Probing the Structure of the HIV-1 Envelope Trimer Using Aspartate Scanning Mutagenesis

In most crystal structures of the soluble ectodomain of the HIV-1 Env trimer, some residues in the fusion and N-heptad repeat regions are disordered. Whether this is true in the context of native, functional Env on the virion surface is not known. This knowledge may be useful for stabilizing Env in its prefusion conformation and will also help to improve understanding of the viral entry process. Burial of the charged residue Asp in a protein structure is highly destabilizing. We therefore used Asp scanning mutagenesis to probe the burial of apparently disordered residues in native Env and to examine the effect of mutations in these regions on Env stability and conformation as probed by antibody binding to cell surface-expressed Env, CD4-induced shedding of HIV-1 gp120, and viral infectivity studies. Mutations that prevent shedding can potentially be used to stabilize native-like Env constructs for use as vaccine immunogens.

ABSTRACT

HIV-1 envelope (Env) glycoprotein gp160 exists as a trimer of heterodimers on the viral surface. In most structures of the soluble ectodomain of trimeric HIV-1 envelope glycoprotein, the regions from 512 to 517 of the fusion peptide and from 547 to 568 of the N-heptad repeat are disordered. We used aspartate scanning mutagenesis of subtype B strain JRFL Env as an alternate method to probe residue burial in the context of cleaved, cell surface-expressed Env, as buried residues should be intolerant to substitution with Asp. The data are inconsistent with a fully disordered 547 to 568 stretch, as residues 548, 549, 550, 555, 556, 559, 562, and 566 to 569 are all sensitive to Asp substitution. In the fusion peptide region, residues 513 and 515 were also sensitive to Asp substitution, suggesting that the fusion peptide may not be fully exposed in native Env. gp41 is metastable in the context of native trimer. Introduction of Asp at residues that are exposed in the prefusion state but buried in the postfusion state is expected to destabilize the postfusion state and any intermediate states where the residue is buried. We therefore performed soluble CD4 (sCD4)-induced gp120 shedding experiments to identify Asp mutants at residues 551, 554 to 559, 561 to 567, and 569 that could prevent gp120 shedding. We also observed similar mutational effects on shedding for equivalent mutants in the context of clade C Env from isolate 4-2J.41. These substitutions can potentially be used to stabilize native-like trimer derivatives that are used as HIV-1 vaccine immunogens.

IMPORTANCE In most crystal structures of the soluble ectodomain of the HIV-1 Env trimer, some residues in the fusion and N-heptad repeat regions are disordered. Whether this is true in the context of native, functional Env on the virion surface is not known. This knowledge may be useful for stabilizing Env in its prefusion conformation and will also help to improve understanding of the viral entry process. Burial of the charged residue Asp in a protein structure is highly destabilizing. We therefore used Asp scanning mutagenesis to probe the burial of apparently disordered residues in native Env and to examine the effect of mutations in these regions on Env stability and conformation as probed by antibody binding to cell surface-expressed Env, CD4-induced shedding of HIV-1 gp120, and viral infectivity studies. Mutations that prevent shedding can potentially be used to stabilize native-like Env constructs for use as vaccine immunogens.

INTRODUCTION

HIV-1 is the causative agent of AIDS. The virus infects target cells through its envelope (Env) glycoprotein. This envelope glycoprotein is synthesized as a gp160 precursor protein inside the cell. During the course of transport to the virion surface, it is cleaved by the host cell protease furin into surface-exposed gp120 and membrane-anchored gp41. The three gp120 and gp41 monomers are associated noncovalently, forming a trimer of heterodimers (1, 2). The integrity of this trimer is important for HIV-1 viral entry into the host cell. Since a significant fraction of gp120 is surface exposed, it is the primary target of the host humoral immune response. gp120 binds to the primary host cell receptor CD4; this induces conformational changes in the envelope that in turn expose the coreceptor binding site for binding to CXCR4/CCR5, ultimately leading to virus-cell fusion (1, 3–6).

On the virion surface, there are also other nonfunctional Env derivatives, such as gp41 stumps and misfolded gp120 termed “junk envelope” (7, 8), that elicit nonneutralizing antibodies. Monomeric gp120, when used as a vaccine, also primarily elicits nonneutralizing immune responses (9–11). In order to drive elicitation of broadly protective neutralizing immune responses, the native trimer is thought to be the most relevant immunogen. Many efforts have been made to make an immunogen that closely mimics the native trimer. Since gp41 is metastable in nature, a disulfide linkage between gp120 and gp41 in combination with the I559P mutation (SOSIP) has been used to stabilize the trimer (12, 13). In an alternative strategy, the gp120 and gp41 subunits were linked with a G4S 20-residue flexible linker to generate the so-called NFL trimer (14–16). A number of other stabilizing mutations have been introduced into these base constructs (17–20). However, so far none of these derivatives have elicited broadly neutralizing antibodies in animal immunizations (21–24). Challenge studies in rhesus macaques have revealed various levels of protection against heterologous pathogenic challenge (25, 26). Recently, a cyclically permuted gp120 trimer derived from the JRFL sequence was shown to confer significant protection against heterologous challenge with the same SHIV162.P3 isolate used in the above studies in nonhuman primates (27). Multiple studies (28–30) have shown that cleaved, cell surface-expressed Env displays some antigenic differences from the BG505 SOSIP.664 soluble gp140 structure that is the template for most immunogen design. A recent single-molecule fluorescence resonance energy transfer (FRET) study also suggested that the native, pretriggered conformation of HIV-1 Env on intact virions differs subtly from those in the various soluble Env derivatives characterized to date (31–33).

