The integrin α₄β₇ forms a complex with cell-surface CD4 and defines a T-cell subset that is highly susceptible to infection by HIV-1

Results

α₄β₇ ^{CD4 T Cells Are Prone to Highly Productive HIV Infection.}

α₄β₇ is activated and expressed at high levels on CD4 ^{T cells in immune-competent tissues of the gut that are undergoing antigen specific activation (}12). In contrast, only a minor subset of freshly isolated circulating peripheral T cells express high levels of α₄β₇, the majority of which appears on the cell membrane in an inactive conformation. However, by culturing peripheral CD4 ^{T cells in vitro in the presence of retinoic acid (RA) one can imprint on them a gut phenotype that includes increased α}₄β₇ expression (13). Such cultures include CD4 ^{T-cell subsets that can be described as α}₄β₇^{, α}₄β₇^{, and α}₄β₇^{. Under the conditions we employ, virtually all α}₄β₇ ^{CD4 T cells can be distinguished by high-level expression of integrin β}₇ (Fig. S1). To better understand the role of α₄β₇ in viral replication, we infected RA-cultured CD4 ^{T cells that included α}₄β₇^{, α}₄β₇^{, and α}₄β₇ ^{cells with HIV-1 SF162, a CCR5-tropic isolate, and virus production was monitored by intracellular staining of HIV-1 Gag p24. Three days post-infection the majority of cells producing relatively high levels of p24 exhibited an α}₄β₇ ^{phenotype (}Fig. 1A). The increased production of p24 in α₄β₇ ^{cells occurred over a range of virus inputs (}Fig. 1B and Fig. S2). Over time the α₄β₇ ^{cells disappeared and viral production shifted to α}₄β₇ ^{and α}₄β₇ ^{subsets (}Fig. 1A). The selective depletion of α₄β₇ ^{cells in infected cultures was striking when compared to uninfected controls (}Fig. 1C). We monitored the expression of CD45RO on α₄β₇ ^{cells and found that it also disappeared (}Fig. S3). It is unlikely that HIV-1 mediates the downregulation of both receptors; therefore, we conclude that the loss of α₄β₇ ^{CD4 T cells was the result of virus-mediated cell death. To confirm the bias toward HIV-1 replication in α}₄β₇ ^{cells we separated the α}₄β₇ ^{cellular subset from the α}₄β₇ ^{subsets (}Fig. 1D, inset) and then inoculated cultures with a low amount of virus (100× less than in Fig. 1A). Viral replication was monitored by p24 antigen ELISA. On the peak day of replication in α₄β₇ ^{cells from three donors we observed an average of 15-fold greater level of replication in the α}₄β₇ ^{subset than in the α}₄β₇ ^{subset (}P = 0.027, Wilcoxon rank-sum test) (Fig. 1D).

Fig. 1.

α₄β₇ ^{CD4 T cells are a preferential target of productive infection. In vitro infection of CD4 T cells after 6 days of culture in the presence of retinoic acid. (}A) Flow cytometric analysis of cell surface integrin β₇ and intracellular Gag p24 expression on and in CD4 ^{T cells 3, 6, and 8 days after inoculation with the R5 HIV-1 SF162. (}B) Percent intracellular Gag p24 expression in α₄β₇ ^{and α}₄β₇ ^{CD4 T cells 3 days post-inoculation using three different amounts of SF162. Values represent the % p24 cells within each subset. (}C) Comparison of integrin β₇ expression on CD4 ^{T cells 8 days post-inoculation with the same cells left uninfected. (}D) p24 antigen ELISA of culture supernatants harvested from α₄β₇ ^{and α}₄β₇ ^{CD4 T-cell subsets 3, 6, and 8 days post-inoculation with SF162. Levels of integrin β}₇ expression on both subsets on the day of inoculation are included (Inset). These results are representative of three independent experiments using different donor CD4 ^{T cells.}

α₄β₇ ^{CD4 T Cells Are CCR5 and Metabolically Activated.}

To better understand the preferential replication of HIV-1 in α₄β₇ ^{cells, we carried out additional phenotypic analyses. α}₄β₇ ^{cells are CD45RO, and unlike the α}₄β₇ ^{and α}₄β₇ ^{cells expressed high levels of CCR5 (}Fig. 2A and B). α₄β₇ ^{cells were strongly reactive for Ki-67, indicative of an activated metabolic state (}Fig. 2C). These cells were also distinct in expressing low levels of CXCR4 and could be further defined as CD45RA^{/CD62L/CCR7, consistent with a memory phenotype (}Fig. S4). We conclude that the rapid infection and high-level production of HIV-1 in α₄β₇ ^{CD4 T cells can be explained by the fact that these cells are metabolically activated, although other factors may also contribute. Thus, high levels of α}₄β₇ demarcate a subset of CCR5^{/CD4 T cells that are an ideal substrate for highly productive HIV-1 infection.}

Fig. 2.

