HIV-1 gp41 Residues Modulate CD4-Induced Conformational Changes in the Envelope Glycoprotein and Evolution of a Relaxed Conformation of gp120

Binding of the HIV envelope glycoprotein (Env) to cellular CD4 and chemokine receptors triggers conformational changes in Env that mediate virus entry, but premature triggering of Env conformational changes leads to virus inactivation. Currently, we have a limited understanding of the network of residues that regulate Env conformational changes. Here, we identify residues in HR1 of gp41 that modulate conformational changes in response to gp120 binding to CD4 and show that the mutations in HR1 and HR2 that confer resistance to fusion inhibitors are associated with gp120 mutations in different regions of Env that confer a more open conformation. These findings contribute to our understanding of the regulation of Env conformational changes and efforts to design new entry inhibitors and stable Env vaccine immunogens.

ABSTRACT

Entry of human immunodeficiency virus type 1 (HIV-1) into host cells is mediated by conformational changes in the envelope glycoprotein (Env) that are triggered by Env binding to cellular CD4 and chemokine receptors. These conformational changes involve the opening of the gp120 surface subunit, exposure of the fusion peptide in the gp41 transmembrane subunit, and refolding of the gp41 N- and C-terminal heptad repeat regions (HR1 and HR2) first into an extended prehairpin intermediate and then into a compact 6-helix bundle (6HB) that facilitates fusion between viral and host cell membranes. Previously, we reported that Envs resistant to HR1 peptide fusion inhibitors acquired key resistance mutations in either HR1 or HR2 that increased 6HB stability. Here, we identify residues in HR1 that contribute not only to fusion inhibitor resistance and 6HB stability but also to reduced reactivity to CD4-induced conformational changes that lead to 6HB formation. While all Envs show increased neutralization sensitivity to mimetic CD4 (mCD4), Envs with either the E560K or Q577R HR1 mutation reduced conformational reactivity to CD4 that resisted viral inactivation and triggering to the 6HB. Using a panel of monoclonal antibodies (mAbs), we further determined that Envs from both HR1 and HR2 resistance pathways exhibit a relaxed trimer conformation due to gp120 adaptive mutations in different regions of Env that segregate by resistance pathway. These findings highlight regions of cross talk between gp120 and gp41 and identify HR1 residues that play important roles in regulating CD4-induced conformational changes in Env.

IMPORTANCE Binding of the HIV envelope glycoprotein (Env) to cellular CD4 and chemokine receptors triggers conformational changes in Env that mediate virus entry, but premature triggering of Env conformational changes leads to virus inactivation. Currently, we have a limited understanding of the network of residues that regulate Env conformational changes. Here, we identify residues in HR1 of gp41 that modulate conformational changes in response to gp120 binding to CD4 and show that the mutations in HR1 and HR2 that confer resistance to fusion inhibitors are associated with gp120 mutations in different regions of Env that confer a more open conformation. These findings contribute to our understanding of the regulation of Env conformational changes and efforts to design new entry inhibitors and stable Env vaccine immunogens.

INTRODUCTION

The HIV-1 envelope glycoprotein (Env) mediates receptor binding and fusion of the virus with host cell membranes. Env is translated as the gp160 polyprotein that is subsequently cleaved by a cellular furin-like protease to the gp120 surface (SU) and gp41 transmembrane (TM) subunits. gp120 and gp41 associate as a non-covalently linked dimer, three of which assemble into a trimer of dimers forming the functional Env spike on the virion and infected cell surface. gp120 binding to both the CD4 primary cellular receptor (1, 2) and either the CXCR4 or CCR5 chemokine coreceptor (3–6) triggers a series of conformational changes that release the membrane fusion function of gp41.

The native Env trimer exists in a metastable state prior to interactions with receptors. In the native conformation, Env predominantly occupies a closed structure, in which the gp120 variable loops and extensive surface glycosylation shield much of the Env core (7–10). Receptor binding opens gp120 to expose the coreceptor binding site (11–13). Receptor and coreceptor binding leads to further conformational changes that enable two heptad repeat (HR) regions in the ectodomain of gp41 to self-assemble into a stable, hairpin-like, six-helix bundle (6HB) structure, the formation of which facilitates membrane fusion (14–17). In the native Env trimer, the C-terminal half of the N-terminal HR (HR1) in gp41 forms an α-helix arranged as a trimeric coiled coil with the other two gp41 protomers of the trimer (18). The N-terminal half of HR1 contains another short helix, with a loop region connecting the two HR1 helical domains. At the extreme N terminus of gp41, the hydrophobic fusion peptide (FP) is partially buried in the HR1 coiled coil. Receptor activation releases gp120 constraints on gp41 so that the entire HR1 region can extend into a single, long, trimeric coiled coil that repositions the FP for insertion into host cell membranes (19–21). This extended Env conformation is often referred to as the prehairpin intermediate (PHI). Resolution of the PHI into the final, stable 6HB draws the membranes together and drives fusion through a series of conformational changes that are poorly understood.

To gain insights into receptor activation of Env and fusion-intermediate conformations, we previously generated a large panel of Envs that are resistant to peptide fusion inhibitors corresponding to the highly conserved HR1 region of gp41 (22–25). According to the dominant negative mode of inhibition, HR1 peptides as monomers or dimers can bind to gp41 HR1, forming a peptide-gp41 coiled coil that traps the PHI (26, 27). In addition, HR1 peptides can oligomerize and bind to gp41 HR2, forming a peptide-gp41 6HB that can also trap the PHI (24, 28). In these resistance cultures, we consistently found that HIV-1 escapes inhibition by HR1 peptide fusion inhibitors through one of two genetic pathways, each defined by a key resistance mutation in HR1 or HR2 (22, 23). Curiously, each pathway was accompanied by a different pattern of gp120 mutations. In the HR1 pathway (pathway 1), mutations in gp120 tended to emerge in residues in the first and second variable regions (V1/V2) and in the vicinity of the CD4 binding site (CD4bs) (23). In the HR2 pathway (pathway 2), gp120 mutations tended to emerge in residues in the third variable region (V3), in the vicinity of the chemokine receptor binding site. These two distinct patterns of associated gp120 and gp41 mutations suggested that they were involved in functional networks controlling Env-mediated virus entry.

