Conserved and Exposed Epitopes on Intact, Native, Primary Human Immunodeficiency Virus Type 1 Virions of Group M

Viruses and virus stock preparation.

A total of 26 HIV-1 group M isolates of clades A to H were studied. The viruses used and the clade to which each belongs are shown in Fig. 1 (see below). The HIV-1 isolates were obtained from patients in Cameroon (CA1, CA4, CA5, CA13, and CA20), Belgium (VI191), Uganda (92UG021 and 92UG001), France (BX08), the United States (MNp, IIIB, and JR-FL), Senegal (SG2728), Zimbabwe (2036), Rwanda (92RW021), Zambia (ZB18), Ivory Coast (CI13), Zaire (MAL), Brazil (93BR019 and 93BR029), Thailand (92TH009, 92TH011, BK131, and CM235), and Gabon (VI525 and VI526). All the viruses have been passaged only in human peripheral blood mononuclear cells (PBMCs) except for MAL and IIIB, which had, in the past, been passaged in a continuous cell line (H9 cells) before being carried in human PBMCs. To prepare virus stocks, 1 ml of p24-positive culture supernatants were used to infect 3-day phytohemagglutinin-stimulated HIV-negative donor PBMCs as previously described (47, 48). After 2 to 3 weeks of culture, the culture supernatant from infected PBMCs was aliquoted (1 ml/tube) and stored in liquid nitrogen until use. The p24 concentration in each virus stock was quantitated using our noncommercial p24 enzyme-linked immunosorbent assay (ELISA) (45).

Human anti-HIV MAbs.

A total of 45 human anti-HIV-1 MAbs against various regions of gp120 (V2, C2, V3, CD4bd, and C5) and gp41 (clusters I and II), produced in our laboratory (9, 14–16, 19, 29, 72), were used. Table 1 lists these MAbs along with the epitopes for which each is specific. These MAbs were derived from the PBMCs of HIV-1 clade B-infected patients (and in one case, clade E [MAb 1342E]) as previously described (14–16). Briefly, patients PBMCs were transformed with Epstein-Barr virus and those that were positive for the production of the desired antibodies were expanded and then fused with the SHM-D33 human X mouse heteromyeloma. The resulting heterohybridomas were rescreened for the production of the desired antibody, and the positive cultures were sequentially cloned at limiting cell concentrations until monoclonality was achieved.

Recombinant human MAb IgG1b12 (anti-CD4bd) was provided by D. Burton and P. Parren (6). Recombinant MAb 2F5 to gp41 cluster II (42) was obtained from the Medical Research Council Reagent Program. These MAbs have been extensively described elsewhere (6, 32, 42). Human MAb 860-55D to parvovirus B19 (12) was used as a negative control.

Virus binding assay.

We have previously described a virus binding assay for the detection of epitopes exposed on the surface of intact, native HIV-1 primary isolates (45). Briefly, ELISA wells were coated at 4°C overnight with 100 μl of MAb at 10 μg/ml. After the wells were washed with phosphate-buffered saline (PBS) and blocked with 0.2 ml of 3% bovine serum albumin (BSA) in PBS (BSA-PBS), virus was added (100 μl/well at 100 ng of p24/ml). After incubation at 37°C for 1 h, the wells were washed with RPMI 1640 to remove unbound virus, the bound virus was lysed with 250 μl of 1% Triton X-100, and the contents of each well were assayed for p24 by a noncommercial ELISA as described below. All experiments were performed in duplicate, and after the broad reactivity of two cluster I gp41 MAbs was identified, one of these (MAb 246-D) was consistently used as a positive control. MAb 860-55D to parvovirus B19 was used as a negative control.

