Crosslinked HIV-1 envelope–CD4 receptor complexes elicit broadly cross-reactive neutralizing antibodies in rhesus macaques

Abstract

The identification of HIV envelope structures that generate broadly cross-reactive neutralizing antibodies is a major goal for HIV-vaccine development. In this study, we evaluated one such structure, expressed as either a gp120–CD4 or a gp140–CD4 complex, for its ability to elicit a neutralizing antibody response. In rhesus macaques, covalently crosslinked complexes of soluble human CD4 (shCD4) and HIV-1_IIIB envelope glycoproteins (gp120 or gp140) generated antibodies that neutralized a wide range of primary HIV-1 isolates regardless of the coreceptor usage or genetic subtype. Ig with cross-reactive neutralizing activity was recovered by affinity chromatography with a chimeric single-chain polypeptide containing sequences for HIV_BaL gp120 and a mimetic peptide that induces a CD4-triggered envelope structure. These results suggest that covalently crosslinked complexes of the HIV-1 surface envelope glycoprotein and CD4 elicit broadly neutralizing humoral responses that, in part, may be directed against a novel epitope(s) found on the HIV-1 envelope.

A safe and effective vaccine against HIV should elicit humoral responses that are effective against a broad spectrum of primary HIV strains (1). Unfortunately, an immunogen capable of generating such antibodies in humans remains elusive. The structural and antigenic characteristics of the free HIV envelope (Env) have been studied intensively in hopes of identifying a configuration that might serve as a subunit immunogen. However, these efforts have produced Env-based immunogens that typically fail to elicit antibodies capable of neutralizing more than a minor fraction of primary isolates (2–9).

The gp120–CD4 complex, which forms during virus attachment to cell surface receptor CD4 (10), has been considered as another potential target for broadly neutralizing antibodies. This complex is antigenically distinct from free gp120 and is essential for viral entry via 7-transmembrane domain (7-TM) chemokine receptors (10). Only a few studies have attempted to characterize immune sera raised by gp120–CD4 complexes for broadly cross-reactive neutralizing antibodies (11–14). These studies showed that broadly neutralizing antibody responses were generated in mice by immunization with mixtures of gp120 and CD4. However, these responses were directed, in part, against the interactive domains of CD4 and gp120 and were not qualitatively different from what are elicited by the uncomplexed antigens (11, 13). In contrast, a gp120–CD4 complex, stabilized by covalent crosslinking, elicited antibody responses in goats that neutralized primary HIV isolates (12). Such an antigen can be useful for characterizing the immunogenic characteristics of HIV Env–CD4 complexes. Accordingly, in this study we evaluated the immunogenicity of crosslinked gp120–CD4 complexes in rhesus macaques, a nonhuman primate commonly used as an animal model for developing strategies to elicit protective anti-HIV immune responses.

Materials and Methods

HIV Isolates.

Virus isolates were obtained from the following sources: IIIB, SF2, 89.6, BaL, and ADA, National Institutes of Health (NIH) AIDS Reagent Repository (Rockville, MD); 2005, 2044, and 2075 (15), Paul Clapham (Windeyer Institute, London); 451, CDC (Atlanta); 22069-06, 036979-07, 22069-05, 22069-03, and 30256 (16), Ruth Connor (Aaron Diamond AIDS Research Center, New York); CM245A, 92UG021W, CM235, DJ-258, ZAM-18, World Health Organization Network for HIV Isolation and Characterization (Geneva); and 168P, 168C (17), Jack Nunberg (University of Montana, Missoula, MT).

Antibodies and Reagents.

Murine anti-CD4 mAbs were obtained from the following sources: Q4120, Q4084, Q4116, and Q425 (18), Quentin Sattentau (Wright–Fleming Institute, London) and the National Institute for Biological Standards and Control Centralised Facility for AIDS Reagents (Herts, U.K.); Sim2 (19), NIH AIDS Reagent Repository; 13B8.2 (20), Coulter; 5A8 (21), Michael Fung (Tanox, Houston); Leu3a (22), Becton Dickenson; OKT4a (23), Ortho Diagnostics; D7324, a purified sheep anti-gp120 IgG, Cliniqa (Fallbrook, CA); soluble four-domain human (shCD4) and rhesus CD4s (24), Werner Meier (Biogen, Cambridge, MA); soluble two-domain CD4, James Arthos [National Institute of Allergy and Infectious Diseases (NIAID)]; biotinylated shCD4, Bartels (Issaquah, WA); Galanthus nivalis lectin (GNA) bound to Sepharose, Sigma. HEK-293, and U373-MAGI-CCR5 (25) cells were obtained from the NIH AIDS Reagent Repository. U373-MAGI-CXCR4cem cells (25) were obtained from Michael Emerman (Seattle).

