Access of Antibody Molecules to the Conserved Coreceptor Binding Site on Glycoprotein gp120 Is Sterically Restricted on Primary Human Immunodeficiency Virus Type 1

Reagents.

The following materials were obtained from the National Institute of Health AIDS Research and Reference Reagent Program (ARRRP): molecular clones of HIV-1 89.6, HxB2, JR-FL, JR-CSF, and ADA; sCD4 (amino acids 1 to 370; contributed by N. Schülke); and recombinant gp120_JR-FL and CD4-immunoglobulin G2 (IgG2; kindly provided by Paul Maddon and William Olson [Progenics, Tarrytown, N.Y.]).

Construction and purification of IgG1 X5.

Fab X5 was isolated from a phage display library constructed from the bone marrow of an HIV-1-seropositive donor as described previously (32). The DNA fragments encoding the heavy chain (Fd fragment) and light chain of X5 were transferred from the pComb3H phagemid vector (1) to the pDR12 mammalian expression vector (4) and transfected into Chinese hamster ovary (CHO) cells by using FuGENE6 transfection reagent (Roche, Indianapolis, Ind.). Stable transfected clones were selected under methionine sulfoximine (MSX; Sigma, St. Louis, Mo.) amplification. The clone with the highest production of IgG1 X5, as determined by enzyme-linked immunosorbent assay (ELISA) with cell culture supernatant, was chosen for scale-up in the CellCube System (Corning, Inc., Corning, N.Y.) and purified by affinity chromatography with protein A (Pharmacia, Uppsala, Sweden). Purified IgG1 X5 was concentrated and dialyzed against phosphate-buffered saline (PBS). Purity was determined by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis, and the concentration was measured by determining the A₂₈₀ value.

Production and purification of antibody fragments.

Antibody Fab fragments were produced by papain digest as described previously (24). Single chain Fv (scFv) X5 was engineered into the pComb3X vector by using antibody specific primers (1), produced in Escherichia coli and purified by nickel chelate chromatography (Qiagen, Valencia, Calif.) according to the manufacturer's instructions. Purified protein was dialyzed with PBS and stored at −80°C until use. Purity was determined by SDS-polyacrylamide gel electrophoresis, and the concentration was measured by determining the A₂₈₀ value. The 17b single-chain Fv plasmid was a gift from Wayne Marasco (Dana-Farber Cancer Institute, Boston, Mass.). In short, the 17b scFv was produced by PCR amplification of the heavy- and light-chain variable-region fragments from the 17b hybridoma cell line and cloned into the prokaryotic expression vector, pHEN (16). From this construct, the 17b heavy chain, the (G₄S)₅ linker, the light-chain variable sequences and a His₆ epitope tag were PCR amplified and subcloned into the inducible expression vector pMT (17) downstream of the Drosophila metallothionein promoter and in frame with the tissue plasminogen activator leader sequence. A stable S2 Drosophila cell line was established by cotransfection with a hygromycin resistance plasmid, pCo-hygro, and selection with hygromycin (300 μg/ml). Production of the 17b single chain was induced by the addition of 750 μM CuSO₄ to the cells at a density of 10⁷cells/ml in serum-free insect medium containing 0.1% pluronic (BASF, Mount Olive, N.J.). Purification of the single chain was performed in a single step by direct passage over a nickel chelating column (Pharmacia). The 33-kDa protein was eluted by 50 mM EDTA, dialyzed, and quantified by A₂₈₀ and SDS gel analysis.

Pseudovirion production.

Plasmids containing the env genes of HIV-1 strains HxB2 (35), JR-CSF (65), ADA (53), ADAΔV1V2 (21), and SOS-JRFL (2) were constructed as described previously. Similarly, the env genes of HIV-1 strains JR-FL, HxB2, and 89.6 were cloned into pSVIIIenv (14) and those of strains SF162, 92HT594, JR-CSF, JR-FL, and ADA were cloned into pSV7d (35). Additionally the env gene of amphotropic murine leukemia virus (A-MLV) was cloned into pSV7d (38). Recombinant pseudovirions were produced as described previously (7, 28, 38, 62, 65).

Neutralization assay.

Neutralization was measured in various luciferase reporter gene assays as described previously (2, 28, 38, 62, 65).

(i) Standard neutralization assay A.

A pseudovirus inoculum, previously determined to yield ∼1,000,000 relative light units, was incubated with a serial dilution of MAb for 1 h at 37°C and added to 2 × 10⁴ U87.CD4.CCR5 or U87.CD4.CXCR4 cells (ARRRP [H. Deng and D. Littman]). After 24 h of incubation, fresh medium was added, followed by incubation for 3 days (37°C, 5% CO₂). Luciferase activity was measured by using the luciferase assay system (Promega, Madison, Wis.) according to the manufacturer's instructions.

