Derivation of simian tropic HIV-1 infectious clone reveals virus adaptation to a new host

Significance

HIV-1, the predominant cause of AIDS in humans, is unable to replicate in most nonhuman primate species. We derived an HIV-1–based molecular clone that can cause AIDS in monkeys after transient CD8⁺ cell depletion during acute infection, and have used this clone to identify the role of species-specific adaptations. In addition to illuminating how viruses colonize new host species, the availability of a cloned, pathogenic simian tropic HIV-1 strain could provide a more relevant animal model for the development and testing of drug and vaccine interventions against HIV-1.

Abstract

To replicate in a new host, lentiviruses must adapt to exploit required host factors and evade species-specific antiviral proteins. Understanding how host protein variation drives lentivirus adaptation allowed us to expand the host range of HIV-1 to pigtail macaques. We have previously derived a viral swarm (in the blood of infected animals) that can cause AIDS in this new host. To further exploit this reagent, we generated infectious molecular clones (IMCs) from the viral swarm. We identified clones with high replicative capacity in pigtail peripheral blood mononuclear cells (PBMC) in vitro and used in vivo replication to select an individual IMC, named stHIV-A19 (for simian tropic HIV-1 clone A19), which recapitulated the phenotype obtained with the viral swarm. Adaptation of HIV-1 in macaques led to the acquisition of amino acid changes in viral proteins, such as capsid (CA), that are rarely seen in HIV-1–infected humans. Using stHIV-A19, we show that these CA changes confer a partial resistance to the host cell inhibitor Mx2 from pigtail macaques, but that complete resistance is associated with a fitness defect. Adaptation of HIV-1 to a new host will lead to a more accurate animal model and a better understanding of virus–host interactions.

The search for effective HIV-1 treatments and vaccines has been hampered by the lack of an animal model that fully recapitulates HIV-1 infection in humans. This is in part because HIV-1 is highly specific for humans and does not generally infect or cause disease in other species. Thus, far, infection of macaques with simian immunodeficiency viruses (SIV) or SHIVs (SIV expressing HIV-1 Env proteins) have been the most widely used models for AIDS research and have yielded important advances; for example, in vaccine evaluation (1). Nevertheless, differences between HIV-1 and SIVs in immunologic and drug targets limit the utility of these models.

To overcome these limitations, we previously developed a virus, stHIV-1 (simian tropic HIV-1), in which 94% of the genome is derived from HIV-1 and incorporates minimal SIV-derived gene coding sequences to overcome species-specific restrictions in pigtail macaques. stHIV-1 achieves high levels of acute viremia in pigtail macaques (2). Using serial in vivo passage, we subsequently adapted stHIV-1 and derived a viral swarm that is able to cause AIDS in pigtail macaques (3). Progression to disease in this model requires CD8⁺ cell depletion during acute infection, whereas animals with intact CD8⁺ cells exhibit high levels of acute viremia that progressively decline. Thus, an adapted HIV-1–based virus is able to cause AIDS in a nonhominid.

Adaptation of HIV-1, and lentiviruses in general, to new hosts requires acquisition of the ability to optimally use host proteins that play key roles in virus replication and the ability to overcome host antiviral proteins. Indeed, we demonstrated that engineering HIV-1 to overcome macaque APOBEC3 proteins was necessary and sufficient to obtain high levels of replication in pigtail macaques (2). Furthermore, during in vivo replication, the HIV-1 Vpu adapted to antagonize macaque tetherin (3). Finally, our more recent studies demonstrated that in vivo adaptation also led to the acquisition of changes in the envelope (Env) glycoprotein that improve its ability to use macaque CD4 (4). Here, we describe the derivation of a cloned stHIV strain that can recapitulate the properties of the viral swarm from which it was derived, in that it causes AIDS specifically in transiently CD8⁺ T-cell-depleted pigtail macaques. The cloned virus allowed us to examine the effect of acquired mutations in the HIV-1 capsid (CA) on interactions with host cell factors and on virus replication in pigtail macaque cells. Our studies reveal the potential role of additional species-specific restriction factors, such as Mx2, in limiting virus replication as it adapts to a new host. Moreover, a cloned stHIV provides a reproducible source of virus for future animal experiments.

Results

Derivation of stHIV-1 Infectious Molecular Clones.

