New SHIVs and Improved Design Strategy for Modeling HIV-1 Transmission, Immunopathogenesis, Prevention and Cure

Simian-human immunodeficiency virus (SHIV) chimeras contain the HIV-1 envelope (env) gene embedded within an SIVmac proviral backbone. Previously, we showed that substitution of Env residue 375-Ser by bulky aromatic residues enhances Env binding to rhesus CD4 and enables primary or transmitted/founder (T/F) HIV-1 Envs to support efficient SHIV replication in rhesus macaques (RMs). Here, we test this design strategy more broadly by constructing and analyzing SHIVs containing ten strategically selected primary or T/F HIV-1 Envs corresponding to subtypes A, B, C, AE and AG, each with six allelic variants at position 375. All ten SHIVs bearing wildtype Env375 residues replicated efficiently in human CD4+ T cells, but only one of these replicated efficiently in rhesus CD4+ T cells. This was a SHIV whose subtype AE Env naturally contained a bulky aromatic His residue at position 375. Replacement of wildtype Env375 residues by Trp, Tyr, Phe or His in the other nine SHIVs uniformly led to efficient replication in rhesus CD4+ T in vitro and in RMs in vivo. Env375-Trp – the residue found most frequently among SIV strains infecting Old World monkeys – was favored for SHIV replication in RMs, although some SHIVs preferred Env375-Tyr, -His or -Phe. Nine SHIVs containing optimized Env375 alleles were grown large scale in primary activated rhesus CD4+ T cells to serve as challenge stocks in preclinical prevention trials. These virus stocks were genetically homogeneous, native-like in Env antigenicity and tier-2 neutralization sensitivity, transmissible by rectal, vaginal, penile, oral or intravenous inoculation routes, and exhibited acute and early replication kinetics that were indistinguishable from HIV-1 infection in humans. Finally, to expedite future SHIV constructions and eliminate short redundant elements in tat1 and env gp41 that were spontaneously deleted in chronically infected monkeys, we engineered a simplified second-generation SHIV design scheme and validated it in RMs. Overall, our findings demonstrate that SHIVs bearing primary or T/F Envs with bulky aromatic amino acid substitutions at position Env375 consistently replicate in RMs, recapitulating many features of HIV-1 infection in humans. We further show that SHIV challenge stocks grown in primary rhesus CD4+ T cells are efficiently transmitted by mucosal routes common to HIV-1 infection and can be used effectively to test for vaccine efficacy in preclinical monkey trials.

SHIVs have a long history dating to 1992 when Sodroski and colleagues first subcloned the tat, rev and env sequences HIV-1 HXB2c into SIVmac239 (21). This clone was further modified by substitution of the env from the dual CCR5/CXCR4 tropic HIV-1 89.6 strain and later adapted by serial passage in RMs, eventually yielding the molecular clone SHIV-KB9 (22). Thus, the earliest SHIVs contained T-cell line adapted, in vivo passaged HIV-1 Envs that were CXCR4 tropic, highly syncytium-inducing and cytopathic, and led to accelerated disease in monkeys. As a consequence, many of the essential features of HIV-1 biology, including cell and tissue tropism, sensitivity to neutralizing antibodies (NAbs), immunopathogenesis, transmission efficiency and natural history, were not faithfully represented in the macaque model (3). Attempts to develop a SHIV infection model that included primary (non-T-cell line adapted) CCR5-tropic Envs were generally met with failure, and when they were successful, such SHIVs often required adaptation 5 by serial monkey passage to achieve consistent replication in vivo (3,(23)(24)(25). In an attempt to better understand restrictions to SHIV infection and replication in RMs, Overbaugh and Sawyer examined the affinity of primary HIV-1 Envs to rhesus CD4 (26, 27). They discovered that the Envs of most primary HIV-1 strains exhibited low affinity for rhesus CD4 and did not support efficient virus entry into rhesus cells. Overbaugh identified a key amino acid at position 39 in domain 1 of rhesus CD4 that differed between human and rhesus CD4 and was largely responsible for the poor binding and infectivity of primary HIV-1 Envs in rhesus cells (27). This presented a major obstacle to new SHIV designs. Hatziioannou identified a mutation at residue 281 in the CD4 binding region of HIV-1 Env that occurred commonly in SHIV-infected RMs, where it could be shown to facilitate virus replication (28). However, unlike the Env375 substitution, the 281 substitution on its own was unable to consistently convert primary or transmitted/founder (T/F) Envs, which fail to replicate efficiently in RMs, to do so. Moreover, the addition of the 281 mutation to SHIV Envs that already contain a rhesus-preferred Env375 allele, did nothing to further enhance virus replication in rhesus animals (29).