(ii) Standard neutralization assay B.

A pseudovirus inoculum was incubated with a serial dilution of MAb for 18 h at 37°C and added to U87.CD4 expressing both CCR5 and CXCR4 cells (ARRRP [H. Deng and D. Littman]). After 72 h of incubation (37°C, 5% CO₂), luciferase activity was measured by using the luciferase assay system (Promega) according to the manufacturer's instructions.

(iii) SOS neutralization assays.

Standard and postattachment neutralization assays using mutant SOS pseudovirions have been described previously (2). Briefly, for standard neutralization, a pseudovirus inoculum normalized for p24 content by ELISA was incubated with a serial dilution of MAb for 1 h at 37°C and added to 2 × 10⁴ U87.CD4.CCR5 cells for 2 h. Unbound virus was removed by changing the medium, and the culture was incubated for an additional 1 h. Alternatively, to measure postattachment neutralization, a pseudovirus inoculum was incubated with 2 × 10⁴ U87.CD4.CCR5 cells for 2 h to form an SOS-arrested intermediate. After replacement of the medium, a serial dilution of MAb was added for 1 h. Cells were subsequently treated with 5 mM dithiothreitol for 10 min to activate the fusion reaction, and the medium was replaced. After an additional 3-day incubation (37°C, 5% CO₂), luciferase activity was measured as described above.

(iv) Temperature-regulated neutralization assay.

A pseudovirus inoculum, previously determined to yield ∼10,000 reverse transcriptase (RT) units, was incubated with a serial dilution of MAb for 1 h at 37°C and added to 6 × 10³ Cf2Th cells expressing CD4 and CCR5 (22). Alternatively, the virus inoculum was incubated for 4 to 5 h at 4°C with the cells prior to washing. Serial dilutions of antibodies were added, and the temperature was raised to 37°C. After a 24-h incubation at 37°C (5% CO₂), fresh medium was added, and the cells were incubated for an additional 3 days. Luciferase activity was measured as described above.

ELISA.

Microtiter plates (Costar, Corning, N.Y.) were coated overnight at 4°C with 5 μg of anti-gp120 antibody D7324 (Aalto BioReagents, Ltd., Dublin, Ireland)/ml in PBS (pH 7.5). Plates were blocked with 3% bovine serum albumin for 1 h at room temperature and then washed with PBS-0.05% Tween 20 (PBST). Lysed pseudovirions (1% Empigen; Sigma) were added, diluted in PBST-1% bovine serum albumin, and incubated for 4 h at room temperature, in the presence or absence of sCD4 (2 μg/ml). Next, serial dilutions of MAb were incubated with the captured proteins, and bound MAb was detected with horseradish peroxidase-labeled goat anti-human IgG F(ab′)₂ (Pierce, Rockford, Ill.) and tetramethylbenzidine substrate (Bio-Rad). The color reaction was stopped after 20 min by the addition of 2 M H₂SO₄, and the absorbance at 450 nm was measured.

Modeling.

The positions of four-domain CD4 structures were generated by superimposing the four-domain CD4 structures (pdb accession numbers 1WIO, 1WIP, and 1WIQ, residues 1 to 363) (58) onto the trimeric model of the gp120-D1D2 complex (27) by using the main-chain atoms of CD4 D1D2 (residues 1 to 178). Root-mean-square (RMS) deviations for the superpositions are ∼1.0 Å for all superpositions. A model having an additional 35° rotation of D3D4 toward the target cell membrane was constructed by using the interactive graphics program “O” (19), by altering the backbone angle between CD4 residues 176 and 177 in 1WIP from ψ, φ values of −96 and 108° for residue 176 to values of −96° and 162°, respectively, and from ψ, φ values of −72° and 154° for residue 177 to values of −153 and 154°, respectively. The nine-amino-acid linker between the last crystallographically ordered residue and the transmembrane-spanning region contains two prolines. Its length was estimated by using calculations of polymer dimensions (48), where an Ala₉ polymer has an average N-terminal and C-terminal distance of 16 Å and a Pro₉ polymer a distance of 21 Å. All superpositions were carried out with the program LSQKAB, which is in the CCP4 suite of programs (6).

Antibody Fab and scFv fragments of CD4i MAbs are more effective in neutralization of HIV-1_JR-CSF than whole IgG1 antibodies.

We recently reported the isolation of a novel CD4i Fab fragment, designated X5, that differed from previously isolated CD4i antibodies, such as 17b or 48d, in that it potently neutralized an array of primary HIV-1 isolates (32). Whole-antibody molecules are generally more effective in neutralization than their corresponding Fab fragments (57). We therefore converted Fab X5 to a whole IgG1 molecule, expressed it in CHO cells and purified it by affinity chromatography. IgG X5 was tested against CCR5-dependent (R5) HIV-1_JR-CSF in a neutralization assay. Surprisingly, the whole antibody neutralized ∼5-fold less effectively than the Fab fragment (Fig. 1A and Table 1) .

