The V2 loop of HIV gp120 delivers costimulatory signals to CD4⁺ T cells through Integrin α₄β₇ and promotes cellular activation and infection

Significance

Understanding the early events in HIV transmission will aid in the development of an efficacious HIV vaccine. Productive infection requires that virions access metabolically activated CD4⁺ T cells. These cells are, in general, limited in number, which contributes to inefficient viral transmission. This report describes a mechanism whereby the HIV gp120 envelope protein can deliver activating signals to CD4⁺ T cells. This activity may increase both productive infection in mucosal tissues around the time of transmission and the formation of viral reservoirs. gp120 mediates activating signals by binding to integrin α₄β₇. Antibodies specific to the V2 domain of gp120 block this interaction and may contribute to the efficacy of an HIV vaccine.

Abstract

Acute HIV infection is characterized by rapid viral seeding of immunologic inductive sites in the gut followed by the severe depletion of gut CD4⁺ T cells. Trafficking of α₄β₇-expressing lymphocytes to the gut is mediated by MAdCAM, the natural ligand of α₄β₇ that is expressed on gut endothelial cells. MAdCAM signaling through α₄β₇ costimulates CD4⁺ T cells and promotes HIV replication. Similar to MAdCAM, the V2 domain of the gp120 HIV envelope protein binds to α₄β₇. In this study, we report that gp120 V2 shares with MAdCAM the capacity to signal through α₄β₇ resulting in CD4⁺ T cell activation and proliferation. As with MAdCAM-mediated costimulation, cellular activation induced by gp120 V2 is inhibited by anti-α₄β₇ monoclonal antibodies (mAbs). It is also inhibited by anti-V2 domain antibodies including nonneutralizing mAbs that recognize an epitope in V2 that has been linked to reduced risk of acquisition in the RV144 vaccine trial. The capacity of the V2 domain of gp120 to mediate signaling through α₄β₇ likely impacts early events in HIV infection. The capacity of nonneutralizing V2 antibodies to block this activity reveals a previously unrecognized mechanism whereby such antibodies might impact HIV transmission and pathogenesis.

In the early (acute) stages of HIV infection, gut tissues are one of the preferential targets for viral replication (1, 2). Infection and consequent damage of gut-associated lymphoid tissues (GALT) are widely considered to play a significant role in the pathogenesis of HIV-1 (3). Within the first weeks of infection, high levels of viral replication typically occur in Peyer’s Patches (PPs) and mesenteric lymph nodes (MLNs) (2), followed by a severe depletion in gut lamina propria (LP) of CD4⁺ T cells (4, 5). Early antiretroviral therapy (ART) intervention leads only to a partial restoration of these LP CD4⁺ T cells (6). Understanding the underlying mechanisms associated with high-level viral replication in PP, MLNs, and additional immunologic inductive sites in the gut, and the nature of the irreversible damage to the LP, are subjects relevant to our understanding of the pathogenic mechanisms of early HIV infection. Such information may provide insight into the complex mechanisms surrounding the rapid establishment of persistent viral reservoirs early in the course of HIV infection that act as barrier to HIV cure.

Integrin α₄β₇, is a cell-surface receptor that facilitates CD4⁺ T cell homing to PPs, MLNs, and LP (7–9). Memory CD4⁺ T cells that express high levels of α₄β₇ (α₄β₇^high) are preferentially infected and depleted in vivo in the early stages of acute HIV infection (10). Analysis of gut biopsies from subjects enrolled in the RV217 acute infection cohort (11) revealed that α₄β₇^high memory CD4⁺ T cells were significantly and selectively depleted in Fiebig stage I/II of HIV infection (within ∼10 to 20 d postinfection). The role of these cells in virus transmission has been the subject of several studies in both human and nonhuman primates (NHP). The frequency of circulating α₄β₇^high memory CD4⁺ T cells in blood varies among individuals, ranging from ∼7 to 20% of total CD4⁺ T cells (12). In adults, these levels are relatively stable over time. In a study of uninfected female subjects enrolled in the CAPRISA 004 study cohort, the risk of HIV acquisition and, upon infection, disease progression, was found to be directly correlated with the frequency of α₄β₇^high memory CD4⁺ T cells (10). Similar findings were demonstrated in rhesus macaques (13, 14). In a study involving subjects who participated in the RV254 acute infection cohort, α₄β₇^high memory CD4⁺ T cells were found to be preferentially targeted during early infection (Fiebig stage II/III) (15). Taken together, these studies suggest that α₄β₇^high memory CD4⁺ T cells are among the first CD4⁺ T cells infected by HIV and, given the role of α₄β₇ in gut homing, the preferential infection of these cells may explain, in large measure, the seeding of gut tissues in the acute phase of infection. Consistent with the concept that infection of α₄β₇^high memory CD4⁺ T cells provides a path for HIV to access GALT, pretreatment of rhesus macaques with an anti-α₄β₇ monoclonal antibody (mAb) protected animals from vaginal challenge with a highly infectious isolate of simian immunodeficiency virus (SIV) (SIVmac251) (16).