In this study, we characterized the functional and structural effects of combinations of Env mutations coselected in our resistance cultures to identify gp120 and gp41 residues that cooperate to regulate receptor use and Env conformational changes. Using a panel of neutralizing and conformation-specific antibodies (Abs), we found that both pathway 1 and pathway 2 Envs evolved relaxed or more-open trimer conformations in their native state, although they did so via different patterns of mutations. Notably, Envs from both pathways showed increased sensitivity to neutralization by a CD4 mimetic compared to the parental wild type, but Envs with mutations in HR1 showed reduced CD4-induced conformational changes in gp41. These studies provide new insights into the regulation of Env conformational changes and suggest ways to modify Env conformations.

RESULTS

gp120 and gp41 mutations influence sensitivity to mCD4 inhibition.

Previously, we isolated four Envs (Table 1) from HIV-1 JR_CSF cultures selected for resistance to a 44-mer peptide corresponding to gp41 HR1 (N44) that segregated into two genetic pathways, defined by either the E560K founder resistance mutation in HR1 (pathway 1) or the E648K founder resistance mutation in HR2 (pathway 2) (23). We reported that these peptide-resistant Envs (Table 1) were more sensitive to neutralization by soluble CD4 (sCD4) than the parental, wild-type (WT) Env. Here, we further investigated the contribution of individual mutations in the peptide-resistant Envs to receptor-induced inactivation of Env using the CD4 mimetic M48-U1 (mCD4) (29). As seen with sCD4 in our previous studies (22, 23), all peptide-resistant Envs showed enhanced sensitivity to mCD4 (Fig. 1A). Three of four peptide-resistant Envs (from culture 1 [C1], C2, and C4) had a 25- to 50-fold increase in sensitivity to mCD4, while C3 Env had a 6.5-fold increase in mCD4 sensitivity, likely reflecting fewer mutations in this Env (T319I and E648K) than in the other peptide-resistant Envs with multiple mutations in gp120 and gp41 (Table 1). Mutations in gp41 had only a modest effect on mCD4 sensitivity (Fig. 1B), as seen previously for sCD4 (22, 23). The E560K and Q577R HR1 mutations each increased mCD4 neutralization approximately 4-fold and 2.5-fold, respectively. In contrast, the E648K HR2 mutation conferred little effect.

TABLE 1.

Mutations in Env clones from HR1 peptide resistance cultures

Pathway	Culture (Env clone)	gp120 mutation(s)	gp41 mutation(s)a
			HR1	HR2	Other
1	C1	L125F, I165K	N554K, E560K, V580L	T641I
	C2	L125F, E429K	E560K, Q577R	T641I
2	C3	T319I		E648K
	C4	S306G, T319I, E322D	Q577R	T641I, E648K	D620N, V833I

^aFounder mutations are marked in bold.

FIG 1.

Mutations alter sensitivity to inhibition by mCD4. Shown are data for sCD4 mimetic M48-U1 inhibition of pseudoviruses bearing WT Env or culture 1 to culture 4 [C1 to C4] N44-resistant Envs (A), Envs with gp41 mutations (B), Envs with pathway 1 gp120 mutations (C), and Envs with pathway 2 gp120 mutations (D). Data points are the averages of results from at least two independent experiments, each conducted in duplicate. Error bars show standard errors of the means (SEM).

Single gp120 mutations also revealed differences between the pathways (Fig. 1C and D). In Envs from pathway 1, L125F and I165K mutations in V1/V2 increased mCD4 sensitivity by ∼1.5 log units, but the E429K mutation had the opposite effect, conferring approximately 2-fold resistance to mCD4 neutralization (Fig. 1C). This mutation occurred only in the presence of other mutations that dominated the phenotype with enhanced mCD4 sensitivity (Table 1 and Fig. 1A). In contrast, Envs with S306G and/or T319I mutations in V3 from pathway 2 had a more modest effect on mCD4 sensitivity, increasing sensitivity 4- and 10-fold, respectively (Fig. 1D). The E322D mutation in V3 had no effect. Together, these findings suggest that the E648K mutation may constrain the evolution of CD4 sensitivity, while Envs with the E560K mutation in HR1 are able to tolerate single gp120 mutations that confer large increases in CD4 sensitivity.

HR1 modulates mCD4-mediated virus entry.

Because the gp120 and gp41 mutations conferred different sensitivities to mCD4, we investigated the effect of these mutations on Env-mediated virus entry. The addition of sCD4 can render HIV-1 competent for fusion for a brief period, but longer incubations can irreversibly inactivate virus, presumably by triggering irreversible conformational changes in Env to fusion-incompetent conformations that may involve the formation of the gp41 6HB. We assessed the contributions of various mutations to CD4-induced conformational changes (Env reactivity) using mCD4 and CCR5⁺ target cells that lacked CD4 (Fig. 2). Following spinoculation onto target cells, pseudoviruses bearing WT Env treated with mCD4 resulted in infectivity only at relatively high concentrations of mCD4 (WT at time zero [T₀]) (Fig. 2A), with half-maximal infectivity at approximately 30 nM mCD4. In contrast, pretreatment with mCD4 for 60 min neutralized infectivity at all mCD4 concentrations (WT Pre) (Fig. 2A). Shortening the mCD4 preincubation time to as little as 15 min still resulted in complete inactivation (data not shown), demonstrating that mCD4 rapidly inactivates pseudoviruses with WT Env.

FIG 2.

mCD4-mediated entry of pseudoviruses into CD4⁻ CCR5⁺ cells. Shown are data for the infectivity of pseudoviruses bearing WT (A) and C1 to C4 (B) Envs spinoculated onto CD4⁻ CCR5⁺ 293T target cells under the conditions of mCD4 added following spinoculation (T0) and mCD4 preincubated with pseudoviruses for 1 h at 37°C prior to spinoculation (Pre). Infectivity was normalized to the maximum fusion observed for each pseudovirus. Data points are the averages of results from at least two independent experiments, each conducted in duplicate. Error bars show SEM.