In a previous study, we showed that aspartate scanning mutagenesis can be a useful tool to probe residue burial in proteins (34), as introduction of Asp at a buried position will significantly destabilize the folded structure. During the course of viral fusion, gp41 undergoes conformational changes in which the NHR (N-heptad repeat) region of gp41 interacts with the CHR (C-heptad repeat) to form a six-helix bundle (6HB) structure that drives the fusion of viral and host cell membrane. The postfusion structure of gp41 is known (35). There has been much recent progress in determining the structure of the Env ectodomain gp140 by cryo-electron microscopy (cryo-EM) and crystallography (36–41) (Fig. 1). In most of these structures, the gp41 region of 547 to 568 and the fusion peptide region of 512 to 517 are disordered (40). In addition, there is no high-resolution structure of the Env ectodomain in the absence of any stabilizing ligands or mutations. The prefusion structure of JRFL gp160 (38) suggests that part of the 547 to 568 region is helical, but this needs confirmation by additional experiments. The above structure is also involved in a complex with the monoclonal antibody (MAb) PGT151, which has recently been suggested to trap Env in a nonnative conformation (31). We therefore mutated individual residues in the above two stretches to aspartate in order to probe the burial of residues in these apparently disordered regions. We also substituted a few residues with the bulky hydrophobic residue tryptophan since this should also perturb the structure when introduced at buried locations (42). In earlier studies, tryptophan scanning mutagenesis was used to identify membrane-facing regions in helical segments of transmembrane proteins (43, 44). We expressed wild-type (WT) and mutated derivatives on the surface of HEK293T cells and monitored their expression and binding to various antibodies by flow cytometry (33, 45). To further examine the generality of results obtained with subtype B JRFL Env, a subset of mutations was also made in Env from the clade C isolate 4-2J.41, as this isolate is also cleaved and binds selectively to neutralizing antibodies when expressed on the mammalian cell surface (46–48). The data identify sites in both fusion peptide and the 547 to 568 region that are likely to be at least partially buried in native Env.

FIG 1.

Structure of BG505 SOSIP gp140. (A) Schematic showing different regions of gp41; dotted lines represent the apparently disordered regions. (B) The left panel shows the crystal structure of BG505 SOSIP gp140 (PDB ID 4TVP) (40) showing the trimer axis (red vertical line). The right panel shows a protomer of the BG505 SOSIP gp140 trimer; the stretch of residues with missing electron density in most gp140 structures is shown as a dotted line, (C) Postfusion structure (PDB ID 1AIK). N-heptad repeat (NHR) in green and C-heptad repeat (CHR) in blue. Left and right panels show views from the side and the bottom, respectively. NHR and CHR form a six-helix bundle structure during viral fusion, where NHR forms a parallel trimeric coiled coil and CHRs bind antiparallelly to the outside of this NHR coiled coil. The N and C termini of one protomer of NHR and CHR are labeled in pink and red, respectively. MPER, membrane-proximal external region; CP, cytoplasmic portion; FP, fusion peptide; TM, transmembrane.

We were able to identify positions where the Env expression and binding to conformation-specific antibodies were not affected by mutation to Asp. During the course of viral fusion, gp120 is shed, and the N-heptad and C-heptad repeats of gp41 form a six-helix bundle which drives viral fusion. We therefore also investigated whether these mutations could prevent gp120 shedding and help retain Env glycoprotein in its native, prefusion form and identified several such mutations that prevent shedding. These mutations can potentially be used to stabilize native-like Env constructs for use as vaccine immunogens.

RESULTS

Expression of different aspartate mutants.

JRFL gp160 (truncated at its cytoplasmic tail: JRFL gp160dCT), when expressed on the HEK293T cell surface, is efficiently processed into gp120 and gp41, and native-like Env oligomers are displayed on the cell surface (45, 49, 50). Previous studies have shown that cytoplasmic tail truncation has minimal effects on Env conformation (32). Cell surface-expressed Env was probed for binding against different neutralizing and nonneutralizing antibodies by fluorescence-activated cell sorting (FACS). The properly cleaved native-like envelope trimer binds selectively to CD4 binding site (CD4bs)-directed neutralizing antibodies such as b12 and significantly less to nonneutralizing antibodies such as b6 (45). JRFLgp160dCT is truncated at the cytoplasmic tail (residue 711) for increased surface expression. The gene in the plasmid is not human codon optimized, so the pcTAT plasmid is cotransfected with the Env plasmid into HEK293T cells to enhance expression.

The disordered stretches in the Env crystal structures comprising residues 512 to 517 (fusion peptide) and 547 to 568 (HR1) were selected for Asp scanning mutagenesis. Following mutagenesis, individual mutants were transfected into 293T cells. Surface expression was measured by binding of the WT and mutants to the 2G12 antibody by FACS as described previously (28) (Fig. 2). Mutants in the fusion peptide (V513D, G514D, and G516D) are expressed at lower levels than in the WT. Although some mutants that lie in the 547 to 567 region, such as G547D, N554D, L555D, A558D, residues 561 to 566, and T569D, are very well expressed, suggesting they are likely to be exposed, several residues in this supposedly disordered stretch, such as I548D, V549D, and L556D, are sensitive to Asp substitution. The three residues that do not fall in this disordered region (L576, A578, and I580) are all buried in published ectodomain structures (40) and are expectedly sensitive to substitution with Asp residues (Fig. 2 and Table 1), validating the approach.

FIG 2.

Surface expression of Asp mutants. Cell surface expression levels were detected by 2G12 antibody binding, the 2G12 binding ratio is calculated with respect to the wild type. HEK293T cells were transfected with constructs. After 48 h, the transfected cells were harvested and probed with 2G12 as the primary antibody and with anti-human phycoerythrin (PE) to quantitate cell surface expression. All the mutants are normalized to the wild type, where the wild type ratio is 1. Vertical bars represent the mean ratio value for each mutant along with standard deviation from three independent experiments analyzed using an unpaired t test (*, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; ns, nonsignificant, P ≥ 0.05).

