Identification of HIV-1 Envelope Mutations that Enhance Entry Using Macaque CD4 and CCR5

Although Rhesus macaques are an important animal model for HIV-1 vaccine development research, most transmitted HIV-1 strains replicate poorly in macaque cells. A major genetic determinant of this species-specific restriction is a non-synonymous mutation in macaque CD4 that results in reduced HIV-1 Envelope (Env)-mediated viral entry compared to human CD4. Recent research efforts employing either laboratory evolution or structure-guided design strategies have uncovered several mutations in Env’s gp120 subunit that enhance binding of macaque CD4 by transmitted/founder HIV-1 viruses. In order to identify additional Env mutations that promote infection of macaque cells, we utilized deep mutational scanning to screen thousands of Env point mutants for those that enhance HIV-1 entry via macaque receptors. We identified many uncharacterized amino acid mutations in the N-terminal heptad repeat (NHR) and C-terminal heptad repeat (CHR) regions of gp41 that increased entry into cells bearing macaque receptors up to 9-fold. Many of these mutations also modestly increased infection of cells bearing human CD4 and CCR5 (up to 1.5-fold). NHR/CHR mutations identified by deep mutational scanning that enhanced entry also increased sensitivity to neutralizing antibodies targeting the MPER epitope, and to inactivation by cold-incubation, suggesting that they promote sampling of an intermediate trimer conformation between closed and receptor bound states. Identification of this set of mutations can inform future macaque model studies, and also further our understanding of the relationship between Env structure and function.


Introduction 39
Entry receptors are a major determinant of the host cell specificity of viruses (1). Within  Deep mutational scanning identifies Env mutations that improve usage of macaque receptors. 92 The Env strain used in our DMS, subtype A BF520.W14M.C2 env (hereafter referred to as 93 BF520), was cloned directly from a Kenyan infant at the timepoint when HIV was first detected 94 following mother-to-child transmission (19). Like Envs from other circulating HIV-1 strains, the 95 wild type BF520 Env supported infection of cells engineered to express human CD4 and CCR5 96 (293T hu ) but was severely restricted in its ability to infect cells expressing macaque receptors 97 (293T rhm ) ( Figure 1

) (3). 98
In our DMS, we screened a library of BF520 Env variants to identify mutations that increased 99 entry into cells expressing macaque receptors. The DMS strategy we employed is outlined in 100 To quantify the effect of each Env mutation on infection of either 293T rhm or 293T hu cells we 128 first inferred the relative preference for all 20 amino acids at each Env site, for each infection, 129 as previously described (21). We next calculated a statistic, S mut/wt , which describes selection on 130 each Env mutant as the log 2 of the ratio of the mutant residue preference to the wild-type was enriched 38-fold relative to the wild-type residue, compared to a 13-fold enrichment for 146 the most strongly selected mutation during 293T hu infection (E482K). This finding likely reflects 147 the fact that wild-type BF520 Env is poorly adapted to infect macaque cells, and therefore 148 mutants in this background have the potential to greatly enhance usage of macaque receptors. 149 Env mutations that had positive S mut/wt values following infection of 293T rhm cells are shown in 150 Figure 3. Surprisingly, while previous studies have predominately identified mutations in Env's 151 gp120 subunit that improve infection of macaque cells (4, 11, 12, 22), the majority of mutations 152 highly enriched in our experiments were located in the gp41 subunit. In fact, all 29 amino acid 153 10 mutations enriched greater than 10-fold over wild-type residues in our DMS were located at 154 just eight sites, all within gp41. Seven of these sites were located within either the N terminal 155 heptad repeat (NHR) or C-terminal heptad repeat (CHR), while the eighth was in the intervening 156 loop region. 157 While the most strongly enriched mutations were in N/CHR regions, we observed modest 158 enrichment of mutations in several other Env regions. At Env sites 58-63, 20 mutations were 159 selected that improved 293T rhm infection 3-to 8-fold compared to wild-type residues (Figure 3). 160 These sites lie within a short helix that was recently found to make contact with a CD4 molecule 161 bound to the inner domain of a neighboring gp120 protomer (26). Our DMS also revealed that 162 293T rhm infection is modestly improved by Env mutations that abolish the putative glycosylation 163 motif at sites 611-613 ( Figure 3). 