Envelope Glycoprotein Binding to the Integrin α₄β₇ Is Not a General Property of Most HIV-1 Strains

ABSTRACT

The HIV-1 surface glycoprotein gp120 has been reported to bind and signal through α₄β₇ by means of a tripeptide motif in the V2 loop that mimics structures present in the natural ligands for α₄β₇, suggesting that α₄β₇ may facilitate HIV-1 infection of CD4⁺ T cells in the gut. Furthermore, immune correlates in the RV144 vaccine efficacy trial generated the hypothesis that V1V2 antibodies to an epitope near the putative α₄β₇ binding motif may play a role in protection against HIV-1 infection. In the interest of developing an assay to detect antibodies that block gp120 binding to α₄β₇, we used retinoic acid (RA)-activated human peripheral blood mononuclear cells (PBMCs) and transfected HEK293T (293T) cells expressing the integrin complex to study the α₄β₇ binding properties of 16 HIV-1 envelope glycoproteins. The natural ligand for α₄β₇, mucosal addressin cell adhesion molecule-1 (MAdCAM-1), bound efficiently to RA-activated PBMCs and transfected 293T cells, and this binding was blocked by antibodies to α₄. gp120 from multiple HIV-1 subtypes bound to RA-activated PBMCs from three donors in a CD4-dependent manner, but little or no α₄β₇ binding was detected. Similarly, little or no binding to α₄β₇ on transfected 293T cells was detected with multiple gp120s and gp140s, including gp120s from transmitted/founder strains, or when gp120 was produced in CHO, 293T, and 293S/GnT1^−/− cells. Finally, we found no evidence that infectious HIV-1 virions produced in either PBMCs or 293T cells could bind α₄β₇ on transfected 293T cells. Infectious HIV-1 virions and most gp120s/gp140s appear to be poor ligands for the α₄β₇ integrin complex under the conditions tested here.

IMPORTANCE Certain HIV-1 gp120 envelope glycoproteins have been shown to bind the gut-homing receptor α₄β₇, and it has been suggested that this binding facilitates mucosal transmission and virus replication in the gut mucosa. Additional evidence has generated the hypothesis that antibodies that bind near the putative α₄β₇ binding motif in the V2 loop of gp120, possibly disrupting gp120-α₄β₇ binding, may be important for HIV-1 vaccines. Our evidence indicates that infectious HIV-1 virions and many gp120s lack detectable α₄β₇ binding activity, suggesting that this homing receptor may play a limited role in direct HIV-1 infection of cells.

INTRODUCTION

Integrins are cell receptors that play important immunomodulatory functions, such as cell adhesion, cellular trafficking, immune responses, as well as control of tumor growth and metastasis (1). These integrin receptors are composed of α and β subunits, and their surface expression plays a key role in the migration of cells to different tissues (2). The α₄ subunit is expressed on T and B lymphocytes, monocytes, natural killer cells, and dendritic cells, where it can associate with β₁ or β₇ subunits (2, 3). The α₄β₇ heterodimer acts as a gut-homing receptor, mediating lymphocyte migration to the intestinal mucosa through interaction with the mucosal addressin cell adhesion molecule-1 (MAdCAM-1), which is predominantly expressed on venules in the gut-associated lymphoid tissue (GALT) and intestinal lamina propria (4–6).

Following mucosal exposure, human immunodeficiency virus type 1 (HIV-1) replicates at low levels due to the insufficient numbers of CD4⁺ T cell targets to infect. Once the virus reaches the GALT, where large amounts of activated CD4⁺ T cells are present, a high level of replication takes place, resulting in immune dysfunction and the massive depletion of CD4⁺ T cells during acute infection (7–12).

The HIV-1 surface gp120 envelope glycoprotein has been reported to bind and signal through the α₄β₇ integrin complex. This interaction is mediated by a tripeptide motif in the V2 loop of gp120 that mimics the binding motif of the natural ligands for α₄β₇ (13). Although binding of α₄β₇ to gp120 is not a prerequisite for HIV-1 entry, it has been suggested that strong α₄β₇ reactivity may provide an increased fitness for mucosal transmission (13, 14). Additional evidence has shown that scaffold proteins containing the HIV-1 V1V2 loop can block α₄β₇-gp120 binding (15). These studies also showed that gp120 binding to α₄β₇ results in the rapid activation of LFA-1, the central integrin that mediates the formation of virological synapses, an event that could also facilitate HIV-1 cell-to-cell transmission (16). These findings have provided a plausible explanation for the massive infection of CD4⁺ T cells in the gut. Consistent with this hypothesis are the results of studies showing that the loss of β₇-expressing CD4⁺ T cells in blood closely parallels the loss of CD4⁺ T cells in the intestine of rhesus macaques after simian immunodeficiency virus (SIV) infection (12). Moreover, administration of an α₄β₇ monoclonal antibody (MAb) to rhesus macaques just prior to and during acute SIV infection resulted in a significant decrease in the plasma and gastrointestinal tissue viral load and a marked reduction in the gut tissue proviral DNA load compared with the loads in control SIV-infected animals (17).