The basis upon which α₄β₇^high memory CD4⁺ T cells appear to be targeted in the early phases of infection is not fully understood. We recently reported that an interaction between MAdCAM and α₄β₇ may contribute to this targeting (17). MAdCAM is an adhesion receptor that is expressed on high endothelial venules (HEVs) that service GALT (7–9). It is also expressed on follicular dendritic cells (FDCs) in MLNs (18). MAdCAM binding to α₄β₇ on CD4⁺ T cells delivers a costimulatory signal to these cells that can substitute for or augment classical CD80/86-CD28 costimulation (19–21). We have previously demonstrated that MAdCAM costimulation can drive HIV replication in primary CD4⁺ T cell cultures derived from people with HIV (17). Retinoic acid (RA), a metabolite of vitamin A, plays an important role in the differentiation of gut lymphocytes, particularly those localized to gut immunologic inductive sites (22). We found that RA augmented MAdCAM costimulation in a way that allowed recently activated naïve CD4⁺ T cells to support HIV replication. Considering that both memory and naïve α₄β₇-expressing CD4⁺ T cells typically engage MAdCAM as they transit from HEVs into PPs and MLNs (and may subsequently encounter α₄β₇ on FDCs in these tissues), we proposed a model in which MAdCAM signaling contributes to the high-level viral replication that typically occurs in GALT in newly infected individuals (17).

α₄β₇ is not an entry receptor, however, the HIV envelope protein can bind to α₄β₇ (23–27).The role of gp120-α₄β₇ interactions in HIV infection is not well understood; however, a number of other viruses have hijacked integrin signaling pathways as a way of enhancing infection (28–31). We have shown that gp120 binding to α₄β₇ transduces intracellular signals (32). In addition, we reported that HIV gp120, like human herpesvirus-8 (HHV-8), induces focal adhesion kinase (FAK) phosphorylation (33, 34). This raises the possibility that gp120, like MAdCAM, can facilitate HIV infection.

gp120 V2 binds to the ligand binding site of α₄β₇ in a manner that is at least partially similar to the manner whereby MAdCAM engages α₄β₇. In particular, V2 only binds to an activated conformation of α₄β₇, and antagonists specifically developed to inhibit MAdCAM binding also block gp120 V2 binding (23, 25). These similarities raise the possibility that gp120, like MAdCAM, can deliver a costimulatory signal to α₄β₇-expressing CD4⁺ T cells and that, like MAdCAM, such signals can drive HIV infection. This would represent a previously unrecognized activity for the V2 domain of gp120. In this report, we investigate the capacity of gp120 V2 to provide a costimulatory signal to CD4⁺ T cells via α₄β₇ that supports HIV infection. In addition, we evaluate the capacity of nonneutralizing anti-gp120 V2 domain mAbs to inhibit this activity.

Results

gp120 V2 Loop Costimulation Induces Activation of CD4⁺ T Cells.

To determine whether the V2 domain of gp120 (V2), similar to MAdCAM, can deliver a costimulatory signal through α₄β_7, we utilized a procedure in which plate-bound anti-CD3 mAb is used to provide the primary signal to CD4⁺ T cells through the T cell receptor. It is well established that gp120 can deliver signals through CD4 and chemokine coreceptors (35–37). These signals are mediated by domains of gp120 that fall outside of V2. In order to precisely determine the activity of V2, we began our analysis by employing a cyclic peptide derived from the V2 loop of a subtype A/E isolate 92TH023 (92TH023 cV2) (gp120 AA 157 to 196 HXB2 numbering) in which N and C termini are joined by a disulfide bond. In a previous study, we carried out a detailed analysis of the physical interaction between this cyclic peptide and α₄β₇ and showed that it retains the basic features of the entire gp120 that mediates binding to α₄β₇ (25). Primary bulk CD4⁺ T cells were added to wells precoated with anti-CD3 or anti-CD3 in combination with 92TH023 cV2. Wells plated with anti-CD3 and recombinant soluble MAdCAM were employed as a positive control. As an initial readout, we evaluated, by flow-cytometry, the cell-surface expression of CD25 along with the nuclear expression of Ki67, two markers of cellular activation and proliferation. After 96 h, CD4⁺ T cells stimulated with anti-CD3 and 92TH023 cV2 showed a significant increase in the frequency of CD25⁺/Ki67⁺ CD4⁺T cells relative to cells stimulated with anti-CD3 alone (Fig.1A). From eight independent donor CD4⁺ T cells, we observed a mean of 16% CD25⁺/Ki67⁺ cells with anti-CD3 alone vs. 47% CD25⁺/Ki67⁺ cells in cultures stimulated with anti-CD3 plus the cV2 peptide. Inclusion of an anti- α₄ antibody (2B4), that inhibits gp120 binding to α₄β₇, significantly reduced this activation. Similarly, the inclusion of an anti-V2 mAb (CH58) that blocks gp120-α₄β₇ interactions also significantly reduced activation. These results indicate that V2 can costimulate primary CD4⁺ T cells in an α₄β₇-dependent manner. We previously reported that RA can impact MAdCAM-mediated costimulation (17). Further analysis regarding the way in which RA affects cV2 signaling through α₄β₇ will be presented below.