α₄β₇ ^{CD4 T cells express high levels of CCR5 and are metabolically active. Flow cytometric analysis of peripheral CD4 T cells cultured in retinoic acid for 6 days followed by staining with fluoresceinated mAbs specific for (}A) β₇ and CD45RO. (B) β₇ and CCR5. (C) β₇ and Ki-67. This is a representative analysis of receptor expression on more than three independent donor CD4 ^{T cells.}

Activated CD4 ^{T Cells Occur at a High Frequency in the α}₄β₇ ^{Subset of Gut Mucosal CD4 T Cells.}

The fact that HIV-1 has the capacity to bind to a receptor that defines a subset of activated targets provides potentially important clues about how HIV-1 uses α₄β₇ in vivo. To that end, we determined whether α₄β₇ ^{CD4 T cells taken directly from gut tissues exhibited a similar phenotype. Practical limitations confined our analysis to rectum and colon biopsies of healthy volunteers where we were able to readily identify α}₄β₇^{/CCR5 CD4 T cells (}Fig. 3A and Fig. S5). In 12 biopsies from three healthy donors, the frequency of α₄β₇ ^{CD4 T cells (relative to total CD4 T cells) ranged from 1.9–12.1% (Avg. = 7.9) (}Fig. 3A and B). On average, 23% of the α₄β₇ ^{CD4 T cells were also CCR5 and Ki-67 (95% CI 18–28%) (}Fig. 3B). In contrast, on average, only 6% (95% CI 4–9%) of the α₄β₇ ^{CD4 T cells were CCR5 and Ki-67. Therefore, the percentage of Ki-67 cells is higher in the α}₄β₇ ^{subset than in the α}₄β₇ ^{subset. Thus, Ki-67 cells were enriched in the α}₄β₇ ^{CD4 cell subset both in vitro (}Fig. 2) and ex vivo (Fig. 3).

Fig. 3.

Activated CD4 ^{T cells in the gut and rectum are enriched in the α}₄β₇ ^{subset. Cells freshly isolated from rectal and colon biopsies were analyzed by flow cytometry. (}A) A representative analysis of CCR5 and Ki-67 expression on β₇ ^{and β}₇^{CD4 T cells. (}B) Summary of Ki-67 and CCR5 expression on β₇ ^{and β}₇^{CD4 T cells isolated from the colon and rectum of three healthy donors. Values are reported as % within each population. Average % expression of CCR5 and Ki-67 expression in all β}₇ ^{and β}₇^{CD4 T-cell samples analyzed is presented along with a significance value (nonparametric Wilcoxon signed-rank test for paired samples). (}C) α₄β₇^{/CD4 T cells are detected in female genital mucosa (Mean 17.5, S.D. 13.4).}

α₄β₇ ^{CD4 T Cells Reside in Female Genital Mucosa.}

The initial events that follow HIV deposition in genital mucosal sites are poorly understood, as are the events that promote HIV-1 invasion of GALT. In considering whether α₄β₇ plays a role in infection of T cells in genital mucosa and/or seeding of GALT, we noted a recent report that described a unique population of α₄β₇ ^{memory T lymphocytes resident in the female reproductive tract (}14). This finding prompted us to determine whether we could identify α₄β₇ ^{CD4 T-cells in female genital mucosa tissue samples. To this end, we analyzed cervical cytobrush samples obtained from eight women visiting a female sex worker clinic in Nairobi, Kenya. The mAb Act-1 was for used to measure the surface expression of α}₄β₇. α₄β₇ ^{CD4 T-cells were detected in all eight samples (mean 17.5%) (}Fig. 3C). The presence of these cells could provide a link between productive infection of CD4 ^{T cells in female genital mucosa and the subsequent spread of HIV-1 to GALT.}

The specific circumstance in which HIV-1 initially infects α₄β₇ ^{CD4 T cells is unknown. It encounters multiple barriers to successful transmission (}15) as it migrates through highly compartmentalized mucosal tissues. Moreover, it has been suggested that dendritic cells facilitate this migration (15). Neither the biopsies that we analyzed nor our culture systems reflect this complexity; however, the capacity of α₄β₇ to capture HIV-1 virions (9) suggests that it can facilitate infection of activated cells in vitro. To this end, we inoculated cultures of α₄β₇ ^{CD4 T cells with low amounts of virus in the presence or absence of the α}₄ mAb 2B4 that efficiently blocks gp120 binding to α₄β₇ (9). Viral input was reduced 100-fold relative to the amount used in experiments reported in Fig. 1A, and unbound virus was rinsed away 3 h post-inoculation. In 2B4-treated cultures, we observed a significant delay in viral spread through the culture relative to control cultures (Fig. 4) such that on day 6 post-inoculation, lower levels of extracellular Gag p24 were observed in 2B4-treated cultures than in control cultures. By day 9 post-infection, the inhibitory activity of 2B4 decreased. These results indicate that 2B4 slows the spread of virus through a culture enriched in α₄β₇ ^{CD4 T cells.}

Fig. 4.

Delay in HIV-1 replication by an α₄ mAb. (A–C) Purified CD4 ^{T cells cultured in retinoic acid and inoculated with a low amount of HIV-1 SF162 in the presence of IgG1 or the α}₄ mAb 2B4 (as indicated). Viral replication was determined by measuring by p24 Gag levels in culture supernatants on days 3, 6, and 9 post-infection. Three representative replication time-courses are presented. (D) Cumulative inhibition of viral replication (percent reduction in p24 Gag) from six independent infections mediated by mAb 2B4 relative to an IgG1 control on days 6 and 9 is presented. 2B4 significantly inhibited viral replication on day 6 (P < 0.001), but not on day 9 (P < 0.139). On average there is an 80% (95% CI: 71–89%) reduction in the level of p24 on day 6 in the infection in presence of 2B4. Error bars represent standard deviation from mean inhibition. Inhibition on day 6 was significantly greater than on day 9 (Wilcoxon signed-rank test).