Expression and Production of HIV Env Glycoproteins, Crosslinked gp120–CD4 Complexes, and SCBaL/M9.

The HIV-1_IIIB gp120 and gp140 were purified from the culture supernatant of chronically infected 6D5 cell line by immunoaffinity chromatography as described (26, 27). HIV-1_BaL gp120 was produced via transient transfection of HEK-293 cells and purified as described (28). Crosslinked HIV-1_IIIB Env–CD4 complexes were generated by incubating equimolar amounts of HIV_IIIB Env with shCD4 for 1 hr at 37°C. The Env–CD4 complexes then were chemically crosslinked with 0.5 mM bis(sulfosuccinimidyl)suberate (BS3, Sigma) as described (12). Complexes were examined by SDS/PAGE to verify complete crosslinking. To produce SCBaL/M9 from FLSC (29), primers encoding the CD4M9 sequence (30) were annealed and cloned into pEF6-FLSC R/T, which differs from FLSC by an arginine-to-threonine substitution at position 508 in gp120 (31). SCBaL/M9 was produced via transient transfection of HEK-293 cells with pEF6-SCBaL/M9 and purified as described (28).

Immunization of Macaques.

The HIV-1_IIIB gp120, gp140, and crosslinked HIV-1_IIIB gp120–shCD4, or HIV-1_IIIB gp140–shCD4 complexes (300 μg) were formulated in 100 μg of QS21 (Antigenics Biopharmaceuticals, Framingham, MA; ref. 32) and inoculated intramuscularly into rhesus macaques. Animals 492 and 493 received gp120–shCD4 complexes, and animals 497 and 498 were immunized with HIV-1_IIIB gp140–shCD4 complexes. Control animals were inoculated with either gp120 (animal 494) or gp140 (animal 495). All animals were inoculated on weeks 0, 4, 8, 12, 16, 20, 44, 94, 122, 178, and 250 with either complexed or noncomplexed Env glycoprotein formulated in QS21 adjuvant (32). Sera were collected on weeks 0, 4, 8, 12, 16, 22, 32, 46, 90, 96, 124, 180, 186, 252, and 254.

HIV-1 Neutralization Assays.

Sera obtained from immunized macaques were tested in six standardized neutralization assay formats routinely performed in four independent laboratories. Sera collected from naive animals were used as controls.

Format 1 (A.D. laboratory).

Anticomplex and control sera were tested in an assay system that uses U373/CD4/MAGI cells expressing either CCR5 or CXCR4 (25) as targets. The cells (10⁴ per well) were attached overnight to a flat-bottom 96-well tissue-culture well. Culture medium then was removed and replaced with 100 μl of fresh medium containing serial dilutions of sera and 50 TCID₅₀ (tissue culture 50% infective dose) per well of primary or laboratory-adapted viruses. After 24 hr, the cells were washed and maintained in fresh culture medium (5–7 days). Infectivity was measured by Galactostar chemiluminescent β-galactosidase assay (Applied Biosystems) according to manufacturer protocol.

Format 2 (R.P. laboratory).

Sera were tested in assays with human peripheral blood mononuclear cells (PBMCs) from HIV-seronegative donors as targets as described previously (12).

Format 3 (C.V.H. laboratory).

Serially diluted sera (starting at 1:10) were assayed by using characterized and cryopreserved PBMCs from HIV-seronegative donors as described previously (17, 33).

Format 4 (D.M. laboratory).

Serially diluted sera were assayed using phytohemagglutinin-activated PBMCs and the indicated primary subtype C R5 viruses as described previously (34).

Format 5 (D.M. laboratory).

A fixed dilution of serum was preincubated with 500 TCID₅₀ of the indicated primary subtype B R5 viruses for 1 hr and then added to phytohemagglutinin-activated PBMCs to achieve a final serum concentration of 1:12. The cells were washed 24 hr later and cultured in fresh medium. Infection was measured after 4 days. Neutralization was determined as the percentage reduction in viral infection versus control assays carried out in the absence of serum.

Format 6 (D.M. laboratory).