The finding that antibodies to the V1V2 region of gp120 inversely correlated with infection risk in the RV144 HIV-1 vaccine efficacy trial (18) has generated the hypothesis that these antibodies may have contributed to protection against HIV-1 infection. A plausible mechanism of protection by plasma anti-V2 antibodies does not appear to be dependent on their neutralizing activity (19). An alternative possible mechanism is that V2 antibodies block virus binding to α₄β₇ integrin in a way that impedes the establishment of chronic infection (20).

Notably, only a small group of gp120s has been shown to possess measurable α₄β₇ binding properties. In addition, removal of potential N-linked glycosylation sites in V1V2 or the use of proteins expressed in glycosylation-deficient mutant cell lines is often required to demonstrate an interaction of α₄β₇ with gp120s (14, 21, 22). More recently, the infectivity of transmitted/founder (T/F) HIV-1 subtype C viruses was found to be unaffected by anti-α₄β₇ antibodies, suggesting that the α₄β₇ binding properties of gp120 seen in vitro are not seen with native Env trimers on infectious viral particles (23). In addition, two independent studies have shown that non-gut-homing resting memory CD4⁺ T cells (integrin β₇ negative) are preferential targets for productive HIV infection (24, 25).

Motivated by these observations and in an attempt to develop an assay to monitor vaccine-elicited antibodies that block α₄β₇-gp120 binding, we used retinoic acid (RA)-treated human peripheral blood mononuclear cells (PBMCs) as well as CD4-negative HEK293T (293T) cells transiently expressing this integrin complex to determine the optimal conditions for gp120 binding to α₄β₇. Our results suggest that α₄β₇ binding is not a general property of most gp120 proteins.

MATERIALS AND METHODS

Cells.

PBMCs from three different healthy HIV-1-naive donors were obtained from the local Red Cross and were purified by Ficoll-Hypaque separation from buffy coats. They were selected for these studies on the basis of the quantity of the cryopreserved stocks and susceptibility to HIV infection. After thawing, the PBMCs were cultured in RPMI containing 20% heat-inactivated fetal bovine serum (FBS) containing 5% interleukin-2 (IL-2; Advanced Biotechnologies, Inc.) and 5 μg/ml phytohemagglutinin (PHA; Sigma) for 24 h. After this period, PBMCs were cultured in RPMI containing 20% FBS and 5% IL-2 without PHA. 293T cells were obtained from the American Type Culture Collection (ATCC) and maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% heat-inactivated FBS, 25 mM HEPES, and 50 μg/ml of gentamicin (Sigma).

Antibodies.

Allophycocyanin (APC)-α₄ and phycoerythrin (PE)-β₇ conjugated MAbs were obtained from BD Pharmingen. Purified CD49d (α₄) MAbs, clones HP2/1 and L25, were obtained from Beckman Coulter and BD Pharmingen, respectively. A recombinant human MAdCAM-1 Fc chimera was obtained from R&D systems. The CD4 MAb SK3¹³ was obtained from BD Pharmingen.

Source of HIV-1 envelope glycoproteins.

gp120s from the subtype B isolate SF162 (SF162gp120) and the subtype C isolate TV-1 (TV-1gp120) were produced in Chinese hamster ovary (CHO) cells by Novartis. HIV-1 gp120s from strains 92UG037 (subtype A), BaL (subtype B), 92US715.6 (subtype B), and 96ZM651 (subtype C) were produced in CHO cells by ABL Inc. HIV-1 AN-1gp120 protein (subtype B) was provided by J. Arthos and was produced in cells of the CHO lec1.1^−/− cell line, a CHO cell derivative that lacks N-acetylglucosamine glycosyltransferase activity (14, 26). HIV-1 gp140 proteins from strains CN54 (subtype C) and UG37 (subtype A) were made in CHO cells by Polymun Scientific Inc. The gp140 glycoproteins from the T/F strain C.1086 (subtype C), Q23.17 (subtype A), and a consensus sequence of group M Env (M-CONS) were made in CHO cells. The gp120s for SF162 (subtype B) and Q23.17 (subtype A) were produced in CHO, 293T, and 293S/GnT1^−/− cells, all of which were prepared at the Duke Human Vaccine Institute, Duke University Medical Center.

T/F HIV-1 Env proteins.

Genes for four HIV-1 T/F gp120s, including three subtype B proteins (B.040, B.65321, and B.6240) and one subtype C protein (C.1086) obtained from subjects acutely infected with HIV-1 by single-genome amplification, were described previously (27, 28). All gp120 genes were codon optimized by converting amino acid sequences to nucleotide sequences by employing the codon usage of highly expressed human housekeeping genes (29) synthesized de novo for expression as gp120 with deletion of either 7 (C.1086Δ7gp120) or 11 (B.040Δ11gp120, B.65321Δ11gp120, and B.6240Δ11gp120) amino acid residues at the N terminus of the mature HIV-1 Env proteins (30, 31). These genes were then cloned into a mammalian expression plasmid, pcDNA3.1/hygromycin (Invitrogen, Grand Island, NY). The recombinant Env glycoproteins were purified from the supernatants of CHO cell cultures transfected with the HIV-1 gp120-expressing pcDNA3.1 plasmids by using DEAE–ion-exchange resins and Galanthus nivalis lectin-agarose (Vector Laboratories, Burlingame, CA) column chromatography and stored at −80°C until use.