CD4 and α₄β₇ Reside in a Complex on the Cell Membrane.

Because α₄β₇ demarcates a highly infectible subset of CD4 ^{T cells, we asked whether it colocalizes with CD4 on gut-derived CD4 T cells, as we previously observed on RA-cultured T cells (}9). We stained freshly isolated cells from gut biopsies with α₄β₇, CD4 and CCR5 mAbs (Fig. 5). α₄β₇ is presented in a strikingly clustered way on these cells; moreover, there was extensive colocalization of α₄β₇ with both CD4 and CCR5. We observed more pronounced CD4 clustering on gut cells than we typically observe on peripheral T cells (Fig. S6). This clustering raised the possibility that these receptors exist in close physical proximity on the membrane surface. The resolution of standard confocal microscopy (≈200 nm) is insufficient to make such a determination. Thus, to determine if α₄β₇ and CD4 reside in close proximity on the surface of a T cell, we analyzed live α₄β₇ ^{CD4 T cells by acceptor-photobleaching Förster resonance energy transfer (FRET) (resolution ≈10 nm) (}16). Cells were stained with the β₇ mAb FIB27 Alexa Fluor 488 (donor) and the CD4 mAb Leu3A Alexa Fluor 568 (acceptor) (Fig. S7A). Donor and acceptor fluorescence levels were recorded both before, and following acceptor-photobleaching. Regions of interest (ROIs) were created around clusters of β₇ and FRET efficiencies were calculated based upon changes in donor fluorescence subsequent to acceptor photobleaching. As a positive control we measured FRET efficiencies generated between two CD4 mAbs, Leu3A Alexa Fluor 568 (acceptor) and OKT4 Alexa Fluor 488 (donor). As a negative control, we measured FRET efficiencies generated between CD4 and CD43, an abundantly expressed anti-adhesion receptor known not to cluster with CD4 on activated cells (17). Twelve cells were analyzed, one of which is presented in Fig. S7A. ROI-3 (Fig. S7AInset) exemplifies the FRET effects we observed between CD4 and β₇, with an efficiency of ≈50%. Upon analyzing up to 19 ROIs on 12 cells, we observed CD4-β₇ FRET effects of ≈34% (Fig. S7B) (P = 0.0001 Wilcoxon signed-rank test for paired samples.) By comparison, we observed CD4-CD4 FRET effects of approximately 40% and CD4-CD43 FRET effects of <5%. These results were confirmed by sensitized-emission FRET (Fig. S8 A and B). The resolution of FRET is approximately 10 nm; therefore, these results indicate that CD4 and α₄β₇ appear on the surface of a cell in extremely close association.

Fig. 5.

On gut α₄β₇ ^{CD4 T cells α}₄β₇ colocalizes with CD4 and CCR5. Freshly isolated gut cells obtained from biopsies taken from healthy donors were stained with the α₄β₇ mAb Act-1 (red), the CD4 mAb OKT4 (purple), and the CCR5 mAb 2D7 (green) and viewed under a confocal microscope. Unstained and individual stains of a representative cell are presented along with digitally defined regions of colocalization between α₄β₇ and CD4 (yellow) and α₄β₇, CD4 and CCR5 (blue). This cell is representative of greater that 60 cells analyzed from four donors.

Considering the proximity of CD4 and α₄β₇ we wondered whether these receptors could be coprecipitated. Because CD4 is normally recruited into the T-cell receptor complex upon T-cell activation, its coprecipitation with α₄β₇ would represent an unexpected pairing on the cell membrane. α₄β₇ ^{CD4 T cells were preincubated on ice and then briefly treated with 3,3′-dithiobis-(sulfosuccinimidylpropionate) (DTSSP), a crosslinking reagent with a spacer arm of approximately 1.2 nm. Cells were lysed, and proteins were precipitated with an anti CD4 mAb. Precipitates were electrophoresed and immunoblotted with β}₇ polyclonal antisera (Fig. 6A). In the presence of DTSSP, β₇ coprecipitated with CD4, indicating that they can reside within ≈1.2 nm of each other. Considering the close proximity of these two receptors, their relative height, and the estimated diameter of an envelope spike (18) (Fig. 6B), it is possible that a gp120 trimer could engage α₄β₇ and CD4 in a near-simultaneous manner. We do not yet know whether such dual ligation takes place on the cell surface.

Fig. 6.

Integrin β₇ coprecipitates with the CD4 receptor. (A) A Western blot stained with a an integrin β₇ polyclonal antisera of cell lysates derived from α₄β₇ ^{CD4 T cells treated or not with the crosslinking reagent DTSSP followed by coprecipitation with protein A agarose beads and the CD4 mAb OKT4 (lane 1), IgG1+DTSSP (lane 2), and OKT4+DTSSP (lane 3). A recombinant soluble α}₄β₇ was run directly as a positive control (lane 4). These results are representative of three independent experiments. (B) A schematic depicting approximate sizes of α₄β₇, CD4, and a gp120 trimer.