164 Given the surprising result that previously uncharacterized mutations in the N/CHR showed the 165 greatest enrichment in the DMS, we sought to validate the effect of these mutations on 166 macaque receptor usage through an independent assay. We generated pseudoviruses bearing a 167 subset of these gp41 mutations in the BF520 Env background, then tested the ability of the 168 pseudovirus mutants to infect 293T rhm cells. Compared to wild-type BF520, all seven gp41 169 mutants assayed in these experiments improved entry into 293T rhm cells (Figure 4a). Similar to 170 their relative enrichment in the DMS, mutations in the CHR were most beneficial to 293T rhm 171 infection. For example, CHR mutations Q652F and Q653L improved infection of 293T rhm cells 9-172 and 7-fold respectively, whereas the most advantageous NHR mutation, I573L, improved 173 infection 4-fold ( Figure 4a). These data thus validate that N/CHR mutations enriched in our DMS 174 11 following infection of 293T rhm cells do indeed enhance entry of BF520 Env into cells bearing 175 macaque receptors. 176

Similar mutations improve usage of both macaque and human receptors 177
Infection of 293T hu cells with the mutant virus pools selected for beneficial mutations at many 178 of the same sites where mutations were enriched following 293T rhm infection ( Figure 5). 179 Mutations at sites 58-61, as well as those in the N611 glycosylation motif, were enriched up to 180 previously been identified to enhance infection of macaque cells (4, 11, 12). Mutations at these 208 sites were adequately represented in the initial mutant plasmid library (data not shown), thus 209 ruling out the possibility that they were not present at the start of our DMS experiment. Several 210 possible hypotheses could explain the apparent lack of selection for these mutants in the DMS. 211 First, these mutations may not have been tolerated in the BF520 Env strain background. If this 212 had been the case, they would have dropped out of the DMS following passage of the mutant 213 13 plasmid library. A second possibility is that these mutations may not have improved usage of 214 macaque receptors in the BF520 strain background. 215 To determine whether either of these possibilities explained the lack of enrichment of these 216 previously identified mutants, we first tested whether we could generate infectious 217 pseudovirus bearing two of these mutations, A204E and S375Y, in the BF520 background. The 218 titers of pseudoviruses bearing the S375Y Env mutation, as determined by infecting TZM-bl 219 cells, were similar to those of pseudoviruses bearing wild-type BF520 (2.6 x 10 6 and 2.8 x 10 6 220 infectious units/ml, respectively). In contrast, replicate transfection supernatants containing 221 A204E mutations were not infectious. We thus conclude that the A204E mutation is 222 incompatible with the BF520 Env, thereby explaining its lack of enrichment in our DMS. 223 We next assessed whether the S375Y mutation in the BF520 background would enhance 224 infection of 293T rhm and 293T hu cells in pseudovirus reporter assays. As shown in Figure 4a, the 225 S375Y mutations increased infection of 293T rhm cells 34-fold compared to wild-type BF520. This 226 was a higher fold increase in infection than any of the mutations that were most enriched in the 227 DMS. In the context of 293T hu infection, S375Y enhanced wild-type infection by 1.7-fold ( Figure  228 4b). These data indicate that the lack of enrichment of S375Y in our DMS cannot be attributed 229 to this mutation being incompatible with the BF520 Env, nor to it impeding 293T rhm infection in 230 this strain background. Additionally, the fact that S375Y was not detrimental to 293T hu infection 231 indicates that the mixture of 293T hu and 293T rhm cells used to establish a mutant genotype-232 phenotype link in our DMS was likely not responsible for selection against this mutant. 233 14 Why then, was 375Y not enriched in our DMS? An examination of our data revealed that the 234 frequency of S375Y, as well as previously identified macaque adaptive mutations G312V and 235 A281T, was lower following 293Trhm and 293Thu than in the mutant plasmid libraries, 236 evidence of selection against these mutations during the experiment. To further explore this 237 finding, we revisited data from a previously conducted DMS experiment in which the same 238 BF520 mutant Env library was passaged and selected in a human cell line expressing human 239 CD4 and CCR5 (SupT1 T cell line; (17)). In these data, we observed a similar drop in frequencies 240 of S375Y, G312V, and A281T that mirrored the depletion of these mutations in the current DMS 241 experiment. Selection against these mutations in both DMS experiments suggests that 242 properties inherent to the mutant plasmid library or the DMS protocol are likely responsible for 243 their lack of enrichment following 293T rhm infection. One potential explanation is that while 244 these mutations enhance entry of pseudovirus into 293T rhm and 293T hu cells, they are 245 detrimental in the context of multi-cycle replication and purged from the mutant pool during 246 the low MOI infection necessary to generate genotype-phenotype linked virions. 247