Fig. 1.

gp120 V2 binding to α4β7 activates CD4+ T cells. (A) Flow cytometric analysis of CD25/-Ki67 double positive primary CD4+ T cells cultured for 96 h in the presence of wells coated with anti-CD3 alone or anti CD3 in combination with: MAdCAM, 92Th023cV2, 92TH023 + a control IgG mAb, 92TH023cV2 + anti α4 mAb (2B4), 92TH023cV2 + anti gp120 V2 mAb (CH58). The y axis indicates the % cells expressing both CD25 and Ki67. Purified bulk CD4+ T cells were obtained from eight independent donors. *P < 0.05 **P < 0.01 (two-tailed parametric paired t test). (B) DI (by CFSE dye dilution) of primary CD4+ T cells isolated from seven independent donors and stimulated with ligands as in A and also with anti-CD3 + MAdCAM. *P < 0.05, **P < 0.01 (two-tailed parametric paired t test). (C) A244 gp120 binding to α4β7 mediates proliferation of CD4+ T cells. Flow cytometric determination of average DI in primary CD4+ T cells from three independent donors cultured for 96 h in the presence of wells coated with anti-CD3 alone or anti-CD3 in combination with A244 gp120, A244 gp120 + a control IgG mAb, A244 + a V2 mAb (CH58), or A244 + CD4 binding site mAb VRC01. **P < 0.01 (two-tailed parametric paired t test).

To corroborate the results presented above, we employed a standard carboxyfluorescein succinimidyl ester (CFSE) dye dilution assay that directly measures cell proliferation. A culture protocol similar to that described above was employed; however, cells were prelabeled with CFSE and harvested at day 5 for flow cytometric analysis. Proliferation is reported as the average number of cell divisions (Division Index [DI]). Results from a representative donor (SI Appendix, Fig. S1) in addition to seven independent donor CD4⁺ T cell cultures (Fig. 1B) are provided. Consistent with the CD25/ Ki67 data presented above, we observed a significant increase in proliferation in the presence of anti-CD3 plus 92TH023 cV2 vs. anti-CD3 alone. Proliferation was significantly inhibited by the addition of either an anti-α₄ mAb or an anti-V2 mAb (CH58), demonstrating that costimulation was mediated by a specific interaction between V2 and α₄β₇. To determine whether this activity was restricted to a V2 peptide derived from HIV 92TH023, we evaluated two similarly constructed cyclic peptides derived from subtypes A (c06980v0c220) and C (BG505) and determined that they too could mediate increases in cellular proliferation (SI Appendix, Fig. S2). Taken together, these results reveal the capacity of the V2 loop of gp120 to drive cellular proliferation. This activity is mediated via α₄β₇ and represents a previously unrecognized activity of gp120 V2.

gp120 Induces Activation and Proliferation of CD4⁺ T Cells.