α₄β₇ ^{CD4 T Cells Are Prone to Highly Productive HIV Infection.}

α₄β₇ is activated and expressed at high levels on CD4 ^{T cells in immune-competent tissues of the gut that are undergoing antigen specific activation (}12). In contrast, only a minor subset of freshly isolated circulating peripheral T cells express high levels of α₄β₇, the majority of which appears on the cell membrane in an inactive conformation. However, by culturing peripheral CD4 ^{T cells in vitro in the presence of retinoic acid (RA) one can imprint on them a gut phenotype that includes increased α}₄β₇ expression (13). Such cultures include CD4 ^{T-cell subsets that can be described as α}₄β₇^{, α}₄β₇^{, and α}₄β₇^{. Under the conditions we employ, virtually all α}₄β₇ ^{CD4 T cells can be distinguished by high-level expression of integrin β}₇ (Fig. S1). To better understand the role of α₄β₇ in viral replication, we infected RA-cultured CD4 ^{T cells that included α}₄β₇^{, α}₄β₇^{, and α}₄β₇ ^{cells with HIV-1 SF162, a CCR5-tropic isolate, and virus production was monitored by intracellular staining of HIV-1 Gag p24. Three days post-infection the majority of cells producing relatively high levels of p24 exhibited an α}₄β₇ ^{phenotype (}Fig. 1A). The increased production of p24 in α₄β₇ ^{cells occurred over a range of virus inputs (}Fig. 1B and Fig. S2). Over time the α₄β₇ ^{cells disappeared and viral production shifted to α}₄β₇ ^{and α}₄β₇ ^{subsets (}Fig. 1A). The selective depletion of α₄β₇ ^{cells in infected cultures was striking when compared to uninfected controls (}Fig. 1C). We monitored the expression of CD45RO on α₄β₇ ^{cells and found that it also disappeared (}Fig. S3). It is unlikely that HIV-1 mediates the downregulation of both receptors; therefore, we conclude that the loss of α₄β₇ ^{CD4 T cells was the result of virus-mediated cell death. To confirm the bias toward HIV-1 replication in α}₄β₇ ^{cells we separated the α}₄β₇ ^{cellular subset from the α}₄β₇ ^{subsets (}Fig. 1D, inset) and then inoculated cultures with a low amount of virus (100× less than in Fig. 1A). Viral replication was monitored by p24 antigen ELISA. On the peak day of replication in α₄β₇ ^{cells from three donors we observed an average of 15-fold greater level of replication in the α}₄β₇ ^{subset than in the α}₄β₇ ^{subset (}P = 0.027, Wilcoxon rank-sum test) (Fig. 1D).

Fig. 1.

α₄β₇ ^{CD4 T cells are a preferential target of productive infection. In vitro infection of CD4 T cells after 6 days of culture in the presence of retinoic acid. (}A) Flow cytometric analysis of cell surface integrin β₇ and intracellular Gag p24 expression on and in CD4 ^{T cells 3, 6, and 8 days after inoculation with the R5 HIV-1 SF162. (}B) Percent intracellular Gag p24 expression in α₄β₇ ^{and α}₄β₇ ^{CD4 T cells 3 days post-inoculation using three different amounts of SF162. Values represent the % p24 cells within each subset. (}C) Comparison of integrin β₇ expression on CD4 ^{T cells 8 days post-inoculation with the same cells left uninfected. (}D) p24 antigen ELISA of culture supernatants harvested from α₄β₇ ^{and α}₄β₇ ^{CD4 T-cell subsets 3, 6, and 8 days post-inoculation with SF162. Levels of integrin β}₇ expression on both subsets on the day of inoculation are included (Inset). These results are representative of three independent experiments using different donor CD4 ^{T cells.}

α₄β₇ ^{CD4 T Cells Are CCR5 and Metabolically Activated.}

To better understand the preferential replication of HIV-1 in α₄β₇ ^{cells, we carried out additional phenotypic analyses. α}₄β₇ ^{cells are CD45RO, and unlike the α}₄β₇ ^{and α}₄β₇ ^{cells expressed high levels of CCR5 (}Fig. 2A and B). α₄β₇ ^{cells were strongly reactive for Ki-67, indicative of an activated metabolic state (}Fig. 2C). These cells were also distinct in expressing low levels of CXCR4 and could be further defined as CD45RA^{/CD62L/CCR7, consistent with a memory phenotype (}Fig. S4). We conclude that the rapid infection and high-level production of HIV-1 in α₄β₇ ^{CD4 T cells can be explained by the fact that these cells are metabolically activated, although other factors may also contribute. Thus, high levels of α}₄β₇ demarcate a subset of CCR5^{/CD4 T cells that are an ideal substrate for highly productive HIV-1 infection.}

Fig. 2.

α₄β₇ ^{CD4 T cells express high levels of CCR5 and are metabolically active. Flow cytometric analysis of peripheral CD4 T cells cultured in retinoic acid for 6 days followed by staining with fluoresceinated mAbs specific for (}A) β₇ and CD45RO. (B) β₇ and CCR5. (C) β₇ and Ki-67. This is a representative analysis of receptor expression on more than three independent donor CD4 ^{T cells.}

Activated CD4 ^{T Cells Occur at a High Frequency in the α}₄β₇ ^{Subset of Gut Mucosal CD4 T Cells.}

The fact that HIV-1 has the capacity to bind to a receptor that defines a subset of activated targets provides potentially important clues about how HIV-1 uses α₄β₇ in vivo. To that end, we determined whether α₄β₇ ^{CD4 T cells taken directly from gut tissues exhibited a similar phenotype. Practical limitations confined our analysis to rectum and colon biopsies of healthy volunteers where we were able to readily identify α}₄β₇^{/CCR5 CD4 T cells (}Fig. 3A and Fig. S5). In 12 biopsies from three healthy donors, the frequency of α₄β₇ ^{CD4 T cells (relative to total CD4 T cells) ranged from 1.9–12.1% (Avg. = 7.9) (}Fig. 3A and B). On average, 23% of the α₄β₇ ^{CD4 T cells were also CCR5 and Ki-67 (95% CI 18–28%) (}Fig. 3B). In contrast, on average, only 6% (95% CI 4–9%) of the α₄β₇ ^{CD4 T cells were CCR5 and Ki-67. Therefore, the percentage of Ki-67 cells is higher in the α}₄β₇ ^{subset than in the α}₄β₇ ^{subset. Thus, Ki-67 cells were enriched in the α}₄β₇ ^{CD4 cell subset both in vitro (}Fig. 2) and ex vivo (Fig. 3).