The amount of p24 captured in the virus binding assay was quantified using human anti-p24 MAb 91-5 (14) in a noncommercial p24 ELISA. For this assay, ELISA wells were coated at 4°C overnight with 100 μl of MAb 91-5 at 0.5 μg/ml. After the wells were washed with PBS-Tween 20 (PBST) and blocked with 0.2 ml of 5% BSA in PBST (BSA-PBST) for 6 h at 37°C, the lysate from the virus binding assay was added (100 μl/well). Recombinant p24 from HIV-1_SF2 expressed in yeast (Chiron Corp., Emeryville, Calif.) was reconstituted to contain 250 pg/ml and then serially diluted twofold from 250 to 15.6 pg/ml, and a 100-μl sample of each concentration was added in duplicate to microtiter wells. The microtiter plates were incubated at 4°C overnight and then washed with PBST, biotinylated anti-p24 MAb 241-D (14, 45) (which recognizes a different epitope from that seen by MAb 91-5) was added, and the mixture was incubated at 37°C for 2.5 h. After subsequent washing, the plates were incubated with streptavidin-alkaline phosphatase (Life Technologies Inc., Gaithersburg, Md.) for 1 h. Color was developed with an amplification system (Life Technologies Inc.), and plates were read at 490 nm. The amount of p24 measured was determined by extrapolating the optical density value of the test sample (lysate) on the curve plotted from the values obtained with the recombinant p24.

Infectivity assay.

To determine whether MAbs bound infectious virions, the virus binding assay described above was used with some modifications. After virus was incubated on MAbs coated on microtiter wells, unbound virus particles were removed by washing four times with RPMI 1640. PBMCs, preincubated for 3 days with phytohemagglutinin (at 0.5 μg/ml) and for 2 days with interleukin-2 (at 20 U/ml), were added to the microtiter wells at 1.5 × 10⁵ cells/well in 50 μl RPMI 1640 and incubated for 2 h in 5% CO₂ at 37°C, after which 150 μl of RPMI 1640 was added. (The RPMI 1640–IL-2 medium was supplemented with 15% fetal calf serum, 3% l-glutamine, 2 μg of Polybrene per ml, 5 μg of hydrocortisone per ml, and antibiotics [penicillin and streptomycin]). The cultures were further incubated in 5% CO₂ at 37°C for 3 weeks and monitored weekly for p24 production by using the noncommercial p24 ELISA described above.

Neutralization.

The ability of MAbs to neutralize HIV-1 isolates was tested in the GHOST cell neutralization assay, as previously described (7). For use in neutralization assays, the titers of virus stocks were determined on GHOST cell lines carrying the coreceptors used by each isolate so as to determine the dilution factor at which virus stocks were able to infect 200 to 1,000 GHOST cells/15,000 total cells at 3 to 5 days postinfection. The appropriate dilution of virus in Dulbecco's modified Eagle's medium was incubated with MAbs (at 10 μg/ml) for 1 h at 37°C. The virus-MAb mixture (110 μl/well) was then added to GHOST cells expressing the appropriate coreceptor (CCR5 or CXCR4) and incubated at 37°C in 7% CO₂ overnight. The virus and antibody were then washed off, Dulbecco's modified Eagle's medium was added, the mixture was incubated again for 3 to 5 days, and the cells were then analyzed for infectivity by fluorescence-activated cell sorting as previously described (7).

Statistical analysis.

We determined whether a particular MAb-virus combination displayed binding by comparing the amount of captured p24 (in picograms per milliliter) to a threshold value (45). The threshold was calculated as the mean plus 6 standard deviations of p24 from virus captured by the negative control anti-parvovirus MAb 860-55D. The threshold value (T) is 11 pg of p24/ml (log₁₀ T = 1.04), and any value exceeding T was considered to indicate positive binding.

Anti-gp120 MAb binding patterns with intact, native primary HIV-1 isolates of clades A to H. (i) anti-V3 MAbs.