Envs from all peptide-resistant cultures (C1 to C4) mediated virus entry in CD4⁻ CCR5⁺ target cells at significantly lower concentrations of mCD4 than for the WT (T₀) (Fig. 2B), achieving half-maximal infectivity at between 0.75 and 2 nM mCD4. Envs from pathway 1 reacted similarly to Envs from pathway 2. Remarkably, pseudoviruses with C1, C2, and C4 Envs maintained significant infectivity following mCD4 preincubation (Pre) (Fig. 2B). Pseudoviruses bearing C1 and C2 Envs retained approximately 50% infectivity following 60-min preincubations with up to 1,000 nM mCD4. The pathway 2 Env responses were variable. Pseudoviruses with C4 Env retained ∼25% infectivity after 60 min, while pseudoviruses with C3 Env were neutralized similarly to WT Env. C3 Env has only two mutations, which are shared with C4 Env, indicating that additional mutations in C4 Env contributed to the maintenance of fusion competence. Because pathway 1 and 2 Envs have little overlap in mutations, these results suggest that either different combinations of gp120 and gp41 mutations or a limited set of overlapping mutations preserved fusion competence after prolonged pretreatment with mCD4.

To investigate residues responsible for the different responses to mCD4, we first studied mCD4 pretreatments using chimeric Envs with mutations from the C1 or C4 Envs in either gp120 or gp41, paired with WT sequences in gp41 or gp120, respectively (Fig. 3A and B). Pseudoviruses with the gp120 mutations from C1 Env infected target cells with low levels of mCD4 (50% fusion at ∼0.6 nM mCD4) but lost infectivity after mCD4 preincubation (Fig. 3A, blue solid and dashed curves, respectively). In contrast, pseudoviruses with gp41 mutations from C1 Env infected target cells only with high mCD4 levels, similar to the WT (50% fusion at ∼30 nM mCD4), but retained infectivity after 60-min mCD4 preincubations (Fig. 3A, red solid and dashed curves, respectively). Chimeric Envs with mutations from the C4 Env (Fig. 3B) also showed that gp41 mutations were responsible for the preservation of fusion competence after prolonged treatment with mCD4.

FIG 3.

HR1 mutations reduce inactivation of pseudoviruses pretreated with mCD4. Shown are data for the infectivity of pseudoviruses bearing Envs with C1 (A), C4 (B), gp120 (C), and gp41 (D) mutations spinoculated onto CD4⁻ CCR5⁺ cells under the conditions described in the legend of Fig. 2, with mCD4 added following spinoculation (T0) and mCD4 preincubated with pseudoviruses for 1 h at 37°C prior to spinoculation (Pre). Data points are the averages of results from at least two independent experiments, each conducted in duplicate. Error bars show SEM.

A subset of single gp120 mutations from both pathways conferred infectivity with lower concentrations of mCD4 (Fig. 3C). Mutations L125F and I165K from pathway 1 Envs and T319I from pathway 2 Envs accounted for most of the increased infectivity using low concentrations of mCD4. The S306G mutation had a more modest effect, lowering the concentration of 50% infectivity to ∼10 nM mCD4, while the E322D and E429K mutations failed to lower the mCD4 requirements. Single gp41 mutations revealed a different pattern (Fig. 3D). The E560K mutation allowed 50% infectivity at a modest mCD4 concentration (∼10 nM mCD4) and ∼25% infectivity following mCD4 preincubation. The Q577R mutation also allowed infectivity at lower mCD4 concentrations and ∼50% retention of infectivity after mCD4 pretreatment. In contrast, the E648K mutation did not lower the mCD4 concentrations needed for infectivity or confer retention of infectivity following mCD4 preincubation. Similarly, T641I behaved like WT Env. These results show that HR1 residues 560 and 577 are responsible for the retention of fusion activity following mCD4 exposure and therefore regulate CD4-mediated triggering (Env reactivity) of Env to a fusion-incompetent conformation.

gp41 HR1 resistance mutations reduce mCD4-induced refolding of gp41 to the 6HB.

To investigate whether Env inactivation by mCD4 was due to a refolding of gp41 to the 6HB, we assessed the binding of a 6HB-specific monoclonal antibody (mAb) (3-17) to mCD4-treated Envs. This mAb, raised in mice against a 6HB recombinant protein (see Materials and Methods), exhibits an ∼900-fold binding preference for the 6HB over individual HR1 or HR2 peptides (Table 2). The level of immunoprecipitation of pseudoviruses bearing WT Env with the 3-17 mAb was low (5.5% ± 2.5% of the input) in the absence of mCD4 but increased nearly 7-fold (37% ± 1.2% of input) after mCD4 treatment (Fig. 4). In contrast, the level of immunoprecipitation of pseudoviruses bearing C1 and C2 Envs was barely above the background (<5%) after mCD4 treatment (Fig. 4A and C). Pseudoviruses with Envs containing E648K were more similar to WT pseudoviruses. Pseudoviruses with C3 Env showed a 5-fold increase of 6HB formation (Fig. 4C), while pseudoviruses with C4 Env showed significant immunoprecipitation of gp41 prior to mCD4 treatment (28%) and an ∼2-fold increase after treatment. The latter finding is consistent with C4 Env conferring intermediate sensitivity to mCD4 inactivation (Fig. 2B) and suggests that the E648K mutation in HR2 may interfere with the ability of the Q577R mutation to reduce the conformational reactivity of gp41. Finally, we confirmed these results with a polyclonal rabbit serum (948) that was previously shown to preferentially bind CD4-activated Env (30) (Table 2 and Fig. 4A and B) to rule out the possibility that C1 to C4 mutations specifically affected the 3-17 mAb epitope. We conclude that the resistance mutations in HR1 make Env less prone to CD4-induced triggering to the 6HB.

TABLE 2.

ELISA binding of antibodies to gp41 peptides

Antibody or serum	ELISA binding titera
	IZN36 + C34 (6HB)	IZN36	C34
3-17	0.027 μg/ml	24.1 μg/ml	UD
948	1:30,300	1:1,180	1:4,680

^aELISA binding for 3-17 is expressed as the Ab concentration at which half-maximal binding occurred (EC₅₀). ELISA binding for serum 948 is expressed as the serum dilution at which half-maximal binding occurred (EC₅₀). UD, binding undetectable at any tested Ab concentration.

FIG 4.

Detection of gp41 6HB after mCD4 treatment. (A) Immunoprecipitation (IP) of pseudoviruses bearing WT and C1 to C4 Envs in the presence (+) or absence (−) of mCD4 with the anti-6HB monoclonal antibody (3-17) (top) or polyclonal rabbit serum (948) (bottom) and analyzed by reducing SDS-PAGE followed by Western blotting (WB) with a gp41 monoclonal antibody (2F5). (B) One-fifth of the input used in immunoprecipitations analyzed in the same Western blot. (C) Quantitation of fold changes in gp41 expression from immunoprecipitation with MAb 3-17. Bars show averages of data from two independent immunoprecipitation experiment with SEM.