Fig. 3.

Activated CD4 ^{T cells in the gut and rectum are enriched in the α}₄β₇ ^{subset. Cells freshly isolated from rectal and colon biopsies were analyzed by flow cytometry. (}A) A representative analysis of CCR5 and Ki-67 expression on β₇ ^{and β}₇^{CD4 T cells. (}B) Summary of Ki-67 and CCR5 expression on β₇ ^{and β}₇^{CD4 T cells isolated from the colon and rectum of three healthy donors. Values are reported as % within each population. Average % expression of CCR5 and Ki-67 expression in all β}₇ ^{and β}₇^{CD4 T-cell samples analyzed is presented along with a significance value (nonparametric Wilcoxon signed-rank test for paired samples). (}C) α₄β₇^{/CD4 T cells are detected in female genital mucosa (Mean 17.5, S.D. 13.4).}

α₄β₇ ^{CD4 T Cells Reside in Female Genital Mucosa.}

The initial events that follow HIV deposition in genital mucosal sites are poorly understood, as are the events that promote HIV-1 invasion of GALT. In considering whether α₄β₇ plays a role in infection of T cells in genital mucosa and/or seeding of GALT, we noted a recent report that described a unique population of α₄β₇ ^{memory T lymphocytes resident in the female reproductive tract (}14). This finding prompted us to determine whether we could identify α₄β₇ ^{CD4 T-cells in female genital mucosa tissue samples. To this end, we analyzed cervical cytobrush samples obtained from eight women visiting a female sex worker clinic in Nairobi, Kenya. The mAb Act-1 was for used to measure the surface expression of α}₄β₇. α₄β₇ ^{CD4 T-cells were detected in all eight samples (mean 17.5%) (}Fig. 3C). The presence of these cells could provide a link between productive infection of CD4 ^{T cells in female genital mucosa and the subsequent spread of HIV-1 to GALT.}

The specific circumstance in which HIV-1 initially infects α₄β₇ ^{CD4 T cells is unknown. It encounters multiple barriers to successful transmission (}15) as it migrates through highly compartmentalized mucosal tissues. Moreover, it has been suggested that dendritic cells facilitate this migration (15). Neither the biopsies that we analyzed nor our culture systems reflect this complexity; however, the capacity of α₄β₇ to capture HIV-1 virions (9) suggests that it can facilitate infection of activated cells in vitro. To this end, we inoculated cultures of α₄β₇ ^{CD4 T cells with low amounts of virus in the presence or absence of the α}₄ mAb 2B4 that efficiently blocks gp120 binding to α₄β₇ (9). Viral input was reduced 100-fold relative to the amount used in experiments reported in Fig. 1A, and unbound virus was rinsed away 3 h post-inoculation. In 2B4-treated cultures, we observed a significant delay in viral spread through the culture relative to control cultures (Fig. 4) such that on day 6 post-inoculation, lower levels of extracellular Gag p24 were observed in 2B4-treated cultures than in control cultures. By day 9 post-infection, the inhibitory activity of 2B4 decreased. These results indicate that 2B4 slows the spread of virus through a culture enriched in α₄β₇ ^{CD4 T cells.}

Fig. 4.

Delay in HIV-1 replication by an α₄ mAb. (A–C) Purified CD4 ^{T cells cultured in retinoic acid and inoculated with a low amount of HIV-1 SF162 in the presence of IgG1 or the α}₄ mAb 2B4 (as indicated). Viral replication was determined by measuring by p24 Gag levels in culture supernatants on days 3, 6, and 9 post-infection. Three representative replication time-courses are presented. (D) Cumulative inhibition of viral replication (percent reduction in p24 Gag) from six independent infections mediated by mAb 2B4 relative to an IgG1 control on days 6 and 9 is presented. 2B4 significantly inhibited viral replication on day 6 (P < 0.001), but not on day 9 (P < 0.139). On average there is an 80% (95% CI: 71–89%) reduction in the level of p24 on day 6 in the infection in presence of 2B4. Error bars represent standard deviation from mean inhibition. Inhibition on day 6 was significantly greater than on day 9 (Wilcoxon signed-rank test).

CD4 and α₄β₇ Reside in a Complex on the Cell Membrane.

