferred by observing adaptive evolution (also called positive selection) signatures in the antiviral genes that are indicative of repeated episodes of Darwinian selection due to ancient viral infections (10). In fact, the exact amino acids under positive selection can describe the sites of host-virus interactions (42). When there are multiple viral antagonists, such detailed evolutionary analyses focused on positive selection can also reveal which type of viral antagonist exerted the greatest selective pressure during the course of primate evolution. Two previous studies using a set of primate sequences primarily from Old World monkeys (OWM) and hominoids found that tetherin has evolved under positive selection (14, 28). Here we examine all three lineages of simian primates (including the New World monkeys [NWM]) with a larger data set that allows us to determine which part of tetherin has been under the strongest positive selection during specific periods in primate evolution. We find that during simian primate evolution, three separate types of antagonists have shaped tetherin, specifically the cytoplasmic tail of Tetherin, with distinct amino acid residues evolving rapidly in New World monkeys versus Old World monkeys and hominoids. Changes in the amino acid under the strongest positive selection correspond exactly to the specificity of Nef.
Consistent with ongoing selective pressure on tetherin genes, we show that both SIVcpz Nef and SIVgor Nef are potent antagonists of chimpanzee and gorilla Tetherins but are unable to antagonize human Tetherin. Conversely, the Vpu proteins of SIVcpz and SIVgor are unable to antagonize Tetherin, while this function has been gained by HIV-1 type M strains. While this article was in preparation, similar results were published by Sauter et al. (40) and Yang et al. (54). We demonstrate that the site of Nef interaction in the cytoplasmic domain of Tetherin is under the strongest selective pressure in hominoids and Old World monkeys. However, a deletion covering this site is fixed in human tetherin. Therefore, cross-species transmission of HIV-1 to humans necessitated a gain of function by Vpu through adaptations in two regions within the N-terminal transmembrane domain of Vpu in order for the virus to downregulate Tetherin and escape its antiviral effects.
Our combined evolutionary and functional studies allow us to reconstruct the host-pathogen interactions that have shaped Tetherin as well as two antagonists encoded by lentiviruses. We propose that the Nef interaction interface of Tetherin was subject to strong positive selection in primates but that this interaction domain was completely lost in the human lineage, potentially to escape antagonism by an ancient Nef-like factor. Subsequently, a modern lentivirus adapted to grow in humans by neofunctionalization of Vpu through evolving changes in the N terminus of Vpu.
The tetherin genes from the following primates were amplified from RNA isolated from cell lines obtained from Coriell Cell Repositories (Camden, NJ): bonobo (Pan panisucus), gorilla (Gorilla gorilla), Sumatran orangutan (Pongo pygmaeus), white-cheeked gibbon (Nomascus leucogenys), agile gibbon (Hylobates agilis), rhesus macaque (Macaca mulatta), patas monkey (Erythrocebus patas), mustached monkey (Cercopithecus cephus), kikuyu colobus (Colobus guereza kikuyuensis), douc langur (Pygathrix nemaeus), Francois' leaf monkey (FLM) (Trachypithecus francoisi), tamarin (Saguinus labiatus), pygmy marmoset (Callithrix pygmaea), white-faced saki (Pithecia pithecia), Bolivian red howler (Alouatta sara), and woolly monkey (Lagothrix lagotricha). African green monkey (Chlorocebus aethiops) Tetherin was amplified by reverse transcription-PCR (RT-PCR) from an RNA extract of COS-7 cells. Tetherin was amplified by RT-PCR with a OneStep RT-PCR kit (Qiagen), and the cDNA derived was directly sequenced. Tetherin was amplified with “forward” primer CD317upstreamForward (5′-CCCCTAACTTCAGGCCAGACTC-3′) or CD317ATGForward (5′-CAGCTAGAGGGGAGATCTGGATG-3′) in combination with “reverse” primer CD317downstreamReverse (5′-CTCACTGACCAGCTTCCTGGG-3′).