Peptide-resistant Envs reveal distinct genetic pathways that confer relaxed conformations of native Env.

The changes in mCD4 sensitivity and conformational reactivity conferred by mutations in the peptide-resistant Envs led us to further investigate how the mutations altered Env structure and neutralization. We tested the neutralization sensitivity of our peptide-resistant Envs, as well as Envs with individual mutations, against a large panel of mAbs to assess the exposure of specific epitopes (Fig. 5 and Table 2). For the 17b mAb that targets a CD4-induced (CD4i) epitope, WT and C3 Envs were resistant to neutralization (Fig. 5A), but both C1 and C2 Envs had 50% inhibitory concentrations (IC₅₀s) of 0.2 μg/ml, and C4 Env was moderately sensitive, with an IC₅₀ of 2.9 μg/ml. Single gp120 mutations conferred various effects. Envs with L125F or I165K mutations displayed significant sensitivity to the 17b mAb, with IC₅₀s of 0.6 μg/ml and 3.2 μg/ml, respectively. In contrast, Env with the E429K (C2) mutation and Envs with T319I (C3), S306G, and E322D (C4) mutations remained resistant (Fig. 5B). Only when T319I, S306G, and E322D were found in combination (Env C4) did Envs show moderate sensitivity to the 17b mAb (Fig. 5A and B). The E560K mutation in HR1 conferred a modest increase in sensitivity, with an IC₅₀ of 13 μg/ml (Fig. 5C), but other gp41 mutations had no effect. The CD4i 412d mAb showed a pattern similar to that of the 17b mAb (Table 3). Another CD4i mAb, 447-52D, which targets a distinct CD4i epitope in the V3 loop (31, 32), neutralized all peptide-resistant Envs (Fig. 5D). Unlike mAb 17b, g120 mutations S306G and T319I (pathway 2), in addition to L125F and I165K, conferred sensitivity to the 447-52D mAb (Fig. 5E). These residues lie within the base of the V3 loop recognized by 447-52D and individually likely affect the V3 structure locally, without globally inducing a more open CD4i-type conformation. gp41 mutations E560K and Q577R showed a modest increase in 447-52D sensitivity (Fig. 5F). Overall, exposure of CD4i epitopes was often associated with sensitivity to mCD4. Pathway 1 Envs were more sensitive to 17b and 412d, and sensitivity could be largely attributed to specific changes (L125F and I165K) in gp120. Pathway 2 Envs were less sensitive to 17b and 412d, and sensitivity appears to be the result of an accumulation of changes rather than the result of specific changes, as no single pathway 2 mutation had an effect on 17b neutralization.

FIG 5.

Effect of mutations on neutralization by various monoclonal antibodies. Shown are data for neutralization of pseudoviruses bearing Envs with the indicated mutations by 17b (A to C), 447-52D (D to F), PGT151 (G and H), and PG9 (I). Data points are the averages of results from at least two independent experiments, each conducted in duplicate. Error bars show SEM.

TABLE 3.

Neutralization sensitivity of HIV-1 Env pseudovirus^a

^aIC₅₀ values for each MAb are presented as micrograms per milliliter. Red, IC₅₀ of <0.15 μg/ml; orange, 0.15 μg/ml ≤ IC₅₀ < 2 μg/ml; yellow, 2 μg/ml ≤ IC₅₀ < 15 μg/ml; green, IC₅₀ of ≥15 μg/ml. nd, not determined.

Next, we investigated whether the mutations conferred changes in sensitivity to broadly neutralizing antibodies (bnAbs) representing each of the five known neutralizing sites in Env (Fig. 5 and Table 3). In contrast to mCD4, mutations in the peptide-resistant Envs did not affect neutralization of the CD4 binding site (CD4bs) VRC01 bnAb, consistent with its ability to bind multiple trimer conformations and consequently neutralize a broad range of viruses. The similar potencies of VRC01 bnAb against the pseudoviruses in this panel also indicate that there are not large variations in mature Env incorporation in the pseudoviruses. However, the peptide-resistant Envs from both pathways showed 3- to 10-fold increases in sensitivity to the less potent and conformationally dependent CD4bs b12 mAb. HR1 mutation E560K and multiple gp120 mutations (L125F, I165K, and T319I) contributed to neutralization sensitivity, indicating that these residues affect the accessibility of the CD4bs. Using the gp120/gp41 interface PGT151 bnAb (Fig. 5G and H), we found that both C3 and C4 Envs (pathway 2) exhibited remarkable resistance to PGT151, while C1 and C2 Envs retained WT levels of sensitivity (Fig. 5G). Single mutations demonstrated that PGT151 resistance mapped to the E648K resistance mutation in HR2 (Fig. 5H). This finding suggests that E648K leads to structural changes of the gp41 base in the protomer interaction site.

Compared to WT Env, all peptide-resistant Envs were less easily neutralized by the PG9 bnAb (Fig. 5I), which targets V1/V2 at the apex of the native Env trimer. C1 and C2 Envs tended to be more resistant than C3 and C4 Envs. Reduced sensitivity to neutralization by the PG9 bnAb appeared to be due to cumulative gp120 mutations rather than one key mutation (Table 3). Because the PG9 bnAb binds the native Env trimer in a closed conformation, our findings indicate that many different gp120 mutations can relax the native Env conformation so that it can dynamically sample conformations that are more open.

Most of the peptide-resistant Envs showed moderate increases in MPER antibody sensitivity, with the exception of C3. Pathway 1 gp120 mutation L125F showed a modest increase in 2F5 sensitivity, and both mutations I165K and T319I showed ∼2-fold increases in sensitivity to 10E8. Single mutations alone were unable to account for the increase in MPER antibody sensitivity seen in C1, C2, and C4 Envs. Single mutations in gp41 resulted in an even more modest effect on MPER antibody sensitivity; the E560K and Q577R mutations in HR1 slightly increased sensitivity to the 2F5 mAb compared to the WT, while the E648K mutation in HR2 resulted in modest resistance to neutralization by the 2F5 mAb. Residue 648 is close to the MPER but has not been reported to be part of the 2F5 epitope. MPER mAbs 4E10 and 10E8 generally showed patterns of changes in sensitivity similar to that of 2F5, despite different baseline neutralization levels (Table 3).