Because α₄β₇ demarcates a highly infectible subset of CD4 ^{T cells, we asked whether it colocalizes with CD4 on gut-derived CD4 T cells, as we previously observed on RA-cultured T cells (}9). We stained freshly isolated cells from gut biopsies with α₄β₇, CD4 and CCR5 mAbs (Fig. 5). α₄β₇ is presented in a strikingly clustered way on these cells; moreover, there was extensive colocalization of α₄β₇ with both CD4 and CCR5. We observed more pronounced CD4 clustering on gut cells than we typically observe on peripheral T cells (Fig. S6). This clustering raised the possibility that these receptors exist in close physical proximity on the membrane surface. The resolution of standard confocal microscopy (≈200 nm) is insufficient to make such a determination. Thus, to determine if α₄β₇ and CD4 reside in close proximity on the surface of a T cell, we analyzed live α₄β₇ ^{CD4 T cells by acceptor-photobleaching Förster resonance energy transfer (FRET) (resolution ≈10 nm) (}16). Cells were stained with the β₇ mAb FIB27 Alexa Fluor 488 (donor) and the CD4 mAb Leu3A Alexa Fluor 568 (acceptor) (Fig. S7A). Donor and acceptor fluorescence levels were recorded both before, and following acceptor-photobleaching. Regions of interest (ROIs) were created around clusters of β₇ and FRET efficiencies were calculated based upon changes in donor fluorescence subsequent to acceptor photobleaching. As a positive control we measured FRET efficiencies generated between two CD4 mAbs, Leu3A Alexa Fluor 568 (acceptor) and OKT4 Alexa Fluor 488 (donor). As a negative control, we measured FRET efficiencies generated between CD4 and CD43, an abundantly expressed anti-adhesion receptor known not to cluster with CD4 on activated cells (17). Twelve cells were analyzed, one of which is presented in Fig. S7A. ROI-3 (Fig. S7AInset) exemplifies the FRET effects we observed between CD4 and β₇, with an efficiency of ≈50%. Upon analyzing up to 19 ROIs on 12 cells, we observed CD4-β₇ FRET effects of ≈34% (Fig. S7B) (P = 0.0001 Wilcoxon signed-rank test for paired samples.) By comparison, we observed CD4-CD4 FRET effects of approximately 40% and CD4-CD43 FRET effects of <5%. These results were confirmed by sensitized-emission FRET (Fig. S8 A and B). The resolution of FRET is approximately 10 nm; therefore, these results indicate that CD4 and α₄β₇ appear on the surface of a cell in extremely close association.

Fig. 5.

On gut α₄β₇ ^{CD4 T cells α}₄β₇ colocalizes with CD4 and CCR5. Freshly isolated gut cells obtained from biopsies taken from healthy donors were stained with the α₄β₇ mAb Act-1 (red), the CD4 mAb OKT4 (purple), and the CCR5 mAb 2D7 (green) and viewed under a confocal microscope. Unstained and individual stains of a representative cell are presented along with digitally defined regions of colocalization between α₄β₇ and CD4 (yellow) and α₄β₇, CD4 and CCR5 (blue). This cell is representative of greater that 60 cells analyzed from four donors.

Considering the proximity of CD4 and α₄β₇ we wondered whether these receptors could be coprecipitated. Because CD4 is normally recruited into the T-cell receptor complex upon T-cell activation, its coprecipitation with α₄β₇ would represent an unexpected pairing on the cell membrane. α₄β₇ ^{CD4 T cells were preincubated on ice and then briefly treated with 3,3′-dithiobis-(sulfosuccinimidylpropionate) (DTSSP), a crosslinking reagent with a spacer arm of approximately 1.2 nm. Cells were lysed, and proteins were precipitated with an anti CD4 mAb. Precipitates were electrophoresed and immunoblotted with β}₇ polyclonal antisera (Fig. 6A). In the presence of DTSSP, β₇ coprecipitated with CD4, indicating that they can reside within ≈1.2 nm of each other. Considering the close proximity of these two receptors, their relative height, and the estimated diameter of an envelope spike (18) (Fig. 6B), it is possible that a gp120 trimer could engage α₄β₇ and CD4 in a near-simultaneous manner. We do not yet know whether such dual ligation takes place on the cell surface.

Fig. 6.

Integrin β₇ coprecipitates with the CD4 receptor. (A) A Western blot stained with a an integrin β₇ polyclonal antisera of cell lysates derived from α₄β₇ ^{CD4 T cells treated or not with the crosslinking reagent DTSSP followed by coprecipitation with protein A agarose beads and the CD4 mAb OKT4 (lane 1), IgG1+DTSSP (lane 2), and OKT4+DTSSP (lane 3). A recombinant soluble α}₄β₇ was run directly as a positive control (lane 4). These results are representative of three independent experiments. (B) A schematic depicting approximate sizes of α₄β₇, CD4, and a gp120 trimer.