TABLE 1.

Summary of the surface expression and antibody binding data for Asp mutants

Residue no.	Mutation	Position in helical wheelc	Surface expression (mean ± SD)d	b12/b6 ratio (mean ± SD)d	Alanine scanning result	Infectivity of Asp mutant (% ± SD)
WT			1.02 ± 0.04	1.01 ± 0.02
512	A512D		1.07 ± 0.14	1.01 ± 0.15		110 ± 3.0
513	V513D		0.68 ± 0.05	0.73 ± 0.07		Noninfective
514	G514D		0.95 ± 0.02	0.99 ± 0.06		Noninfective
515	I515D		1.14 ± 0.18	0.69 ± 0.04		36.65 ± 7.2
516	G516D		0.72 ± 0.07	0.82 ± 0.03		23.04 ± 3.9
517	A517D		0.95 ± 0.09	0.86 ± 0.06		3.4 ± 0.60
547	G547D	c	1.05 ± 0.02	0.92 ± 0.02	WT likeb	65.34 ± 12.13
548	I548D	d	0.59 ± 0.12	0.53 ± 0.03	WT likeb	∼2
549	V549D	e	0.56 ± 0.03	0.84 ± 0.13	WT likeb	∼1
550	Q550D	f	0.84 ± 0.09	0.69 ± 0.05	WT likeb	7.84 ± 1.73
551	Q551D	g	0.92 ± 0.05	0.77 ± 0.11	WT likeb	Noninfective
552	Q552D	a	0.78 ± 0.18	0.76 ± 0.06	nf foldinga and Imp associationb	Noninfective
553	N553D	b	0.71 ± 0.02	0.76 ± 0.05	Reduced cleavageb	122.93 ± 11.8
554	N554D	c	1.26 ± 0.24	0.95 ± 0.09	WT likeb	2.7 ± 0.53
555	L555D	d	0.97 ± 0.01	0.54 ± 0.02	Reduced cleavageb	Noninfective
556	L556D	e	0.37 ± 0.06	0.47 ± 0.07	Imp folding and associationa ,b	Noninfective
557	R557D	f	0.85 ± 0.24	0.92 ± 0.03	WT likeb	Noninfective
558	A558D	g	0.99 ± 0.05	0.9 ± 0.08	Imp associationb	Noninfective
559	I559D	a	0.76 ± 0.01	0.58 ± 0.05	Imp associationb	Noninfective
560	Q560D	b	0.78 ± 0.04	0.84 ± 0.12	nf foldinga	35.4 ± 2.68
561	A561D	c	0.93 ± 0.1	0.9 ± 0.16	Imp associationb	Noninfective
562	Q562D	d	1.04 ± 0.15	0.69 ± 0.05	Imp foldinga	Noninfective
563	Q563D	e	0.94 ± 0.07	0.9 ± 0.1	WT likeb	Noninfective
564	R564D	f	0.94 ± 0.27	0.78 ± 0.05	WT likeb	Noninfective
565	M565D	g	1.21 ± 0.33	0.98 ± 0.06	Imp processingb	Noninfective
566	L566D	a	0.84 ± 0.06	0.65 ± 0.05	Imp association and processingb	Noninfective
567	Q567D	b	0.58 ± 0.03	0.58 ± 0.03	Imp association and processingb	Noninfective
	Q567W	b	0.93 ± 0.04	0.44 ± 0.10		Noninfective
568	L568D	c	0.67 ± 0.04	0.67 ± 0.04	Imp associationb	Noninfective
	L568W	c	0.89 ± 0.04	0.75 ± 0.04		Noninfective
569	T569D	d	1.06 ± 0.06	0.62 ± 0.04	nf associationa	Noninfective
576	L576D	d	0.35 ± 0.08	0.29 ± 0.05	nf association,a Imp processingb	Noninfective
578	A578D	f	0.26 ± 0.08	0.29 ± 0.05		Noninfective
	A578W	f	0.93 ± 0.03	0.86 ± 0.06		Noninfective
580	I580D	d	0.29 ± 0.06	0.31 ± 0.05	nf associationa ,b	Noninfective

^aAla scanning mutants (73) based on viral entry that have impaired (Imp) folding (5 to 40% entry and <25% gp41), nonfunctional (nf) folding (<5% entry and <25% gp41), or nonfunctional (nf) association (<5% entry with a gp120/gp41 ratio of >0.5) (73).

^bSee Pacheko et al. (65).

^cPositions of mutants in the helical wheel diagram of the NHR in the postfusion 6HB.

^dThe mean surface expressions and mean b12/b6 ratios shown are each from three independent experiments and are given with standard deviation.

Expression of tryptophan mutants.

Q567, L568, and A578 were also mutated to Trp to compare with results of Asp substitutions. Q567W has similar expression as the wild type, and L568W and A578W have nearly comparable expression (Table 1). A578 is known to be buried in the Env structure (40). The lack of effect seen with the A578W substitution suggests that W is not as good a probe of residue burial as D, confirming the results seen from previous saturation mutagenesis studies (51, 52).

Conformational integrity of mutants.