In the experiments described above, we employed a gp120 V2 cyclic peptide to determine precisely the capacity of this specific domain of gp120 to bind to and signal through α₄β₇. This approach excluded interactions of CD4 and CCR5 receptors with other domains of gp120. After establishing that a V2 loop alone can impart a signal to CD4⁺ T cells, we asked whether the proliferation-inducing capacity of the V2 domain is retained in the context of the entire gp120 protein. We measured cellular proliferation by CFSE dye dilution of primary CD4⁺ T cells cultured in plates coated with a recombinant gp120 termed A244 (subtype A/E) that has the capacity to engage CD4 and CCR5 in addition to α₄β₇ (25). A culture protocol similar to that described above was employed. As expected, anti-CD3 alone induced relatively low levels of cellular proliferation (Fig. 1C). When A244 gp120 was included, proliferation increased. gp120-mediated proliferation was not inhibited by a control immunoglobulin G (IgG) mAb. However, when an anti-V2 mAb was included, proliferation was inhibited to levels comparable to stimulation with anti-CD3 alone. These data indicate that costimulation of CD4⁺ T cells by gp120 can be mediated by an interaction with the V2 domain. However, these results do not rule out the possibility that other components of the gp120 protein can costimulate CD4⁺ T cells leading to cellular proliferation. Indeed, CD4 was the first identified T cell costimulatory receptor (38). In this regard, when VRC01, an anti CD4 binding site (CD4bs) mAb was included, we also observed inhibition of cell proliferation. Inhibition by VRC01 may reflect the capacity of CD4 to function as a costimulatory receptor.

Nonneutralizing HIV V2 Domain Antibodies Inhibit gp120-Mediated Costimulation.

Reduced risk of infection in the RV144 vaccine trial has been linked to nonneutralizing antibodies specific to the V1/V2 region of gp120 (39). Sieve analysis of the viral quasi-species replicating in vaccinees who became infected focused the site of interest more precisely on the V2 domain (40), around the region that binds to α₄β₇ (Fig. 2A). We and others have identified nonneutralizing V2 mAbs, including mAbs isolated from an RV144 vaccinee, that block V2 binding to α₄β₇ (25, 41–43). We next asked whether such nonneutralizing V2 mAbs could block V2-mediated activation/proliferation of CD4⁺ T cells. Four V2 mAbs that block V2 binding to α₄β₇ and one V2 mAb that does not block α₄β₇ binding were employed (25, 43, 44) (Fig. 2B). The four blockers included mAb CH58 that was derived from a RV144 vaccine recipient along with mAbs CAP228 3D1, 16H, and 19F that were derived from an HIV subtype C-infected subject participating in the CAPRISA 002 cohort study. mAb CAP228 9D, which was also derived from the same subject, binds around the RV144 sieve site in V2 but fails to block gp120 binding to α₄β₇. An anti-integrin α₄ mAb (2B4) was employed as a specificity control. CD4⁺ T cells were stimulated as described above. Activation/proliferation was assessed on day 4 by flow cytometric measurement of the frequency of CD25⁺/Ki67⁺ cells. In multiple independent donor CD4⁺ T cell cultures, each of the four α₄β₇-blocking V2 mAbs significantly inhibited activation while CAP228 9D, the mAb that failed to block the binding of gp120 to α₄β₇, failed to block activation (Fig. 2B). The anti-α₄ mAb also inhibited V2-mediated activation, while a control mAb showed minimal inhibition. We next carried out a similar analysis but employed A244 gp120 in place of the cV2 peptide. We again found that the V2 mAbs that block binding to α₄β₇ inhibited CD4⁺ T cell activation (Fig. 2C). Noting that, although CAP228 16H and 19F were isolated from a subtype C-infected subject, they appeared to inhibit activation more efficiently than CH58; we determined, using surface plasmon resonance assays, that their affinity for the V2 loop of 92TH023 was similar to that of CH58 (SI Appendix, Fig. S3), suggesting that the epitopes for each of these mAbs share common structural elements.

Fig. 2.

Nonneutralizing HIV V2 mAbs inhibit gp120-mediated costimulation. (A) The sequence of the V2 domain of HIV 92TH023. RV144 sieve residues are highlighted in red. Residues implicated in binding to α4β7 are highlighted in blue. Residues in contact with V2 mAbs CH58, CAP228 16H, and CAP228 3D1 are listed below (along with corresponding HIV isolate). The size of the amino acids is proportional to the contact surface area. For CAP228 19H and CAP228 9D, the approximate location of the epitope in V2 is indicated by a solid bar. (B) Flow cytometric analysis of the expression of Ki67/CD25 double positive CD4+ T cells from multiple donors stimulated with anti-CD3, anti-CD3 + 92T H023cV2 in the absence or presence of V2 mAbs CH58, and CAP228 16H, CAP228 9D, CAP228 16H, and CAP228 19F. Individual donors are color coded. Not all donors were treated with all six mAbs. An anti α4β7 mAb (2B4) and an irrelevant IgG mAb are included as specificity controls. The y axis indicates the % of cells double positive for Ki67 and CD25. **P < 0.01, ***P < 0.001, ****P < 0.0001 (two-tailed parametric paired t test). (C) Flow cytometric analysis of the expression of Ki67/CD25 double positive CD4+ T cells as in B with A244 gp120 in place of 92TH023cV2. *P < 0.05, **P < 0.01 (two-tailed parametric paired t test).