We have previously reported the generation of an HIV-1–based virus that can replicate in pigtail macaques, and its diversification and adaptation via animal-to-animal passage in transiently CD8⁺ cell-depleted pigtail macaques (3). In the prior study, at the fourth animal passage, one animal (designated P4-C) progressed to AIDS at week 28 postinoculation (p.i.) (Fig. 1A). Subsequent passage of blood from P4-C to non-CD8⁺ cell-depleted animals (P5-A and P5-B) lead to high levels of acute viremia, but virus replication was subsequently controlled (3). In contrast, passage to transiently CD8⁺ cell-depleted pigtail macaques (P5-C and P5-D) led to rapid progression to AIDS. On the basis of this finding, we sought to identify infectious molecular clones (IMCs) that would recapitulate the phenotype obtained with the viral swarm from the blood of animal P4-C.

Fig. 1.

Adaptation of stHIV to pigtail macaques. (A) Outline of serial animal-to-animal passage of stHIV. All animals except P1A, P5A, and P5B, subjected to transient CD8⁺ cell depletion at the time of inoculation. Daggers indicate euthanasia resulting from progression to AIDS. Animal colors match those in B. (B) Phylogenetic analysis of the 5′ half of the viral genome (5′LTR-Vpr) sequences found in the stHIV-1–infected macaques indicated. Circles correspond to individual sequences, and colors to the animal and week p.i. indicated. stHIV/AD8 is the parental virus used for initiation of passage. (A and B) 5′ genome halves tested in subsequent studies and stHIV-A19 the clone ultimately selected for in vivo studies. (C) As in B, with phylogenetic analysis of the 3′ half of the viral genome (Vpr-3′LTR) sequences in the stHIV-1–infected macaques indicated. Numbers (1–20) indicated 3′ genome halves tested in subsequent studies.

We monitored virus evolution over the course of infection in P4-C during acute viremia at week 2 (wk2) p.i. in blood, and at necropsy at wk28 p.i. in blood and tissues. To identify virus clones that were transmitted from P4-C to the subsequent passage recipients, we also analyzed sequences from P5 animals during acute viremia (wk2 p.i.). Viral genomes were amplified in two halves; the 5′ half of the viral genome exhibited limited diversity (Fig. 1B). In addition to amino acid changes in CA, we noted frequent changes in matrix at positions 12 and 30 and Pol, one in protease, one in reverse transcriptase, and two in integrase (SI Appendix, Fig. S1). In contrast, the 3′ half of the virus, particularly Env, was more variable (Fig. 1C).

Full-length viral genomes were reconstructed (Fig. 2A). On the basis of the extent of sequence variation, we selected 2 representative variants for the 5′ half (named A and B; Fig. 1B) and 20 variants for the 3′ half (1–20; Fig. 1C). Each 5′ half was combined with each 3′ half to generate 40 full-length viral clones named A/B-1 through A/B-20 (Fig. 2A). Each clone was evaluated for virus infectivity, virion-associated replication (RT) activity, and ability to replicate in pigtail and human peripheral blood mononuclear cells (PBMC), and the highly susceptible human T-cell line MT4 engineered to express CCR5 (MT4-R5). With the exception of clones expressing the 3′ half 9, which did not produce infectious particles and thus did not replicate in any cell line (Fig. 2B and SI Appendix, Fig. S2 A–C), all clones produced infectious particles and replicated to various extents in MT4-R5 (SI Appendix, Fig. S2C). Although the majority of clones also replicated well in huPBMC (SI Appendix, Fig. S2B), greater variation in their ability to replicate in pgtPBMC was evident (Fig. 2B). Viruses with the 5′ half B replicated less well than viruses with the 5′ half A. The replication deficit in 5′ B-half-containing viruses was more pronounced in huPBMC (SI Appendix, Fig. S2B) than pgtPBMC (Fig. 2B), suggesting that the 5′ B half likely maintains some species-specific advantages. Several 3′ genome halves, including 13, 14, 18, and 19, conferred high-level replication capacity when paired with either A or B 5′ half in pgtPBMC (Fig. 2B) and with A in huPBMC (SI Appendix, Fig. S2B). Thus, certain combinations of adaptive mutations in both the 5′ and 3′ half of the virus could confer high levels of replication in pigtail macaque cells without affecting their ability to function in human cells.

Fig. 2.