DNA sequences were aligned by ClustalX (49) and were edited by hand based on amino acid sequences. A phylogeny of tetherin genes was constructed from DNA sequences with ClustalX by the neighbor-joining method using the Jukes Cantor method of correction and with MrBayes by the maximum-likelihood method. The two methods yielded trees with identical topologies. Maximum-likelihood analysis was performed with CODEML from the PAML suite of programs (55). To detect selection in Tetherin, we fitted the data to site-specific models (nonsynonymous [NS] sites) that disallowed (NSsites model 7) or permitted (NSsites model 8) positive selection and compared the results by likelihood ratio tests. Consistent results were obtained when M1 (a two-state neutral model) was compared with M2 (a selection model allowing a nonsynonymous/synonymous change [dN/dS] ratio of >1). Sequence alignments were obtained when the data were fitted with an F61 model of codon frequency, and consistent results were obtained when the data were fitted with an F3 × 4 model of codon frequency. We calculated the global ratios of dN to dS by a free-ratio model that allows dN/dS to vary along individual branches.
HIV-1 group M Vpu sequences and SIVcpz Vpu sequences were obtained from the HIV sequence database (http://www.hiv.lanl.gov), aligned by ClustalX (49), and edited by hand. A total of 1,271 sequences representing HIV-1 group M Vpu were analyzed to generate the consensus. Sequence logos of the alignment were plotted using WebLogo (http://weblogo.berkeley.edu) (7).
293T cells were maintained in a solution containing Dulbecco's modified Eagle's medium, 8% bovine growth serum, and 1% penicillin-streptomycin under a 5% CO2 atmosphere at 37°C. SupT1 cells were maintained in a solution containing RPMI 1640 medium, 10% fetal bovine serum, and 1% penicillin-streptomycin.
Human, chimpanzee, gorilla, C. cephus, and Francois' leaf monkey Tetherins were cloned from the cDNAs of the respective species and were ligated into a pLPCX lentiviral expression vector as untagged constructs, as described previously (24). Codon-optimized HIV-1 Lai Vpu was a gift from Stephan Bour (32), and codon-optimized HIV-1 Q23-17 Vpu, SIVcpzUS Vpu, SIVcpzUS Nef, SIVcpzTan3.1 Vpu, SIVgor Vpu, and SIVgor Nef were synthesized (GenScript). HIV-1 Lai Nef was cloned from the HIV-1 Lai proviral plasmid (34); HIV-1 Q23-17 Nef was cloned from the HIV-1 Q23-17 provirus, a gift from Julie Overbaugh (36); and SIVcpzTan3.1 Nef was cloned from the SIVcpzTan3.1 proviral plasmid, obtained from the NIH AIDS Research and Reference Reagent Program (48). The Vpu and Nef constructs were ligated into a pCDNA3.1 expression vector.
Chimeras between HIV-1 Q23-17 Vpu and SIVcpzUS Vpu were constructed with a QuikChange II XL site-directed mutagenesis kit (Stratagene) or by overlapping PCR approaches. For expression studies, a hemagglutinin (HA) epitope tag was fused to the C termini of chimeric Vpu constructs. The ancestral human Tetherin was constructed by overlapping PCR to insert 5 amino acids (aa) in the cytoplasmic tail domain (residues 14 to 18) and to substitute E for K at aa 19.
293T cells were cotransfected with 200 ng of an HIV-1 reporter provirus (HIV1VpuFSLuc2), 25 ng of a Vpu or Nef construct, and native Tetherin as indicated in the figures. The total amount of DNA in all transfections was maintained constant with appropriate empty vectors. The HIV-1 double mutant reporter virus (HIV1VpuFSLuc2) expresses a frameshift mutation in Vpu and a luciferase gene inserted into the Nef open reading frame (24). Forty-eight hours after transfection, an infectivity assay was performed with SupT1 cells at 2.5 × 105/ml in 96-well plates as described previously (24).