Transient expression of α₄β₇ integrins in 293T cells.

HEK293T cells (grown to 80% confluence in a 75-cm² flask) were transfected with 6 μg each of the cDNA expression clones for the α₄ and β₇ human integrin subunits (Origene Technologies, Inc.) using the Fugene reagent (Promega). Prior to use for gp120 binding and infectious virion capture studies, cells were assessed for α₄β₇ surface expression by flow cytometry at 48 h posttransfection, using APC- or PE-conjugated integrin-specific MAbs.

Biotin labeling of proteins.

HIV-1 gp120 and gp140 envelope glycoproteins and the recombinant human MAdCAM-1 Fc chimera proteins were labeled with biotin using a Biotin-XX microscale protein labeling kit (Molecular Probes/Invitrogen). In brief, 50 μg of protein was labeled with Biotin-XX sulfosuccinimidyl ester in a 50-μl reaction mixture for 15 min at 20°C, according to the instructions provided by the manufacturer. Unbound biotin was removed from the sample using a spin filter packed with Bio-Gel P-6.

Assay to measure α₄β₇ binding to gp120.

PBMCs were cultured for 7 days in RPMI containing 20% heat-inactivated FBS, 5% IL-2 (Advanced Biotechnologies, Inc.), and 10 nM RA (all-trans; Sigma). Expression of α₄β₇ integrins was monitored by flow cytometry with integrin-specific MAbs. Similarly, 293T cells transiently transfected with α₄β₇ cDNAs were removed from flasks using a nonenzymatic cell dissociation solution (Millipore) and tested for integrin expression by flow cytometry. The binding assay was performed as previously described by Arthos et al. (13). Cells (2 × 10⁶) were washed and resuspended in 200 μl of binding buffer (HEPES-buffered saline supplemented with 100 μM Ca²⁺, 1 mM Mn²⁺, and 0.5% bovine serum albumin). For blocking experiments, cells were preincubated for 10 min at 4°C with saturating concentrations of either anti-α₄ MAbs, unlabeled MAdCAM-1, or a combination of them. Saturating concentrations were predetermined by binding studies using serial dilutions of each of the MAbs. Biotinylated gp120 and MAdCAM-1 (400 ng) were added, and the mixture was incubated at 4°C for 15 min. Cells were washed and incubated with PE-streptavidin (BD Pharmingen) in binding buffer at 4°C for 10 min. After two final washings, cells were fixed in 1% formaldehyde–phosphate-buffered saline and analyzed by flow cytometry using a BD FACSCalibur analyzer (BD Biosciences). Between 12,000 and 100,000 live events were acquired per sample. Compensation and analysis were performed by using FlowJo software (Tree Star).

Infectious virions binding to α₄β₇-expressing 293T cells.

293T cells transfected with the cDNA expression vectors for α₄ and β₇ integrins were seeded at 50,000 cells per well in 96-well culture plates at 1 day posttransfection and incubated overnight at 37°C. The cells were then incubated with serial dilutions of infectious molecular clones of HIV-1, clones R2184c4.IMC.LucR (subtype CRF01_AE) and WEAU3-3.IMC.LucR (subtype B) (32), for 1 h at 37°C. Control wells containing nontransfected 293T cells were treated in the same manner. The cells were gently washed three times with culture medium, overlaid with10,000 TZM-bl cells per well, and incubated for 48 h at 37°C. Infection of TZM-bl cells was measured with a Britelite luminescence reporter gene assay system (PerkinElmer Life Sciences).

RESULTS

Retinoic acid upregulates the α₄β₇ integrin complex on human PBMCs.

Dendritic cells from mesenteric lymph nodes and Peyer's patches produce the vitamin A metabolite retinoic acid, which enhances the expression of α₄β₇ and CCR9 on T cells and imprints them with gut tropism (33). To study the α₄β₇ reactivity of different HIV-1 gp120s, human PBMCs from three donors were cultured in RA for 7 days. After this treatment, the level of α₄β₇ expression was determined by flow cytometry using APC- or PE-conjugated MAbs specific for the human integrins. As shown in Fig. 1A, untreated PBMCs from the three donors expressed different levels of both integrins. After culturing in RA, the levels of α₄ remained mostly unchanged, but the level of β₇ was increased for all three donors, particularly donors 2 and 3. Using Quantibrite PE beads, we estimated that for donor 3, the number of β₇ molecules per cell (inferred with the β₇-PE MAb) increased from 590 in untreated cells to 3,367 in RA-treated cells. Next, we tested the binding of the natural ligand for α₄β₇, MAdCAM-1, in the absence and presence of two anti-α₄ antibodies, CD49d (anti-VLA-α₄) clone L25 and CD49d clone HP2/1, which have been reported to block the gp120-α₄β₇ interaction (13). RA-treated PBMCs from donor 3 readily bound biotinylated MAdCAM-1, and this binding was blocked to near background levels by either anti-α₄ MAb at a concentration of 100 μg/ml (Fig. 1B).