In addition to native Env, there are other nonnative forms of envelope present on the cell surface, such as monomeric gp120 and gp41 stumps (13). To examine the conformational integrity of the mutants with respect to the wild-type native-like trimer, cells expressing Env were stained with CD4bs-directed neutralizing (b12) and nonneutralizing antibody (b6) for JRFL gp160dCT and VRC01 and F105 for clade C Env 4-2J.41 (46, 47). Native-like trimeric Env is known to bind better to neutralizing antibodies relative to nonneutralizing ones. At 48 h after transfection, cells were harvested and stained with CD4 binding site-neutralizing (b12 or VRC01) and nonneutralizing (b6 or F105) antibodies to screen for an Env which behaves similarly to that of the wild type, using an already established FACS method (28, 45, 53). The b12 and b6 binding of all mutants were normalized with respect to the wild type. The b12/b6 ratio provides information on the ability of the unliganded native trimer to adopt a wild type-like conformation, as nonnative (Fig. 3 and Table 1) and cleavage-defective forms of Env have been shown previously to have lower b12/b6 ratios than the WT (28, 45, 49).

FIG 3.

Conformational integrity of mutants. This is determined by measuring the ratio of the binding of the neutralizing antibody b12 to the binding of nonneutralizing b6 to HIV-1 Env displayed on HEK293T cells. HEK293T cells were transfected with constructs. After 48 h, the transfected cells were harvested and probed with 2G12 as the primary antibody for expression and with anti-human PE as a secondary antibody All of the mutant values are normalized to the wild type, where the wild type ratio is 1. Vertical bars represent the mean value for each mutant along with standard deviation from three independent experiments analyzed using an unpaired t test (*, P < 0.05; **, P < 0.01; ***, P < 0.001; **** P < 0.0001; ns, nonsignificant, P ≥ 0.05).

Among all the mutants in the fusion peptide 512 to 517 region, V513D and I515D have a lower b12/b6 ratio than the wild type (Fig. 3 and Table 1). The Asp mutants at residues 547, 549, 554, 557, 558, 560, 561, 563, and 565 behaved similarly to the wild type and apparently did not perturb native trimer formation. However, mutants I548D, L555D, L556D, I559D, and Q562D and residues 566 to 569 have a lower b12/b6 ratio than the wild type (Table 1). L576D, A578D, and I580D mutations also significantly perturb the native trimer conformation and appear to have the highest destabilization among all NHR mutants (Fig. 3 and Table 1). Unlike L568D and A578D, L568W and A578W behaved nearly similarly to the wild type. We examined the cleavage efficiency of all mutants expressed on the cell surface by using PGT151 antibody (54, 55), as PGT151 selectively binds to properly formed and cleaved trimers. Almost all mutants were found to be well cleaved on the cell surface (Fig. 4). We also examined the effects of a few mutants in the fusion peptide (FP) (V513D, G514D, and A517D) and the N-heptad repeat (HR1) (Q552D, L555D, and Q563D) regions in clade C Env 4-2J.41. Since this isolate is not neutralized by b12, we could not use the b12/b6 ratio to probe the conformational integrity of the cell surface-expressed Env. Instead, we used the VRC01/F105 ratio, since VRC01 and F105 serve as neutralizing and nonneutralizing antibodies for this isolate (47). The values of this ratio for selected Asp mutants in strain 4-2 J.41 are shown in Fig. 5A, and the overall trends are similar to those observed for the b12/b6 ratio in corresponding mutants in subtype B JRFL Env (Fig. 3).

FIG 4.

Cleavage efficiency of mutants expressed on the cell surface. Cleavage was monitored by PGT151 binding. Most mutants showed similar cleavage efficiency to that of the WT.

FIG 5.

Conformational integrity and sCD4-induced gp120 shedding of clade C Env 4-2 J.41 expressed on the surface of HEK293T cells. (A) Mutants of clade C Env were transfected into HEK293T cells and probed for binding to VRC01 and F105 monoclonal antibodies (MAbs). (B) Transfected cells were incubated with sCD4 to induce gp120 shedding (46, 47).

Some mutants diminish gp120 shedding upon treatment with soluble CD4.

During the course of viral fusion, gp120 binds to the primary host cell receptor (CD4). This leads to conformational changes in gp120 and gp41. gp120 is shed off from the virus, and gp41 then drives viral fusion with the host membrane (50, 56). In the postfusion conformation, the NHR and CHR of gp41 form a six-helix bundle (6HB) structure in which the NHR trimer is surrounded by three CHR helices (Fig. 1). The mutation V570D has been shown to be 6HB destabilizing by using a yeast surface two-hybrid display system (57, 58). The same mutation was shown using mammalian cell surface display to prevent gp120 shedding (28). We attempted to identify other gp41 mutations that would prevent shedding, since these can be incorporated into Env immunogens to restrict their conformational variability.

A few mutations were selected from the NHR of gp41 in the 547 to 567 stretch that showed good surface expression and, more importantly, a b12/b6 ratio of ≥0.7 (Table 1), indicating a WT-like conformation. The mutant Env plasmids were cotransfected along with the pcTAT plasmid in HEK293T cells. After 48 h, the transfected cells were harvested, 1 × 10⁶ cells were incubated with 50 μg/ml of soluble CD4 (sCD4), and cells were washed and then incubated with 2G12 at various concentrations (shown here at 10 μg/ml). gp120 shedding from the cell surface was measured by 2G12 binding. The mutants G547D, V549D, Q551D, N553D, N554D, and Q560D could not prevent gp120 shedding (Fig. 6 and Table 2). However, other mutants, namely Q552D, R557D, A558D, A561D, Q563D, R564D, and M565D, helped in preventing gp120 shedding (Fig. 6A and Table 2). Other mutants that lie in the “a” and “d” positions of the NHR postfusion coiled coil (35, 57) and also in the apparently disordered 547 to 569 stretch in the BG505gp140 ectodomain crystal structure (40) were also examined, and these mutants (L555D, I559D, Q562D, L566D, and T569D) prevented gp120 shedding. The previously characterized mutant V570D was taken as a positive control. All of the above mutants behaved in a similar way to V570D (28, 57). Overall, the ability to prevent shedding in the 547 to 570 region is largely restricted to residues in the stretch from 555 to 570, suggesting that these residues become buried in gp41 intermediates that form during the shedding process (Table 2). We also examined shedding effects in subtype C 4-2J.41 Env. Since this strain is not neutralized by b12, we used VRC01 instead (46–48). Overall, the results for this subtype C Env (Fig. 5B) were similar to those observed with JRFL Env.