To analyze the expression of Tetherin on the cell surface, 293T cells were cotransfected with 250 ng native human Tetherin in the presence of 750 ng of the respective Vpu constructs. Twenty-four hours posttransfection, cells were gently rinsed with 1× Dulbecco's phosphate-buffered saline (DPBS) and were resuspended in DPBS containing 4% calf serum. Cells were incubated first with an anti-human BST-2 (anti-human Tetherin) antibody obtained from the NIH AIDS Research and Reference Reagent Program (30) and subsequently with an anti-rabbit antibody conjugated with fluorescein isothiocyanate (FITC) (BD Pharmingen). Unbound antibodies were washed away, and labeled cells were resuspended in DPBS containing 4% calf serum. Flow cytometry was performed with a BD FACSCalibur platform and CellQuest software (BD). To analyze the downregulation of Tetherin expression on the cell surface by Vpu, cell events were plotted by the forward scatter (FSC) parameter versus the FL-1 parameter (Tetherin). To analyze the downregulation of CD4 expression by Vpu, 293T cells were cotransfected with 125 ng of the bicistronic pIRES-eGFP2 vector, expressing human CD4 and green fluorescent protein (GFP), and 1 μg of the respective Vpu constructs. Twenty-four hours posttransfection, cells were incubated with an allophycocyanin (APC)-conjugated mouse anti-human CD4 antibody (BD Pharmingen) or an APC-conjugated mouse IgG1(κ) isotype control antibody (BD Pharmingen). For analysis, cell events were plotted by the FL-1 (GFP) parameter versus the FL-4 parameter (APC). To analyze the downregulation of CD4 expression by Vpu, GFP-positive events were gated, and the percentages of CD4-positive event counts in the presence or absence of the various Vpu constructs were compared.
Forty-eight hours posttransfection, cells were analyzed by Western blotting as described previously (24). An HA-specific antibody (HA.11 mouse immunoglobulin G; Babco) at a 1:1,000 dilution and a rabbit anti-actin antibody (Sigma-Aldrich) at a 1:2,000 dilution were used. The primary antibodies were detected with a horseradish peroxidase-conjugated goat anti-mouse or anti-rabbit antibody, respectively.
The sequences of the 20 primate tetherin genes have been entered into the GenBank database under accession numbers NM_004335, GQ864267, and HM136905 to HM136922.
HIV-1 Vpu potently antagonizes Tetherin, whereas SIVmac or SIVsm does not encode a Vpu protein and instead uses Nef to antagonize Tetherin (15, 31). In order to understand the evolutionary pressures on both viral antagonists and on hominoid Tetherin, we tested the capabilities of Vpu and Nef from multiple strains of HIV-1, SIVcpz (the precursor virus of HIV-1), and the closely related SIVgor for antagonistic activity against Tetherin from their own host species. 293T cells were used for our experiments, because they express very small endogenous amounts of human Tetherin (31). 293T cells were cotransfected with an HIV-1 reporter virus lacking Vpu and Nef, along with constant amounts of different Vpu or Nef constructs, and increasing amounts of plasmids encoding different tetherin genes. The release of the infectious HIV-1 reporter virus was assayed by infecting SupT1 cells as previously described (24).
Consistent with previous observations, we found that human Tetherin potently inhibits virus release and that this restriction is effectively antagonized by Vpu but not by Nef proteins from either the Lai (clade B) or the Q23-17 (clade A) strain of HIV-1, even at high levels of Tetherin (Fig. (Fig.1A,1A, compare filled circles with open circles). In contrast, chimpanzee Tetherin is antagonized by Nef, but not at all by Vpu, from SIVcpzTan3.1 and only very slightly by Vpu from SIVcpzUS (Fig. (Fig.1B).1B). The potency of Nef antagonism differs between SIVcpz strains; SIVcpzTan3.1 Nef did not antagonize chimpanzee Tetherin as efficiently as did SIVcpzUS Nef. Nonetheless, Nef is the primary antagonist in SIVcpz (Fig. (Fig.1B,1B, compare open circles with filled circles). As with SIVcpz, SIVgor Nef, but not SIVgor Vpu, is a potent antagonist of gorilla Tetherin (Fig. (Fig.1C,1C, compare open circles with filled circles). Although SIVcpz and SIVgor Vpu proteins were poorly active against their respective hosts' Tetherins, they were still able to downregulate the levels of human CD4 expression efficiently (Fig. (Fig.1D).1D). These results demonstrate that primate lentiviruses have selectively evolved antagonists against their hosts' Tetherins. SIVcpz and SIVgor use Nef as a potent antagonist against Tetherin, whereas HIV-1 accomplishes that task with Vpu. Similar results have recently been reported by others (40, 54).