FIG 1.

RA upregulation of α₄β₇ integrins on three donor human PBMCs and MAdCAM-1 binding. (A) Expression of α₄β₇ integrins on PBMCs from three donors (D1, D2, and D3) after RA treatment was analyzed by flow cytometry. Cells were cultured with and without RA for 7 days in medium containing IL-2 and PHA. Cells were stained with an APC-conjugated anti-α₄ MAb or PE-conjugated anti-β₇ MAb, washed, and fixed with formaldehyde before analysis with a BD FACSCalibur analyzer. The mean fluorescence intensity (MFI) obtained from the analysis of the data with FlowJo software is shown. α₄ and β₇, the anti-human integrin MAbs used. (B) Binding of MAdCAM-1, the natural ligand for the α₄β₇ integrin complex, was measured by flow cytometry using RA-treated PBMCs from donor 3 incubated with a biotinylated recombinant human MAdCAM-1 Fc chimera and PE-avidin at 4°C. In parallel assays, RA-treated PBMCs were preincubated with the anti-α₄ MAbs HP2/1 and L25 prior to the addition of MAdCAM-1. Black bars, mean fluorescence intensity obtained by analysis with FlowJo software.

CD4-dependent binding of gp120 to RA-treated human PBMCs.

To establish the integrity of our experimental system, we first aimed to demonstrate gp120 binding to CD4 on the surface of the donor PBMCs to be used in subsequent experiments. As shown in Fig. 2, biotin-labeled SF162gp120 and TV-1gp120 both bound to PBMCs from all three donors with the same magnitude, despite different levels of α₄β₇, and this binding could be completely blocked by the CD4 MAb Leu3a.

FIG 2.

CD4-specific binding of gp120 to RA-treated human PBMCs. Biotinylated gp120 from HIV-1 isolates SF162 (subtype B) and TV-1 (subtype C) was incubated with RA-treated PBMCs at 4°C for 15 min in HEPES buffer containing Ca²⁺ and Mn²⁺. Bound gp120 was detected with PE-avidin and flow cytometry analysis. The binding of gp120 to RA-treated PBMCs preincubated with a saturating concentration (25 μg/ml) of the mouse anti-human CD4 MAb Leu3a was also measured in this assay. Assays were performed with PBMCs from three donors. Black curves in the histograms, unstained PBMCs; gray curves, PBMCs incubated with avidin-PE only; red curves, PBMCs incubated with biotinylated gp120 from the indicated HIV-1 isolate.

Attempts to detect α₄β₇-specific binding of SF162gp120 and TV-1gp120 to RA-treated human PBMCs.

It was previously reported that gp120 binding to α₄β₇ can be blocked with anti-α₄ MAbs HP2/1 and L25 (13). We tested whether these MAbs and the α₄β₇ natural ligand MAdCAM-1 would block the binding of SF162gp120 and TV-1gp120 to donor 3 PBMCs. As shown in Fig. 3A and B, Leu3a clearly competed with both gp120s for binding to PBMCs. Marginal inhibition of SF162gp120 binding was seen in the presence of MAdCAM-1 (5% inhibition) and a combination of MAdCAM-1 plus L25 (27% inhibition), whereas no inhibition was seen in the presence of either HP2/1 or L25. We saw no decrease in TV-1gp120 binding in the presence of HP2/1, L25, MAdCAM-1, or combinations of these potential blocking agents.

FIG 3.

Lack of α₄β₇-specific binding of gp120 from two different HIV-1 isolates and effect of subsaturating concentrations of Leu3a. (A and B) Binding of biotinylated SF162gp120 (A) and TV-1gp120 (B) to RA-treated PBMCs from donor 3 with and without preincubation with either unlabeled MAdCAM-1, the anti-α₄ MAbs HP2/1 and L25, or the anti-CD4 MAb Leu3a at a saturating concentration or combinations of these reagents. (C) Titration of Leu3a for suboptimal inhibition of gp120 binding to donor 3 RA-treated PBMCs. (D) Effect of subsaturating concentrations of Leu3a on the binding of SF162gp120 or TV-1gp120 to donor 3 RA-treated PBMCs in the presence and absence of either HP2/1, L25, or MAdCAM-1. A subsaturating concentration of 0.08 μg/ml of Leu3a was chosen for both gp120s. Binding was assessed by flow cytometry. The mean fluorescence intensity obtained for each reagent and combinations is shown. Black bars, SF162gp120; gray bars, TV-1gp120. The same avidin-PE control was used for both gp120s in panels C and D (gray bar with a black border).