NHR/CHR mutants enhance usage of macaque receptors by Env strain BG505 248
We next sought to determine whether the mutations enriched in the DMS were advantageous 249 solely in the context BF520 Env, or instead, reflected a more widely accessible mutational 250 strategy through which Env could evolve to utilize macaque receptors. To investigate this, we 251 introduced seven N/CHR mutants enriched in our DMS into the BG505.W6M.C2.T332N env 252 (hereafter referred to as BG505) (23), then tested the ability of these mutants to enhance 253 pseudovirus infection of 293T rhm cells. BG505 is a subtype A Env (23), 15.8% diverged from 254 BF520 at the amino acid level, and used extensively as a SOSIP trimer for Env structural studies 255 (24-28), and as a vaccine immunogen (28, 29). Recent DMS efforts have also revealed that the 256 site-specific amino acid preferences and mutational tolerance of BG505 Env have diverged from 257 BF520 (15). 258 Infection of 293T rhm cells with wild-type BG505 also revealed this strain to be ~7-fold less 259 effective at entering cells bearing macaque receptors compared to wild-type BF520. At an MOI 260 of 5, only 0.12% of 293T rhm cells were infected following wild-type BG505 infection (Figure 4c) 261 compared to 0.87% following wild-type BF520 infection ( Figure 4a). 262 Four of the seven N/CHR mutants introduced into BG505 improved the ability of this Env to 263 infect 293T rhm cells (Figure 4c). The magnitude of this effect was smaller than in the BF520 264 strain however, with the most advantageous mutant, (Q653L) conferring a 2.5-fold 265 improvement over wild-type BG505. Generally, mutations that conferred the greatest 266 improvement to macaque receptor usage in the BF520 strain background also had the largest 267 effect in BG505. For example, mutations Q652F and Q653L, which conferred the greatest 268 benefit to infection in the BF520 background were also the most advantageous in the BG505 269 strain. Only one mutation, R557L, benefited BF520 infection of 293T rhm cells, but was 270 detrimental to infection of these cells in BG505. Taken together, these data indicate that while 271 some N/CHR mutations have strain specific effects, the benefit conferred by larger-effect 272 mutations is not restricted to the BF520 background. 273

epitopes. 275
We next sought to investigate the mechanism through which the N/CHR mutations enhanced 276 usage of macaque receptors. First, we tested whether these mutants conferred a CD4-277 independent phenotype by assaying whether they allowed entry of BF520 pseudovirus into 278 293T cells expressing only the macaque CCR5 receptor (293T rhmCCR5 ). No infection was observed 279 by either wild-type or mutants Envs in this assay, thereby ruling out CD4-independence as the 280 mechanism through which N/CHR mutants enhance 293T rhm infection (figure 4d). 281 We next used a panel of monoclonal antibodies (mAbs) and inhibitory mimetic drugs to 282 determine whether N/CHR mutations caused alterations in Env trimer structure that would 283 inform on a mechanism through which they improved entry into 293T rhm cells. In these 284 experiments, we assessed neutralization susceptibility of three mutants that were highly 285 enriched in our DMS: I573L (NHR), Q653L (CHR) and N656Y (CHR). Neutralization of each 286 mutant was assessed in both the BF520 and BG505 Env strains so as to inform on the 287 background dependence of any structural effects. Compared to wild-type Envs, introduction of 288 N/CHR mutants had little effect on neutralization sensitivity to antibodies targeting the CD4bs, 289 trimer apex, V3 loop, or six-helical-bundle epitopes ( Table 1). All tested mutants did however 290 increase sensitivity to the MPER antibodies 4E10 and 10E8 in both BG505 and BF520 291 backgrounds (Table 1). This effect was strongest for CHR mutants Q653L and N656Y, which 292 increased sensitivity to MPER antibodies by 15-and 8-fold, in BF520 and BG505 backgrounds, 293 respectively. In the BF520 strain, NHR/CHR mutants also modestly increased sensitivity to T-20, 294 a fusion-inhibitor drug that binds a hydrophobic pocket at the N-terminus of the NHR helix. 