Sera were tested in assays with CD4⁻ BC7 (35) cells and the CD4-independent virus, HIV-1_IIIB8x. Sera were preincubated with 500 TCID₅₀ of HIV-1_IIIB8x for 1 hr and then added to BC7 cells. After 1 hr, the cells were washed and fresh medium was added. Infection was measured after 4 days (36).

HIV Env and CD4 ELISAs.

For direct ELISAs, HIV-1_IIIB gp140 was coated onto plastic as described (12). The HIV-1_IIIB gp120 Env capture ELISAs were carried out with native and denatured Env captured onto plastic assay wells by using polyclonal sheep antibody D7324 as described (37, 38). To create complexes, a saturating concentration (1 μg/ml) of shCD4 (Biogen) was added to captured native gp120. For the CD4 ELISA, soluble rhesus or human four-domain (D1D4) or two-domain (D1D2) antigens (Biogen) were adsorbed overnight to plastic at 1 μg/ml. A goat anti-human CD4 (GIBCO/BRL) antibody was used as a positive control for binding. For the CD4 capture ELISA, human D1D4 CD4 (5 μg/ml) was captured via either OKT4 or Q4120 adsorbed to the solid phase. In experiments where BaLgp120 was used as a competitor, test sera were preincubated with 100 μg/ml of BaLgp120 for 1 hr before addition to the plates. Bound antibodies were detected by horseradish peroxidase-conjugated goat or rabbit antibodies raised against the appropriate Ig (Kirkegaard & Perry Laboratories). Plates were developed by using a 3,3′,5,5′-tetramethyl benzidine peroxidase reagent (Kirkegaard & Perry Laboratories). For the CD4 capture ELISA, bound antibody was detected with horseradish peroxidase-labeled goat anti-human IgG F(ab)2 (Jackson ImmunoResearch).

Macaque Ig ELISA.

To detect macaque Ig, goat anti-human IgG(Fc) (Kirkegaard & Perry Laboratories) cross-reactive with macaque IgG was adsorbed to plastic. Serial dilutions of test fractions were added to the plate, and captured Ig detected with horseradish peroxidase-conjugated goat antiserum was raised against macaque IgG (Kirkegaard & Perry Laboratories). Standard curves were generated by using purified macaque IgG and used as a basis for quantitation.

Affinity Chromatography of Macaque Immune Sera.

Affinity columns were prepared by coupling 4 mg of purified SCBaL/M9, BaLgp120, or BSA (Sigma) to 2 ml of activated Sepharose 4B (Pierce) according to manufacturer protocol. Columns were equilibrated in PBS at room temperature before use. Typically, equal portions of the anti-Env–CD4 sera from the four animals were combined, filtered, and loaded for at least 1 hr at room temperature. The unbound (flow-through) fraction was collected, and the column was washed extensively with PBS. The column was eluted with 0.2 M glycine, pH 2.8, in ≈1-ml fractions, which then were supplemented with FBS to a final concentration of 10% and dialyzed overnight at 4°C against RPMI medium 1640.

Results

Immunization of Macaques with Crosslinked gp120–CD4 Complexes.

Based on our previous protocols (12), rhesus macaques were immunized with either free HIV Env or crosslinked complexes containing either gp120 or gp140 (Env–CD4 complexes). As expected, antibodies reactive with gp140 and Env–CD4 complexes were detected in immune sera from all animals; sera from the animals immunized with Env–CD4 complexes (492, 493, 497, and 498) also reacted with shCD4 (Fig. 3, which is published as supporting information on the PNAS web site, www.pnas.org). Titers against the antigens concurrently reached a peak after the fifth inoculation (bleed 5, week 22). Because the titers did not increase after a subsequent boost at week 44, the animals were rested for 48 weeks and then boosted at week 94. Serum samples (bleed 9) were collected 2 weeks later. These samples exhibited 2–12-fold higher ELISA titers compared with the bleed 5 samples and therefore were used to characterize the anticomplex antibody responses in greater detail. Additional sera were obtained as needed by boosting the animals at weeks 122, 178, and 250 and collecting samples 2 weeks later. All ELISA titers in these later sera were not significantly different from the bleed 9 samples (Fig. 3).

Neutralization of Primary Isolates by Anti-Env–CD4 Sera.