To determine the specificity of Tetherin antagonists in the event of cross-species transmission, we tested a panel of Vpu and Nef proteins from HIV-1, SIVcpz, and SIVgor against human, chimpanzee, and gorilla Tetherins by using an infectivity assay as a measure of viral release (Fig. (Fig.2A).2A). The relative infectivity of the released virus was also plotted on a radar chart in order to visualize the ability of Vpu (Fig. (Fig.2B)2B) or Nef (Fig. (Fig.2C)2C) to counteract each species' Tetherin. HIV-1 Vpu potently antagonized human, chimpanzee, and gorilla Tetherins (Fig. (Fig.2A,2A, blue) while HIV-1 Nef was inactive (Fig. (Fig.2B,2B, blue). However, Vpu encoded by SIVcpz or SIVgor could not antagonize the Tetherins of closely related species (Fig. (Fig.2A,2A, red and green). Nef encoded by SIVcpz or SIVgor was able to counteract both chimpanzee and gorilla Tetherins but not human Tetherin (Fig. (Fig.2B,2B, red and green). These results show that differences between human Tetherin, on the one hand, and chimpanzee and gorilla Tetherins, on the other, have determined the landscape of species-specific antagonism by SIVcpz and SIVgor Nef proteins.
Tetherin has a unique topology consisting of an N-terminal cytoplasmic tail, a transmembrane domain, and a coiled-coil domain followed by a C-terminal glycosylphosphatidylinositol (GPI) anchor (22, 35). Previous findings have demonstrated that the cytoplasmic tail domain of Tetherin harbors the sites of interaction with SIVsm Nef, while HIV-1 Vpu interacts with the transmembrane domain of Tetherin (14, 15, 28). Although recent studies have shown that tetherin has evolved under positive selection among primates (14, 28), both of these studies had a limited diversity of sequences in their data sets; hence, they lacked the power to detect which part of tetherin was under the strongest positive selection and, therefore, which viral antagonist has exerted the most selective pressure during primate evolution. Thus, we sequenced the tetherin coding sequence (approximately 555 bp) from 20 primate genomes representing 33 million years of evolution, including 7 hominoids, 8 Old World monkeys (OWM), and 5 New World monkeys (NWM). The phylogeny constructed from the primate tetherin sequences was congruent with the generally accepted primate phylogeny (37), confirming that the sequences are orthologous (see Fig. S1 in the supplemental material). There was no evidence of recombination as ascertained by a GARD analysis (21). Using a maximum-likelihood approach with CODEML from the PAML suite of programs (55), we compared the likelihood of tetherin evolution under models that disallowed (NSsites model 7) or permitted (NSsites model 8) positive selection. In agreement with previous studies, we found that tetherin had evolved under positive selection (Fig. (Fig.3A).3A). To determine the selective pressures across the different primate lineages, we performed a free ratio analysis of primate tetherin that allows for variation in the ω (dN/dS) ratios for each lineage. Several branches of the phylogeny, mainly the branches leading up to the New World monkeys, Old World monkeys, and hominoids, showed dN/dS ratios of >1 (see Fig. S1 in the supplemental material), suggesting an ancient positive selection in primate lineages.
To determine which domains are responsible for the signatures of positive selection, we performed PAML analyses on the separate domains of tetherin: the cytoplasmic tail domain (aa 1 to 25), the transmembrane domain (aa 26 to 53), and the remaining extracellular domains, consisting of the coiled-coil domain and the GPI anchor (aa 54 to 185). We found that the cytoplasmic tail and transmembrane domains have evolved under positive selection, whereas the extracellular domains have not (Fig. (Fig.3A3A).
Next, we sought to identify the specific residues that have been subjected to positive selection. Three amino acid residues (codons 9, 17, and 43) were found to evolve under positive selection with strong confidence (posterior probability, >0.95) as determined by PAML (Fig. 3B and C). This was confirmed by random-effect likelihood (REL) analyses (Fig. (Fig.3D)3D) (20). Previous analyses have suggested that a number of additional residues within Tetherin's cytoplasmic tail and transmembrane domain have evolved under positive selection, although the statistical power of those studies was hampered by a small data set (14, 28). However, we find no detectable signal for positive selection when we omit residues 9, 17, and 43 of primate Tetherin (Fig. (Fig.3E).3E). Therefore, the use of the more-extensive sequence divergence in our data set allows us to conclude that only three codons (residues 9, 17, and 43) of Tetherin display evidence of recurrent positive selection in the primates. Residue 9 evolved rapidly primarily in the NWM lineage (although some OWM also are divergent at this position) but is highly conserved in hominoids. Indeed, NWM Tetherins primarily display a signature of positive selection at residue 9, whereas OWM and hominoid Tetherins display a signature of positive selection at residues 17 and 43 instead. Importantly, we find that residue 52 (the boxed isoleucine in boldface in Fig. Fig.3B),3B), which is critical for resistance against HIV-1 Vpu (14), has not evolved under positive selection (Bayes factor, 17.3).