Nonneutralizing SIV V2 Domain Antibodies Inhibit gp120-Mediated Costimulation.

Following the RV144 vaccine trial, an SIV vaccine recapitulated in nonhuman primates some of the features of the RV144 trial in humans. In these studies (14, 45, 46), protection from infection was correlated with nonneutralizing V2 antibody responses that targeted the SIV gp120 V2 domain. Several anti-SIV V2 mAbs that map to this region (mAbs ITS03, ITS09.01, and NCI09) blocked SIV gp120 binding to α₄β₇ (25, 47) (Fig. 3A). Similar to the HIV V2 mAbs that block α₄β₇, these mAbs are nonneutralizing. Interestingly, this region of SIV V2 shares sequences with HIV V2, suggesting a common α₄β₇ binding motif. Moreover, ITS03 cross-reacts with A244 gp120. This conservation is described elsewhere (25). We asked whether these mAbs could inhibit SIV gp120-mediated activation/proliferation of CD4⁺ T cells. Stimulations were carried out as described above with an SIV Mac251 M766 gp120 in place of HIV gp120. mAbs ITS03, ITS09.01, and NCI09 all strongly inhibited gp120-mediated activation, consistent with their ability to block α₄β₇ binding to SIV gp120 (Fig. 3B). Taken together, these results indicate that nonneutralizing SIV and HIV gp120 V2 mAbs that interfere with gp120-α₄β₇ interactions can inhibit gp120 V2-mediated activation of CD4⁺ T cells.

Fig. 3.

Nonneutralizing SIV V2 mAbs inhibit SIV gp120-mediated costimulation. (A) A sequence alignment of the HIV subtype B consensus V2 and SIV Mac251 M766 V2 domains. Shown below are linear peptides that react with SIV mAbs NCI09, ITS03, and ITS09.01. (B) Flow cytometric analysis of the expression of Ki67/CD25 double positive CD4+ T cells from three donors stimulated with anti-CD3, anti-CD3 + SIV Mac251 M766 gp120 in the absence or presence of SIV V2 mAbs NCI09, ITS03, or ITS09.01. An anti α4β7 mAb (2B4) and an irrelevant IgG mAb are included as specificity controls. The y axis indicates the % of cells double positive for Ki67 and CD25. *P < 0.05, **P < 0.01 (two-tailed parametric paired t test).

V2-Mediated Costimulation in Combination with Retinoic Acid Supports HIV Infection.

The results presented above demonstrate that the V2 domain of gp120 can provide costimulation to CD4⁺ T cells through α₄β₇. We previously reported that MAdCAM costimulation could support HIV replication in cells derived from HIV+ patients (17). With this in mind, we asked whether V2 stimulation was also able to facilitate HIV infection. CD4⁺ T cells from healthy donors were stimulated with anti-CD3 in the absence or presence of MAdCAM or a cV2 peptide (Fig. 4). Because RA enhances MAdCAM-mediated viral replication, we also included a MAdCAM + RA and a cV2 92TH023 + RA condition. The R5-tropic HIV isolate BG505 was added 96 h later, and infection was evaluated 6 d postinfection by intracellular staining for HIV gag p24. Consistent with our previous report, cells from most donors stimulated with MAdCAM and MAdCAM + RA supported infection (Fig. 4A). In a minority of the donors, 92TH023 cV2 also supported a modest but significant increase in infection relative to anti-CD3 alone. However, the effect of 92TH023 cV2 costimulation on infection was significantly enhanced when RA was included in cultures such that all donors now supported infection. Combining anti-CD3 + RA in the absence of cV2 did not enhance infection relative to anti-CD3 alone. The RA effect on cV2-mediated infection was unexpected, as we typically do not detect an increase in the percentage of Ki67⁺/CD25⁺ cells in anti-CD3 + cV2 + RA vs. anti-CD3 + cV2 in the absence of RA. To better understand the mechanism by which RA increased infection, we examined the responses of donor CD4⁺ T cells from six subjects to cV2 stimulation ^+/− RA using a panel of surface markers relevant to viral replication (Fig. 4B). As noted above, no significant difference in CD25⁺/Ki67⁺ was observed. However, the inclusion of RA with anti-CD3 and cV2 resulted in a significant increase in the surface expression of CD38 over that observed with anti-CD3 and cV2 in the absence of RA. It also increased the number of α₄β₇^high CD4⁺ cells by more than twofold. RA also increased the level of CCR5 expression. Thus, RA increased the expression of three markers that are linked to HIV infection. Two of these markers, CCR5 and α₄β₇, are receptors that bind directly to gp120. In summary, the V2 domain of gp120 can deliver a costimulatory signal to CD4⁺ T cells that promotes HIV infection. RA, which does not increase proliferation, enhances this effect.