Bleed 9 sera were evaluated for HIV-1-specific neutralizing activity in six standardized assay formats routinely used in four independent laboratories (Table 1). One format used U373 MAGI cells expressing CD4 and either CXCR4 or CCR5 as target cells (25, 29), four used mitogen-activated PBMCs (12, 17, 33, 34), and one used a CD4-independent virus, IIIB8x (36), and a CD4-negative, CXCR4-positive cell line (35). As shown in Table 1, the anti-Env–CD4 sera exhibited a broad pattern of neutralization with four noteworthy characteristics. First, the sera effectively neutralized an array of primary subtype B viruses including X4, R5, and dual-tropic X4/R5 isolates, although in any given assay format the neutralization potencies of the anti-Env–CD4 sera varied among isolates. The neutralizing potencies of the anti-Env–CD4 sera were relatively consistent between the PBMC-based formats (compare CM235 in Table 1, Formats 3 and 4) but were higher when U373 cells were used as targets for infection. Specifically, the ID₉₀ values for HIV_BaL in the U373 MAGI cell-based assay (Table 1, Format 1) was 3-fold higher than the ID₅₀ values for the same isolates in a PBMC-based assay (Table 1, Format 2). Second, the anti-Env–CD4 sera neutralized primary viruses regardless of coreceptor usage and genetic subtype. Third, in PBMC-based assays the neutralizing activity of the anti-Env–CD4 sera seemed to be biased toward primary isolates. Three of the anti-Env–CD4 sera (492, 497, and 498) neutralized the primary isolate 168P but not its laboratory-adapted derivative, 168C (17). The opposite trend was seen with the anti-Env sera, which neutralized only the laboratory-adapted virus (Table 1). The anti-Env–CD4 sera were ineffective also against the laboratory-adapted HIV_IIIB strain in PBMC-based assays, even though its Env was used to construct the immunogen (Table 1). This failure to neutralize HIV_IIIB contrasted with the potent neutralization of this strain by sera raised against Env alone (Table 1). Unlike HIV_IIIB, however, the related CD4-independent virus, IIIB8x, was neutralized by anti-Env–CD4 sera in a CD4-negative cell line (Table 1, Format 6). The inherently greater sensitivity of CD4-independent strains to neutralization versus conventional HIV isolates (39) may explain this discordance. Fourth, there were no obvious distinctions between the neutralizing potencies of sera raised against Env–CD4 complexes prepared with gp120 versus gp140.

Table 1.

Neutralization of HIV-1 isolates by macaque anti-Env–CD4 antisera

Virus	Macaque no.
	492	493	494	495	497	498	Naïve
Format 1 (ID90)
Subtype B
IIIB (TCLA-X4)	20	10	100	100	20	20	<10
SF2 (TCLA-X4)	30	35	250	80	30	15	<10
451 (X4)	45	20	<10	<10	80	60	<10
2005 (X4)	30	10	<10	<10	60	200	<10
2044 (X4)	40	30	<10	<10	<10	40	<10
2075 (X4/R5)	80	30	<10	<10	<10	40	<10
89.6 (X4/R5)	20	15	<10	<10	30	30	<10
BaL (R5)	20	20	<10	<10	40	25	<10
ADA (R5)	12.5	12.5	<10	<10	25	15	<10
Format 2 (ID50)
Subtype B
IIIB (TCLA-X4)	<5	<5	>80	>80	<5	<5	<5
451 (X4)	27	>80	<5	<5	16	17	<5
22069-06 (R5/X4)	15	<5	<5	<5	12	9	<5
036979-07 (R5/X4)	10	46	<5	<5	9	8	<5
22069-05 (R5/X4)	<5	37	<5	<5	8	7	<5
22069-03 (R5)	42	12	<5	<5	28	65	<5
BaL (R5)	<5	19	<5	<5	8	6	<5
Subtype D
92UG021W (X4)	34	9	<5	<5	>40	8	<5
Subtype E
CM245A (X4)	24	>40	<5	<5	>40	8	<5
Format 3 (ID50)
Subtype A
DJ-258 (R5)	30	<10	<10	<10	15	11	<10
Subtype B
30256 (R5)	30	160	<10	<10	15	15	<10
Subtype C
ZAM-18 (R5)	22	<10	<10	<10	25	15	<10
Subtype E
CM235 (R5)	25	<10	<10	<10	40	40	<10
Format 4 (ID80)
Subtype B
168P	20	40	<12	<12	48	28	<12
168C	<12	32	172	168	<12	<12	<12
Subtype C
S080 (R5)	36	108	<12	<12	108	324	<12
S021 (R5)	12	36	<12	<12	36	36	<12
S009 (R5)	12	36	<12	<12	36	12	<12
S018 (R5)	<12	12	<12	<12	<12	12	<12
DU123 (R5)	108	36	<12	<12	108	108	<12
DU151 (R5)	36	12	<12	<12	36	12	<12
DU422 (R5)	108	36	<12	<12	12	<12	<12
Subtype E
CM235	20	44	<12	<12	28	24	<12
Format 5 (% reduction at 1:12)
P15 (R5)	98	96	0	0	98	98	11
PVO (R5)	99	99	31	63	99	99	37
H69172 (R5)	74	88	0	0	87	62	0
Format 6 (ID80)
IIIB(8x) (X4)	120	171	NT	NT	153	258	<12