Taken together, these antibody neutralization results indicate that the peptide-resistant Envs evolved relaxed Env conformations but did so via different genetic pathways. Envs from both pathways have trimers that were more sensitive to mCD4 neutralization. For pathway 1 Envs, key changes in gp120, including L125F or I165K, along with the E560K mutation in HR1, conferred substantial mCD4 sensitivity. For pathway 2 Envs, mCD4 sensitivity was associated with an accumulation of changes in V3 that appeared to have an additive effect on mCD4 sensitivity. Envs from both pathways also had exposed CD4i epitopes, demonstrated by 17b, 412D, and 447-52D mAb neutralization. Finally, compared to WT Env, Envs from both resistance pathways were less sensitive to the PG9 mAb, which targets V1V2 at the trimer apex. However, pathway 1 Envs were more resistant to the PG9 mAb than pathway 2 Envs. These distinct patterns of gp120 mutations associated with HR1 and HR2 resistance mutations suggest long-range functional interactions between these gp41 HR1 and HR2 mutations and specific regions in gp120. In particular, pathway 2 Envs with the founder E648K resistance mutation in HR2 appear to disfavor V1/V2 mutations that have large effects on mCD4 sensitivity.

DISCUSSION

We describe two patterns of functional interactions between gp120 and gp41 residues in a panel of Envs resistant to peptide fusion inhibitors: one involves the gp41 E560K resistance mutation in HR1 along with mutations in V1/V2 and the hydrophobic core of gp120, and the other involves the gp41 E648K resistance mutation in HR2 along with mutations in V3. While both patterns of gp120 mutations conferred enhanced sensitivity to mCD4 inhibition and relaxed gp120 conformations, the HR1 and HR2 mutations conferred different degrees of CD4-induced conformational reactivity in Env. The E648K HR2 mutation maintained wild-type sensitivity to mCD4 inhibition and mCD4-induced triggering to the 6HB, but the E560K and Q577R HR1 mutations each increased sensitivity to mCD4 inhibition and reduced mCD4-induced triggering to the 6HB (Fig. 2 and 4). These distinct pathways provide new insights into different ways in which the Env trimer can sample more open conformations to facilitate interactions with CD4 while regulating receptor-induced conformational changes.

The finding of reduced gp41 conformational reactivity to CD4 conferred by the HR1 mutations initially seemed at odds with our previous findings that the E560K and Q577R HR1 resistance mutations stabilize the 6HB (Fig. 6B) (22, 23, 33). This paradox may be reconciled if the HR1 resistance mutations also affect earlier conformational states of Env. Several recent high-resolution structures reported to represent native-like HIV-1 Env trimers provide clues to potential stabilizing interactions involving the HR1 gp41 resistance mutations. Although most early structures derived from BG505 SOSIP soluble trimers lack resolution of portions of gp41, including HR1 and HR2 (9, 18), residue 577 in the “e” position of the central HR1 coiled coil is resolved in most structures (Fig. 6A, Q577 in red). In the postfusion 6HB structure, residues in the “e” position of the heptad interact with the HR2 helix (Fig. 6B) (34, 35). However, in the prefusion trimer (e.g., PDB accession numbers 5CEZ and 5FUU), Q577 points away from the trimer axis, where it can potentially participate in polar interactions with HR1 residues 552 to 569 in an adjacent protomer (36, 37). These residues reside in a flexible loop in a few trimer structures (e.g., PDB accession number 5CEZ) but are unresolved in others (e.g., PDB accession numbers 4TVP and 4NCO). This poorly resolved region contains the key pathway 1 HR1 E560K mutation as well as Gln at position 567 in the JR_CSF strain that is within H-bond range of residue 577 from the adjacent gp41 protomer in the BG505 SOSIP structure (PDB accession number 5CEZ) (Fig. 6C). Thus, the Q577R mutation may allow the formation of an additional interaction(s) between adjacent gp41 protomers that may favor the native conformation of gp41. Interestingly, the Q577R mutation also emerged in resistance to HR2 peptide fusion inhibitors (38).

FIG 6.

Model of gp41 and gp120 interactions. (A) gp120-gp41 monomer from the BG505 SOSIP trimer structure (PDB accession number 5CEZ). gp120 is shown in orange, gp41 HR1 is in light green, and gp41 HR2 is in blue. Key gp41 positions E560, Q577, and E648 are shown in magenta, red, and bright green, respectively. (B) gp41 monomer from the postfusion 6HB structure (PDB accession number 1IF3), colored as described above for panel A. (C) Closeup of the N-HR1 region from the SOSIP trimer structure under PDB accession number 5CEZ. One protomer of gp41 is shown in blue, and the others are in gray; gp120 is in beige. Residue Q577 is highlighted in red, and the potential interaction with Q567 from an adjacent gp41 monomer is indicated. (D) JR_FL trimer structure (PDB accession number 5FUU), stabilized by two PGT151 Fabs. A single gp120/gp41 protomer is colored as described above for panel A; the remaining gp120/gp41 protomers are shown in beige and gray. PGT151 Fabs are in black. (E) Closeup of the N-HR1 region of gp41 from the structure reported under PDB accession number 5FUU. Each gp41 protomer is colored separately; the unliganded protomer is shown in blue, and the PGT151-liganded protomers are shown in white and charcoal. (Top) Side view of the unliganded (blue) protomer showing ladder-like packing between the N-HR1 and C-HR1 regions. (Bottom) Side view of a PGT151 liganded protomer showing extension of N-HR1 up and away from the C-HR1 coiled coil. (F) Model of key V1/V2 and V3 gp120 mutations in trimer structure. One protomer of gp120 (PDB accession number 4TVP) is shown as a ribbon diagram with colored domains: V1/V2 is in green, the V3 loop is in pink, and the remainder is in gray. The other two gp120 protomers are shown in beige space-fill. Key V1/V2 residues L125 and I165 are highlighted in red. Key V3 residues S306 and T319 are highlighted in yellow.