We noted from studies by Finzi and Sodroski (30) that residue 375 in the CD4 binding pocket of primate lentiviral Envs was under strong positive evolutionary pressure across the broad spectrum of primate lentiviruses. These investigators further showed that substitution of 375-Ser (found in most HIV-1 group M viruses) by 375-Trp (found in most SIV strains from lower primates) favored an HIV-1 Env conformation that was closer to the CD4-bound state (31-34).
Based on these findings, we hypothesized that residue 375 might act as a "molecular switch" conferring enhanced Env affinity to rhesus CD4 (35) and a lower energetic barrier to conformational change following CD4 binding (31, 34, 36, 37) when the naturally-occurring Ser 6 or Thr residues were substituted by bulky aromatic residues like Trp. In testing this hypothesis, we discovered that substitution of a single residue, 375-Ser, in primary or T/F HIV-1 Envs by Trp, Phe, Tyr, His or Met resulted in SHIVs that exhibited enhanced binding to rhesus CD4, increased infection of primary rhesus CD4 + T cells in culture, and consistent infection and replication by SHIVs in RMs in vivo (35). Importantly, these amino acid substitutions at residue 375 did not alter the tier 2 neutralization phenotype of the primary Envs nor did they appreciably alter their sensitivity to bNAbs that targeted any of the canonical bNAb recognition sites, including CD4bs, V2 apex, V3 high mannose patch or membrane proximal external region (35). Thus, it became possible, for the first time, to prospectively design SHIVs that expressed particular primary or T/F Envs, including those that elicited bNAbs in HIV-1 infected humans, and to explore parallels in the immune responses of rhesus monkeys and humans to essentially identical Env immunogens (38). This Env∆375 design strategy also made possible the development of SHIVs to evaluate preclinical efficacy of novel active or passive vaccination regimens against challenge by viruses bearing homologous or heterologous primary Envs (7)(8)(9)(10). Here, we extend this work by constructing ten new SHIVs, each containing a strategically selected primary HIV-1 Env, that we then validate for retention of native antigenicity, tier 2 neutralization sensitivity and efficient replication in human and rhesus CD4 + T-cells in vitro and in RMs in vivo. We next describe the development and characterization of a panel of nine SHIV challenge stocks, each containing a unique tier 2 primary HIV-1 Env and grown large scale in primary rhesus CD4 + T-cells, for distribution as challenge strains for active or passive vaccine protection trials. We show that these SHIVs can be efficiently transmitted by different mucosal routes (rectal, vaginal, penile or oral) and that current vaccination regimens and passively administered bNAbs can prevent transmission of these viruses at neutralization titers similar to those reported in the recently concluded human Antibody-Mediated Prevention (AMP) trials (11). Finally, we describe a new second-generation design strategy that simplifies SHIV construction and eliminates extraneous tat1 and env sequences, thereby making the rhesus-SHIV infection model a more readily accessible and useful research tool.

Results
Ten primary HIV-1 Envs were chosen for SHIV constructions (Table 1A). These Envs were selected based on their genetic subtypes, biophysical properties, derivation from primary or T/F virus strains, and in some cases, prior development as candidate vaccine strains for human clinical trials (see Table 1 for Env features and relevant literature citations). Env subtypes included A, B, C, AE and AG, which complement subtype A, B, C and D SHIVs that we reported previously [see (35, 38); Table 1B). All ten of the new SHIVs contained Envs from tier 2 viruses except for Q23.17 Env (39), which has been variably classified as tier 1b or 2 (40-42). Seven of the new SHIVs contained Envs from T/F strains of HIV-1. The 1086 Env (43) corresponds to a vaccine strain employed in the HVTN 703 efficacy vaccine trial (44)(45)(46), and the B41 Env was developed as a SOSIP trimer for potential human immunizations (47). The Ce1176 Env is from a widely used global test panel for bNAb detection (41). Env RV217.40100 is a new subtype AE T/F strain (48,49) and Envs CH1012 and CH694 are T/F strains that elicited potent bNAbs in their respective human hosts (50,51). Envs T250, ZM233, WITO, Q23.17 and CAP256SU were shown previously to bind unmutated common ancestors (UCAs) of human V2 apex targeted bNAbs (52)(53)(54). Thus, the Envs selected for new SHIV constructions exhibited unique pedigrees complementary to 8 previous SHIV designs (35, 38, 55-64) that made them desirable for downstream investigations related to HIV-1 transmission, prevention, immunopathogenesis or cure.