Discussion

We have reported previously that HIV-1 gp120 binds to and signals through an activated/extended form of α₄β₇ that is present on α₄β₇ ^{CD4 T cells. Others have reported that during the acute phase of SIV infection, peripheral α}₄β₇ ^{CD4 T cells are preferentially infected (}19), and, in humans, circulating α₄β₇ ^{CD4 T cells are preferentially depleted during the acute phase of infection (}20). In this report, we demonstrate that a subset of CD4 ^{T cells that can be defined by high levels of α}₄β₇ expression are prone to highly productive HIV-1 infection. Multiple factors are likely to contribute to the preferential replication of HIV in α₄β₇ ^{CD4 T cells. In particular, a large fraction of these cells are metabolically active as measured by Ki-67, express high levels of CCR5, and α}₄β₇ resides on the cell membrane in close association with CD4. We conclude that these factors provided the selective pressures that resulted in HIV-1 evolving a specific affinity for α₄β₇. Sexual transmission of HIV is inefficient (11, 21). Any replication advantage that HIV-1 can gain during the early stages of infection can have a significant impact on the efficiency of transmission. Although resting cells are thought to be the first cells infected, activated cells are critical to propagation and dissemination following sexual transmission (5, 6, 22). In this regard α₄β₇ mediates the migration of CD4 ^{T cells between Peyer's patches, mesenteric lymph nodes, and lamina propria (}12, 23), sites that represent a “target rich” environment for HIV-1. Among all lymphoid tissues in humans, it is in these tissues that the majority of activated CD4 ^{T cells reside (}24). To the extent that HIV-1 can rapidly gain access to these sites, the probability of establishing an irreversible infection is increased. However, even in these tissues, before the onset of viral dissemination, fully activated CD4 ^{T cells are in the minority (}22). Thus, the specific affinity of gp120 for α₄β₇ likely reflects the fact that α₄β₇ is presented on CD4 ^{T cells that are a desirable target for productive infection.}

The CD4 receptor-independent nature of gp120-α₄β₇ interactions is important in understanding the role that this receptor plays in infection. As noted above, a subset of α₄β₇ ^{CD4 T cells are distinct in expressing high levels of CCR5 and relatively low levels of CXCR4 (note that sexual transmission strongly favors R5 viruses). High-level expression of CCR5 on CD4 T cells is, to a degree, a marker of activation in the gut, and one might consider then that HIV-1 could use CCR5 as a means of discriminating between active and resting cells. However, unlike α}₄β₇, CCR5 is effectively hidden from HIV until after a virion engages a CD4 receptor. The CD4 receptor exhibits near-uniform expression on CD4 ^{T cells regardless of whether they are metabolically active or resting, or whether they are CCR5 or CCR5. Once a virion engages CD4 receptors, conformational changes in the envelope spike effectively commit the virion to infect that cell regardless of its level of CCR5 expression, metabolic state, or any other property that might facilitate productive infection. In contrast, the interaction of the HIV envelope with α}₄β₇ is CD4-independent. Rather than being hidden from the virus, as is CCR5, the extended form of α₄β₇ is estimated to rise >20 nM from the cell membrane (25). By comparison, CD4 rises only approximately 7 nM from the cell surface (26). Thus, α₄β₇ represents a structurally prominent gp120-binding receptor on the surface of a highly susceptible subset of CD4 ^{T cells that HIV-1 can engage independently of CD4.}

We show that α₄β₇, CD4, and CCR5 colocalize on the cell membrane. Moreover β₇ is sufficiently close to CD4 that these two receptors can be coprecipitated. This study demonstrates that these two receptors appear together in a complex on the surface of a CD4 ^{T cell. Although many details remain unknown, we speculate that virions (or envelope spikes on infected cells) are likely to first encounter the activated/extended form of α}₄β₇ and subsequently engage CD4. This scenario is consistent with a recent structural analysis of the HIV envelope that places the V1V2 domain of gp120 near the apex of the trimeric spike (27). An LDV tripeptide in the gp120 V2-loop mediates binding to α₄β₇ (9), placing this critical contact site at or near the tip of the spike, a position well suited for an initial engagement with α₄β₇. It is noteworthy that such positioning may provide opportunities for therapeutic or vaccine-induced intervention.

Two issues surrounding the interaction between HIV-1 and α₄β₇ require further investigation. First, what role does gp120 signal transduction through α₄β₇ play in infection? Such signals may facilitate the infection of activated cells but may also promote infection of suboptimally activated α₄β₇ CD4 ^{T cells. In this regard, we have previously reported that among the subsets of CD4 T cells, the α}₄β₇ ^{subset is distinct in responding to gp120-mediated signals by rapidly activating LFA-1. This is potentially important because LFA-1 plays a central role in stabilizing virological synapses and plays a key role in the preferential infection of memory CD4 T cells (}28). Yet, the specific sequence of events that follow cell-free or budding virion engagement of α₄β₇ and how those events facilitate infection remain to be determined. Second, activated/extended α₄β₇ is found principally in Peyer's patches, mesenteric lymph nodes, and lamina propria, all of which play central roles in the early stages of infection following sexual transmission. We do not yet know in which of these tissues the interaction between HIV-1 and α₄β₇ is most relevant. Moreover, α₄β₇ ^{CD4 T cells have also been found in genital mucosa (}3, 14), and we detected α₄β₇ ^{CD4 T in cervical cytobrush samples. In this regard, the genital mucosa does not contain well-organized immune-inductive sites; instead, it relies upon α}₄β₇ ^{CD4 T cells trafficking from more organized immune-inductive sites, including Peyer's patches (}29). Thus, we cannot exclude the possibility that the initial and most relevant interaction between HIV-1 and α₄β₇ occurs in the genital mucosa in the earliest days following exposure. Whether this interaction involves free virions, infected cells, or virions sequestered by dendritic cells (15, 30) remains to be determined.

In conclusion, we have demonstrated that α₄β₇ demarcates a subset of CD4 ^{T cells that are prone to highly productive infection. In rectum and colon biopsies, the highest frequency of metabolically activated CD4 T cells appears in the α}₄β₇ ^{population. Although there are other markers of activation on CD4 T cells, HIV-1 has likely acquired α}₄β₇ -specificity because this receptor lies in close physical proximity to the CD4 receptor and because it mediates the trafficking of CD4 ^{T cells to the gut tissues that are an ideal environment to firmly establish infection after sexual transmission. Noting that sexual transmission occurs inefficiently, we propose that, by interacting with α}₄β₇, HIV-1 increases the likelihood of successful infection following exposure at mucosal surfaces. In future studies, it will be important to determine at which point during sexual transmission the interaction between HIV-1 and α₄β₇ is most relevant, and in addition whether targeting this interaction can contribute to strategies designed to prevent infection.