An alternative view of the N-terminal half of HR1 comes from the JR_FL Env structure (PDB accession number 5FUU) that shows the trimer binding to two PGT151 mAbs, resulting in an asymmetric trimer (Fig. 6D) (36). Each gp41 protomer reveals a different position and conformation of the N-terminal portion of HR1 (Fig. 6E). In Fig. 6E, the unliganded protomer is highlighted in blue, and both protomers bound to mAb PGT151 are shades of gray. While the C-terminal HR1 coiled coil that forms the basis of the Env trimer axis is relatively unperturbed, there is asymmetry of the surrounding regions of gp41. In the unliganded protomer of gp41, N-terminal HR1 residues 547 to 566 form a curved α-helix that packs tightly into the groove formed by the C-terminal HR1 trimeric coiled coil (residues 569 to 596). This groove is lined by the residues occupying “g” and “c” heptad positions of the same protomer and the “b” and “e” heptad positions from its clockwise adjacent protomer. This positions the N-terminal HR1 residues H564, E560, R557, N554, and Q550 in range to directly interact with the C-terminal HR1 coiled coil of the same and adjacent protomers (Fig. 6E, top). Both the E560K and Q577R mutations conferred partial resistance to mCD4-mediated irreversible inactivation (Fig. 3D), indicating that these residues play a key role in Env conformational reactivity to CD4. Modeling suggests that E560K could allow an additional H-bond to the backbone of residue 577 in the adjacent gp41 protomer. Together, the E560K and Q577R mutations could allow multiple additional H-bonds between the N- and C-terminal halves of HR1, between E560K and Q577R and between Q577R and the backbones of Q563 and Q567 in adjacent protomers (Fig. 6E, top), potentially stabilizing gp41 in this tightly packed HR1 conformation.

The other gp41 protomers that are bound to PGT151 display the N-terminal HR1 residues as a straight α-helix, tilted away from the trimer axis and toward the same protomer's gp120 domain (Fig. 6E, bottom) (36). In this conformation, not only is the position of the fusion peptide more exposed, but the stabilizing interactions between the N- and C-terminal HR1 helices described above are not possible. This alternate conformation requiring stabilization by the PGT151 bnAb may represent an intermediate on the pathway toward HR1 extension and formation of the extended gp41 prehairpin intermediate. Further characterization of the native conformation of the Env trimer, particularly gp41, will be necessary to properly order early conformational changes (39). Altogether, our modeling suggests that the decrease in mCD4-mediated triggering of Env to the 6HB conformation in the peptide-resistant Envs may be due to stabilization of the native gp41 conformation via intra- and interprotomer interactions by HR1.

An alternative or additional explanation for reduced mCD4-induced triggering to the 6HB could be that the mutations result in less-efficient contacts between gp120 and gp41 that fail to efficiently transduce receptor-induced conformational changes in the gp120 subunit to the gp41 subunit. The ambiguities in the various gp41 HR1 structures make determining the precise effects of mutations E560K and Q577R difficult at this time. Nonetheless, our functional data indicate that in addition to stabilizing the gp41 postfusion 6HB, these HR1 mutations also reduce native trimer reactivity to CD4-induced Env conformational changes. The finding that the E648K HR2 resistance mutation, which also increases the stability of the 6HB, does not change in CD4-induced conformational reactivity (Fig. 3D) confirms the importance of the HR1-mediated interactions in regulating Env conformational reactivity. We note that residue 560 is immediately adjacent to residue 559 that was previously identified as contributing to the stabilization of the engineered SOSIP trimer. Unlike SOSIP mutation I559P (39), the E560K and Q577R mutations maintain Env fusion activity and function, which allowed us to directly test their effects on Env reactivity to CD4. Altogether, these results highlight the important role that the regulation of the interactions between the N-terminal and C-terminal halves of HR1 has on the conformational triggering of gp41.

In addition to the gp41 mutations, each fusion inhibitor resistance pathway was associated with a distinct set of gp120 changes (Table 1) involving mutations L125F, I165K, and E429K in V1/V2 and the gp120 core (pathway 1) or mutations S306G, T319I, and E322D in V3 (pathway 2). gp120 mutations in both pathways were responsible, either alone or in combination, for increasing sensitivity to many neutralizing antibodies and mCD4. We modeled these residues in the closed, native-like Env trimer structures (Fig. 6F) (PDB accession number 5CEZ) (37) and a recent structure of the open, liganded trimer bound to sCD4 and the 17b mAb (PDB accession number 5VN3) (40). Despite the coevolution of mutations in gp120 and gp41 that affected mCD4 sensitivity and gp41 conformational reactivity, we note that in neither pathway were changes observed in topographical layers 1 to 3 of the gp120 inner domain, previously identified as being key determinants of regulating gp120-gp41 interactions and reactivity to CD4 and the coreceptor (41). Instead, the changes occurred entirely in regions of gp120 that appear to be important for receptor/coreceptor interactions and the maintenance of a closed trimer structure.

Pathway 1 mutations L125F and I165K are located in the V1/V2 region of native gp120 (Fig. 6F) (PDB accession number 4TVP) (18). Residue I165 is in close proximity to the interface between the adjacent gp120 protomer in the native trimer (37). In the liganded structure, the interfaces between gp120 protomers are broken as each gp120 structure opens and extends outwards. Residues L125 and E429 wind up in or lining the binding surface of CD4, while I165 falls into an extended unresolved portion of the V1/V2 loop (40). We found that L125F and I165K mutations strongly influenced sensitivity to mCD4 and CD4bs antibody neutralization. Their position in the structure suggests that they contribute to V1/V2-V3 and gp120 interprotomer interactions, with the mutations allowing the premature opening of the Env trimer apex, confirmed by the loss of sensitivity to neutralization by the PG9 mAb (Fig. 5I and Table 3). Mutation L125F was previously reported to be an important determinant of intraprotomer interactions between V1/V2 and V3 in strain JR_FL, allowing the exposure of both V2 and V3 neutralizing epitopes (42). Those studies also identified residue I184 as a potential interprotomer V1/V2 interaction partner with residue 165, consistent with our work identifying I165 as a key determinant of trimer opening in JR_CSF Env. In contrast, mutation of neither L125 nor I184 resulted in broad exposure of CD4i epitopes in JR_FL Env (42, 43), but L125F and I165K result in CD4i epitope exposure in the context of JR_CSF Env. These results suggest that these related strains respond to V1/V2 alterations differently, emphasizing the complexity and potential strain dependence of the coordinated interactions that stabilize the native trimer.