To assess whether the observed phenomenon was a general characteristic of CD4i MAbs, we tested whole antibody and antibody Fab fragments of other CD4i MAbs in neutralization against HIV-1_JR-CSF. The Fab fragments of MAbs 17b and 48d were made by papain digestion. Both Fab fragments were more effective in neutralization compared to their corresponding whole antibody molecules, although the difference appeared less significant for 48d (Fig. 1B and C and Table 1). In addition, single-chain Fv (scFv) variants of X5 and 17b were constructed and tested for neutralization. The smaller scFvs were even more effective than the corresponding Fab fragments (Fig. 1A and B).

Lack of HIV-1_JR-CSF neutralization by IgG is not due to a loss of structural integrity of the IgG molecule.

To assess whether the inability of the whole antibody molecules to neutralize HIV-1_JR-CSF was caused by diminished recognition of gp120_JR-CSF, we compared the binding of IgG with Fab fragments to monomeric gp120_JR-CSF in ELISA in the presence or absence of sCD4. In contrast to the neutralization results, the IgG molecules of X5 and 17b had an approximately two- to threefold-higher apparent affinity, probably due to the bivalency of the whole IgG molecule (Fig. 2). Fab X5 obtained by papain digestion of IgG X5 had a similar apparent affinity to E. coli-expressed Fab X5 (data not shown). No differences in apparent binding affinities were observed between Fab and IgG 48d. All antibodies displayed an ∼10-fold increase in apparent affinity for gp120 in the presence of sCD4. Although affinity to monomeric gp120 does not correlate with neutralization, this finding does demonstrate that the lack of neutralizing activity was not due to a loss of structural integrity of the whole antibody molecule.

Antibody fragment size-dependent neutralization is a function of viral isolate.

To determine whether the observed phenomenon was virus isolate specific, we assessed a panel of HIV-1 isolates for sensitivity to CD4i antibody (fragment)-mediated neutralization (Table 1). HIV-1_JR-FL and HIV-1_ADA, both R5 primary isolates, showed a broadly similar tendency to be better neutralized by smaller antibody fragments, with the exceptions that the 17b Fab and scFv fragments were equally effective against HIV-1_JR-FL and that HIV-1_ADA was somewhat better neutralized by the 48d whole IgG than by the Fab fragment. HIV-1_ADA was overall somewhat less sensitive to CD4i MAb neutralization than was HIV-1_JR-CSF. The relatively neutralization-resistant isolate HIV-1_YU2 (29, 52) was weakly neutralized by scFv CD4i MAbs, but infectivity was enhanced by CD4i whole IgG1 molecules (data not shown).

In agreement with other studies (54, 60), the T-cell line-adapted, CXCR4-dependent (X4) HIV-1 strain HxB2 was relatively sensitive to all three tested CD4i MAbs as whole IgG molecules. In this case, the Fab fragments were generally somewhat less effective than whole IgG. The scFv fragment of X5 was more effective than the Fab fragment, whereas for 17b the two fragments were approximately equally effective (Table 1).

Two dualtropic (R5X4) primary isolates were evaluated, HIV-1_89.6 and a dualtropic variant of HIV-1_JR-CSF, designated HIV-1_JR-CSF/IR (P. Poignard et al., unpublished data). When HIV-1_89.6 neutralization by X5 was evaluated on CXCR4-expressing U87 cells, whole IgG and antibody fragments displayed similar potency (Fig. 3B and Table 1). For HIV-1_JR-CSF/IR evaluated on CXCR4-expressing U87 cells, X5 scFv was clearly more effective than the Fab fragment and the whole IgG1 molecule (Fig. 3D). Next, we evaluated X5-mediated neutralization of HIV-1_89.6 and HIV-1_JR-CSF/IR on CCR5-expressing U87 cells. For HIV-1_89.6, neutralization was achieved at somewhat lower antibody concentrations on CCR5-expressing U87 cells than on CXCR4-expressing cells, with whole IgG and Fab neutralization somewhat less effective than with scFv (Fig. 3A). When HIV-1_JR-CSF/IR neutralization by X5 was evaluated on CCR5-expressing U87 cells, whole IgG and antibody fragments displayed similar potency that was comparable to whole IgG and Fab neutralization on CXCR4-expressing U87 cells (Fig. 3C). Thus, the precise pattern of neutralization potency displayed by the CD4i MAb X5 and antibody fragments is dependent upon isolate and coreceptor usage.