A total of 19 anti-V3 MAbs were tested for their abilities to bind to 26 HIV-1 group M isolates of clades A to H, giving 494 anti-V3 MAb-virus test combinations. The log transformations of p24 values (in picograms per milliliter) are shown in Fig. 1, and the number of anti-V3–virus test combinations showing different levels of binding is shown in Table 2. Figure 1a shows the anti-V3 MAbs in rank order from the strongest average binding to all 26 isolates to the weakest average binding. (All subsequent groups of MAbs in Fig. 1 are similarly displayed.) Of the 494 individual anti-V3–virus combinations tested, 218 (44%) displayed significant levels of virus binding (as defined in Materials and Methods), with log p24 values from bound virus being greater than 1.04 (11 pg/ml). The remaining 56% of the test combinations were negative, with log p24 values from bound virus being less than 1.04. Each anti-V3 MAb was capable of binding to at least 4 of the 26 isolates. The most reactive anti-V3 MAb, 447-52D, bound to 24 of the 26 viruses (92%) of all the clades examined, although binding to the viruses of clades G and H was weak (Fig. 1a). In addition, MAbs 419-D, 694/98D, 838-D, 412-D, and 1006-15D bound to 39 to 69% of the isolates tested. These results suggest that the epitopes to which the first six anti-V3 MAbs shown at the top of Fig. 1a are directed are well exposed and conserved on 39 to 92% of the isolates of all the HIV-1 clades examined. These results also confirm previous observations (17, 45, 73) that the V3 loop is neither type-specific nor cryptic, as suggested by others (4, 49, 70). While the six most reactive anti-V3 MAbs bound to several isolates (10 to 24 viruses), other MAbs such as the last six anti-V3 MAbs listed in Fig. 1a failed to bind to most of the isolates or bound poorly and sporadically to only a few isolates. This suggests that these latter MAbs may be specific for “private” epitopes, which are more restricted in their representation on viruses within the HIV-1 family.

The converse analysis shows that every isolate could be bound by one or more anti-V3 MAbs (Fig. 1a). Some viruses among clades A, B, C, and D could bind to the majority of the anti-V3 MAbs. For example, viruses CA1 (clade A), CA5 (clade B), SG2728 (clade C), and MAL (clade D) bound to the majority of anti-V3 MAbs (Fig. 1a), indicating that several epitopes in the V3 region are exposed on some isolates belonging to different clades. However, some other viruses in the same clades (A, B, C, and D) and most viruses in clades E, F, G, and H, were bound by only a minority of anti-V3 MAbs.

Anti-V3 MAbs–virus binding did not correlate with the virus clade or virus phenotype (syncytium or non-syncytium inducing), as has previously been suggested on the basis of peptide binding (17, 73). As previously shown, the binding of MAbs to virus also did not correlate with the presence or absence of the exact sequence of the core epitope on the virus (reference 45 and data not shown).

(ii) Anti-C5 MAbs.

Four anti-C5 MAbs were tested with the 26 viruses. Two patterns of binding were observed: two MAbs bound well and across almost all the clades (MAb 670-D and 1331A) and two MAbs did not bind or bound weakly (858-D and 989-D). Those four MAbs, 670-D, 1331A, 858-D, and 989-D, bound to 21 (81%), 18 (69%), 4 (15%), and 6 (23%) of the isolates, respectively (Fig. 1a). The core epitope to which 670-D is directed (amino acids [aa] 503 to 509) is different from that recognized by 1331A, 858-D, and 989-D (aa 495 to 516) (Table 1). Because these last three MAbs recognize the same linear epitope and have similar relative affinities (unpublished data), the different pattern of binding of 1331A compared to 858-D and 989-D might be due to recognition of different conformational aspects of C5. Indeed, the conformation dependence of MAb 1331A has recently been described (24), and we have previously shown that epitope conformation may contribute as much as 90% of the binding energy of even those MAbs for which linear core epitopes can be identified (13). Thus, the data are clear that MAbs 670-D and 1331A recognize a complex conformational region of the C terminus of gp120 that is highly conserved and well exposed on the surface of virions.

(iii) Anti-CD4bd MAbs.

A total of six anti-CD4bd MAbs were tested for their abilities to bind to the 26 HIV-1 isolates. Of the eight HIV-1 clades (A to H) tested, these MAbs consistently bound to most isolates of clade D, with binding levels up to 208 pg/ml (2.3 log units) (Fig. 1b). However, binding by anti-CD4bd MAbs to isolates in clades A, B, C, E, F, G, and H was lacking in the majority of combinations and rarely exceeded weak binding. Of all six anti-CD4bd MAbs examined, only one MAb, IgG1b12, bound consistently to the isolates tested. This MAb (IgG1b12) bound to 22 of 26 isolates (85%) with moderate to strong binding (Fig. 1b). MAb 654-D was the weakest in binding and bound to only 4 of 26 isolates (15%) tested. Overall, the binding of anti-CD4bd MAbs was generally poor (Table 2). The binding patterns observed with the anti-CD4bd MAbs suggest that the epitopes to which MAbs 654-D, 448-D, 559/64, 588, and 1202 are directed are not exposed on the surfaces of most of the 26 viruses tested. The epitope to which MAb IgG1b12 is directed is better exposed and conserved on the surfaces of viruses in several clades.