The design strategy for constructing SHIVs is illustrated in Fig. 1A. This construction scheme allowed for the complete extracellular gp140 region of Env plus the transmembrane segment and 9 aa of the cytoplasmic tail (nucleotides 1-2153; HXB2 numbering) to be PCRamplified en bloc and subcloned into a chimeric T/F SIVmac766-HIV-1 proviral backbone (35). If sequences were available for vpu in the source material, then the homologous vpu-env gp140 gene segment was amplified and subcloned into the proviral vector, since homologous vpu-env sequences could potentially enhance the efficiency of Env translation. Env 375 codon substitutions corresponding to Trp, Phe, Tyr, His or Met were introduced by site-directed mutagenesis into each SHIV construct, which was then prepared as a large-scale DNA stock and sequence confirmed. Genome sequences for all SHIVs were contributed to GenBank ( Table 1).
For each of the ten primary HIV-1 Envs, six variants containing the different Env375 alleles were made bringing the total number of newly constructed SHIVs to 60. In the course of SHIV constructions, we noted that certain aspects of the design scheme were inefficient, especially the requirement for multiple PCR amplifications and ligations (see Methods). We also found in SHIV infected RMs that redundant HIV-1 tat1 and env gp41 sequences of 68 and 21 bp in length, respectively, that were generated as a consequence of the original cloning strategy underwent spontaneous deletion [Figs. 1A and S1; (35, 38)]. Thus, we modified the SIVmac766 backbone vector and amplification primers to simplify the PCR amplification step and eliminate the redundant sequences (Fig. 1B). We used this new design strategy to reclone SHIV.CH505, in order to perform a head-to-head comparison of viruses expressed from this new vector compared with 9 the original SHIV design, and to clone a new SHIV containing the HIV-1 CH694 Env. Plasmid DNA for all SHIVs was transfected into 293T cells and virus-containing supernatants were characterized for p27Ag content and infectivity on TZM-bl cells. For all SHIVs, p27Ag concentrations ranged from 200-2000 ng/ml. One nanogram of p27Ag is equivalent to approximately 10 7 virions, so SHIV titers were estimated to range from 2x10 9 to 2x10 10 virions per ml. We confirmed these titers by quantifying vRNA and assuming vRNA molecules per virion.
Infectivity titers on TZMbl cells ranged from 2x10 5 to 2x10 6 per ml, corresponding to an IU to particle ratio of approximately 10 -4 . This ratio is typical for 293T-derived HIV-1 and SIV virions (35), and 100-fold lower than for virus stocks propagated in primary rhesus CD4 + T cells where between 1 in 100 and 1 in 50 virions are typically infectious on TZMbl cells [ Table 2 and (35)].
For SHIVs bearing the 10 new HIV-1 Envs, we evaluated the replication efficiency of each of them containing six different Env375 residues in primary activated human and rhesus CD4 + T cells in vitro (Fig. 2). With the exception of SHIV.AE.40100, which naturally contains the positively charged, aromatic residue Env375-His, none of the SHIVs containing wild-type Ser or Thr residues at position Env375 replicated appreciably in rhesus CD4 + T cells (Fig. 2). Conversely, all 10 SHIVs with wild-type Env375 residues replicated efficiently in primary activated human CD4 + T cells.
This latter result -efficient replication of SHIVs containing wildtype 375 alleles in human CD4 + T cells -was an expected finding but was nonetheless critical to demonstrate, since it confirmed that the chimeric SHIVs that we made were capable of supporting replication. We next asked if substitution of the wildtype Env375 allele by one or more aliphatic or aromatic residues (Met, Trp, Phe, Tyr or His) would support SHIV replication in rhesus CD4 + T cells. The answer was affirmative for SHIVs expressing each of the ten HIV-1 Envs (Fig. 2). The differences in virus replication in rhesus CD4 + T cells between SHIVs expressing wild-type Env375 residues and those expressing bulky aromatic residues was generally quite large, oftentimes resulting in >100-fold differences in p27Ag concentration in culture supernatants at multiple time points throughout the infection (Fig. 2). Among the six different Env375 alleles that were tested, Env375-Trp most consistently supported SHIV replication in rhesus CD4 + T cells: it was effective in all 10 HIV-1 Env backbones. Env375-Tyr was the second most favored residue followed by Env375-His or -Phe. It is notable that Trp is also the most conserved Env375 allele across the broad evolutionary spectrum of primate lentiviruses excluding humans and great apes (30). These results thus corroborate and extend a substantial body of scientific literature indicating that SHIVs bearing primary (non-adapted) wildtype HIV-1 Envs rarely replicate efficiently in rhesus cells (1-3, 27-29, 35, 38, 65, 66) and that this restriction can be lifted by substituting a single amino acid at position Env375. In our combined studies [this manuscript plus (35, 38)], we replaced wildtype Env375 residues in 16 primary HIV-1 Envs -15 of which could not support SHIV replication in RMs -and found in all instances that this substitution alone led to efficient SHIV replication in rhesus animals.