Residue 17, which lies within the cytoplasmic tail domain, shows the strongest recurrent signature of positive selection during primate evolution (Fig. (Fig.3D).3D). Previous work has shown that amino acids in this part of Tetherin are important for the ability of Nef to antagonize Tetherin (15, 56). Thus, we wanted to determine if changes at this single position, amino acid 17, are responsible for the species specificity of Nef. To do this, we analyzed the specificity of Nef against Tetherins from different species that differ at amino acid 17.
We have shown previously that although SIVagm is unable to effectively antagonize endogenous levels of African green monkey (AGM) Tetherin, SIVagm Nef can potently antagonize chimpanzee Tetherin (24). Importantly, chimpanzee and AGM Tetherins differ at residue 17, which is a cysteine in AGM and a tryptophan in chimpanzees. We altered residue 17 in AGM Tetherin from cysteine (as encoded in AGMs) to tryptophan (as encoded in chimpanzees) and tested it against SIVagm Nef (Fig. (Fig.4A).4A). Both AGM Tetherin constructs were able to inhibit virus release and were potently antagonized by SIVmus Vpu, as previously shown (24). Although SIVagm Nef had a minor effect against AGM Tetherin, a single substitution at residue 17 (AGM C17W) conferred a dramatic susceptibility to SIVagm Nef antagonism (Fig. (Fig.4A,4A, right).
Next, to determine if changes in residue 17 could confer resistance against current SIV Nef antagonists, we performed a reciprocal mutation of residue 17 in chimpanzee Tetherin from tryptophan (as encoded in chimpanzees) to cysteine (as encoded in AGMs) and tested it against SIVcpz Nef. Both chimpanzee Tetherin constructs potently inhibited virus release and were potently antagonized by HIV Vpu. However, although SIVcpz Nef potently antagonized chimpanzee Tetherin, the construct with the single substitution at residue 17 (Chimp W17C) displayed markedly increased resistance to SIVcpz Nef antagonism (Fig. (Fig.4B,4B, right). These results demonstrate that the exact amino acid under positive selection in Tetherin is the determinant for Nef antagonism.
The congruence of the 5-amino-acid deletion in human Tetherin with the site that displayed the highest degree of adaptation in other primates suggests that the deletion was itself driven by the need to escape antagonism by a Nef-like factor early in the human lineage. To test this hypothesis, we reconstructed human Tetherin as it likely existed prior to the deletion event (Fig. (Fig.4C,4C, Anc. Human) (the DDIWK deletion was restored, and the glutamine, the amino acid following the deletion, was replaced with a lysine) and tested a panel of primate lentiviral Vpu and Nef proteins against it. The ancestral human Tetherin effectively restricted virus release but was antagonized by HIV-1 Vpu. However, the ancestral human Tetherin was antagonized by Nef proteins from SIVcpz and SIVgor, whereas the extant human Tetherin (after the deletion) was resistant to both Nef proteins (Fig. (Fig.4C).4C). We also introduced the same 5-amino-acid deletion into chimpanzee Tetherin (Chimp ΔDDIWK) and found that the deletion conferred resistance against SIVcpz Nef (Fig. (Fig.4B).4B). Therefore, we conclude that an ancient Nef-like factor likely drove both the positive selection of residue 17 in the cytoplasmic tail domain of Tetherin in simian primates and the deletion of this residue in humans.