295 Again, the effect was largest for CHR mutants, which increased T-20 sensitivity 2-to 4-fold. 296 Prior to engagement of host receptors, it is thought that Env predominately maintains a closed 298 conformation in which MPER and NHR residues are poorly exposed and not accessible to 299 interactions with mAbs or small molecule inhibitors (30). Receptor binding initiates formation 300 of the pre-hairpin intermediate conformation, in which both the MPER epitopes, as well as the 301 T-20 binding site in the NHR are exposed (31-33). Several recent publications have described 302 Env mutations that promote unliganded trimer sampling of an intermediate conformation, 303 termed "state 2", between closed (state 1) and receptor bound (state 3) conformations (34-37). 304 State 2 is characterized by a heightened intrinsic reactivity to ligand binding, increased capacity 305 to infect cells expressing low levels of CD4 or CCR5 receptors, and increased exposure of MPER 306 and NHR residues. Given that the N/CHR mutants we identified increase exposure of MPER 307 epitopes, and in strain BF520, the NHR helix, we hypothesized that they may increase reactivity 308 to macaque receptors by promoting sampling of a state 2 conformation. 309 Envs frequently sampling state 2 conformations display increased sensitivity to inactivation 310 during cold incubation (34, 38). To investigate whether N/CHR mutants identified in our DMS 311 also increased sensitivity to cold-inactivation, we assessed infectivity of wild-type and mutant 312 BF520 pseudoviruses following incubations on ice. Introduction of Q653L, N656Y and I573L into 313 BF520 decreased Env infectivity following ice incubation ( Figure 6). This effect was more 314 pronounced for Q653L and N656Y than for I573L, paralleling the relative effect sizes of these 315 mutants observed during 293T rhm infection and in neutralization sensitivity assays. It thus 316 appears that the degree to which N/CHR mutants enhance usage of macaque receptors is 317 18 correlated to increased cold-sensitivity and exposure of MPER epitopes on the unliganded Env 318 trimer. These observations are consistent with the hypothesis that N/CHR mutations promote 319 increased Env reactivity through increased sampling of a state 2 like conformation. 320

Discussion 321
We identified mutations in several regions of an Env from a transmitted HIV-1 virus that led to 322 increased entry using macaque CD4 and CCR5 receptors. Mutations within gp120 modestly 323 improved usage of macaque receptors, but mutations with the largest beneficial effect were 324 located in gp41, most notably in the N/CHR regions. N/CHR mutations enhanced entry into cells 325 bearing macaque receptors by a second transmitted Env strain, indicating that their effects 326 were not strain specific, but they did not confer a CD4 independent phenotype. While N/CHR 327 mutations have previously been identified that heighten fusogenic activity (39, 40) and increase 328 infectivity of cells expressing low levels of CD4 and CCR5 receptors (34, 41), none have been 329 shown to enhance entry of cells bearing macaque receptors. Many of the N/CHR mutations we 330 identified also enhanced entry into cells bearing human CD4 and CCR5 receptors, but the 331 magnitude of this effect was smaller than during infection of cells bearing macaque receptors. 332 N/CHR mutations that enhanced usage of macaque receptors increased sensitivity to 333 neutralization by MPER antibodies, and, in the BF520 background, to inhibition by the fusion 334 inhibitor T-20. The MPER epitope (sites 671-683) and T-20 binding pocket (sites 547-556) are 335 located immediately N, and C terminal, respectively, to the NHR and CHR mutants enriched in 336 our DMS (36, 37). Our neutralization data thus indicate that these mutations cause Env 337 conformation changes that are confined to regions of the trimer directly adjacent to the 338 mutated residues. The mechanism by which these mutants enhance usage of macaque 339 receptors thus appears to be distinct from that of mutations A204E and G312V, which 340 neutralization data suggest disrupt quaternary Env contacts and promote a more open trimer 341 20 conformation (9). The mechanism also appears to be distinct from mutations at site 375 and 342 281, which are located within the CD4 binding site region, and minimally impact Env 343 susceptibility to neutralizing antibodies (10, 11). 344 Viruses with mutations in NHR and CHR also demonstrated increased sensitivity to inactivation 345 by cold. This phenotype has been linked to Envs frequently sampling state 2 conformations, in 346 which the thermodynamic barrier between unliganded and CD4-bound states is lowered (34, 347 35). Envs displaying a state 2 phenotype are also known to display heightened sensitivity to 348 MPER antibodies (34-37), and in some cases, increased exposure of the NHR helix (34). Our 349 observations that N/CHR mutations increase sensitivity to MPER antibodies, and in the BF520 350 background modestly increase T-20 sensitivity, are consistent with the hypothesis that these 351 mutations promote a state 2 conformation. 