To extend these findings to in vivo analyses, we inoculated 41 RMs intravenously in groups of 3 to 6 animals each, with SHIVs containing one of the ten selected HIV-1 Envs and an equal mixture of the six Env375 alleles (Table S1 and Fig. 3). We used this experimental design for two reasons: First, because target cell availability is not limited in the initial two weeks of infection during which time virus titers increase exponentially (67-70), we could use deep sequencing of plasma vRNA/cDNA to directly compare the relative replication rates of the six Env375 allelic variants in an in vivo competitive setting. Second, it would be impractical and prohibitively expensive to test 60 SHIVs individually in 60 different monkeys, and even if this could be done, the results would be confounded by monkey-specific variables such as MHC class I and II recognition. Each of the 41 RMs that we inoculated with a SHIV Env375 mixture became productively infected after a single challenge (Fig. 3). In most animals, peak viremia occurred at day 14 post-SHIV inoculation and plasma virus load setpoints were reached 16-24 weeks later.
Animals treated with anti-CD8 mAb at the time of SHIV inoculation developed significantly higher peak and setpoint viremia titers compared with untreated animals (p<0.01 for both). A subset of animals was treated with anti-CD8 mAb at setpoint, 20-50 weeks after infection; most of these animals exhibited increases in virus titers. We performed next generation sequencing (NGS) on plasma samples taken 2 and 4 weeks post-infection to determine the relative replication rates of the different Env375 allelic variants (Fig. 3). We expected that differences in infectivity of the Env375 variants would be reflected in the plasma virus quasispecies by two weeks postinoculation since the combined half-lives of circulating virus and the cells producing it is <1 day (71), resulting in multiple rounds of de novo virus infection and replication during this early interval. This was indeed the case. Overall, there was a good correlation between Env375 residues that supported SHIV replication in vitro and in vivo. For example, in all ten different Env backgrounds, Env375-Ser failed to support SHIV replication in primary rhesus CD4 + T cells in vitro ( Fig. 2) and the same was true in RMs in vivo (Fig. 3). Conversely, Env375-Trp supported SHIV replication in all ten Env backgrounds in vitro and was a predominant allele supporting efficient SHIV replication in 7 of 10 Env backgrounds in vivo. There were some differences in Env 375 residues that best supported SHIV replication in vitro versus in vivo. For example, for SHIVs bearing ZM233 and CH0694 Envs, 375-Trp supported efficient virus replication in vitro but not in vivo, where 375-Tyr was dominant. And the Env375-His allele, which is naturally present in most subtype AE viruses including the AE.40100 strain, supported efficient SHIV.AE.40100 replication in rhesus CD4 + T cells in vitro but not in vivo. Taken together, the findings indicate that substitution of wildtype Env375 alleles in primary HIV-1 Envs with Trp, Tyr or His results in SHIV chimeras that replicate efficiently in RMs. However, since it is impossible to predict with certainty which Env375 allele will best support in vivo replication of a SHIV bearing any particular HIV-1 Env, an in vivo competition experiment similar to that illustrated in Fig. 3 must be conducted.
We also compared the relative replication efficiency of SHIV.CH505.375H generated by the first and second generation construction strategies (Fig. S2). We showed previously that in animals infected by viruses produced from the first generation design, that redundant HIV-1 tat1 and env gp41 sequences (68 and 21 bp, respectively) were spontaneously deleted following prolonged in vivo replication [ Fig. S1; (35,38)]. This suggested a fitness disadvantage for viruses containing the redundant sequences, leading us to hypothesize that animals infected by an equal mixture of the viruses derived from the two designs would show preferential replication by viruses lacking the redundant sequences. This was indeed the case (Figs. S1A and S1B). At three weeks post-infection, viruses lacking the redundant sequences comprised >95% of the plasma virus quasispecies, and by week 8, they comprised >99% of plasma virus.
To be a relevant model for HIV-1 vaccine studies, SHIV Envs should exhibit clinically relevant antigenic profiles, neutralization sensitivity phenotypes, and coreceptor usage indistinguishable from the primary HIV-1 Envs from which they were derived. We evaluated the neutralization sensitivity patterns of Envs expressing the wild-type Env375 allele compared with Envs expressing one or more of the alternative Env375 alleles that were found to support 13 replication in rhesus CD4 + T cells in vitro (Figs. 2) and in RMs in vivo (Figs. 3). SHIVs were analyzed using polyclonal anti-HIV-1 sera and a battery of monoclonal antibodies (mAbs) that bind canonical bNAb epitopes, linear V3 epitopes or CD4-induced (CD4i) epitopes (Fig. 4). Linear V3 and CD4i epitopes are generally concealed on native Env trimers from primary viruses (40, [72][73][74], and thus mAbs targeting these epitopes typically fail to neutralize primary virus strains. Conversely, neutralization by linear V3 or CD4i mAbs is generally an indication of a non-native "open" trimer structure typical of laboratory-adapted viruses. In none of the ten primary Env backbones that we tested did Env375 substitutions result in neutralization by linear V3 or CD4i mAbs (Fig. 4). Nor did Env375 mutations alter the neutralization sensitivity of these Envs to HIVIG B, HIVIG C or a high titer, broadly neutralizing HIV-1 infected patient plasma specimen CH1754.