Replication of stHIV IMCs in pig-tailed macaque lymphocytes in vitro. (A) Schematic representation of adapted viral clone generation. Brackets indicate the PCR products amplified from animal samples. Two 5′ halves, A and B, were combined with 20 3′ halves, 1–20, to generate 40 full-length viral molecular clones. (B) In vitro replication of stHIV molecular clones. PBMC from two pigtail macaque donors (red and black lines) were inoculated with 200 pg RT per 10⁶ cells for each virus. Replication was measured by RT in samples collected longitudinally.

Replication of stHIV Molecular Clones in Vivo.

Of the clones that replicated with the highest efficiency in pgtPBMC, we selected two (stHIV-A18 and stHIV-A19) whose envelope genes were divergent and were represented in recently transmitted virus populations in two different animals (P5-A and P5-D; Fig. 1C). To test the ability of these clones to replicate in vivo, two naive pigtail macaques were inoculated i.v. with a mixture containing equivalent titers of the two clones. These animals developed peak plasma viral loads of 6.1 × 10⁵ and 1.2 × 10⁶ viral RNA copies/mL, although replication was eventually controlled in both animals (Fig. 3A). Peripheral CD4⁺ T-cell counts declined during the first 2 wk p.i. and then transiently recovered, but then gradually declined despite low levels of viral replication (SI Appendix, Fig. S3A). Notably, the stHIV-A19 clone was the only virus detected by SGA at 1 wk p.i. in plasma. Therefore, we used the stHIV-A19 clone alone to inoculate four naive pigtail macaques, three of which were subjected to CD8⁺ cell depletion (Fig. 3 B and C). Although acute viremia reached >10⁶ viral RNA copies/mL at week 2 p.i. in the non-CD8⁺ cell-depleted animal, virus replication was subsequently reduced and persisted at 10² to 10³ RNA copies/mL, similar to the two animals infected with the stHIV-A18/stHIV-A19 mixture (Fig. 3B). Accordingly, CD4⁺ T cells declined during the first few weeks p.i., but stabilized and gradually increased thereafter (SI Appendix, Fig. S3B). In sharp contrast, all CD8⁺ cell-depleted animals developed high levels of acute viremia (Fig. 3C). Despite transient and partial recovery of CD4⁺ T cells in two of three animals, high-level viremia persisted, and CD4⁺ T cells were nearly completely depleted in all animals by 27 wk p.i. (SI Appendix, Fig. S3C). All animals had to be euthanized within 25–35 wk p.i. because of their progression to AIDS. Thus, clone stHIV-A19 recapitulates the phenotype observed using the viral swarm in the blood of P4-C and can consistently cause AIDS in CD8⁺ cell-depleted pigtail macaques.

Fig. 3.

Replication of stHIV IMCs in vivo. (A) Plasma viremia in two pigtail macaques (red and black) inoculated i.v. with 1.5 × 10⁵ i.u. of each of stHIV-A18 and stHIV-A19. (B) Plasma viremia in a pigtail macaque inoculated i.v. with 3 × 10⁵ i.u. of stHIV-A19. (C) Plasma viremia in three pigtail macaques (red, black, and green) inoculated i.v. with 3 × 10⁵ i.u. of stHIV-A19 and subjected to transient CD8⁺ cell depletion at the time of inoculation.

Evolution of the HIV-1 CA During Adaptation in Pigtail Macaques.