(iv) Anti-C2 MAbs.

Overall, binding of the two anti-C2 MAbs (847-D and 1006-30D) to the 26 HIV-1 isolates was weak (Fig. 1b). Of the 52 MAb-virus test combinations, 24 (46%) were positive for binding, although they were weak (range, 12 to 43 pg/ml = 1.08 to 1.6 log units) (Table 2 and Fig. 1b). They bound to isolates of clades B, C, D, E, F, and G but not to isolates of clades A and H. Since only two MAbs to C2 exist and could be tested, the data reflecting the poor binding of virions to these MAbs must be taken as preliminary, suggesting that C2 is not well exposed on isolates of different clades but that this region may be conserved among some of the isolates of different clades.

(v) Anti-V2 MAbs.

Five anti-V2 MAbs were examined for their abilities to bind to 26 isolates, resulting in 130 MAb-virus test combinations. Overall, binding of these anti-V2 MAbs to the HIV-1 isolates was weak and sporadic (Fig. 1b). A total of 30 MAb-virus test combinations (23%) bound poorly (Table 2). However, five and one MAb-virus test combinations showed moderate and strong binding, respectively (Table 2). This category of MAbs bound most frequently to viruses of clades C and D. A total of 130 anti-V2/virus combinations were tested, and the results suggest that the epitopes to which these MAbs are directed are generally not well exposed on the surfaces of virions.

(vi) Anti-gp41 MAbs.

A total of 11 anti-gp41 MAbs (five to cluster I and six to cluster II) (Table 1) were used to study the exposure and conservation of gp41 epitopes on HIV-1 isolates belonging to different clades. Interestingly, the two categories of anti-gp41 MAbs demonstrated distinct patterns of binding. MAbs to cluster I bound strongly or moderately across all the HIV-1 clades examined, while MAbs to cluster II bound only moderately or poorly and sporadically to the isolates (Fig. 1c and Table 2).

Of the 130 anti-gp41 cluster I MAb-virus test combinations examined, 50% showed moderate to strong binding patterns (Table 2). In particular, MAb 246-D (to cluster I) bound well to all the isolates tested, while MAb 240-D also bound well to 24 of the 26 isolates (Fig. 1c). Indeed, MAbs 246-D and 240-D bound better than any other anti-HIV envelope MAbs tested, including those to the V3 and C5 regions of gp120. The other three cluster I MAbs (50–69, 1367, and 181-D) also bound the majority of isolates, although the binding was often weak to moderate. As in the case with the anti-C5 antibodies, this could be due to differences between the anti-cluster I MAbs in their requirements for conformational or linear structures.

Unlike the anti-gp41 cluster I MAbs, the majority of cluster II MAbs did not bind (47%) or bound only poorly (39%) to the isolates examined (Fig. 1c and Table 2). Of the 156 cluster II-virus combinations, only 20 (13%) and 2 (1%) combinations bound moderately and strongly, respectively (Table 2). Two of the MAbs (98-6 and 1342) failed to bind to most of the isolates. The data suggest that the cluster II region may be concealed in most virions, although binding of this class of MAbs was strongest to clade D viruses, as was the case for other classes of MAbs with marginal binding, such as anti-CD4bd and anti-V2 MAbs.

Determination of the ability of virions bound to various MAbs to infect human PBMCs.