Fig. 4.

V2 costimulation in combination with RA supports HIV infection. (A) Flow cytometric analysis of intracellular staining of CD4+ T cells from eight donors with an anti HIV p24 mAb. Cells were stimulated with anti-CD3 alone or in combination with MAdCAM or 92TH023cV2 ^+/−RA as indicated. The y axis indicates the % p24+ cells 6 d postinfection (PI). Individual donors are color coded. *P < 0.05, **P < 0.01, ***P < 0.001 (two-tailed parametric paired t test). (B) Flow cytometric analysis of the expression of Ki67/CD25 (Upper Left), β7 high (Upper right), CCR5 (Lower Left), and CD38 (Lower Right) on CD4+ T cells from six donors. Cells were stimulated with anti-CD3 alone or in combination with 92TH023cV2 ^+/−RA as indicated. The y axis indicates the frequency of CD25 and Ki67, β7 high, or CD38 positive cells. For CCR5, the mean fluorescence intensity (MFI) is reported. *P < 0.05, **P < 0.01, ***P < 0.001 (two-tailed parametric paired t test).

Nonneutralizing Anti-V2 Loop Antibodies Block V2-Dependent HIV Infection.

We next tested the ability of the nonneutralizing anti-V2 mAbs to block viral infection induced by costimulation mediated by gp120 V2 + RA. CD4⁺ T cells from four healthy donors were stimulated with anti-CD3 + 92TH023 cV2 + RA in the absence or presence of nonneutralizing anti-V2 mAbs, and subsequently inoculated with HIV as described above. Additional controls included cultures treated with two integrin mAbs, 2B4 (anti α₄) and vedolizumab (anti α₄β₇), that block gp120 binding to α₄β₇ (25) or the CD4bs broadly neutralizing gp120 mAb VRC01. As expected, the VRC01 mAb neutralized the virus (Fig. 5). Both vedolizumab and the anti-α₄ mAb inhibited infection. CAP228 3D1, the nonblocking V2 mAb, failed to suppress infection, while the three V2 mAbs that block V2 binding to α₄β₇ suppressed viral replication in at least three of four donors. In summary, anti α₄, β₇ and nonneutralizing V2 mAbs inhibited HIV infection. We conclude that this inhibition results from their capacity to block V2-mediated costimulation via α₄β₇, a process that leads to cellular activation, proliferation, and increased permissiveness to HIV infection.

Fig. 5.

Nonneutralizing V2 mAbs block V2-dependent infection. Flow cytometric analysis of intracellular staining of CD4+ T cells from four donors with an anti HIV p24 mAb. Cells were stimulated with anti-CD3 + 92TH023cV2 + RA in the absence or presence of five V2 mAbs: CH58, CAP228 3D1, CAP228 9D, CAP228 16H, and CAP228 19F. A nonspecific IgG mAb was included as a reagent control, an anti α4 mab and vedolizumab as α4β7 specificity controls, and VRC01 as a positive control for the inhibition of infection. The y axis indicates the % p24+ cells 6 d PI.

Discussion

In a previous report, we found that MAdCAM costimulation of CD4⁺ T cells supports HIV replication (17) in an α₄β₇-dependent manner. Considering that MAdCAM is highly expressed in the gut, we suggested that this costimulatory activity could contribute to the high levels of viral replication that occur in GALT during the early phases of HIV infection. We now extend this observation and demonstrate that the V2 domain of HIV and SIV gp120 can also costimulate CD4⁺ T cells via α₄β₇ in a manner that induces cellular activation and enhances HIV/SIV infection. Infection driven by V2 costimulation is enhanced by RA, a vitamin A metabolite that is generated by dendritic cells in gut immunologic inductive sites (22). We show that combining RA with V2 signaling through α₄β₇ up-regulates the expression of both α₄β₇ and CCR5. This up-regulation likely explains the RA-mediated enhancement of viral replication that we observed. Our demonstration that gp120 V2 can increase CD4⁺ T cell activation represents a previously unrecognized feature of gp120. Because this activity is mediated by α₄β₇, it is most likely relevant in the context of gut tissues. Thus, we propose that the costimulatory activity of α₄β₇ on CD4⁺ T cells mediated by both MAdCAM and gp120 V2, in combination with RA, contributes to HIV gut tropism.