Although the binding of gp120 to α₄β₇ is CD4 independent, previous studies have demonstrated a close association of CD4 and α₄β₇ on the surface of T cells (34). To investigate the possibility that gp120 engages α₄β₇ and CD4 in a cooperative manner, we tested the α₄β₇ reactivity of the gp120s with RA-treated PBMCs preincubated with a subsaturating concentration of the anti-CD4 MAb Leu3a. As shown in Fig. 3C, a subsaturating concentration of Leu3a that partially blocked the binding of SF162gp120 and TV-1gp120 to PBMCs was identified. Using a combination of a subsaturating concentration of Leu3a and saturating amounts of α₄ MAbs or MAdCAM-1, the binding of SF162gp120 was reduced either 6% (MAdCAM-1 plus Leu3a) or 16% (α₄ MAbs plus Leu3a) relative to the level of binding seen in the presence of the subsaturating concentration of Leu3a alone (Fig. 3D). Similarly, TV-1gp120 binding was reduced either 10% (Leu3a plus L25) or 17% (Leu3a plus MAdCAM-1) relative to the level of binding seen in the presence of the subsaturating concentration of Leu3a alone. These results again indicate that binding of these gp120s to the integrin complex is negligible.

Binding of gp120s from multiple HIV-1 isolates to RA-cultured PBMCs.

Because we found very little binding of SF162gp120 to α₄β₇, we extended our studies to include an additional panel of gp120s from HIV-1 subtypes A, B, and C. As a positive control for α₄β₇ binding, we included the deduced ancestral AN-1gp120 (26) that had previously been shown to bind α₄β₇ (13, 14). With the exception of AN-1gp120, all Env glycoproteins exhibited clear CD4-dependent binding to RA-treated PBMCs; this was demonstrated by specific inhibition of binding to near background levels by Leu3a (Fig. 4). Little or no additional blocking of binding was seen with combinations of the anti-α₄ MAb L25 or HP2/1 and Leu3a. Binding of AN-1gp120 to the surface of these PBMCs was blocked very poorly by Leu3a, suggesting that binding was mostly CD4 independent and presumably due in part to α₄β₇ binding. The integrity of α₄β₇ expression on these cells was confirmed by showing strong binding of MAdCAM-1 and the inhibition of this binding by HP2/1.

FIG 4.

CD4-dependent binding of gp120 from multiple HIV-1 subtypes to human PBMCs. Biotinylated gp120s from HIV-1 subtypes A, B, and C were tested for binding to RA-treated PBMCs by flow cytometry. In addition, PBMCs preincubated with a saturating concentration of Leu3a prior to the addition of biotinylated gp120 were tested in parallel. Biotinylated MAdCAM-1 was used as a positive control to confirm the presence of the α₄β₇ integrins on the PBMCs. Black curves in the histograms, unstained PBMCs; gray curves, PBMCs incubated with avidin-PE only; red curves, PBMCs incubated with gp120 from HIV-1 isolates 92US715.6 (subtype B), 92UG37 (subtype A), BaL (subtype B), AN-1 (deduced ancestral subtype), SF162 (subtype B), and 96ZM651 (subtype C), in which binding was detected with avidin-PE. The results also correspond to those for PBMCs incubated with biotinylated MAdCAM-1, in which binding was detected with avidin-PE.

AN-1gp120 binding to α₄β₇ on RA-treated PBMCs.

We next tested the ability of MAdCAM-1 and the anti-α₄ MAbs L25 and HP2/1 to block the binding of biotinylated AN-1gp120 to RA-treated PBMCs. As shown in Fig. 5, AN-1gp120 bound to PBMCs weakly, and this weak binding could be partially blocked by all three reagents and was blocked to an extent greater than that seen with Leu3a, indicating that this gp120 binds α₄β₇, as previously reported (13, 14). Notably, a substantial amount of binding could not be blocked by any of these reagents, suggesting that a certain level of binding to unknown surface constituents occurs. As before, the integrity of α₄β₇ expression was confirmed by showing nearly complete inhibition of MAdCAM-1 binding to PBMCs in the presence of HP2/1.

FIG 5.

Weak inhibition of AN-1gp120 binding to RA-treated PBMCs in the presence of anti-α₄ MAbs and MAdCAM-1. Binding of biotinylated AN-1gp120 to RA-treated human PBMCs preincubated with either Leu3a, L25, HP2/1, or MAdCAM-1 was tested by flow cytometry. Binding of biotinylated MAdCAM-1 to the same PBMCs preincubated with HP2/1 was used as a control. Black bars, the mean fluorescence intensity obtained by analysis with FlowJo software for each reagent.

Testing of gp120 binding to 293T cells expressing high levels of α₄β₇ integrins.

To study gp120 binding to α₄β₇ in the absence of CD4, we cotransfected cDNA expression vectors for the human α₄ and β₇ integrin subunits into 293T cells. As shown in Fig. 6, high levels of each of the α₄β₇ integrin subunits were detected on the surface of 293T cells by flow cytometry at 48 h posttransfection. Furthermore, these cells readily bound biotinylated MAdCAM-1, and this binding was blocked by pretreatment of the cells with the anti-α₄ MAbs HP2/1 and L25, confirming the integrity of the α₄β₇ complex. Despite the high levels of α₄β₇ expression, no significant binding of biotinylated SF162gp120 could be detected.

FIG 6.