These results suggest that the Envs bearing residue 375 substitutions retained their native or near-native conformation. These Envs also retained their antigenicity with respect to bNAb epitope presentation since mAbs targeting CD4bs, V2 apex, V3 high mannose patch, and MPER sites exhibited similar neutralization patterns against wild-type and Env375 substituted variants.
It is notable that the contours of the neutralization curves, the IC50, IC80 and IC90 values, and the steep sigmoidal inflections were generally indistinguishable between wildtype Envs and Envs bearing residue 375 substitutions. SHIV.Q23.17 demonstrated neutralization sensitivity patterns to the bNAb mAbs, the three polyclonal anti-HIV IgG and plasma reagents and the mAbs targeting linear V3 or CD4i epitopes that were similar to the other nine SHIVs, thus supporting a tier 2 status for this virus. We also tested SHIV.CH505.375H derived by first and second generation design schemes for sensitivity to HIV-1 bNAbs, linear V3 targeted mAbs, HIVIG-C and the anti-HIV-1 broadly neutralizing polyclonal plasma CH1754: the two virus preparations showed 14 indistinguishable neutralization sensitivity patterns (Fig. S2). Finally, the SHIVs containing the ten new HIV-1 Envs were tested for coreceptor usage by analyzing their sensitivity to AMD-3100 (a CXCR4 inhibitor) and Maraviroc (a CCR5 inhibitor). Maraviroc, but not AMD-3100, inhibited the entry of all 10 SHIVs in the TZM-bl entry assay (Fig. S3), thus demonstrating CCR5-dependent entry. Altogether, the results indicate that Env375 substitutions did not appreciably alter the antigenicity, tier 2 neutralization sensitivity or CCR5 tropism of any of the ten SHIVs.
SHIVs intended for use as challenge strains in preclinical vaccine trials can be generated from 293T cells by transfection of proviral DNA or by virus passage and expansion in primary human or rhesus CD4 + T cells. Each approach has potential advantages and disadvantages (3,75).
We chose to prepare challenge stocks by infecting primary, activated rhesus CD4 + T cells with molecularly cloned virus derived from 293T cell transfections and then expanding the virus as rapidly as possible so as to minimize chances for culture adaptation. By this means, we could ensure that the viral envelopes of challenge stocks contained exclusively rhesus (not human) membrane-associated proteins and that glycosylation patterns would be of rhesus (not human) origin. We selected nine SHIV strains for large scale expansion in rhesus cells and these are listed in Table 2. These SHIVs were chosen to be representative of global HIV-1 diversity, including subtypes A, B, C, D, and AG, and to include SHIVs bearing BG505.N332, CH505 and 1086 Envs, which correspond to vaccine candidates in current or recent human clinical trials. Our aim was to generate large numbers of identical replicates of each SHIV stock (>1,000 vials per SHIV), which could then be characterized biophysically for genetic composition, particle content, infectivity, antigenicity and neutralization sensitivity and cryopreserved in vapor phase liquid nitrogen (<160 o C) for subsequent distribution as validated, standardized SHIV challenge stocks. Thus, we inoculated cultures of 100-200 million primary, activated, rhesus CD4 + cells pooled from three naïve Indian RMs at a multiplicity of infection (MOI) of approximately 0.01 with genetically homogeneous, sequence-confirmed, 293T transfection-derived virus stocks. For SHIV.Ce1176, we infected primary rhesus cells with an equal mixture of Env375-His, Phe and Trp alleles, and for SHIV.T250 we infected cells with an equal mixture of Env375-His, Tyr and Trp alleles, because each of these alleles in these two Env backgrounds had shown preferential replication in different animals (Fig. 3). The other SHIV challenge stocks were generated with single Env375 alleles ( Table   2). On days 7 and 14 post-SHIV inoculation, we added new media and approximately 100-200 million fresh, uninfected rhesus CD4 + T cells from three different naïve RMs so as to expand cell numbers and culture volumes while maintaining cell concentrations between 1-2 million per milliliter. Beginning on day ~10 post-SHIV inoculation, we collected the total volume of culture supernatant and replaced it with a greater volume of fresh medium. This complete media collection and replacement was then repeated every 4 days through day 21. By this means, we could collect as much as 2.5 liters of culture medium containing each SHIV over a period of approximately 21 days. Each supernatant collection was centrifuged twice at 2500 rpm (1000g) for 15 minutes to remove any residual cells or cell debris and then immediately frozen in bulk at -80 o C. Supernatants were not filtered so as to retain the highest possible infectivity titers. Thus, most of the virus that was collected and frozen during the 18-21 day culture period was <4 days old and underwent only one freeze-thaw cycle prior to final vialing. After all supernatant collections had been made, they were thawed at room temperature, combined in a sterile 3 liter flask to ensure complete mixing, and then aliquoted into as many as 2,500 cryovials, generally at 1 ml per vial. The vials were then transferred to vapor phase liquid nitrogen for long-term storage. 