FIG 6.

Effect of mutations on soluble CD4 (sCD4)-induced shedding of gp120, as determined by fluorescence-activated cell sorting (FACS). The mutant proteins were expressed on the cell surface. Cells were incubated with sCD4 to induce gp120 shedding. The cells were washed once with FACS buffer, and the gp120 remaining on the cell surface was quantified by staining cells with 2G12 at a concentration of 10 μg/ml. The ratio of mean fluorescent intensities (MFIs) from sCD4-treated and untreated cells for each mutant is shown for (A) mutants that did not prevent shedding and (B) mutants that prevented shedding. V570D was taken as a positive control since it is known to prevent shedding (28).

TABLE 2.

Comparative summary of accessibilities of mutants in cryo-EM prefusion and postfusion structures of gp41 and effect of Asp mutants on gp120 shedding^a

Residue no.	Position in helical wheel 6-helix bundle	Residue	Prefusion accessibilities in 5FUU (%) for protomer:			Postfusion accessibilities in 1AIK (%)	Prevention of sCD4-induced gp120 shedding
			1	2	3
547	c	GLY	17.5	77.9	80.9	13.1	No
548	d	ILE	80.7	28.8	50.6	10.9	No
549	e	VAL	43.3	57.2	62.4	6.1	No
550	f	GLN	24.4	61.5	73.5	50.1	No
551	g	GLN	61.4	94.1	57.4	0	No
552	a	GLN	66.7	38.3	40.7	0	Yes
553	b	ASN	17.8	98.5	57.8	47.3	No
554	c	ASN	45.8	83.2	80	0	No
555	d	LEU	74	9.1	19.1	0.1	Yes
556	e	LEU	17.7	36.6	28.1	4.4	Yes
557	f	ARG	12	95.5	73.1	49.1	Yes
558	g	ALA	89.3	17.4	10.9	0.8	Yes
559	a	ILE	48	17.2	32.3	0.2	Yes
560	b	GLU	22.8	35.7	41.7	27.7	No
561	c	ALA	53.2	56.2	52.9	3.2	Yes
562	d	GLN	76.2	8.5	5.4	1	Yes
563	e	GLN	15.4	19.5	15.5	1	Yes
564	f	ARG	19.7	31.9	30.4	54.6	Yes
565	g	MET	55.3	47.2	47.9	0.1	Yes
566	a	LEU	16	3.6	5.7	1.4	Yes
567	b	GLN	60.5	43.5	5	37.5	Yes
568	c	LEU	23.8	21.1	38.5	18	No
569	d	THR	24.3	56	39.9	0.9	Yes
570	e	VAL	2	39.8	19.9	3.2	Yes

^aThe cryo-EM structure is in complex with PGT 151. The MAb binds asymmetrically to D and F chains of gp41; because of this, the three gp140 interfaces are not equivalent. Consequently, the unliganded protomer (protomer 1) is found to be conformationally variable compared to the liganded protomers, as is clearly evident from the accessibility values. The N-heptad repeat residues lie at different positions in the helical wheel in the postfusion conformation (PDB ID 1AIK) and have different percent accessibilities. There is no clear correlation between residue accessibility in the postfusion structure and ability to prevent sCD4-induced shedding.

Mutants that prevent sCD4-induced gp120 shedding showed enhanced recognition of CD4i antibodies 17b and 19e with gp41 cluster 1 region remain occluded upon sCD4 induction.

Mutants that prevent gp120 shedding upon sCD4 induction may or may not bind to sCD4. To further examine this, cells expressing a subset of mutants shown in Fig. 6 were incubated with sCD4 and probed for binding to 17b (1, 59) and 19e (60–62), which bind to CD4-induced (CD4i) epitopes (Fig. 7). The wild type and all of the mutants bound to 17b and 19e. The binding mean fluorescent intensity (MFI) for mutants that prevented gp120 shedding (A558D and V570D) increased by more than 2-fold after sCD4 induction compared to that of the mutants that could not prevent gp120 shedding (V549D, Q551D, and Q560D). This suggests that mutants that prevent gp120 shedding still bind to sCD4 and that this results in exposure of CD4i epitopes (Fig. 7A and B). D49 antibody binds to the gp41 cluster 1 region, which is normally exposed upon gp120 shedding (28, 63, 64). Expectedly, following incubation with sCD4, mutants which prevent gp120 shedding do not show any increased binding to D49, unlike what is observed for mutants that do not prevent shedding (Fig. 7C).

FIG 7.

Probing binding of mutant Env proteins to 17b, 19e, and D49 following incubation with sCD4. All of the constructs were transfected into HEK293T cells. After 48 h, transfected cells were harvested, incubated with sCD4 for 2 h, and washed to remove shed gp120, then incubated with (A) 17b, (B) 19e, or (C) D49 for 1 h and washed and probed for antibody binding with phycoerythrin-labeled secondary antibody using FACS. A second set of cells was similarly probed for antibody binding without sCD4 incubation. The MFI ratio of sCD4-treated/untreated cells was calculated.

Pseudoviral infectivity of all Asp mutants.