Binding of SF162gp120 and MAdCAM-1 to α₄β₇ expressed on 293T cells. 293T cells were transfected with expression vectors for the cDNA of the human α₄ and β₇ integrins. The cell surface expression levels of the integrins were detected by flow cytometry 48 h later (top; blue and red for α₄ and β₇, respectively). (Middle) Binding of biotinylated SF162gp120 and MAdCAM-1 to transfected cells (red plots); (bottom) specificity of binding determined by testing the ability of HP2/1 and L25 to block the binding of biotinylated MAdCAM-1. Black curves, untreated cells; gray curves, cells incubated with avidin-PE only.

AN-1gp120 binding to 293T cells expressing high levels of α₄β₇ integrins.

We next tested the binding of AN-1gp120 to 293T cells expressing α₄β₇. As shown in Fig. 7, AN-1gp120 readily bound to these cells, and this binding could be partially blocked by either MAdCAM-1 or the anti-α₄ MAbs, suggesting a modest level of α₄β₇-specific binding. Nonetheless, as seen in experiments with RA-treated PBMCs as described above, a substantial level of binding was not blocked by these reagents, indicating that this gp120 binds other cell surface constituents that remain to be identified.

FIG 7.

Binding of AN-1gp120 to α₄β₇ expressed on 293T cells. Flow cytometry was used to test the binding of biotinylated AN-1gp120 to 293T cells transiently transfected with the human α₄ and β₇ cDNAs. In parallel, AN-1gp120 was tested for binding to α₄β₇-expressing cells preincubated with either L25, HP2/1, or MAdCAM-1. Biotinylated MAdCAM-1was used as a control to monitor α₄β₇ expression levels. Black bars, the mean fluorescence intensity obtained from the analysis of the data with FlowJo software.

Binding of gp140 Env glycoproteins to 293T cells expressing the α₄β₇ complex.

The recombinant uncleaved gp140 glycoproteins from the subtype C/B strain CN54, subtype A strains 92UG037 and Q23.17, subtype C strain C.1086, and M-CONS (a consensus sequence of group M Env) (35) were biotinylated and tested for binding to CD4-negative 293T cells transiently expressing the α₄β₇ integrin complex. As shown in Fig. 8A, some of these gp140 Env products bound only marginally to 293T cells expressing high levels of the α₄β₇ complex, whereas AN-1gp120 and MAdCAM-1 showed much more substantial binding. The marginal α₄β₇ binding observed for some of the gp140s did not change with pretreatment of the cells with the α₄ MAbs (not shown).

FIG 8.

Binding of gp140 and T/F gp120s to α₄β₇ expressed on 293T cells. (A) The gp140s of HIV-1 strains CN54 (subtype C), 92UG037 (UG37; subtype A), M-CONS (a consensus sequence of group M Env), the transmitted/founder subtype C strain C.1086, and Q23.17 (subtype A), all produced in CHO cells, were biotinylated and tested for binding to α₄β₇-expressing 293T cells by flow cytometry. The mean fluorescence intensity representing the binding of the different gp140s to the integrin-expressing cells, in which binding was detected by avidin-PE, is shown. (B) gp120s from the transmitted/founder subtype B isolates B6240D11 (B6240), B040D11 (040), and 63521D11 (63521) and from the subtype C isolate C.1086 were biotinylated and tested in parallel for binding to α₄β₇-expressing cells by flow cytometry. Black bars, the mean fluorescence intensity obtained by analysis of the data with FlowJo software. The basal background level of fluorescence determined for cells incubated only with avidin-PE is also shown.

Binding of gp120 from transmitted/founder HIV isolates to α₄β₇-expressing cells.

It was previously suggested that the genotype of T/F HIV gp120s promotes α₄β₇ binding. We tested the integrin binding capacity of four T/F gp120s, three subtype B gp120s (B.040Δ11gp120, B.65321Δ11gp120, and B.6240Δ11gp120) and one subtype C gp120 (C.1086Δ7gp120). As shown in Fig. 8B, some of the T/F gp120s bound marginally to α₄β₇-expressing cells. Pretreatment of the cells with α₄ MAbs or the integrin natural ligand had very little or no effect on the observed binding. In contrast, binding of AN-1gp120 and MAdCAM-1 to the same α₄β₇-expressing cells was substantially higher and was partially inhibited by a combination of the two α₄ MAbs and by MAdCAM-1.

α₄β₇ binding of gp120 produced in different cell lines.

During mucosal transmission of HIV-1 subtype C, the T/F viruses have gp120s with a reduced number of N-linked glycosylation sites and shorter V1V2 variable loops. It has been suggested that a lower number of N-linked glycosylation sites gives a transmission advantage by facilitating α₄β₇ binding (14). Because in our assays a gp120 lacking complex glycans (AN-1) was the only glycoprotein with detectable α₄β₇ binding, we next compared the integrin binding capacity of other gp120s grown in three different cell lines that possess different glycosylation machineries. SF162 and Q23.17 gp120s produced in CHO, 293F, and 293S/GnT1^−/− cell lines were tested for binding to α₄β₇-expressing 293T cells. As shown in Fig. 9A, only the gp120s produced in CHO cells demonstrated binding at levels that were above the background levels. This marginal binding could not be blocked by a combination of HP2/1 and MAdCAM-1 (Fig. 9B). As a control, the combination of HP2/1 and MAdCAM-1 showed partial blocking of binding of AN-1gp120 produced in CHO lec1.1^−/− cells (Fig. 9B).