Sera obtained from macaques immunized with the indicated immunogens were tested in six standardized neutralization assay formats (Materials and Methods). Naïve sera collected from unimmunized animals were tested as controls. The HIV_IIIB, HIV_SF2, and IIIIB(8x) were T cell line-adapted viruses and are indicated as TCLA. All the other viruses shown were passaged and titered only in primary human PBMCs and were designated accordingly as primary isolates. The values represent the reciprocal of the highest final serum dilutions interpolated from the dose-response curves as inhibiting 90 (ID₉₀), 80 (ID₈₀), or 50% (ID₅₀) of viral growth relative to control assays carried out in the absence of test serum as indicated. The > symbol indicates cases where neutralization values were greater than the highest dilution of serum tested. The < symbol indicates no neutralization at the highest concentration (noted) of serum tested. The averages of triplicate or quadruplicate assays are shown. 

Notably, the anti-Env–CD4 antisera failed to neutralize SHIV_89.6, SHIV_89.6P, and SHIV_KU2 in the human PBMC-based assays and SIV_mac239 in assays with either human or macaque PBMCs (data not shown). Thus, neutralizing activity was not universal for all primate lentiviruses or their chimeric derivatives.

Immunoreactivity of Anti-Env–CD4 Sera.

ELISAs revealed three noteworthy differences between the anti-Env–CD4 versus anti-Env responses. First, anti-Env–CD4 sera (492, 493, 497, and 498) were biased toward immunoreactivity with nondenatured versus denatured HIV_IIIB gp120 (Fig. 1A). In contrast, anti-Env sera (494 and 495) reacted similarly with both antigen forms. Second, anti-Env–CD4 sera did not inhibit the interaction of gp120 with shCD4 significantly (ID₅₀ values ≤ 1:20), whereas both anti-Env sera effectively blocked the CD4–gp120 interaction with ID₅₀ values of 1:200–1:700 (data not shown). Third, the anti-Env–CD4 sera were reactive with shCD4 but had a marked specificity for the D1D4 shCD4 versus D1D2 (Fig. 1B). Serum from animal 493 was not reactive with D1D2. None of the anti-Env–CD4 sera recognized soluble rhesus CD4 (D1D4) in ELISA (Fig. 1B) and did not bind to primary macaque CD4⁺ T cells in flow-cytometric assays (data not shown). In accordance, the CD4⁺ T cell counts of the Env–CD4-immunized and the Env-immunized animals remained equivalent throughout the study protocol (data not shown). To map the anti-human CD4 binding in greater detail, competition ELISAs were carried out with a variety of anti-CD4 mAbs with established HIV-1-neutralizing activity. Anti-Env–CD4 sera from macaques 493, 497, and 498 did not compete significantly with any of the mAbs even at low dilutions, although serum from animal 492 marginally interfered with all the mAbs to varying degrees (Table 2, which is published as supporting information on the PNAS web site). Notably, competition against Leu-3a may have been partially nonspecific, as it was similarly observed with the anti-Env sera (Table 2).

Figure 1.

Analyses of macaque immune sera by ELISA using HIV Env and CD4 proteins. Macaques were immunized and sera were collected as described in Materials and Methods. Serum samples designated bleed 9 were analyzed by ELISA with HIV_IIIB gp120–CD4 complexes (□), denatured HIV_IIIB gp120 (), or nondenatured HIV_IIIB gp120 (■) (A) or by ELISAs performed with D1D4-soluble rhesus CD4 (□), D1D2 shCD4 (), or D1D4 shCD4 (■) (B) as described in Materials and Methods. Serum from a naïve animal was tested as a control. Reciprocal end-point binding titers reflect serum dilutions giving absorbance values equal to background ± 3 standard deviations. Mean reciprocal end-point titers for duplicate assays are shown. The data are from a representative experiment repeated several times with similar results.