352 Mapping of the N/CHR mutations identified here onto the structure of the unliganded BG505 353 SOSIP trimer reveals that NHR mutations are located within the central core of the trimer, while 354 CHR mutations are positioned at the trimer base. Mutations in both regions are located at sites 355 that make inter-and intra-protomer contacts with adjacent gp120 and gp41 subunits (Figure 7). 356 Given that previous reports have described N/CHR mutations that modulate trimer reactivity 357 through alterations of inter-protomer contacts (37, 42, 43), it is possible that the mutations 358 identified here promote a state 2 phenotype by disrupting the stabilizing interactions between 359 protomors within the unliganded trimer. If N/CHR identified here do indeed promote a state 2-360 like conformation, this would lower the thermodynamic barrier between unliganded and CD4 361 bound Env states, which could allow relatively weak interactions between Env and macaque 362 21 CD4 to initiate the conformational changes leading to exposure of the co-receptor binding site. 363 Subsequent binding of the co-receptor would lead to the cascade of Env rearrangements that 364 results in membrane fusion and cell entry. While speculative, this model of macaque receptor 365 usage is in line with a recent report of retroviral host-range expansion associated with envelope 366 mutations that promote a receptor-bound conformation (44). 367 We note that although the NHR and CHR mutations identified in this work generally enhance 368 usage of both macaque and human receptors, they occur at sites that are highly conserved 369 among circulating HIV-1 strains and lentiviruses (43, 45). Selective pressures exerted by the 370 human immune system could explain the discrepancy between the enrichment of these 371 mutations in our experiments and their lack of representation in natural isolates. Specifically, 372 our data suggest that N/CHR mutations that enhance entry also make the virus more 373 susceptible to neutralization by MPER antibodies. As the MPER epitope is highly conserved (46) 374 and functionally constrained (15), mutations that exposed this region are predicted to be 375 strongly selected against. Thus, in an established HIV-1 infection, the advantage that N/CHR 376 mutations have during entry may be balanced by the necessity of evading the immune 377 response. 378 While the most strongly enriched mutations were in N/CHR regions, we observed modest 379 enrichment of mutations in several other Env regions, including sites 58-63 and 611-613. 380 Enriched mutations at sites 58-63 are located within the recently discovered CD4-contacting 381 helix, in which several residues make contact with CD4 bound to the inner domain of an 382 adjacent gp120 protomer (47). Mutations in this region, when introduced into SOSIP trimers, 383 22 have also been shown to affect reactivity to CD4 engagement and conformational dynamics 384 (29). The mutations enriched in our DMS may therefore improve entry by either strengthening 385 contacts with CD4, or increasing Env reactivity to CD4 engagement. The role of mutations at 386 sites 611-613, which abolish the putative N611 glycosylation motif, is less clear. Previous 387 studies have found that removal of certain N-linked glycosylation sites in SIV gp120 enhance 388 replication in macaque macrophages (48) and increase exposure of CD4-contact residues (49), 389 but the relevance of these observations to the loss of the N611 glycan in the context of 293T rhm 390 infection is unclear. Interestingly, the magnitude of mutant enrichment within both the N611 391 glycan region, and the CD4-contacting helix was similar between 293T rhm and 293T hu selections, 392 suggesting that these mutations function by a mechanism that is similarly advantageous to both 393 human and macaque infection. 394 Our DMS did not select for mutations that have previously been found to improve usage of 395 macaque CD4 and CCR5. This observation can be attributed in part to incompatibilities between 396 these mutants and the BF520 Env background, as was observed for A204E. For other mutations, 397 such as S375Y that clearly improved entry of BF520 pseudovirus into 293T rhm cells, the lack of 398 enrichment is harder to explain. One possibility is that these mutations are detrimental in the 399 BF520 background during multi-cycle replication and were removed from the library during the 400 initial low MOI passage. Regardless of the underlying cause, depletion of these mutations 401 indicates that our DMS did not quantify the effects of every possible BF520 Env mutation on the 402 usage of macaque receptors. Future studies that employ different experimental methods are 403 therefore likely to discover additional Env mutants that enhance entry into macaque cells.