FIG 9.

The α₄β₇ reactivity of gp120s produced in three different cell lines. (A) Biotinylated SF162 and Q23.17 gp120s produced in CHO, 293F, and 293S/GnT1^−/− cell lines were tested in parallel for binding to 293T cells transiently expressing α₄β₇ integrins. (B) The same gp120s described for panel A were produced in CHO cells and tested for binding to α₄β₇-expressing 293T cells in the presence and absence of a combination of anti-α₄ MAb HP2/1 and MAdCAM-1. Binding of biotinylated MAdCAM-1 to α₄β₇-expressing 293T cells in the presence and absence of the anti-α₄ MAb HP2/1 was used as a control. Black bars, the mean fluorescence intensity obtained by analysis with FlowJo software. The basal background level of fluorescence determined for cells incubated only with avidin-PE is also shown.

Binding of infectious virions to α₄β₇-expressing 293T cells.

We next tested whether infectious virus produced in either human PBMCs or 293T cells was capable of binding the α₄β₇ integrin complex on transfected 293T cells. For these experiments we used as molecular clones two HIV-1 isolates carrying a Renilla luciferase (Luc) reporter gene (Env.IMC.LucR viruses): the subtype AE isolate R2184c4.IMC.LucR and the subtype B isolate WEAU3-3.IMC.LucR. As shown in Fig. 10, some infectious virus was retained on 293T cells in a virus dose-dependent manner, but no measurable differences in the amount of virus that bound to nontransfected 293T cells and α₄β₇-expressing 293T cells were observed for either virus. Since these cells do not express CD4 or any known coreceptors for HIV-1, the nonspecific binding may be attributed to any number of cell adhesion molecules that are known to bind HIV-1 (36).

FIG 10.

Capture of HIV-1 infectious virions by α₄β₇ expressed on 293T cells. Serial dilutions of the HIV-1 infectious molecular clones R2184c4.IMC.LucR (A) and WEAU3-3.IMC.LucR (B) were incubated with either α₄β₇-expressing 293T cells or mock-transfected 293T cells for 1 h at 37°C. Cells were washed and overlaid with TZM-bl cells. Activation of the Luc reporter gene in the TZM-bl cells was measured 48 h later using a Britelite luminescence reaction kit. Circles, virus grown in PBMCs and incubated with mock-transfected 293T cells; squares, virus grown in 293T cells and incubated with mock-transfected 293T cells; triangles, virus grown in PBMCs and incubated with α₄β₇-expressing 293T cells; inverted triangles, virus grown in 293T cells and incubated with α₄β₇-expressing 293T cells. RLU, relative light units.

DISCUSSION

We aimed to develop an assay to detect antibodies that block gp120 binding to the α₄β₇ integrin complex. The first step in developing such an assay was to identify optimal conditions that reproducibly generate an adequate signal for gp120-α₄β₇ binding. Intestinal dendritic cells and stromal cells produce retinoic acid, which has been shown to act as an extrinsic factor that controls the expression of α₄β₇ and the generation of gut-homing T cells (33, 37, 38). In our experiments, we reproduced this finding in vitro by culturing human PBMCs in the presence of RA for 7 days. This treatment imprints the gut-homing phenotype, as PBMCs from three donors showed increased surface expression of the β₇ integrin and they efficiently bound the natural ligand for this complex, MAdCAM-1. The latter event indicates that the α₄β₇ complex expressed on these cells is in the extended/activated conformation, as previously described for this complex (34, 39). Indeed, our assay conditions generated a strong binding signal for the natural ligand, MAdCAM-1, where the majority of this binding could be inhibited by the anti-α₄ MAbs L25 and HP2/1, confirming the specificity of MAdCAM-1 binding to the α₄β₇ complex on RA-treated PBMCs.

Using these optimal conditions with gp120 glycoproteins from two HIV-1 strains (SF162 and TV-1) and RA-treated PBMCs from three donors expressing the α₄β₇ complex, we detected strong CD4-dependent binding but very little or no α₄β₇-dependent binding to these cells. Moreover, we saw no evidence of increased binding of these gp120s or four additional gp120s to α₄β₇ when the anti-α₄ MAbs and MAdCAM-1 were tested in the presence of subsaturating concentrations of Leu3a, suggesting no cooperative effect between CD4 and the integrin complex, which are thought to localize in close proximity on the cell membrane (34, 40).