In a separate study, with a similar reporter gene neutralization assay in a different laboratory, a panel of HIV-1 isolates was assessed for sensitivity to CD4i antibody (fragment)-mediated neutralization (Table 2). In these experiments, whole IgGs of CD4bs-specific MAbs were included for reference. In addition, pseudovirions expressing the A-MLV env gene were included as negative control. As for the experiments described in Table 1, the R5 primary isolates HIV-1_JR-FL, HIV-1_JR-CSF, and HIV-1_ADA, were sensitive to neutralization by both X5 and 17b scFv fragments and relatively resistant to Fab X5, IgG X5, and IgG 17b. Another R5 primary isolate, HIV-1_SF162, was sensitive to CD4i IgG and antibody fragments, but this isolate was also sensitive to IgG b6, a generally weakly neutralizing anti-CD4bs MAb. Primary dualtropic R5X4 isolate HIV-1_92HT594 was neutralized by MAb X5 and Fab and scFv antibody fragments and by 17b scFv fragment but was resistant to 17b whole IgG (Table 2).

CD4i antibody fragments neutralize HIV-1 subsequent to CD4 binding.

The CD4-induced conformational changes in gp120 include displacement of the V1/V2 variable loops, thus exposing the coreceptor binding site and the CD4i epitopes (31, 47, 60). To investigate the role of the V1/V2 variable loops in CD4i antibody- and antibody fragment-mediated neutralization, we compared the neutralizing activities of different 17b antibody forms against wild-type ADA (WT) (Fig. 4A) and V1V2-deleted ADA (ΔV1V2) (Fig. 4B). Removal of the V1/V2 loops resulted in increased sensitivity to neutralization by MAb 17b, as has been previously reported (5, 21). However, 17b scFv and Fab fragments neutralized the ΔV1V2 virus more efficiently than the whole antibody, suggesting that some steric constraints on MAb 17b neutralization operate even in the absence of the V1/V2 loops.

Membrane fusion mediated by the HIV-1 envelope glycoproteins is a temperature-dependent process (12, 13). To investigate the sequence of events for MAb 17b-mediated neutralization, we tested the ΔV1V2 virus in a temperature-dependent neutralization assay. Virus and target cells were preincubated at 4°C and washed, and then 17b scFv, Fab, or whole IgG was added. The fusion reaction was resumed when the temperature was raised to 37°C. Under these conditions, compared to the standard neutralization assay (Fig. 4B), the neutralizing efficiency of the scFv was only minimally changed or slightly improved, the efficiency of the Fab fragment was slightly reduced, and the whole IgG molecule was markedly less efficient (Fig. 4C). This suggests that the antibody fragments can neutralize virus after it has attached to the target cell, whereas the whole antibody molecule is much less effective in this context. In other words, it seems likely that neutralization is most efficiently mediated by CD4i antibody binding to its epitope after CD4 engagement and that this is sterically restricted for the whole antibody but not for the fragments.

Finally, we used an activatable fusion intermediate system to look further at the role of CD4i antibodies and fragments. Introduction of an intermolecular disulfide bond between gp120 and gp41 subunits has been shown to stabilize their association (designated SOS Env) (3). Incorporation of SOS Env into pseudovirions produces particles that bind to target cells but do not fuse (2). The addition of a reducing agent activates the fusion reaction. Initially, the neutralization of reduced SOS pseudovirions of HIV-1_JR-FL (SOS_JR-FL) was compared to wild-type HIV-1_JR-FL (WT_JR-FL) pseudovirions under standard assay conditions. With the exception of whole IgG 17b, all CD4i antibody fragments could neutralize both SOS and WT virus to a similar extent (Fig. 5A to D). In addition, whole IgG b12 and IgG 2F5, which were included as control antibodies, could efficiently neutralize both viruses under standard conditions (Fig. 5E and F). We next used SOS_JR-FL to examine the ability of CD4i antibody fragments to neutralize under postattachment conditions. SOS_JR-FL pseudovirions were incubated with cells, unbound virus was washed away, and antibody fragments were added to cells with adsorbed SOS_JR-FL pseudovirions. All CD4i antibody fragments could still efficiently neutralize SOS_JR-FL under these conditions, in contrast to whole IgG 17b. Whereas neutralization of SOS_JR-FL by whole IgG b12 was much weaker under postattachment conditions, neutralization by whole IgG 2F5 was comparable, a finding consistent with the concept that b12 neutralizes by inhibiting virus attachment to cells and that 2F5 neutralizes in a postattachment event (2, 37). This provides further evidence that smaller antibody fragments, but not whole antibody, can gain access to virions attached to target cells.

A previous study suggested that the CCR5 binding site is engaged in the SOS-arrested intermediate (2). However, we show here that, on the SOS-arrested intermediate, the CD4i antibody fragments have access to their epitope, which overlaps the coreceptor binding site. One possible explanation for this apparent paradox is that not all CCR5 sites are engaged in the bound SOS intermediate, at least until dithiothreitol is added, or that the association of virus with CCR5 can be displaced by high concentrations of CD4i antibodies.