Adaptation of HIV-1 to macaques was made possible because pigtail macaques lack a TRIM5 protein that targets the HIV-1 CA (5), which allowed us to avoid manipulation of the Gag protein in the stHIV constructs. Nevertheless, the HIV-1 CA interacts with several other host proteins that either facilitate or inhibit viral replication (6–9), and we were intrigued to observe sequence changes in the CA sequences in the 5′ half clones A and B obtained from P4-C (SI Appendix, Fig. S1). Furthermore, although some changes were conserved in the majority of clones sequenced from the subsequent passage, several changes were clearly not fixed, and a total of eight positions varied in the CA sequences analyzed (Fig. 4 and SI Appendix Fig. S4 A and B). Of these, six were at the N terminus of the protein in regions that should be potentially accessible to host cell proteins (Fig. 4A and SI Appendix, Fig. S4B). Four of the changes were in the cyclophilin A (CypA) binding loop, V86A, H87P, I91A/N/T, and A92P (Fig. 4A), although none of these residues appears to be directly involved in interactions with CypA (10). Residues V86, H87, and I91 have been reported to affect sensitivity/resistance to various primate TRIM5α or TRIMCyp proteins (11). Of the other two N-terminal changes, T107 is in helix 6 (Fig. 4A) in a pocket that is bound by multiple host proteins, as well as CA-targeting drugs (8, 9, 12, 13), and G116 is in loop 3 and has been reported to affect sensitivity to TRIM5α and Mx2 (11, 14). Variation at positions 87, 91, and 116 is observed across clade B HIV-1 CA sequences, whereas positions 86, 92, and 107 are highly conserved (Fig. 4B). Frequently, however, the sequence changes that arose in pigtail macaques were rarely or never seen in CAs derived from clade B HIV-1–infected humans (Fig. 4B). For example, although I and V are common at position 91 in HIV-1–infected humans (∼40–50%) A, N, and T are rarely seen in these CA sequences (∼0.3–2%). Notably, A, N, or T were present at position 91 in almost all CA sequences recovered from stHIV-infected pigtail macaques (Fig. 4B). Even more strikingly, T107 is highly conserved, and the T107I mutation that occurred in the stHIV passage experiment was completely absent in 3,112 clade B (Fig. 4B) and 1,392 clade C (SI Appendix, Fig. S4C) HIV-1 CA sequences analyzed.

Fig. 4.

Variation in selected CA amino acid positions. (A) Positions of variable CA amino acid in the 3D structure of the mature HIV-1 CA hexamer. The side view of three CA hexamers is shown as a ribbon diagram. Residues 86, 87, 91, 92, and 116 are colored bright green, 107 dark green, and 180 and 213 as magenta spheres. The positions of residues 107 and 180 are highlighted with arrows. (B) Amino acid variation at the HIV-1 CA position indicated found across clade B sequences (Left) and stHIV sequences from animals P4C and P5A-D (Right). For each position, the amino acid found at the highest percentage of sequences in clade B is represented in dark green. Bright green indicates the amino acid found at the second-highest frequency and so on through yellow, salmon, red, gray, and black. Blue diagonal bar represents an amino acid sequence not found in clade B HIV-1 CAs analyzed herein. Identity of the most frequent amino acid is shown.

Finally, two changes were observed in the C terminus of CA (CTD), which plays a critical role in Gag oligomerization and virus assembly, as well as the formation of the CA lattice in mature particles (15). Both residues are close to the CTD–CTD interface in adjacent hexamers (Fig. 4A and SI Appendix, Fig. S4B): E180 in the dimer and E213 in the trimer interfaces (15). E180D was present in all of the CAs analyzed here, and although E213D was present in the stHIV-A19 clone, it was not always found in P5-derived CAs (SI Appendix, Fig. S3A). Although both changes conserve the residue charge, they alter the amino acid side chain, and it is unclear what their effect on CTD-mediated interactions, CA assembly, and core stability would be. Of note, although D180 is found quite frequently in clade B and C HIV-1 CAs, E213 is almost perfectly conserved (Fig. 4B and SI Appendix, Fig. S4C), even in HIV-2 and SIV_MAC CAs. Nevertheless, E213G can be selected for in vitro in the presence of a specific family of CA assembly inhibitors (16).

To determine the effects of these changes in virus replication, we selected CA sequences isolated from P4 and P5 animals that contained representative combinations of mutations (SI Appendix, Fig. S4) or the unadapted CA sequence from HIV-1 (NL4.3), and introduced them into the backbone of the adapted stHIV-A19 clone. Replication of these viruses was determined in pigtail and human PBMC and MT4-R5. With the exception of the virus expressing CA-B, which had somewhat delayed replication kinetics, all viruses replicated with comparable efficiencies in MT4-R5 cells (SI Appendix, Fig. S4D). In contrast, differences in replication kinetics were observed in PBMC, particularly those from pigtail macaque donors, and discrepancies were observed in the abilities of each CA to support replication in huPBMC (SI Appendix, Fig. S4E) compared with pgtPBMC (Fig. 5). For example, the replication of viruses encoding certain CA sequences (e.g., CA-1, CA-3) was more rapid than the virus encoding the initial NL4-3 CA in pgtPBMC (Fig. 5); however, this advantage was not observed in huPBMC (SI Appendix, Fig. S4E). Conversely, the replication kinetics of viruses expressing CA-8 and CA-9 in huPBMC were equivalent to most other viruses, but slower in pgtPBMC (Fig. 5 and SI Appendix, Fig. S4E). Finally, the virus expressing CA-B replicated poorly, but was especially impaired in huPBMC (SI Appendix, Fig. S4E). This latter finding is consistent with the observation that the replication of stHIV IMCs that included the 5′ B genome half was overall lower than IMCs with the 5′ A genome half (Fig. 2B and SI Appendix, Fig. S2 B and C), and suggests that the differences in replicative capacity between the 5′ A and B IMCs viruses were largely attributable to differences in their CA proteins.