Pathway 2 S306, T319I, and E322D mutations work differently than the pathway 1 mutations. In isolation, these changes had only modest effects on neutralization by most mAbs and a CD4 mimetic (Fig. 5 and Table 3). The exception is the large increase in sensitivity to the 447-52D mAb conferred by the S306G and T319I mutations (Fig. 5E). S306 and T319 are at either side of the V3 base recognized by 447-52D (31) (Fig. 6F, yellow residues), suggesting that these mutations result in a limited exposure/relaxation of V3 alone that may enhance exposure to the coreceptor binding site. This possibility is supported by the position of these residues in the liganded trimer, where S306 is located directly in the 17b binding site, while T319 and E322 are within the unresolved portion of the V3 loop (40). These conclusions are consistent with previous work showing that it is possible to trigger V3 loop exposure independently of V1/V2 exposure and that V1/V2 likely acts as a clamp restraining the unliganded trimer in the closed state (42–49).

None of the gp120 mutations that we describe altered irreversible inactivation by mCD4 treatment (Fig. 3C). Rather, the gp120 mutations lowered the concentration of mCD4 required for mediating fusion without rendering them CD4 independent, consistent with their contribution to a more open or relaxed gp120 trimer conformation. Pathway 1 mutations L125F and I165K and pathway 2 mutation T319I had the greatest effect (Fig. 3C) and correlated well with sensitivity to mCD4 neutralization (Fig. 1C and D). This finding suggests that the increase in neutralization sensitivity is largely due to increased binding and affinity for CD4 rather than irreversible inactivation by conformational triggering of gp41.

The metastable Env trimer has been reported to sample distinct conformational states: state 1 (closed, native), state 2 (relaxed, intermediate), and state 3 (open) (47, 50–52). Primary HIV-1 Envs predominantly occupy state 1 until CD4 engagement, which shifts the trimer toward states 2 and 3 and triggers the cascade of conformational changes leading to fusion. Our findings highlight ways in which Env can alter coupling between gp120 and gp41 conformational changes. The L125F and I165K mutations in V1/V2 and the S306G and T319I mutations together in V3 conferred relaxed trimer conformations that appear to correspond to shifts from state 1 to state 2 or 3. Consistent with this suggestion, these changes also reduce the concentration of CD4 required to mediate fusion.

Our antibody studies indicate that pathway 1 and 2 Envs exist in nonidentical conformations. This suggests that either (i) pathway 1 and 2 Envs are in different states or (ii) each state may contain multiple quasiequivalent conformations sampled by Env, rather than a single, defined, intermediate conformation along a pathway from native “closed” to liganded “open” trimers. The HR1 E560K and Q577R mutations reduce CD4-induced triggering of gp41 and may compensate for the gp120 mutations that facilitate CD4 binding. The coevolution of HR1 mutations that reduce conformational triggering with gp120 mutations that open gp120 suggests that the virus balances efficient receptor use and Env reactivity through coordinated changes in distant domains of both gp120 and gp41, a phenomenon that has been observed previously (49, 53, 54). This observation highlights potential flexibility in gp120 and gp41 interactions that may argue for an expansion of the 3-state model to allow for a greater degree of movement between gp120 and gp41 subunits as well as between trimer protomers as observed by fluorescence resonance energy transfer (FRET) (55).

In summary, our analysis of genetic pathways of HR1 peptide fusion inhibitor resistance provides new insights into Env conformational changes that mediate virus entry. HR1 E560K and Q577K mutations confer not only resistance to fusion inhibitors but also significant resistance to CD4-mediated inactivation via reduced conformational triggering of gp41 to the 6HB. Interprotomer gp41-gp41 interactions in HR1 may contribute to this reduced conformational reactivity. Correspondingly, Envs with the E560K mutation tolerate gp120 mutations that relax the native trimer through the release of V1/V2, which facilitates binding to CD4. In contrast, Envs with the E648K mutation have only modest resistance to CD4-mediated conformational triggering to the 6-HB. Because HR2 resistance mutations do not significantly alter Env reactivity to CD4, gp120 mutations that have a more localized effect on V3 and coreceptor binding site accessibility appear to be favored over individual mutations that dramatically shift CD4bs accessibility and opening of the crown of the Env trimer. These results provide new insights into the regulation of Env conformational changes and suggest potential ways to alter the coupling of trimer opening and receptor binding with gp41 conformational changes. Residues 560 and 577, identified here as having a novel role in stabilizing a prefusion conformation of gp41, may also prove useful for engineering Env immunogens that resist conformational flipping to postfusion Env conformations.

MATERIALS AND METHODS

Cell culture.

293T and U87 cells expressing CD4 and CCR5 (U87-CD4-R5) were provided by Dan Littman (New York University, NY). 293 cells were passaged in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS), penicillin, streptomycin, l-glutamine, 1 mM HEPES, and nonessential amino acids (DMEM⁺). U87-CD4-R5 cells were passaged in DMEM⁺ with the addition of 300 μg/ml G418 sulfate and 1 μg/ml puromycin.

Plasmids and constructs.

Env expression vector pCMV/R, Env-deficient HIV genome plasmid pCMVΔ8.2, and luciferase reporter plasmid pHR′-Luc were kindly provided by Gary Nabel (National Institutes of Health, Bethesda, MD). The wild-type JR_CSF Env expression plasmid (pVRC JR_CSF WT) was created as described previously (22, 23, 33), by inserting the Env gene into pCMV/R via NotI and EcoRV restriction sites. Expression plasmids for N44-resistant culture Envs (C1 to C4) were cloned in the same manner. Expression plasmids for gp120 and gp41 single mutants were generated from pVRC JR_CSF WT with the QuikChange II site-directed mutagenesis kit (Agilent Technologies, Santa Clara, CA). Site-directed mutagenesis and the lack of secondary mutations were confirmed by sequencing the full-length Env gene from recovered clones.

Pseudovirus generation.

Approximately 3 × 10⁶ 293T cells were seeded onto a 10-cm dish 18 h prior to transfection. Transfection mixtures were comprised of 0.25 μg pVRC Env and 3.0 μg each of pCMVΔ8.2 and pHR′CMV-Luc, and optionally 1 μg furin expression plasmid, combined with Fugene 6 (Promega, Madison, WI) according to the manufacturer's protocols. Fresh medium was exchanged at 18 to 24 h posttransfection. Pseudoviruses were harvested from culture supernatants at 48 h posttransfection, filtered through a 0.45-μm-pore-size low-binding filter, and immediately stored at −80°C. Pseudovirus lysates were checked by Western blotting using gp41 and p24 antibodies to ensure that levels of Env incorporation into the pseudoviruses were not more than 4-fold different from those for pseudoviruses with WT Env.