The infectivity of all the JRFL Asp mutants was measured. Pseudoviruses were produced from HEK293T cells. Equivalent p24 levels of the WT and all of the mutants were used to infect TZM-bl cells. The effects of Asp mutations on cell surface Env probe the corresponding residue burial in the context of a native-like trimer. In contrast, effects on pseudoviral infectivity probe burial at all stages of the infection process. All of the mutants that prevented gp120 shedding were found to be noninfectious (Fig. 8 and Tables 1 and 2). This might be because either shedding is a prerequisite for downstream steps that lead to fusion or the mutations prevent the formation of a crucial intermediate in the fusion process. However, some of the mutants that did not prevent gp120 shedding, like those at 548 and 549, were also found to be noninfectious, likely because these residues are buried in the postfusion structure (Table 2). In the fusion peptide region, 513, 514, and 517 show the largest decreases in infectivity, whereas 512, 515, and 516 are relatively unaffected. This is a surprising result, suggesting that the entire fusion peptide is not buried during various stages of the fusion process. Overall, Asp mutations have stronger effects on infectivity than previously observed for Ala mutants (65) at most positions, consistent with the greater burial penalty of Asp versus Ala.

FIG 8.

Pseudoviral infectivity in TZM-bl cells for aspartate mutants in the fusion peptide and N-heptad repeat regions of gp41. Equivalent p24 levels of virions were used to measure infectivity.

DISCUSSION

In the case of HIV-1 Env, both structures of the postfusion six-helix bundle (35) and, more recently, crystal structures of gp140 ectodomain constructs are available (37, 38, 40, 41) (Fig. 1). The latter are thought to closely resemble the native prefusion conformation of the Env ectodomain on the virion. In most prefusion structures, there is missing electron density between residues 512 and 517 of the fusion peptide and in the N-heptad repeat stretch from residues 547 to 568 (Fig. 1). All existing structures were stabilized by MAbs and engineered disulfides, and most contain an I559P mutation. There is currently no structure of native, unliganded Env. In addition, recent single-molecule FRET experiments suggest that native Env as it exists on the virion may have a conformation subtly different from that seen in existing structures of soluble gp140 ectodomains (31). Hence, we have used aspartate scanning mutagenesis as an alternate approach to probe residue burial in these apparently disordered regions, in the context of viral Env expressed on the mammalian cell surface. It is known that aspartate, being a charged residue, is destabilizing and is poorly tolerated at buried positions, and conversely, that is well tolerated at exposed nonactive site residues (34), with the obvious exception of a disulfide-bonded Cys residue. Hence, Asp scanning mutagenesis can be used to probe residue burial. We also mutated a few residues with tryptophan to probe the effect of a bulkier, hydrophobic group on conformation and function (42, 66).

We introduced Asp mutations at a number of non-Cys positions in an Env-expressing plasmid with a cytoplasmic tail truncation (JRFLgp160dCT) as well as a subset of these mutations in Env from the clade C Env from isolate 4-2J.41. The JRFLgp160dCT-bearing mutations were cotransfected with the pcTAT plasmid into HEK293T cells. After 48 h of transfection, cells were harvested and probed with monoclonal antibody 2G12 for surface expression, with MAbs b12 and b6 for clade B JRFL and VRC01 strains, and with F105 for clade C 4-2 J.41 to assess conformational integrity compared to that of the wild type. A few of the mutants, namely L556D, L576D, A578D, and I580D, had very minimal expression on the cell surface (Fig. 2 and Table 1). These mutations also had an impact on the native conformation of Env. The b12/b6 ratio was lowest for L576D, A578D, and I580D (Fig. 3). These three residues acted as a positive control for method validation since they are known to be buried in the gp140 ectodomain structure (40). Other mutants, such as V513D and I515D in the fusion peptide and numerous residues in the 547 to 570 stretch, show significant perturbations in expression and/or conformation relative to those of the WT. These observations suggest that the fusion peptide is not fully exposed and that the 547 to 568 stretch is not disordered in native, membrane-bound Env. A cryo-EM structure of a soluble B41SOSIP ectodomain trimer reveals that the fusion peptide becomes ordered upon b12 binding (41). In this structure, residues 513 and 516 are partially buried, consistent with the reduced b12/b6 ratio and as seen for Asp substitutions at these positions. The same residues also show reduced surface expression in the absence of b12, indicating that they are partially buried in unliganded Env as well (Fig. 3 and Table 1). gp41 of the Env glycoprotein is metastable in nature. The accessibility values of these residues in a recent cryo-EM structure of JRFL gp160dCT in complex with PGT151 were taken for comparison (38). The structure is in complex with PGT151, and the stoichiometry of PGT151 to trimer is 2:1, i.e., 2 fragment antigen-binding regions (Fabs) bind per trimer. Hence, the interfaces of all the protomers are not equivalent, which is clearly evident from Table 2. Furthermore, a recent FRET study (31) suggests that PGT151 may trap Env in a conformation different from the predominant “state 1” conformation present on the virion. The Env structure is highly dynamic, and depending on the conformation, the fusion peptide may be either exposed or sequestered in the trimeric core on native Env on the virion surface (41, 67). In the closed HIV-1 ectodomain structures, the fusion peptide is often found to be disordered (68, 69). The fusion peptide-specific antibody VRC34 binds to the prefusion trimer, contacting the heavy chain at residues 517, 519, and 520. From mutational antigenic profiling studies, it was shown that escape variants are located primarily in the stretch from 512 to 518, of which Asp at 514 and 516 were escape variants (68, 70–72). Interestingly, many viral strains show partial neutralization by VRC34 (68). This is consistent with the fusion peptide having multiple conformations with various degrees of exposure on native Env. Asp substitutions at positions 512, 515, and 516 show relatively small decreases in infectivity, while those at 513, 514, and 517 show large decreases. This periodicity may be indicative of an amphiphilic helical structure of this stretch during fusion. In two Ala mutagenesis studies (65, 73), some mutants in the 547 to 567 stretch were reported to have folding and association defects, consistent with our results (Table 1). Compared to aspartate mutations at 568 and 578, tryptophan mutations at A578 and L568 have comparable expression and a comparable b12/b6 ratio to those of the wild type, confirming that D is superior to W in probing residue burial (Table 1).