Fig. 5.

Replication of stHIV-A19 with different CA mutants in pigtail macaque and human PBMC. Replication of stHIVA19 viruses encoding selected combinations of CA mutations in PBMC from two (red and black) pigtail macaque donors. Cells were inoculated with 200 pg RT per 10⁶ cells for each virus, and replication was measured by RT in samples collected longitudinally.

Interaction of Adapted CA Variants with Host Cell Factors.

Collectively, the CA mutants analyzed contained eight amino acid substitutions (Fig. 4), most of which lie in CA regions predicted to be exposed on the outer surface of the assembled CA lattice and would be available to interact with cell host factors that influence virus replication. Given the position of the majority of these substitutions, and the lack of an HIV-1–restricting TRIM5 protein in pigtailed macaques, we asked whether they affected sensitivity to another CA-targeting antiviral proteins Mx2 (6, 7) or altered functional interactions with other host cofactors.

PgtMx2 variants differed from human Mx2 at 17- and 18-aa positions in the N-terminal domain, which is crucial for antiviral specificity (17) (SI Appendix, Fig. S5A). Mx2 expression was induced by IFNα in pgtPBMC to a similar degree as in huPBMC (SI Appendix, Fig. S5B). To test the Mx2 sensitivity of macaque-adapted CA variants, we generated HeLa cells expressing doxycycline-inducible human or macaque Mx2 proteins, with huMx1 and empty vector-containing cells as controls (SI Appendix, Fig. S5C). We infected doxycycline-treated cells with HIV-1–based vectors carrying a green fluorescent protein reporter and encoding the various CA variants. These experiments were performed in dividing and aphidicolin-arrested cells, as growth arrest has been shown to enhance huMx2 activity in some cells (7, 18). As expected, induction with doxycycline did not affect infection of the huMx1 or empty vector control cell lines, whereas the virus encoding NL4-3 CA was inhibited by all Mx2 proteins to a similar extent (Fig. 6). Notably, various degrees of resistance to Mx2 were observed for viruses encoding the macaque-adapted CA proteins. Of the 11 CA variants tested, seven exhibited some level of resistance to Mx2, and four of those conferred greater resistance to macaque as opposed to human Mx2 proteins (Fig. 6). The largest degree of Mx2-resistance was exhibited by CA-1, CA-2, and CA-B. Indeed, CA-B exhibited almost complete resistance to pgtMx2 variants, suggesting that amino acid changes there may have been selected because they allow CA-B to resist inhibition by the macaque proteins. In addition, cell cycle arrest had diminished effects on Mx2 sensitivity of the majority of the macaque-adapted CA proteins (Fig. 6).

Fig. 6.

CA mutant sensitivity to Mx2. Susceptibility of HIV-1 CA mutants to Mx2-mediated inhibition. Infectivity of virions encapsidating an HIV-1 green fluorescent protein reporter vector and encoding selected combinations of CA mutations in cells expressing the Mx1/2 variant indicated or empty vector (control) under the control of a doxycycline inducible promoter. Assays were performed in the presence and absence of doxycycline (dox) and presence or absence of aphidicolin (Aph).