A direct in vivo demonstration that α₄β₇ -mediated costimulation, either by V2 or MAdCAM, facilitates acute infection would be challenging. However, there are findings from studies of infected subjects that support this hypothesis. Sivro and colleagues reported that α₄β₇^high CD4⁺ T cells are significantly and selectively depleted from gut tissues as early as Fiebig stage I (∼6 to 10 d postinfection), which led them to conclude that α₄β₇-expressing CD4⁺ T cells are among the earliest targets of productive infection (10). Guzzo and colleagues reported that virions generated in the early phase of infection incorporate α₄β₇ (48), indicating that they are derived from cells expressing α₄β₇. A recent report found that α₄β₇^high CD4⁺ T cells were preferentially infected in Fiebig II/III (15). We reported that pretreating macaques with an anti-α₄β₇ mAb prior to vaginal challenge with SIV Mac251 protected a significant number of animals from infection (16). Although we do not know the precise mechanism that led to protection in this study, these findings strongly implicate α₄β₇-expressing CD4⁺ T cells as an important early target in vaginal transmission. Of particular interest and left unexplained in that study was the effect of anti-α₄β₇ mAb pretreatment on the animals that were not protected and developed infection. Viremia in these animals was delayed and gut viral DNA was significantly reduced compared to animals that received control antibody (Ab). In addition, CD4⁺ T cell numbers were maintained for an extended period of time in both the blood and GALT. In the present study, we demonstrate that the same anti-α₄β₇ mAb used in that study blocks both MAdCAM and gp120 V2 costimulation. This leads us to favor the idea that the capacity to block costimulation may have contributed to the delay in viremia and maintenance of normal CD4⁺ T cell numbers that we previously observed (16).

gp120 V2 shares key features with MAdCAM in its binding to α₄β₇. Both direct and allosteric α₄β₇ antagonists designed to block MAdCAM also block gp120 V2 (25). In addition, in order to bind α₄β₇, gp120, like MAdCAM, requires that α₄β₇ adopt an activated and extended conformation. This limitation (the inability to bind to inactive conformations of α₄β₇) indicates that only a subset of α₄β₇-expressing CD4⁺ T cells, i.e., those that display an active gut-homing phenotype, will engage gp120 V2. As such, the selective engagement and costimulation of cells that present an activated conformation of α₄β₇ favor the infection of cells trafficking to GALT.

A subset of nonneutralizing gp120 V2 antibodies blocked V2-mediated costimulation of CD4⁺ T cells and viral replication. The epitopes recognized by these antibodies partially overlap epitopes targeted by several well-characterized broadly neutralizing V2 mAbs. However, these nonneutralizing mAbs recognize a helix/loop structure that is distinct from the β-barrel conformation recognized by V2-specific broadly neutralizing V2 mAbs (49, 50). The context in which V2 alternatively adopts helix/loop vs. a β-barrel conformation is unknown. However, the biological relevance of the helix/loop is clear, as nonneutralizing mAbs that recognize a helix/loop structure are elicited not only by vaccination, but also by infection (43, 44, 51–53). mAb CH58, one of the mAbs that blocked V2-dependent viral infection, was derived from an RV144 vaccinee. It recognizes a sequence associated with reduced risk in the RV144 trial (54, 40). Because CH58 and related mAbs show only weak neutralizing activity, much attention has been directed toward the potential of these mAbs to provide protection from infection via antibody effector functions such as antibody-dependent cellular cytotoxicity (53). However, our findings reported here reveal a different type of antiviral activity that is not associated with antibody effector activities, but instead involves the capacity of these antibodies to block the viral envelope from enhancing the susceptibility of target cells to productive infection.

Combining V2 costimulation with RA failed to increase cellular proliferation relative to V2 costimulation alone. However, the addition of RA did increase HIV infection. It also mediated increased expression of CD38, CCR5, and β₇, all of which are associated with increased susceptibility to infection. As such, we speculate that within immunologic inductive sites in the gut, gp120 costimulation in the presence of RA could increase the number of susceptible cells, resulting in high levels of viral replication in the absence of an increase in cellular proliferation. Much remains to be learned about the overall impact of RA on HIV replication in GALT, particularly the way in which RA alters the differentiation program of α₄β₇-costimulated CD4⁺ T cells.