16 By this means, we could ensure that all vials were virtually identical in their contents. Between 192 and 2,224 vials per SHIV, each containing between 0.25 and 1.0 ml of challenge stock, were cryopreserved ( Table 2). Validation analyses were done on thawed cryovial samples to ensure results would be representative of all cryopreserved samples. Challenge stocks were free of bacterial or fungal contamination based on culture on thioglycolate broth. p27Ag concentrations ranged from 73 to 634 ng/ml and vRNA concentrations ranged from 5.0 x 10 8 to 4.1 x 10 9 vRNA /ml. Infectivity was tested on TZM-bl cells where it ranged from 1.5 x 10 5 to 3.2 x 10 7 IU/ml, and on primary rhesus CD4 + T cells where it ranged between 1.9 x 10 3 to 4.1 x 10 6 IU/ml. The genetic composition of the SHIV challenge stocks was analyzed by single genome sequencing of 3' halfgenomes to validate the authenticity of each stock and to determine if there was evidence of selection in vitro (Fig. 5A). Stocks of SHIV.Ce1176 and SHIV.T250 were sequenced by Illumina deep sequencing to determine the relative proportion of the different Env375 alleles in the final challenge stocks (Fig. 5B). Envelope sequence mean and maximum diversity averaged 0.05% (range 0.03-0.13%) and 0.30% (range 0.15-0.42%), respectively in the nine challenge stocks.
Mutations across the complete gp160 were essentially random in all challenge stocks except in a secondary expansion of SHIV.CH505. This challenge stock was prepared by infecting naïve rhesus CD4 + T cells with virus from the first expansion of SHIV.CH505 in an attempt to expand sequence diversity and increase infectivity titers. Maximum sequence diversity and maximum sequence divergence from the T/F sequence were 0.35% and 0.29% for stock #2 compared with 0.15% and 0.08%, respectively, for stock #1. p27Ag and vRNA concentrations and infectivity titers on TZM cells were similar for stocks #1 and #2 and infectivity titers on primary rhesus CD4 T cells were about 3-fold higher for stock #2 compared with stock #1.
HIV-1 strains produced in primary human CD4 + T cells, compared with the same viruses produced in 293T cells, have been reported to exhibit variably greater resistance to neutralizing antibodies (76,77). These differences have been attributed to differences in Env content, cell adhesion molecules, surface glycan composition or other factors (75). We tested five SHIVs -BG505, CH505, CH848, B41, D.191859 -produced in primary rhesus CD4 + T cells and in 293T cells for sensitivity to 19 neutralizing mAbs that targeted CD4bs, V3 glycan, V2 apex, MPER, surface glycan, CD4i or linear V3 epitopes (Fig. 6). None of the viruses, regardless of cell derivation, were sensitive to the four mAbs that targeted CD4i or linear V3 epitopes, indicating that they retained a native-like closed Env trimer regardless of the cell of origin. SHIVs produced in 293T cells and primary rhesus cells also exhibited similar overall patterns of sensitivity to the other 15 mAbs in that if a SHIV was sensitive (or resistant) to neutralization by a particular mAb, then this was true regardless of its cell of origin. However, as reported for HIV-1 strains, we observed enhanced resistance to some mAbs by some SHIVs grown in primary rhesus cells compared with 293T cells.
The properties of rhesus CD4 + T cell grown SHIV challenge stocks summarized in Table 2, especially their consistently high virus titers and infectivity measurements, suggested that these virus strains might be suitable for mucosal transmission studies and to assess the preclinical 3 that were infected intravenously. Although our intrarectal AID50 titration experiment for SHIV.BG505.N332 involved a small number of animals (n=12) and was subject to stochastic effects related to intrarectal virus inoculation, we could nonetheless estimate the AID50 of this stock to be approximately 1:120 (1 ml) for atraumatic IR challenge. This result was corroborated in the control (sham-treated) arm of a preclinical trial assessing the protective efficacy of BG505 SOSIP vaccine elicited neutralizing antibodies against a homologous SHIV.BG505.N332 challenge (7).