Attempts were made to assign the neutralizing specificity by using gp120 or shCD4 to adsorb cognate neutralizing antibodies in solution. However, each antigen abrogated the majority of neutralizing activity (data not shown) as did single-chain gp120–CD4 complexes (29) containing either human or rhesus CD4 sequences (data not shown). Because these experiments did not assign the neutralizing specificity clearly, we fractionated the neutralizing antibodies by affinity chromatography with a gp120-based antigen. To incorporate CD4-induced epitopes into the affinity matrix without introducing CD4 into the antigen, we constructed a single-chain chimeric complex (SCBaL/M9) containing a mimetic peptide (CD4M9) that was reported to interact with the CD4-binding site on gp120 (30, 40). As predicted, the purified SCBaL/M9 chimera (Fig. 4, which is published as supporting information on the PNAS web site) exhibited enhanced exposure of CD4-induced gp120 epitopes (Fig. 5, which is published as supporting information on the PNAS web site) and direct binding to CCR5 with efficiency equivalent to a single-chain complex containing CD4 D1D2 sequences (Fig. 6, which is published as supporting information on the PNAS web site). Notably, the SCBaL/M9 and the CD4M9 peptide alone were not reactive with antisera against shCD4 or mAbs against the CDR2 region of CD4 (data not shown). The CD4M9 peptide alone did not adsorb the neutralizing activity of anti-Env–CD4 sera in solution (data not shown), showing that the SCBaL/M9 does not react with the anti-CD4 activity in the neutralizing sera.

The neutralizing activity of pooled anti-Env–CD4 sera was depleted greatly after being passed over an SCBaL/M9 affinity column (Fig. 2A). The neutralizing titer (ID₅₀) of the pooled sera applied to the column was 1:262 against HIV₂₀₄₄, but only 1:8 in the unbound flow-through fraction, after correcting for dilution introduced by the chromatography procedure. Reactivity against HIV_BaL gp120 and SCBaL/M9 was also diminished extensively in the flow-through, whereas anti-CD4 reactivity remained virtually unchanged. As expected, the immunoreactive profile of anti-Env–CD4 sera passed over a mock-BSA column under identical conditions was essentially the same as the unabsorbed, pooled sera (Fig. 2A).

Figure 2.

Fractionation of neutralizing Ig by affinity chromatography on SCBaL/M9. (A) Sera from the four Env–CD4-immunized animals (492, 493, 497, and 498) were pooled and passed over a SCBaL/M9 affinity column or a mock column prepared with BSA. A flow-through (FT) fraction containing unbound material was collected from each column. Serial dilutions of the flow-throughs and serum pool were analyzed in parallel in ELISAs with HIV-1_BaL gp120 (□) and SCBaL/M9 (■) captured via sheep polyclonal anti-gp120 Ig (37, 38), in shCD4 ELISAs (), and in neutralization assays using U373 MAGI cells and a primary X4 virus, HIV-1₂₀₄₄ (). To compare the assay curves, the data obtained with the flow-through fractions were corrected for the dilution factor introduced by the chromatography procedure. The ELISA assay values represent the corrected reciprocal end-point serum dilution that produced an absorbance value equal to background ± 3 standard deviations. For neutralization assays, the assay value represents the corrected reciprocal serum dilution that produced 50% neutralization (ID₅₀) of macaque Ig in the sample. The percentage neutralization was calculated versus naïve macaque serum tested in parallel. Infection in wells with naïve sera showed no difference to those without macaque serum. All ELISAs were carried out in triplicate; the bars indicate standard deviation. Neutralization assays were carried out in triplicate or quadruplicate; the mean values from three neutralization experiments are shown. Bars indicate standard errors of the mean. (B) Ig eluted from the SCBaL/M9 affinity matrix was tested in the U373 MAGI cell assay (see Materials and Methods) by using the TCLA HIV-1_SF2 (○) and primary strains HIV-1_BaL (▵), HIV-1₂₀₄₄ (), HIV-1_89.6 (□), HIV-1_SF162 (⋄), HIV-1_92UG021 (■), HIV-1_22069–05 (▴), and HIV-1₂₀₇₅ (●). Matching serial concentrations of naïve macaque Ig were used as controls. Percentage neutralization by the SCBaL/M9 affinity-purified Ig was calculated relative to matching concentrations of naive Ig, which did not inhibit infection at any concentration tested. The average values obtained from triplicate assays are shown. (C) Ig recovered from either SCBaL/M9 (425 μg/ml) or HIV_BaL gp 120 (400 μg/ml) affinity matrices was analyzed in ELISAs with HIV_BaL gp120 (□) or SCBaL/M9 () captured via sheep polyclonal anti-gp120 Ig (37, 38) The Ig was also tested in ELISAs with shCD4 either directly adsorbed to the solid phase () or captured with mAb Q4120 (). Neutralizing activity was assessed in U373 MAGI cell assays with the primary X4 virus, HIV-1₂₀₄₄ (■). The ELISA assay values represent the reciprocal end-point concentration of macaque Ig (μg/ml) that produced an absorbance value equal to background ± 3 standard deviations. For neutralization assays, the assay value represents the concentration of macaque Ig (μg/ml) that produced 90% neutralization (IC₉₀). All ELISAs represent single experiments performed in duplicate, and neutralization assays represent single experiments in triplicate. Mean values of all assays are shown. Error bars indicate standard deviation in the ELISAs and standard errors of the mean in the neutralization assays.