The trimerization of Env is largely mediated by gp41 and assisted by the V1V2 loop of gp120 (39). gp120 sheds during the course of viral fusion. We examined if Asp mutants that behaved like the wild type in terms of the surface expression and the b12/b6 ratio could prevent sCD4-induced gp120 shedding. Mutants which prevented shedding were largely found in the stretch from between residues 555 and 566 (Fig. 6 and Table 2). It is believed that the mutations that are able to prevent gp120 shedding (28, 74–76) destabilize the formation of the six-helix bundle (6HB) or other fusion-intermediate conformations. However, it might also be possible that such mutants prevent binding to sCD4. In the sCD4-bound trimer structure, the coordinates for the above-mentioned NHR residues are missing, and V570D has an accessibility of 61% (PDB ID 5U1F) (77). Hence, it is unlikely that these mutations prevent CD4 binding; instead, it is more likely that they prevent subsequent conformational changes that result in shedding. To further probe whether such mutants bind to sCD4, sCD4-induced binding to antibodies 17b and 19e, which bind to CD4-induced epitopes, was monitored. Mutants that prevented shedding showed increased binding to 17b and 19e in the presence of CD4 relative to that of the WT (1, 59, 61, 78). This is likely because gp120 is not shed, but it also confirms that these mutants are indeed competent to bind CD4. Furthermore, the mutants that prevent gp120 shedding do not show any increase in binding to D49, an antibody that binds to the cluster 1 region that gets exposed after gp120 shedding (63, 64). In contrast, enhanced D49 recognition is seen for mutants that do not prevent shedding, as gp120 is shed and the epitope is exposed for binding (Fig. 7C).

No correlation with the ability to prevent shedding was seen with either location in the helical wheel or accessibility in the postfusion structure (Table 2 and Fig. 9). The residues that prevent gp120 shedding are not found to be more conserved than residues that do not prevent shedding (Fig. 10). The data suggest that each residue in the 555 to 566 stretch is likely to be largely buried in one or more of the intermediates on the fusion pathway; consequently, Asp mutations in this stretch destabilize such intermediates relative to the native state. Such mutations may therefore stabilize the trimer in the prefusion conformation. The mutants L555D, I559D, Q562D, L566D, T569D, L576D, and I580D are also not infective, which is consistent with their positions in the NHR postfusion structure (Fig. 8), as all of them lie in “a” and “d” position of the NHR coiled coil and are buried in the postfusion conformation (35, 65). Overall, the reduced infectivity of Asp mutants either occurs at positions that prevent shedding or at those that are buried in the postfusion structure. This in turn suggests that all Asp mutants that prevent shedding occur at positions that are buried at some stage in the fusion process. A small subset of Asp mutants was made in clade C 4-2J.41 Env. The overall results were very similar to those observed for corresponding positions in clade B JRFL Env (Fig. 3, 4, and 5).

FIG 9.

Structural information from postfusion structure. Residues mapped on postfusion structure (PDB ID 1AIK). (A) Residues that prevent sCD4 gp120 shedding, (B) Residues that do not prevent gp120 shedding, and (C) N-heptad repeat (NHR; green) and C-heptad repeat (CHR; blue) arranged in coiled coil fashion. Residues that prevent gp120 shedding mainly lie between the NHR-NHR interface at positions “a” and “d” and at the NHR-CHR interface positions “e” and “g.” There are also residues that lie in exposed positions at the NHR that prevent shedding (Tables 1 and 2).

FIG 10.

Web logo showing residue conservation for the disordered stretches of residues. Residues that prevent gp120 shedding are not more conserved than residues that do not prevent gp120 shedding. The residues highlighted with an asterisk (*) prevent sCD4-induced gp120 shedding.

Proline mutations at L555, I559, L566, T569, I573, and V580 were previously made in SOS gp140 to understand the effect of proline substitution, as it is known to cause a kink in a helix and can therefore be useful to probe helicity. L555P expressed poorly. I559P, I573P, and V580P have expression comparable to that of the wild type (76). In the same study (76), the effect of mutating residues near position 559 in JRFL SOS gp140 was addressed. L556P, R557P, A558P, I559P, and M565P have similar expression to that of the wild type. Mutations E560P, A561P, Q562P, Q563P, and R564P have higher expression than that of the wild type. Residues in this apparently disordered stretch were further mutated to proline in BG505 and C16055 Env (20), and it was found that S553P, N554P, L555P, Q562P, and Q563P bound better to trimer-specific broadly neutralizing antibodies (bNAbs) than to nonneutralizing antibodies. L555P has increased yield and more ordered trimers than the wild type. The accessibility of residues probed by our experimental method correlates only moderately with the accessibilities in the recent cryo-EM structure of JRFL gp160dCT at most positions (38). For example, residues 555, 566, and 567 are positions at which at least one residue in the protomer has an accessibility of <10% and are sensitive to Asp substitution (Fig. 3). However, several residues (Fig. 3) that are sensitive to Asp substitution, such as 548, 556, 564, and 570, are exposed in all protomers, and conversely, some apparently buried residues (562 and 570) are relatively insensitive to mutation. In the existing ectodomain structures, a proline in this region of the protein is present. While it is possible that this might be responsible for the apparent disorderedness of this stretch (79), this is unlikely because there are several studies that show that Pro can be incorporated even in helical stretches without much structural rearrangement and at a moderate energetic cost (80–83). In summary, using Asp scanning mutagenesis, we have probed residue burial in two apparently disordered stretches. The data identify sites in both regions that are likely to be buried in native Env in the virion context. The data suggest that residues in the 555 to 566 stretch are likely to be buried in one or more fusion intermediates. Asp substitutions that prevent shedding without affecting the expression or the b12/b6 binding ratios can potentially be used to stabilize Env in native-like conformations for structural studies and vaccine applications.