Discussion
In recent years, there have been notable advances in HIV prevention and cure research (79)(80)(81)(82)(83) yet the goals of effective vaccination and cure -even a "functional" cure -seem far in the distance. Increasingly, experimental medicine trials in humans have been pursued as a strategy to accelerate translational research (82), but at the same time, there remain untapped opportunities and needs for animal models to complement and synergize with human studies to hasten progress. Different scientific questions demand different model systems, ranging from transgenic or humanized mice to outbred small and large animals. Aside from the great apes, which are endangered and thus precluded from laboratory investigation, the rhesus macaque monkey (Macaca mulatta) is most similar to humans in its immune repertoire (84,85). At the same time, Env is the target of an array of neutralizing antibodies and cytotoxic T-cells that cause it to evolve continuously in order to escape recognition that would otherwise lead to virus 20 elimination (38, 86,87). Env accomplishes the latter by means of highly evolved properties, including occlusion of trimer-interface epitopes (88), epitope variation (89), conformational masking (90) and glycan shielding (91). Although HIV-1 Env is notorious for its variability and global diversity (www.hiv.lanl.gov), it is nonetheless constrained in its potential for immediate or near-term evolution due to the myriad of essential biological functions encoded in its sequence (38,(92)(93)(94). These constraints can be lifted, however, by in vitro cultivation (66, 95) or extensive passage in unnatural animal hosts (1)(2)(3)22). The implication of these observations is that the most relevant HIV-1 Envs for studies of vaccine-elicited protection, passively acquired antibody protection, or curative intervention are primary or T/F Envs from viruses that are responsible for clinical transmission and the establishment of persistent infection in humans (7)(8)(9)(10)96). T/F Envs express the precise primary, secondary, tertiary and quaternary protein structures that are essential for transmission and T/F Envs are the ones that a vaccine-elicited bNAb response must recognize if it is to be protective (38, 73, 82). Envs derived from short-term virus cultures in human lymphocytes or Env sequences derived from plasma vRNA/cDNA are a first approximation to T/F Envs but they may differ in important but unrecognized features. Envs derived from extensively passaged virus cultures are less likely to reflect the biologic and antigenic properties of T/F viruses. In this context, 7 of the 10 new SHIVs described in the current study, and 12 of 16 SHIVs that we have reported overall (Table 1), were constructed using T/F Envs. The remainder was constructed using primary Envs.
A recent study by Keele and colleagues (29) aimed to create new subtype C T/F SHIVs using 20 South African subtype C T/F Envs and either of two strategies to enhance replication in primary RM CD4 + T cells. One of these strategies was the same EnvΔ375 design employed here and the other was an EnvΔ281 approach reported elsewhere (28). Because the O'Brien study pooled SHIVs for competitive replication analyses in RMs, a precise determination of the proportion of wild-type HIV-1 Envs that could support SHIV replication in monkeys could not be made. However, in the instances where EnvΔ375 substitutions were made and the resulting SHIVs were tested individually, EnvΔ375 substitutions were successful in conferring replication competence to SHIVs in rhesus cells. The addition of EnvΔ281 was neither additive nor synergistic. In our studies described here ( Table 1)  Moreover, in vitro measures of virus content, infectivity and replication in cell culture did not always predict in vivo outcomes, lending a measure of uncertainty to SHIV design and analysis.