Amino acid 43 in the transmembrane region is also under positive selection, albeit the signal is much weaker than that for amino acid 17 (Fig. (Fig.3D).3D). Because the determinants for the action of HIV-1 Vpu on Tetherin lie in the transmembrane region of Tetherin (14, 28, 38), we wanted to determine whether amino acid 43 was driven to positive selection by Vpu. To do this, we chose to examine C. cephus and Francois' leaf monkey (FLM) Tetherins, which differ at amino acid 43 by an isoleucine-to-leucine change (Fig. (Fig.3B).3B). Human, C. cephus, and FLM Tetherins were able to potently restrict virus release (Fig. (Fig.4D,4D, left). HIV-1 Vpu was not able to antagonize either C. cephus (24) or FLM (data not shown) Tetherin, consistent with an inability of HIV-1 Vpu to antagonize Tetherin from other Old World monkeys (24, 40, 54). Likewise, the Vpu from SIVmus was unable to antagonize human Tetherin (Fig. (Fig.4D,4D, right). However, SIVmus Vpu could antagonize both C. cephus and FLM Tetherins equally well (Fig. (Fig.4D,4D, right). This indicates that, in contrast to our findings for Nef (or a Nef-like factor), Vpu has not driven positive selection either in hominoids or in Old World monkeys.
Since HIV-1 Vpu, but not SIVcpz Vpu, is a potent antagonist against Tetherin (Fig. (Fig.2A),2A), we sought to determine how the Vpu protein of HIV-1 type M might have evolved in order to antagonize human Tetherin. We constructed a panel of chimeric proteins between HIV-1 and SIVcpz Vpu (Fig. (Fig.5A,5A, left). The chimeric Vpu proteins were all expressed at similar levels in transfected cells (Fig. (Fig.5A,5A, right).
We found that although full-length SIVcpz Vpu was unable to counteract human Tetherin, substitution of the N-terminal transmembrane domain from HIV-1 Vpu for that of SIVcpz Vpu conferred the ability to counteract Tetherin (Fig. (Fig.5B).5B). This was specific to the transmembrane domain of Vpu, since substitution of the cytoplasmic domain (α-helix) leading up to the β-TrCP binding motif (DSGxxS) was insufficient to rescue the restriction phenotype (data not shown). Two regions in the transmembrane domain of Vpu were necessary for Tetherin antagonism activity: amino acids 1 to 8 and amino acids 14 to 22. Replacement of either region alone was insufficient to counteract Tetherin [Fig. [Fig.5B,5B, HIV(1-8) and HIV(14-22)]. Further chimeras within each region yielded intermediate phenotypes, suggesting that these regions harbored several minor determinants (data not shown). More importantly, SIVcpz Vpu was able to completely rescue the Tetherin restriction phenotype when it encoded both regions 1-8 and 14-22 from HIV (Fig. (Fig.5B5B).
The ability of each Vpu to antagonize Tetherin function correlated with an effect on Tetherin expression on the cell surface (51) in a cotransfection experiment (Fig. (Fig.5C).5C). However, here, it is clear that there is a threshold effect, since each of the Vpu proteins has some effect on cell surface Tetherin levels, but those that rescue virus infectivity (Fig. (Fig.5B)5B) are more effective than those that do not (Fig. (Fig.5C).5C). As a control, we examined each of the chimeric proteins for its ability to downregulate the expression of human CD4 on the cell surface (Fig. (Fig.5D).5D). All the chimeric Vpu proteins (including the singly substituted 1-8 or 14-22 region of HIV-1 in SIVcpz [Fig. [Fig.5D])5D]) effectively downregulated CD4 expression, consistent with the evidence that the ability of Vpu to modulate CD4 levels is separable from its viral release activity (44). Thus, adaptations in two regions within the N-terminal transmembrane domain conferred on HIV-1 Vpu the specific ability to antagonize Tetherin.
Since the difference between the abilities of HIV-1 Vpu and one strain of SIVcpz Vpu to antagonize human Tetherin mapped to two regions of the transmembrane domain, we next examined how likely it was that other strains of SIVcpz Vpu might also encode HIV-1-like activity. Thus, we compared an alignment of Vpu proteins from multiple strains of SIVcpz against the Vpu from HIV-1 group M strains. In contrast to the highly conserved transmembrane domain of HIV-1 group M Vpu, the N-terminal transmembrane domain was divergent across SIVcpz strains, as might be expected based on their more ancient divergence (Fig. (Fig.6).6). Although none of the Vpu proteins of SIVcpz strains were identical to the consensus Vpu of HIV-1 group M, among several strains of SIVcpz (LB7, CAM5, EK505, and MB66) that include the closest relatives of HIV-1 group M (18), the SIVcpz strain LB7 would be predicted to require seven minimal adaptations within the two critical regions of Vpu (amino acids 1 to 8 and 14 to 22) in order to gain the ability to antagonize human Tetherin.