To determine whether the CA substitutions altered interaction with other cellular factors, we exploited truncated or TRIM5 N-terminal domain fusion derivatives of CPSF6, cyclophilin A, and NUP153. These proteins are known to inhibit HIV-1(NL4.3) infection through binding to incoming capsids (8, 13). Only CA-B conferred partial resistance to inhibition by the truncated CPSF6 protein (SI Appendix, Fig. S6A), suggesting that CPSF6 interaction is preserved in the majority of the adapted CA sequences. Variants CA-1 and CA-2 exhibited >10-fold resistance to omkTRIMCyp, suggesting altered configuration of the CypA binding loop; however, none of the CA variants acquired sensitivity to pgtTRIMCyp (SI Appendix, Fig. S6B). All variants exhibited sensitivity to TRIM5-NUP153 fusion proteins, even when the NUP153 portion was mutated (S1441N) to mimic the pigtailed macaque sequence, suggesting conserved configuration of the FG-repeat Nup-binding pocket (SI Appendix, Fig. S6B).

siRNA-mediated knockdown of cellular factors implicated in HIV-1 nuclear import in HeLa cells revealed that infection by most CA variants remained sensitive to depletion of TNPO3, RanBP2, NUP153, and NUP155, again suggesting preserved interactions with nucleoporins (SI Appendix, Fig. S7A). However, the degree to which CsA influenced the phenotypic effect of the knockdowns varied for several CA variants, indicating a tendency for macaque adaptation to influence the configuration of the CsA binding loop. The most notable differences were observed for CA-B, which conferred decreased sensitivity to NUP153 and TNPO3 knockdown and increased sensitivity to CPSF6 and NUP155 knockdown compared with NL4.3 CA (SI Appendix, Fig. S7A). In HT1080 cells, where CsA reduces infectivity, CA-B viruses also displayed the greatest differences compared with NL4.3 CA (SI Appendix, Fig. S7B). In pigtail macaque fibroblasts, unlike human cells, CsA had negligible effects on HIV-1 infectivity, with minor differences between CA variants (SI Appendix, Fig. S8).

Overall, the changes in interaction with host proteins other than Mx2 were mostly minor, and primarily involved alteration in the modulating effect of CsA/CypA. The exception to this conclusion was the CA-B variant that conferred the most notable changes in phenotype with the highest levels of resistance to pgtMx2 proteins and altered interaction with nuclear import-associated cellular host factors. These changes were accompanied by a replication deficit in both human and macaque cells (Fig. 5 and SI Appendix, Fig. S4 D and E).

Discussion

A useful animal model for HIV-1 infection requires the repeated generation of uniform viral stocks for animal challenges. Based on the analysis of stHIV sequences in infected pigtail macaques (3), we generated multiple infectious virus clones, representative of the swarm of circulating viruses. Using in vitro and in vivo analysis, we identified one such clone, stHIV-A19, that recapitulates the phenotype of the viral swarm (Fig. 1A) and consistently causes AIDS-like disease in pigtail macaques transiently depleted of CD8⁺ cells. The availability of this clone considerably facilitated the determination of the phenotypic effects of mutations that were acquired during in vivo adaptation.

Monitoring virus adaptation to a new host can illuminate changes that are driven by species-specific differences in cellular factors that aid or curtail virus replication. Examples include the changes in stHIV Vpu and Env that adapted to antagonize macaque tetherin and use macaque CD4, respectively (3, 4). These changes should perhaps be expected, given the evolutionary history of Vpu (19) and the plasticity of Env. We were more surprised to identify changes in HIV-1 CA, given the absence of a TRIM5 protein that targets HIV-1 CA in pigtail macaques (5) and the extreme genetic fragility of HIV-1 CA (20). Nevertheless, CA performs multiple functions in HIV-1 replication and interacts with multiple cellular proteins during transit of the viral core through the target cell cytoplasm and nuclear pores (reviewed in ref. 21). Interestingly, several CA amino acid changes selected during adaptation in pigtail macaques are rarely found in clade B or C HIV-1 CA sequences in humans (Fig. 4B and SI Appendix, Fig. S4C), suggesting the existence of species-specific selective pressures. Certain selected CA sequences acquired partial resistance to macaque Mx2. Species-specific differences in the N terminus of Mx2, a determinant of CA recognition (17), may have driven the acquisition of these changes. Notably, one adapted CA (CA-B) was fully resistant to pgtMx2 (Fig. 6), but exhibited impaired replication, suggesting that complete resistance to Mx2 incurs a replication fitness cost. CA-B was unique in containing a T107I substitution. Mutations at T107 have been reported to inhibit formation of CA cores, particle production, and infectivity (22), and were not found in CA sequences from 3,112 clade B HIV-1–infected humans (Fig. 4B). T107 is located in a pocket in CA that affects dependence on CPSF6, NUP153, and TNPO3 (8, 9, 23). Positive selection has been reported for proteins that potentially interact with this pocket, such as NUP153 (24) and Mx2 (18), that could potentially select for this particular mutation in macaques. CA-B expressing viruses maintain sensitivity to a TRIM5-NUP153 fusion protein and NUP153 depletion at somewhat lower levels than the NL4.3 CA virus (SI Appendix, Figs. S6 and S7). A caveat of these experiments is that they were performed in human cells, as they are not technically feasible in macaque cells, and species-specific differences in multiple cellular factors could ultimately affect CA interactions.