The EnvΔ375 strategy alleviates much of this uncertainty and unpredictability as demonstrated by the following results: i) Env375 substitutions alone were sufficient to enhance Env affinity to rhesus CD4, reduce the energetic threshold for downstream Env transitions following CD4 binding, and convey efficient infectivity to the virus in primary rhesus CD4 + T cells in vitro and in vivo; ii) the Env375 substitution strategy worked consistently; every attempt that we ( Table 1), Keele (29) and Barouch (97) have made to engineer a T/F or primary HIV-1 Env SHIV by residue 375 substitution has succeeded in producing a chimeric virus that replicates efficiently in RMs; iii) the ability of such EnvΔ375 SHIVs to replicate in vivo was, in each case, predicted by efficient replication in primary, activated rhesus CD4 + T cells in vitro; this is a different result from what has been reported for other SHIVs (1-3, 65) and we suspect that our simple EnvΔ375 design scheme, our protocol for rhCD4+ T cell activation, and our method for infecting these cells in tissue culture are responsible for the differences; iv) the antigenicity and tier 2 neutralization sensitivity of wildtype HIV-1 Envs was closely mirrored by EnvΔ375 mutants expressed from 293T cells or as infectious SHIVs from primary rhesus CD4 T cells; v) the genetic diversity of each SHIV infection stock was very low when virus was expressed either from 293T cells or from primary rhesus CD4 + T-cells; vi) transmission efficiency of SHIVs across rhesus rectal, vaginal, penile and oral mucosa, and intravenously, mirrored the transmission efficiency of HIV-1 in humans; vii) acute and early SHIV replication dynamics in RMs measured by plasma vRNA replicated what has been seen in humans, including a 7-14 day eclipse period before vRNA is detectable in plasma, an exponential increase in plasma virus load to a peak approximately 14-28 days post-infection, establishment of setpoint viremia two or more months later, and immunopathogenesis leading to clinically-defined AIDS in a subset of animals (69,70,73); viii) SHIV infected RMs consistently elicited autologous, strain-specific NAbs, and in some cases bNAbs, with kinetics similar to HIV-1 infected humans (35, 38); ix) molecular pathways of SHIV Env evolution in RMs closely mirrored evolution of homologous HIV-1 Envs in humans, including precise molecular patterns of Env-Ab coevolution leading to Nab escape, and in some animals, the development of bNAbs (38). The latter results speak to the native-like structure of SHIV Envs and to homologies and orthologies in human and rhesus immunoglobulin gene repertoires (38, 85). Altogether, the findings highlight the reproducibility and relevance of the SHIV EnvΔ375 infected RM as fold difference in AID50 between IR and IVAG challenge routes is consistent with previous findings with SHIV and SIVs (3,98,99) and is similar to estimates of relative infectivity in humans exposed to receptive anal intercourse versus receptive vaginal intercourse (78). We also titrated SHIV.CH505 challenge stocks for AID50 in RMs following intrarectal or intravaginal inoculation. In independent studies with a total of 21 RMs, Klatt [(100) and unpublished data] and Haynes estimated the AID50 following IR challenge of naïve RMs to be approximately 1:80 (1 ml), while Felber and colleagues (8) found the AID50 of this stock following IVAG challenge to be approximately 1:2 (1 ml) ( Table 2). These findings again demonstrate reproducibility in AID50 titers in different primate centers and in monkeys from different breeding colonies as well as a 30-40 fold difference in infectivity between IR versus IVAG challenge routes. Previously, we estimated the AID50 for SHIV.D.191879 for IVAG inoculation to be approximately 1:3 (1 ml) (101).
Here, we could not estimate an AID50 for penile transmission by the SHIV.D.191879 challenge stock since 2 of 2 animals became infected after a single inoculation (Fig. 7A), but the findings suggest that the AID50 titers of this stock for penile transmission are likely to be sufficient for it to be used as a challenge stock in preclinical prevention trials once formal titering is completed. Altogether, the findings of this study suggest that the SHIVs listed in Table 1  the vector sequence, respectively. We synthesized two fragments that contain these two enzyme sites and the genes in between. We eliminated the redundant tat1 and env gp41 sequences and replaced the vpu-env and env genes with a linker fragment that carries two BsmBI restriction enzyme sites (Fig. 1). The BsmBI site appended at the N-terminus of the linker recognizes the reverse complementary DNA strand and creates a 3' overhang; the one added at the C-terminus recognizes the positive strand DNA and creates a 5' overhang. This design results in two different sticky ends, which allows unidirectional cloning of the insert into the backbone. BsmBI is a Type IIS restriction enzyme that cleaves outside of its recognition site and thus the enzyme recognition sequence does not remain after ligating the insert into the backbone (Fig. 1). The two synthesized fragments were then cloned into original SHIV backbone separately using the BstBI and XhoI sites. mixture of Env375-His, Tyr and Trp alleles, because these alleles in these two Env backgrounds had shown differential replication in different animals (Fig. 3). The other SHIV challenge stocks were generated with viruses containing single rhesus-preferred Env375 alleles ( Table 2). Total volume of the SHIV-cell mixture was typically 10-30 ml, depending of the infectivity titers of the 293T virus stock. The SHIV-cell mixture was transferred to a T75 flask, which was fixed to a rotating wheel or rocker so that leakage or spillage was not possible. This apparatus was then       Table 2) and in RMs in vivo (Fig. 3).        inhibitors Maraviroc and AMD-3100, respectively. HIV-1 YU2 Env utilizes CCR5 exclusively for cell entry and HIV-1 SG3 Env utilizes CXCR4 exclusively. Maraviroc and AMD-3100 selectively abrogated cell entry of these viruses, as expected. Entry of the ten SHIVs bearing wildtype or rhesus-preferred Env375 alleles was inhibited by >99% by Maraviroc but minimally or not at all by AMD-3100, indicating a dependence on CCR5 for cell entry.                 * B41 Env is also designated as 9032.08_A1 (73).
^ SHIV.BG505 exists in two versions with and without an asparagine and potential N-linked glycan at Env residue 332 (35).