Host proteins locked in genetic conflict often display signatures of positive selection. Here, we find that tetherin has been evolving under positive selection in primates, and the highest recurrent signal of positive selection during primate evolution corresponds to the amino acid that is a determinant for Nef, but not Vpu, antagonism of Tetherin. Chimpanzee and gorilla Tetherins are antagonized by SIVcpz Nef and SIVgor Nef but not by the Vpu proteins encoded by these two viruses. However, because of a unique 5-amino-acid deletion in the cytoplasmic tail domain of human Tetherin that eliminates the site under the greatest positive selection in tetherin, the cross-species transmission of SIVcpz to humans involved the evolution of Vpu to counteract human Tetherin through adaptations in two regions of the N-terminal transmembrane region of HIV-1 Vpu.
Our study differed from previous analyses of positive selection on Tetherin (14, 28, 33) in having less intraspecies sampling but more sampling of deeper, interspecies divergences. Our conclusions differ qualitatively from those reached previously. In particular, because of the increased statistical power of our data set, we can rule out any significant contribution to the positive selection of Tetherin outside of residues 9, 17, and 43. Intriguingly, this signal has not remained uniform among simian primates. New World monkey Tetherin appears to have recurrently evolved at residue 9, which shows some variability among Old World monkeys but is fixed in hominoids. In contrast, residues 17 and 43 appear to be rapidly evolving in Old World monkeys and hominoids but not in New World monkeys. This makes a strong case that mutually exclusive types of antagonists have shaped Tetherin in the 3 lineages of simian primates. This suggests that an unknown antagonist with a specificity unlike that of Nef or Vpu was the primary driver of Tetherin evolution in New World monkeys. It is somewhat surprising that only three residues have been under positive selection, considering that Tetherin restricts a broad diversity of viruses (16, 17, 24, 31, 39) and that recent work shows that a Tetherin with substantial changes in the transmembrane and coiled-coiled domains can still act to block virus release (35). However, there may be additional constraints on tetherin evolution due to some other function of Tetherin (5). In addition, many viruses may have evolved alternative strategies that circumvent Tetherin without necessitating a direct antagonistic interaction. For example, viruses that target induction of the interferon (IFN) pathway (2) might not need to antagonize Tetherin directly if they can prevent its induction following infection.
A single substitution at residue 17 of AGM Tetherin conferred susceptibility to SIVagm Nef, whereas the reciprocal substitution at residue 17 of chimpanzee Tetherin conferred resistance to SIVcpz Nef. Thus, the amino acid under the most intense selective pressure is responsible for both the gain and the loss of antagonism by Nef. These results argue strongly that a Nef-like factor is responsible for the recurrent viral escape from Tetherin during simian primate evolution. By “Nef-like” factor, we mean a viral antagonist with exactly the same specificity toward Tetherin as that of Nef, if not Nef itself. This line of reasoning suggests that more ancient lentiviruses have been in primate populations before the currently known primate lentiviruses (45). This hypothesis is supported by the recent discovery of much older endogenous lentiviruses (11, 12). We cannot, however, rule out the possibility that unrelated viruses encode a Tetherin antagonist with exactly the same specificity as the Nef from primate lentiviruses.
In contrast to the positive selection observed on amino acid 17 that correlates with Nef function, amino acid 52 in the transmembrane domain, which encodes much of the susceptibility of primate Tetherin to Vpu antagonism (14), is not subject to positive selection (Fig. (Fig.3B)3B) and indeed is conserved as a isoleucine (in all Old World monkeys) or a threonine (in all hominoids). This is highly reminiscent of residue 128 in APOBEC3G, which encodes its susceptibility to Vif antagonism (3, 43) but was not found to be evolving under positive selection in primates (41). Thus, as with the APOBEC3G-Vif interactions, we argue that Vpu antagonism by recent lentivirus proteins did not drive tetherin evolution. We also found no evidence that the single residue under positive selection in the transmembrane domain (amino acid 43) was driven by a factor similar to SIVmus Vpu. These results argue that, unlike the Nef-like factor that drove selection on amino acid 17, Vpu and Vpu-like factors (i.e., factors with the same specificity as modern Vpu proteins) are too recent to have had an effect on tetherin evolution.