Ig recovered from the SCBaL/M9 affinity column was tested against a variety of primary isolates and the TCLA HIV_SF-2. As shown in Fig. 2B, the Ig neutralized all the primary viruses in a concentration-dependent manner, although differences in sensitivity to neutralization were evident. Similar to the anti-Env–CD4 sera, the TCLA virus was neutralized less potently. In control experiments, the elution buffer, dialyzed and tested in parallel with the Ig, did not inhibit infection (data not shown). No neutralizing activity was recovered from a mock-BSA column eluted under identical conditions (data not shown). In later experiments, we found that Ig recovered from an affinity matrix constructed with free HIV_BaL gp120 also neutralized the heterologous primary HIV-1 isolate, 2044, with a potency that was equivalent to the Ig fractionated by the SCBaL/M9 column (Fig. 2C). Thus, a CD4-triggered structure was not necessarily required to capture the neutralizing antibodies.

As shown in Fig. 2C, the equal concentrations of Ig eluted from either a gp120 or SCBaL/M9 affinity column exhibited similar levels of immunoreactivity with HIV_BaL gp120 and SCBaL/M9. However, both Ig preparations also exhibited weak reactivity against shCD4 that was either adsorbed directly to the solid phase (Fig. 2C) or captured with mAb Q4120, which reacts with the gp120-binding domain (18) of CD4. Therefore, a limiting concentration of Ig was retested with the mAb Q4120-captured shCD4 ELISA in the presence of exogenous gp120, which was prevented from interacting with CD4 by the capture mAb. The apparent binding of SCBaL/M9 and gp120 affinity-purified Ig was reduced by 90 and 80%, respectively, in the presence of free gp120. Anti-CD4 antiserum used as control confirmed that the presence of free gp120 did not influence the amount of shCD4 that remained captured (data not shown). Thus, the apparent shCD4 binding in the ELISA was explained by promiscuous binding of anti-gp120 antibodies.

Discussion

It is predicted that CD4 binding might expose otherwise cryptic epitopes, or alter the immunogenicity of extant epitopes, on gp120 (11–14, 41). In either case, the end result could enhance the potential for gp120 to elicit broadly cross-reactive HIV-neutralizing antibodies. We explored these possibilities by immunizing macaques with covalently crosslinked complexes of HIV_IIIB gp120 or gp140 and shCD4. Although crosslinking can alter the immunogenicity of epitopes with chemically reactive moieties, it maintains the integrity of the complex structure.

Analyses of the anticomplex response revealed that the immunogenicity of both gp120 and CD4 is altered in the context of a preformed complex. The anti-Env–CD4 sera were strongly immunoreactive with human CD4 domains D3 and D4 (Fig. 1B) in contrast to immune responses elicited by free sCD4, which are typically without apparent bias for specific domains (18, 42). Poor responses against the N-terminal portion of CD4 were reflected also in the inability of the sera to prevent gp120–CD4 binding (data not shown). In the case of the HIV Env, the anti-Env–CD4 sera preferentially reacted with native rather than denatured antigen (Fig. 1A) in contrast with the anti-Env sera.