HIV replication in gut tissues is a central feature of HIV pathogenesis. The results described in the present study support a key role for the V2 domain of gp120 in the early stages of HIV infection in gut tissues. MAdCAM and gp120 V2-mediated costimulation through α₄β₇ provide potential mechanisms for high levels of viral replication in gut tissues. The capacity of signal transduction through α₄β₇ to increase the susceptibility of cells to infection underscores the advantage for HIV of acquiring a specific affinity for α₄β₇. Such an activity may be particularly relevant following transmission when the number of activated target cells in mucosal tissues is often limited (55, 56). Finally, the capacity of nonneutralizing V2 mAbs to disrupt V2-mediated costimulation represents an additional activity of gp120 antibodies that may have relevance in the development of more effective HIV vaccines.

Methods

Human Blood Samples and Primary Cell Preparation.

CD4⁺ T cells were obtained by negative selection (>95% purity) (Stem Cell Technologies) from de-identified freshly isolated peripheral blood mononuclear cells (PBMCs), collected from healthy donors through an NIH Department of Transfusion Medicine (DTM)-approved protocol (Institutional Review Board of the National Institute of Allergy and Infectious Diseases [NIAID]).

CD4⁺ T Cell Activation and Proliferation Assays.

Activation and CFSE proliferation assays were carried out as previously described (17). Briefly, 96-well flat bottom cell culture-treated plates were first coated overnight at 4 °C with either 200 ng of anti-CD3 (clone OKT3, eBioscience), followed by 200 ng of MAdCAM-Fc Chimera (R&D Biosystems), or A244 gp120 (Global Solutions for Infectious Diseases [GSID]) or an N-terminal biotinylated cyclic peptide derived from HIV isolate 92TH023 (JPT Peptide Technologies) in 100 µL HEPES-buffered saline (HBS) at 37 °C for 1 h. For the biotinylated cyclic V2 peptide, wells were precoated with 200 ng with neutravidin at 4 °C overnight. In certain wells, retinoic acid (Sigma Aldrich) was also added at a concentration of 10 nM. For activation assays, cells were stained for activation markers CD25, Ki67, and CD38 on day 4. Flow cytometry analysis was carried out with FlowJo software.

Antibody Binding.

The following antibodies were used for flow cytometry: anti-integrin β7 PE (clone FIB27, Biolegend), anti-CD4 allophycocyanin (APC) (clone RPA-T4, BD Pharmingen), anti-CD25 fluorescein isothiocyanate (FITC) (clone M-A251, BD Pharmingen, San Diego, CA), anti-CD38 APC (HB-7, Biolegend), anti-CD45RO BV421 (clone UCHL1, BD Pharmingen), anti-Ki67 Alexa Fluor 647 (clone B56, BD Pharmingen), and anti-p24 FITC (clone KC67, Beckman Coulter). For intracellular Ki67 staining, FOXP3 buffer set and manufacturer’s protocol were utilized (eBioscience). For intracellular p24 staining, a BD Cytofix/Cytoperm kit was used following the manufacturer’s recommendations. Data were collected on a FACSCanto II (BD Biosciences) and analyzed using FlowJo and GraphPad Prism software. The anti-gp120 V2 CH58 human mAb isolated from RV144 vaccinated individuals and VRC01 were provided by the NIAID AIDS reagent program. CAP228 3D1, 9D, 16H, and 19F mAbs were generated in the laboratory of Dr. Lynn Morris (CAPRISA). The anti-α₄β₇ heterodimer mAb (clone Act-1) was provided by the HIV nonhuman primate repository. The Synagis IgG mAb (IgG control) was provided by Dr. Barton Haynes. The anti-α4 mAb (clone 2b4) was purchased from R&D Biosystems. Surface plasmon resonance analysis of anti-V2 loop mAb affinity was carried out on a Biacore 3000 (GE Life Sciences) as previously described (25).

Viral Infection.

Freshly isolated CD4⁺ T cells were cultured at 37 °C, 5% CO₂ for 4 d in the presence of ligand (as described above). Following stimulation, cells were infected with 2 µL of HIV-containing infectious cell supernatant (0.16 ng/µL, p24). After 12 h, plates were washed and resuspended in Roswell Park Memorial Institute (RPMI) (10% FBS) and cultured for 5 more days. On day 6 postinfection, cells were stained with Live/Dead Aqua (Invitrogen), anti-CD4 APC, and intracellular anti-p24 Gag antigen. Cells were analyzed by flow cytometry, and infection was detected by the percentage of cells positive for p24 Gag antigen. The virus utilized for the in vitro experiments was prepared by cloning the full-length gp160 from the SF162 viral isolate (accession number EU123924) into an NL4.3 viral backbone. Viral stocks were produced by transient transfection of 293T cells and then passaged for one round through PBMCs.

Data Availability.

All study data are included in the article and SI Appendix.