A second assay was developed with CD4-negative 293T cells transiently expressing the activated form of the α₄β₇ integrin complex. High levels of surface expression of the integrins could be demonstrated by staining with MAbs to α₄ and β₇. Similar to the findings of the PBMC assay, a strong MAdCAM-1 binding signal was obtained with these transfected 293T cells, and this binding was nearly completely blocked by anti-α₄ MAbs, confirming the activated form of the integrin complex. Using these transfected 293T cells, we detected very little or no α₄β₇ binding activity for another 11 gp120s/gp140s from multiple HIV-1 subtypes. Importantly, all gp120s used in these assays contain the LDV/LDI tripeptide. This tripeptide motif has been shown to mediate the binding of MAdCAM-1 to α₄β₇ (13). It was previously reported that the gp120 of HIV-1 SF162 binds α₄β₇ (14, 34). In our assays, this gp120 exhibited very weak binding, regardless of whether the glycoprotein was produced in CHO, 293T, or 293S/GnT1^−/− cells (the last cell type lacks N-acetylglucosamine glycosyltransferase activity). AN-1gp120 produced in CHO lec 1.1^−/− cells (which also lack N-acetylglucosamine glycosyltransferase activity), which has been shown to exhibit high levels of binding to α₄β₇ (13, 14), exhibited the best α₄β₇-specific binding activity in our assays, although this binding activity remained relatively low.

In contrast to the other 15 glycoprotein tested here, AN-1gp120 is not from a natural isolate (26) and was produced in CHO lec1.1^−/− cells. Glycoproteins produced in this glycosylation-deficient cell line are devoid of complex carbohydrates and are enriched with oligomannose-type glycans (14). It has been previously reported that this modified glycan profile results in a 100-fold increase in α₄β₇ binding over that for AN-1gp120 produced in CHO cells (14). Thus, AN-1gp120 produced in CHO lec1.1^−/− cells may be structurally different from the gp120s/gp140s used in our study. Support for this conclusion comes from two lines of evidence. First, we detected very little blocking of AN-1gp120 binding to RA-treated PBMCs by the anti-CD4 MAb Leu3a, whereas binding of CHO-derived gp120s was strongly blocked by this MAb, suggesting that the AN-1gp120 used here is far less CD4 dependent than the gp120 from most other strains of the virus. Second, anti-α₄ MAbs and MAdCAM-1 only partially competed the binding of AN-1gp120 to α₄β₇-expressing cells under conditions in which the binding of the natural ligand, MAdCAM-1, to the same cells was nearly completely blocked by the same MAbs. The latter observations suggest that AN-1gp120 binds to a molecule other than CD4, in addition to the α₄β₇ complex. In fact, in our binding assay, AN-1gp120 bound 293T cells in the absence of CD4 and α₄β₇ expression (data not shown).

Binley et al. (41) examined the role of complex carbohydrates in HIV-1 infectivity and resistance to antibody neutralization. Using HIV-1 isolates grown in glycosylation-deficient cell lines, those studies demonstrated that the antennae of N-glycans serve to protect the integrity of both the V3 loop and the CD4 binding site, while the stems regulate the native trimer conformation. Those studies also showed that N-glycan removal can sometimes compromise viral infectivity. N-glycans also constitute a protective shield against antibodies, where their removal makes viruses more susceptible to neutralization by certain antibodies (42). It is possible that the modified N-glycan moieties on AN-1gp120 result in a change in protein conformation that exposes the V2 loop LDV/LDI in a conformation optimal for α₄β₇ reactivity while reducing CD4 binding. A possible explanation for why the gp120s produced in 293S/GnT1^−/− cells did not bind α₄β₇ could be that CHO lec1.1^−/− cells carry an additional glycosylation defect (43). Our attempts to produce gp120s in CHO lec1.1^−/− cells were unsuccessful. Because of this, we cannot rule out the possibility that the production of gp120 in CHO lec1.1^−/− cells is a strict requirement for efficient α₄β₇ binding.

HIV infection in most cases is established by one or a very few founder viruses, and the gp120s of these T/F viruses have V1-V4 domains with reduced numbers of N-linked glycosylation sites (44, 45). Based on the findings of Nawaz et al. (14) that T/F gp120s have an increased affinity for α₄β₇ and that this affinity is influenced by N-linked glycosylation sites in C3/V4, we tested four T/F gp120s expressed in CHO cells. In our assay, these T/F gp120s had poor or no binding to α₄β₇-expressing cells compared to the level of binding of AN-1gp120 or MAdCAM-1 under similar conditions. We also could not detect any specific binding of infectious virions to α₄β₇-expressing cells. This lack of an integrin binding capacity of native envelope trimers on infectious virions parallels the results that we obtained with purified gp120/gp140 glycoproteins and are in agreement with recent findings that anti-α₄β₇ integrin antibodies cannot block the infectivity of transmitted/founder and chronic subtype C HIV-1 isolates (23).

In conclusion, under optimal conditions for α₄β₇ binding to its natural ligand, MAdCAM-1, we detected very little or no α₄β₇-specific binding activity of multiple gp120s and gp140s, including gp120s from T/F strains of HIV-1. Moreover, we found no evidence that infectious HIV-1 virions produced in either PBMCs or 293T cells are capable of binding α₄β₇. Robust α₄β₇ binding assays for gp120 may require the use of special envelope glycoproteins that have specific and as yet poorly defined glycan moieties, such as those that arise in CHO lec1.1^−/− cells. They may also require glycoproteins with certain N-linked glycans removed altogether (14). Except for AN-1gp120, which showed the best binding to α₄β₇ in our assays, we used only naturally occurring gp120s and gp140s. We encourage additional studies to clarify the possible role of the α₄β₇ integrin complex in direct HIV-1 infection of cells.