Infectivity and neutralization assays.

U87-CD4-R5 cells were seeded onto white 96-well plates at a concentration of 1.7 × 10⁵ cells/ml, 100 μl per well, 24 h prior to infection. Serial 2-fold dilutions of pseudoviruses in medium plus 8 μg/ml Polybrene were plated with positive (WT JR_CSF) and negative (mock) controls and incubated at 37°C. At 48 h postinfection, cells were lysed with cell culture lysis reagent, and luciferase activity was assayed with luciferase assay system kit reagents (Promega, Madison, WI) according to the manufacturer's instructions.

For neutralization assays, a standard infectious dose (100,000 relative light units [RLU]) of each pseudovirus was mixed with 2-fold serial dilutions of mAbs or 4-fold serial dilutions of sCD4/mimetic CD4 M48-U1, incubated at 37°C for 1 h, before inoculating target cells and scoring for infectivity as described above. Neutralization dose-response curves were calculated by nonlinear regression using Prism 6 (GraphPad Software Inc., La Jolla, CA). IC₅₀s represent the averages of data from at least two independent experiments, conducted in duplicate. mAbs VRC01 (56), PG9 (57), PGT151 (58), 2F5 (59), 4E10 (59), 17b (60), 412d (61), and 447-52D (62) were kind gifts of Peter Kwong (NIH VRC, Bethesda, MD). mAbs b12 (63) and 10E8 (64) were obtained from the NIH AIDS reagent program. CD4 mimetic M48-U1 was the kind gift of Loïc Martin (Commissariat a l'Energie Atomique, Gif-sur-Yvette, France).

mCD4-mediated entry.

293T cells were transfected with the pcCCR5 expression plasmid (293T-transR5) (3, 65) and seeded onto 96-well plates. Pseudoviruses were spinoculated onto 293T-transR5 cells at 2,000 × g for 1 h at 37°C before the addition of medium containing serial dilutions of mCD4 (T₀). Plates were incubated for 48 h at 37°C and then lysed and assayed for luciferase activity as described above. To determine the degree of CD4-mediated inactivation, pseudoviruses were preincubated for 1 h at 37°C with serial dilutions of mCD4 prior to spinoculation onto 293T-transR5 cells (Pre), before measuring infectivity. Infectivity versus the mCD4 concentration was plotted for each treatment and normalized against the maximum infectivity seen in the T₀ treatment for each pseudovirus.

Isolation and characterization of 6HB mAb.

A gp41 recombinant protein, JR_CSF N34-C28 (amino acid sequence SGIVQQQNNLLRAIEAQQHMLQLTVWGIKQLQARSGGRGGWMEWEKEIENYTNTIYTLIEESQIQQEK), designed to mimic the 6HB by linking 34 residues in HR1 to 28 residues in HR2 with a flexible linker (SGGRGG), was produced in Escherichia coli (GenScript, NJ). JR_CSF N34-C28 was used as an immunogen in BALB/c mice to raise a 6HB-specific mAb (Envigo, Madison, WI). Hybridoma supernatants were screened by an enzyme-linked immunosorbent assay (ELISA) for binding to the immunogen, an HR1 peptide stabilized as a trimeric coiled coil (IZN36 peptide; amino acid sequence IKKEIEAIKKEQEAIKKKIEAIEKEISGIVQQQNNLLRAIEAQQHLLQLTVWGIKQLQARIL), an HR2 peptide (C34 peptide; amino acid sequence WMEWEKEIENYTNTIYTLIEESQIQQEKNEQELL), and a mixture of the IZN36 and C34 peptides preincubated in solution for 30 min at 37°C to allow the formation of the 6HB. MAb 3-17 was selected for binding with high affinity in ELISAs to the immunizing antigen and a mixture of IZN36 and C34 peptides but only weakly to IZN36 or C34 peptides alone.

ELISAs were carried out on Immulon 2HB plates coated overnight with gp41 peptides, IZN36, C34, or a mixture of IZN36 and C34. Serial 5-fold dilutions of MAb 3-17 or rabbit serum 948 were applied to ELISA plates after blocking in 5% nonfat dry milk powder solution (NFDM) and incubated for 1 h at 37°C. After primary incubation, ELISA plates were washed four times in TBST (150 mM NaCl, 20 mM Tris [pH 8.0], 0.05% Tween 20) wash buffer. Secondary anti-mouse–horseradish peroxide (HRP) (for 3-17) or anti-rabbit–HRP (for serum 948) Ab was added at a 1:5,000 dilution in blocking buffer, and the mixture was incubated for 30 min at room temperature (RT). Plates were washed four times after secondary Ab incubation and then developed with the SureBlue peroxidase substrate (SeraCare). Reactions were stopped by the addition of 1.0 M sulfuric acid after 15 to 30 min, and the optical density (OD) at 450 nm was recorded on a Opsys MR microplate reader (Dynex). ELISA binding curves and 50% effective concentration (EC₅₀) values were calculated by nonlinear regression using Prism 6 (GraphPad Software). Reported values include average values of data from at least two independent experiments, performed in duplicate.

Immunoprecipitation and Western blotting.

Pseudoviruses were pretreated with or without 1,000 nM mCD4 prior to incubation with 50 μg MAb 3-17 or isotype control antibody for 60 min at 37°C. Pseudoviruses were then pelleted at 20,000 × g for 2 h at 4°C and resuspended overnight in NP-40 lysis buffer (NP-40 LB) (50 mM Tris [pH 8.0], 150 mM NaCl, 1% NP-40) at 4°C. Following resuspension, the viral lysate was incubated with protein G Dynabeads (Thermo Fisher) and equilibrated in NP-40 LB for 30 min at room temperature, prior to washing four times in NP-40 LB and resuspension in 2× Laemmli sample buffer (Bio-Rad). Samples were analyzed by SDS-PAGE under nonreducing conditions and transferred to a 0.2-μm nitrocellulose membrane. The blot was probed with anti-gp41 MAb 2F5 (1 μg/ml in TBST with 1% NFDM), washed four times, incubated with goat anti-human HRP-conjugated secondary Ab (1:5,000 in TBST), and washed an additional four times before development with the LumiGLO Reserve chemiluminescent substrate (KPL). Images were captured with a G:Box gel imaging system (Syngene), and band intensities were evaluated from raw image files by densitometry using ImageJ software.