In addition to T107I, H87P mutations may also influence macaque Mx2 susceptibility (Fig. 6). H87P and G116A mutations are well documented to occur in the presence of specific HLA alleles in humans (25), and it is likely that HLA divergence between humans and macaques might have driven some CA changes. Finally, mutations at conserved positions in the C terminus of CA, E180/213D, could influence the stability of the CA lattice and indirectly affect its interactions with host factors, possibly, but not necessarily, as a complement to selective pressures on the CA N terminus. Overall, adaptation to a new host might select for multiple solutions to a new environment, and alternative combinations of substitutions occurred in CA during adaptation, in most cases without substantially altering CA properties in human cells.

Although we have focused on the viral CA, additional mutations throughout the stHIV genome likely contribute to its adaptation to macaques. For example, although CA-B in the context of stHIV-A19 was attenuated in replication (Fig. 4), these effects were less pronounced in the context of stHIV-B19 (Fig. 2B), perhaps because of the presence of additional mutations in the B 5′ half from which CA-B was originally derived. Selection for substitutions at certain positions was expected. Specifically, substitutions at matrix residue 30 have been documented in multiple instances, for example, when SIVcpz/HIV-1 were transmitted from one species to another (26), but the role of this repeatedly observed mutation remains unclear. Similarly, selection for the A281T/V/I mutation in all Envs analyzed here was consistent with our finding that adaptive mutations at this position improve the use of macaque CD4 (4). High levels of polymorphism were also seen in Nef, but deciphering their role is complicated, given the lack of reagents to test protein function against macaque proteins and the overlap between Nef and the 3′ LTR. In sharp contrast, no sequence differences between the parental stHIV and stHIV-A19 were seen in the Vpr protein (SI Appendix, Fig. S1), suggesting that the function of this viral protein does not greatly depend on the whether the host species is human or macaque.

The generation of an animal model based on HIV-1 rather than SIV will overcome several limitations of current models. The cloned virus described here should ultimately provide a better tool to test therapy and vaccine candidates in macaques. In addition, the availability of a pathogenic, macaque-tropic, HIV-1 molecular clone will allow the study of genetic determinants that contribute to adaptation, replication, and pathogenesis in vivo. Finally, this virus will be a unique tool in studies of HIV-1 persistence, as it is the only virus available for experiments in macaques that lacks Vpx, a protein that could affect the composition of the latent reservoir (27), and in which the transcriptional machinery is derived exclusively from HIV-1.

Methods

Animal Experiments and Viral Load.

Protocols involving Pigtail macaques (Macaca nemestrina) have been previously described (3); for details, see SI Appendix, Supplementary Methods. Pigtail macaques (Macaca nemestrina) were housed at the National Institutes of Health and cared for in accordance with American Association for the Accreditation of Laboratory Animal Care (AAALAC) standards in an AAALAC-accredited facility (Animal Welfare Assurance number A4149-01). The study was conducted under protocols approved by the Institutional Animal Care and Use Committee of the National Cancer Institute (Protocols AVP-023 and AVP-049) and adhered to the standards of the NIH Guide for the Care and Use of Laboratory Animals (28) in accordance with Animal Welfare Act regulations.

Single-Genome Amplification and Cloning.

A limiting dilution, single-genome amplification approach was used to amplify viral genes, as previously described (3) (SI Appendix, Supplementary Methods).

GenBank Accession Numbers MK490580–MK490663.

Full-length molecular clones were assembled by amplifying selected 5′ and 3′ SGA products (Fig. 2A). Cloning details and primer sequences are provided in the SI Appendix, Supplementary Methods.

Cells and Viruses.

Culture of cell lines and PBMC has been previously described (4). Derivation of Mx2 and other cell lines, generation of viral stocks, and infectivity experiments have been previously described (7, 13). RT assays have been previously described (4) (for details, see SI Appendix, Supplementary Methods).