Because models of the HIV-1 envelope structure indicate that the gp120 protein shields the gp41 molecule (34, 68, 70), it was intriguing that the anti-gp41 MAbs (246-D and 240-D) specific for the apex of gp41 bound to the isolates better than did all the other 45 anti-gp120 and anti-gp41 MAbs examined. To verify that these two anti-cluster I MAbs bound to infectious virions and that virus binding to these MAbs was not the result of exposure of gp41 due to gp120 shedding, we examined whether the virus that was bound to MAb 246-D on microtiter wells was infectious. In addition, we determined whether virus bound to anti-C5 MAb 1331A was infectious. As negative controls, anti-cluster II MAb 1342 and anti-parovirus MAb 860-55D were also tested. Thus, as described in Materials and Methods, after microtiter wells were coated with these MAbs, virus was added and unbound virus was washed off. PBMCs were then added to the microtiter wells and incubated in 5% CO₂ at 37°C for a 3-week period during which p24 antigen production was monitored weekly. Two viruses (CA5 [clade B] and 93BR019 [clade F]) were used in these experiments. As expected, both viruses bound to MAb 246-D and/or to MAb 1331A with levels ranging from 84 to 197 pg/ml but did not bind to MAbs 1342 and 860-55D (Fig. 1 and 2). When PBMCs were added to wells after unbound virus had been thoroughly washed away, infection of PBMCs was demonstrated by the presence of increasing p24 production over 3 weeks when the viruses bound to MAbs 246-D and 1331-A. In cultures where nonbinding MAbs (1342 and 860-55D) were used, no infection was detected (Fig. 2). These results indicate that the MAbs displaying virion binding do indeed bind to intact, native, infectious virions.

Role of epitope conformation in MAb-virus binding.

In our previous study, we examined the ability of anti-V3 MAbs to bind to HIV-1 isolates bearing the exact or altered versions of the core epitope of the MAbs (45). The results clearly indicated that binding of MAbs to virions did not require the presence of the exact core epitope in the viral sequence, suggesting that conformation plays a critical role in the binding of MAbs to the viruses. In the present study, we examined the presence of the exact sequence of the core epitopes in the C2, C5, and gp41 regions of several isolates and the ability of the MAbs to bind to the individual virus isolates (Table 3). The results again indicated that the exact sequence of the core epitope is not required for a MAb to bind to virus. Conversely, the presence of the core epitope does not always result in binding, although binding was noted in 81% of the cases (39 of 48) where the core epitope was present (Table 3). These results, in addition to the patterns of binding observed with MAbs to discontinuous epitopes (Table 1), suggest that both the conformation and exposure of an epitope play critical roles in the binding process.

The conservation of the cluster I epitopes of gp41 is again evident from Table 3, which shows that 10 of 13 of the viruses examined possessed one or more of the epitopes at the apex of this immunodominant region despite the assignment of these viruses to six clades. The binding data reveal that the two MAbs which target this region (MAbs 240-D and 246-D) are highly reactive with a broad array of viruses (Table 3 and Fig. 1c), with or without the exact epitope, suggesting again that there is a conserved conformation in this region and that this region is consistently exposed on intact virions.

Correlation between binding of MAbs to HIV-1 isolates and neutralization of the homologous isolates.

The ability of a MAb to bind a virus isolate and to neutralize the homologous isolate was examined. This is critical to identify immunogenic epitopes that are exposed and conserved on the surface of different viruses that induce antibodies that may be relevant to protection.

A total of eight MAbs were tested for their abilities to neutralize five viruses. These MAbs were selected to include five that bound well to different isolates and three that bound poorly, if at all, to these viruses. The ability of these MAbs to neutralize these viruses was tested with the MAbs at a concentration of 10 μg/ml (the same concentration used in the virus binding assay).

Overall, the majority of MAb-virus combinations tested did not display neutralization (Table 4). Only two of the eight MAbs (447-52D and IgG1b12) achieved 90% neutralization and only one additional MAb (908-D) achieved 50% neutralization at 10 μg of MAb/ml. In most cases, binding of a MAb to a virus isolate did not correlate with the ability of the antibody to neutralize the homologous strain. Of the 27 MAb-virus combinations that showed positive binding, only 7 (26%) were associated with the ability to achieve 50% neutralization of the homologous virus and only 4 (15%) achieved 90% neutralization. Conversely, however, if a MAb neutralized a virus, it always bound to that virus, and if a MAb did not bind to a virus, it did not neutralize it.