Human Tetherin is unique among primates due to a 5-amino-acid deletion (including residue 17) in the cytoplasmic tail domain of Tetherin that has been fixed in the human population (NCBI single-nucleotide polymorphism [SNP] database). This deletion was most likely adaptive, since it includes the amino acid under the strongest selective pressure in Old World monkeys and hominoids. It is not clear why a deletion event to escape Nef-like antagonism would occur exclusively in human Tetherin, whereas other primates instead show adaptive evolution at residue 17. It is possible that this domain also encodes another function in addition to viral restriction, pressure for which was relaxed in the human lineage, or that there was a lack of preexisting SNPs in the human population at the time of selection. Alternatively, simultaneous selective pressure by multiple Nef-like proteins may have driven a deletion instead of an amino acid substitution in human Tetherin.
Due to the unique deletion in the cytoplasmic tail domain of human Tetherin, the Nef protein of the HIV-1 precursor SIVcpz was ineffective as a Tetherin antagonist following cross-species transmission to humans. We propose that as a result of the selective pressure exerted by Tetherin, the precursor virus evolved Vpu to counteract human Tetherin through adaptations in two regions of the N-terminal transmembrane domain of Vpu (Fig. (Fig.5B).5B). Therefore, the most parsimonious explanation points to sequence variations of Vpu (Fig. (Fig.6)6) in SIVcpz that predisposed certain strains to be more adaptable, in particular the strain that gave rise to the pandemic HIV-1 group M. Our findings support and extend the conclusions of Sauter et al., who performed an extensive study of HIV-1 Vpu proteins from groups M, N, and O, and proposed that the ability both to antagonize Tetherin and to degrade CD4 facilitated the pandemic spread of group M HIV-1 (40). We show that gain-of-function antagonism of human Tetherin occurred through adaptations in the N-terminal transmembrane domain of Vpu (Fig. (Fig.5B).5B). In our characterization, we found that the determinants were situated in two regions. An alignment of Vpu from multiple strains of SIVcpz shows that several strains of SIVcpz (LB7, CAM5, and EK505) would require fewer changes. Furthermore, based on the highly divergent sequences currently available, it is highly possible that a strain(s) of SIVcpz responsible for the cross-species transmission has yet to be sequenced.
Another implication of these findings is that while the evolution of SIVcpz Nef was originally constrained in its role as an antagonist of Tetherin, the evolution of HIV-1 Vpu to antagonize Tetherin may have allowed HIV-1 Nef to evolve novel functions in humans and to contribute to the pathogenicity of the virus. For example, Nef has been implicated in the enhancement of viral infectivity, the downregulation of CD4, major histocompatibility complex class I (MHC-I), and other cell surface receptors, and the remodeling of the actin cytoskeleton (1, 8, 19, 25, 46). It will be informative to observe if some of these functions have evolved in the transition of SIVcpz to humans in ways that increase HIV-1 pathogenicity (19). In summary, our combined evolutionary and functional findings reveal an intriguing cascade of known and unknown antagonists and their evolutionary steps that have led to the differential restrictive ability of Tetherin against lentiviruses in simian primates.
The following reagents were obtained through the NIH AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH: SIVcpzTAN3.1 (catalog no. 11498) from Jun Takehisa, Matthias Kraus, and Beatrice Hahn and the anti-Bst-2 antibody (catalog no. 11722) from Klaus Strebel and Amy Andrew. We thank Stephan Bour for HIV-1 Lai Vpu, Julie Overbaugh for the HIV-1 Q23-17 proviral construct, the FHCRC Genomics and Flow Cytometry Shared Resources, and Nels Elde, Maulik Patel, and Semih Tareen for comments on the manuscript.
This work was supported by NIH grant R37 AI30937 (to M.E.) and an NSF Career grant (to H.S.M.). E.S.L. is supported by the Fred Hutchinson Cancer Research Center Interdisciplinary Fellowship. H.S.M. is an Early-Career Scientist of the Howard Hughes Medical Institute.
Published ahead of print on 5 May 2010.
†Supplemental material for this article may be found at http://jvi.asm.org/.