MATERIALS AND METHODS

Cell lines and antibodies.

HEK293T cells were used for mammalian cell surface expression. These cells were grown in Dulbecco’s modified Eagle’s medium (DMEM), 10% fetal bovine serum (FBS), and 1% antibiotics (penicillin-streptomycin and amphotericin). TZM-bl is an engineered HeLa cell line that expresses CD4 and CXCR4 receptors and was grown in the same medium described above. MAb 2G12 was used to probe surface expression. MAbs b12 and b6 were used for determining the conformational integrity of cell surface-expressed JRFL Env gp160dCT. VRC01 and F105 were used for 4-2J.41 Env (47). sCD4 is used for the shedding experiment.

Constructs.

The pSVIII JRFL gp160dCT expression plasmid (45) was used for cell surface expression of JRFLgp160dCT, and mutations were incorporated by site-directed mutagenesis. HXBC2 numbering was followed, and mutations were confirmed with DNA sequencing. JRFL gp160 dCT has a cytoplasmic tail truncation at residue 711, and the Env gene present is nonhuman codon-optimized viral sequence. Hence pcTAT, which codes for HIV-1 Tat protein, is transfected along with the psVIII expression plasmid. A subset of mutants was made in the context of clade C Env 4-2J.41 and probed using cell surface expression to see if the effects are strain specific (47).

Transient transfection of Env plasmid into HEK293T cells.

One day prior to transfection, 3 × 10⁶ 293T cells were seeded in a T75 culture flask. After 24 h, the cells were transfected with pSVIII JRFLgp160dCT and 4-2J.41 Env expression plasmids encoding Env (WT and mutants) and pcTAT in a ratio of 4:1. The transfection mixture was prepared in serum-free medium with DNA:polyethyleneimine (PEI) in a ratio of 1:4. The transfection mixture was incubated for 10 min. After incubation, serum-containing medium was added, and the mixture was added into the flask containing 293T cells.

Staining of surface-expressed Env for FACS analysis.

After 48 h of transfection, the cells were harvested with phosphate-buffered saline (PBS) containing 5 mM EDTA and washed with FACS buffer (PBS, 5% FBS, and 0.02% azide). The harvested cells (4 × 10⁵ cells per tube) were stained with the desired antibody for 1 h at 4°C. The antibody-cell mixture was washed with FACS buffer (PBS [pH 7.4], 5% fetal bovine serum, and 0.02% azide). Anti-human IgG phycoerythrin (PE; Sigma) at a 1:100 dilution was added and incubated for 1 h at 4°C, followed by a wash with FACS buffer (45). The stained cells were analyzed on a FACS analyzer (BD Accuri). On forward scatter (FSC)-side scatter (SSC) plots, the cells were gated to discriminate between dead cells, doublets, and live or single cells. The MFI values were obtained from the gated single cell population. In each experiment, unstained controls, secondary antibody controls, and untransfected cells with primary antibody controls were prepared at the same time as test samples. Each FACS experiment was repeated independently (with independent transfection experiments) to check for consistency of results. BD Accuri software was used to analyze the data.

sCD4-induced gp120 shedding experiment.

At 48 h after transfection, cells were harvested and washed with FACS buffer. Harvested cells (1 × 10⁶ cells per tube) were incubated with or without sCD4 at 50 μg/ml concentration for 2 h at 4°C with occasional mixing. To remove shedgp120, cells were washed with FACS buffer. Cells were then incubated with 2G12 or VRC01 antibody at various concentrations (shown here are 10 μg/ml of 2G12 and 30 μg/ml of VRC01 for JRFL gp160dCT and 4-2J.41 Env, respectively) (47) for 1 h at 4°C. After a wash, cells were incubated with anti-human IgG-PE (Sigma) to stain 2G12 and VRC01 antibody-bound cells. After a wash, cells were analyzed on a FACS analyzer (BD Accuri). BD Accuri software was used to analyze the data (28).

Pseudoviral generation and infectivity assay.

HEK293T cells were transfected with a plasmid containing virus backbone lacking the Env gene (pSG3Δenv) and a mutant Env plasmid (pSVIII JRFL gp160dCT) in a 1:3 ratio using PEI (catalog no. 23966; Polysciences, USA). After 48 h of transfection, the supernatant was collected and stored at −80°C. For testing infectivity, the viral dilutions were mixed with 10,000 TZM-bl cells plated in a 96-well flat-bottomed plate and incubated at 37°C for 48 h (84–88). After incubation, 100 μl of medium was removed from each well, and 80 μl of britelite plus reagent (Perkin Elmer) was added. After 2 min of incubation at room temperature, cells were lysed by gentle pipetting. The 100-μl volume of lysed cells was transferred into a black plate, and luminescence was measured in a Victor X2 luminometer (Perkin Elmer). The infectivity was plotted in relative luminescence units (RLUs).

sCD4-induced gp120 binding to 17b, 19e, and D49.

At 48 h after transfection, cells were harvested and washed with FACS buffer. Harvested cells (1 × 10⁶ cells per tube) were incubated with or without sCD4 at 50 μg/ml concentration for 2 h at 4°C with occasional mixing. To remove shedgp120, cells were washed with FACS buffer. Cells were then incubated with 17b, 19e, and D49 antibody at 10 μg/ml for 1 h at 4°C. After a wash, cells were incubated with anti-human IgG-PE (Sigma) to stain 2G12 antibody-bound cells. After a wash, cells were analyzed on a FACS Analyzer (BD Accuri). BD Accuri software was used to analyze the data (28).