The most noteworthy aspect of the anti-Env–CD4 response was its broadly cross-reactive neutralization profile. The sera from all four complex-immunized macaques (492, 493, 497, and 498) neutralized a wide variety of primary HIV isolates irrespective of the coreceptor usage or genetic subtype of the target virus (Table 1). However, the anti-Env–CD4 sera were less effective against the TCLA strains, including the source of the immunogen (HIV_IIIB), which appeared to be highly resistant to neutralization in the PBMC-based assays. Notably, a similar bias toward the neutralization of primary viruses was observed in our previous studies using goat sera raised against crosslinked gp120–CD4 complexes (12). The anti-Env–CD4 sera also were unable to efficiently neutralize several SHIVs (data not shown) possibly because the characteristics of their Envs are similar to TCLA viruses (43). These results stand in stark contrast to the macaque responses raised against free HIV Env under identical conditions, which neutralized only closely related TCLA viruses.

The affinity chromatography experiments with either SCBaL/M9 or gp120 (Fig. 2) strongly indicated that the broadly neutralizing activity in the anti-Env–CD4 sera is primarily directed against the HIV Env. Neutralizing activity was adsorbed from the anti-Env–CD4 sera by an SCBaL/M9 or affinity matrix and was recovered with the eluted Ig (Fig. 2B). Notably, Ig recovered from an HIV-1_BaL gp120 column also neutralized the heterologous primary HIV₂₀₄₄ strain (Fig. 2C) as well as a primary isolate of HIV_BaL (data not shown). Other attempts to assign neutralizing specificity by treating sera with free gp120, SCBaL/M9, or shCD4 to adsorb reactive antibodies were difficult to interpret, because all antigens reversed the majority of neutralizing activity (data not shown). Possible reasons for such findings were that the soluble antigens formed complexes with cellular or viral antigens; that the sera were unable to neutralize shCD4-triggered gp120 structures; or that shCD4 collaborated with some anti-gp120 antibodies to abrogate binding of the neutralizing antibodies. In any case, specificity of the neutralizing antibodies for the HIV Env coincides with the overall serum neutralization pattern. The anti-Env–CD4 sera were most potent against primary versus TCLA isolates in PBMC-based assays. For example, the primary 168p isolate was susceptible to neutralization, whereas the TCLA derivative 168c was resistant (Table 1). Anti-CD4 mAbs, such as Leu3a and OKT4a, do not show such a distinction (23, 44–48). Further, anti-Env–CD4 sera failed to neutralize HIV_IIIB in the PBMC-based assays yet potently neutralized the CD4-independent derivative, IIIB8x, in a CD4-negative cell line (Table 1).

However, it remains possible that under certain conditions a CD4-reactive component of the anti-Env–CD4 response might contribute more extensively to the overall neutralizing activity against some isolates. In agreement, anti-CD4 antibodies were components of the neutralizing activity detected in anti-Env–CD4 sera raised in goats (12), although they were not primarily responsible for the selective action against primary isolates. In the case of the macaques, the presence of neutralizing anti-CD4 antibodies in the anti-Env–CD4 sera was suggested by competition ELISA (Table 2), which revealed that anti-Env–CD4 serum from one animal (492) marginally competed with a panel of neutralizing murine anti-CD4 mAbs. However, serum 492 had the same neutralization profile as the other three anti-Env–CD4 sera (Table 1), which did not compete significantly with the mAbs (Table 2). It is remotely possible that the sera contain other anti-CD4 antibodies directed against as-yet-undefined neutralizing epitopes on CD4 that are used selectively by primary versus TCLA isolates, although such a scenario is inconsistent with the understood usage of CD4 by TCLA and primary isolates (49, 50). Thus, it seems reasonable to posit that the contribution of anti-CD4 antibodies to the overall anti-Env–CD4 neutralizing response is minor and dispensable. Future analyses with a wider variety of complex forms and relevant animal models might address this possibility.

Overall, our findings suggest that gp120–CD4 complexes and their structural analogues might serve as models for HIV vaccine development. Because the major neutralization epitopes presented by the gp120–CD4 complexes seem to be located on the HIV Env, complex-based immunogens may elicit broadly cross-reactive antibodies in humans as well as macaques. An unanswered question is whether the neutralizing titers we elicited are likely to provide antiviral immunity, although it is encouraging that a recent study (51) indicates that neutralizing titers (inhibiting 50% of replicate cultures) as low as 1:38 are protective in a primate model. In any case, it now remains to be determined whether neutralizing anticomplex responses can be enhanced with different Env sequences, adjuvants, or immunization methods to maximize their efficacy.

Abbreviations

Env

envelope

shCD4

soluble human CD4

PBMC

peripheral blood mononuclear cell