Methods

Cells and Reagents.

Freshly isolated PBMCs were obtained from healthy donors and separated by Ficoll-Hypaque. Purified CD4 ^{T cells and NK cells were obtained by negative selection using magnetic beads (StemCell Technologies). Cultured CD4 T cells were activated with OKT3, IL2 (20 IU/mL) and RA (10 nM) unless otherwise specified. Separation of α}₄β₇ ^{vs. α}₄β₇ ^{subsets was carried out using negative selection with mAbs CCR7 (150503) and CCR5 (2D7) followed by magnetic selection with anti-mouse beads (Miltenyl). Gut-derived cells were obtained from healthy donors under a National Institute of Allergy and Infectious Diseases-approved IRB protocol. Approximately 20 punch biopsies were obtained from each specified tissue/site and treated with collagenase (Sigma) using standard procedures. Cytobrush samples were obtained as described in ref.} 31 under a study protocol approved by the Research Ethics Boards at the University of Toronto and the Kenyatta National Hospital (Nairobi, Kenya). Retinoic acid was obtained from Sigma Chemical. Integrin mAbs were purchased from Serotech, Chemicon and R&amp;D Systems, while other surface marker antibodies were purchased from BD Biosciences and eBioscience. The chicken polyclonal antisera to β₇ was purchased from Lifespan Biosciences. Act-1 was provided by Dr. Stephen Shaw (National Cancer Institute). The Ki-67 mAb was purchased from BD and was used in conjunction with BD permeabilization buffer. Some mAbs were directly conjugated with Alexa Fluor reagents (Invitrogen). Expression and purification of recombinant envelope proteins were described in refs. 9 and 32. The full-length chimeric proviral clone AD8-SF162 (33) was provided by Ronald L. Willey.

HIV Infection.

AD8-SF162 was produced by transient transfection into 293T fibroblasts. Viral stocks were normalized by p24 antigen ELISA (Perkin-Elmer). Purified CD4 ^{T were cultured in medium containing the anti CD3 mAb OKT3 and supplemented with RA and IL2 for 24 h, then rinsed and cultured for an additional 5 days in RA and IL2 before infection. Four hours post virus inoculation cells were washed and resuspended in fresh media containing RA and IL2. Viral replication was measured by intracellular p24 staining (mAb RD1, Beckman Coulter) after fixation and permeabilization using cytofix-cytoperm (BD Biosciences) or by p24 GAG antigen ELISA of culture supernatants.}

Flow Cytometry Binding Assays.

Cells were stained with fluorescein-labeled mAbs using standard procedures (4 °C, 30 min) preceded by Fc receptor blocking with both mouse and human IgG. Staining buffer was freshly prepared and contained10 mM HEPES, 150 mM NaCl (HBS Buffer) with 100 μM CaCl₂ and 1 mM MnCl₂ unless otherwise specified. Buffer without divalent cations contained 10 mM EDTA. Staining with gp120 was carried out with biotinylated gp120 (E-Z link biotinylation reagent Pierce) followed by PE-conjugated neutravidin (Pierce). Data were obtained using a BD FACSCalibur.

Confocal Microscopy.

Freshly isolated gut-derived cells and RA-stimulated peripheral blood CD4 ^{T cells were surface stained with the α}₄β₇ specific mAb Act-1 followed by goat anti mouse IgG Alexa Fluor 568, the CD4 mAb OKT4 Pacific Blue (eBioscience), and the CCR5 mAb 2D7 Alexa Fluor 488. Stained cells were fixed with 2% paraformaldehyde mounted onto slides and coverslipped with prolong gold (Invitrogen). Cells were imaged with a Leica SP 5 confocal microscope using a 63×/1.4 objective. Fluorochromes were excited with the 405-, 488-, and 561-nm laser lines. Fluorescence emission was collected as follows to minimize crosstalk: 415–465 nm pacific blue, 493–545 nm Alexa Fluor 488, and 575–620 nm Alexa Fluor 568. All confocal images were processed and cropped using Imaris imaging software (Bitplane AG).

CD4/β₇ CoPrecipitation.

Purified RA/IL2 cultured CD4 ^{T cells (20 × 10) were washed, resuspended in PBS and then treated for 30 min with 2 mM DTSSP (Pierce) at 24 °C. The reaction was stopped with 20 mM Tris, pH 7.5, and lysed in buffer containing 1% Triton X-100. Lysates were incubated with either OKT4 or mouse IgG1 as specified, and precipitated with Protein G beads (GE Healthcare). Precipitated proteins were electrophoresed on a 6% polyacrylamide gel under reducing conditions and immunoblotted with an anti β}₇ polyclonal antisera (Lifespan Technologies).

Data Analysis.

Statistical analysis was performed with STATA/IC 10.0 (Stata Corporation). Considering that the sample size was small and data included more than a single biopsy from the same donor (non-independent), statistical comparisons were performed using the non-parametric Wilcoxon signed-rank test for paired sample. P < 0.05 was considered significant.

Supplementary Material