South African HIV-1 subtype C transmitted variants with a specific V2 motif show higher dependence on α4β7 for replication.

Products Related

Product# 1039 HIV-1 YU2 (M Tropic)Envelope Glycoprotein gp41

Product# 1081 HIV-1 gp120 (ADA)

Product# 1011 HIV-1 gp120 (subtype C)

Product# 1031 HIV-1 gp120 (YU2)



The integrin α4β7 mediates the trafficking of immune cells to the gut associated lymphoid tissue (GALT) and is an attachment factor for the HIV gp120 envelope glycoprotein. We developed a viral replication inhibition assay to more clearly evaluate the role of α4β7 in HIV infection and the contribution of viral and host factors.


Replication of 60 HIV-1 subtype C viruses collected over time from 11 individuals in the CAPRISA cohort were partially inhibited by antibodies targeting α4β7. However, dependence on α4β7 for replication varied substantially among viral isolates from different individuals as well as over time in some individuals. Among 8 transmitted/founder (T/F) viruses, α4β7 reactivity was highest for viruses having P/SDI/V tri-peptide binding motifs. Mutation of T/F viruses that had LDI/L motifs to P/SDI/V resulted in greater α4β7 reactivity, whereas mutating P/SDI/V to LDI/L motifs was associated with reduced α4β7 binding. P/SDI/V motifs were more common among South African HIV subtype C viruses (35%) compared to subtype C viruses from other regions of Africa (<8%) and to other subtypes, due in part to a founder effect. In addition, individuals with bacterial vaginosis (BV) and who had higher concentrations of IL-7, IL-8 and IL-1α in the genital tract had T/F viruses with higher α4β7 dependence for replication, suggesting that viruses with P/SDI/V motifs may be preferentially transmitted in the presence of BV in this population.


Collectively, these data suggest a role for α4β7 in HIV infection that is influenced by both viral and host factors including the sequence of the α4β7 binding motif, the cytokine milieu and BV in the genital tract. The higher frequency of P/SDI/V sequences among South African HIV-1 subtype C viruses may have particular significance for the role of α4β7 in this geographical region.

Electronic supplementary material

The online version of this article (doi:10.1186/s12977-015-0183-3) contains supplementary material, which is available to authorized users.

Keywords: HIV entry, α4β7, Tripeptide-binding motif, Bacterial vaginosis, Cytokines

Go to:


The primary site of HIV replication following infection is the gut associated lymphoid tissue (GALT) [1, 2]. In order to migrate to the GALT and other tissues, leukocytes engage with adhesion molecules expressed on the surface of vascular endothelial cells. One of these receptors is the integrin α4β7 that binds to monomeric gp120 [3]. Since the genital mucosa does not contain organised immune-inductive sites, it relies on α4β7+ T cells to traffic from other sites including the Peyer’s patches [4]. Therefore, the ability of α4β7+ T cells to home to secondary lymphoid tissues and the GALT [5, 6], coupled with their presence at the site of sexual transmission of HIV [7, 8] and co-expression with multiple HIV susceptibility markers [9], suggests that the initial and most relevant site for the gp120-α4β7 interaction is the genital mucosa.

The natural ligands of α4β7 (MAdCAM-1, VCAM-1 and fibronectin [10]) all bind through structurally homologous binding motifs that comprise three residues with a central aspartic acid; Leu-Asp-Thr (LDT), Ile-Asp-Ser (IDS) and Leu-Asp-Val (LDV), respectively [11]. The principal contact sites for these natural ligands are present on the α4-chain [12]. By blocking α4β7 activity with inhibitory antibodies [13], Arthos et al. showed that gp120 binds to α4β7 in a manner that mimics the natural ligands [3]. The V2 domain of gp120 contains a similar tri-peptide motif at position 179–181 (HXB2 numbering) with the aspartic residue at position 180 being 98% conserved across all HIV isolates [3, 14]. The ITGA4 gene that encodes the α4 subunit shows no polymorphisms in humans and did not correlate with HIV transmission or disease progression [15]. Nevertheless, there appears to be significant variation in the levels of α4β7 reactivity among viruses from different individuals [3]. This suggests that it is the contact residues in gp120 that influence α4β7 affinity. This is bolstered by data that showed differences in the sequence of the α4β7 tri-peptide motif were linked to the differential dissemination potential of distinct HIV-1 genetic forms in China [16]. Recently, Tassaneetrithep et al. described a tri-peptide sequence just upstream of the α4β7 motif as a determinant of integrin binding [17], suggesting that additional viral properties play a role in reactivity with α4β7.

Although gp120 binds α4β7 this interaction is not essential for viral entry, unlike CD4 and CCR5 [3]. Rather, α4β7 is thought to act as an attachment factor, offering a selective advantage for HIV entry by lowering the entropic barrier that slows the ligation of envelope spikes to CD4 and CCR5 [18]. Thus, the gp120–α4β7 interaction may be particularly important during the earliest stages of HIV infection. CD4+ T cells expressing high levels of α4β7 are more susceptible to HIV-1 infection partly because this subset also expresses high levels of CCR5 [9]. This phenotype extends to sites of initial HIV infection such as blood, rectum, colon and genital mucosa of the female reproductive tract [7–9]. However, other studies have failed to confirm any impact of α4β7 on replication in vitro [19–21]. Despite this controversy, when healthy macaques were treated with an anti-α4β7 mAb (Act-1), they were protected from transmission by low-dose SIVmac251 challenge [22]. This antibody also reduced viremia and proviral DNA in the GALT in a high dose challenge model although it did not extend to protection [23]. In addition, a recent study has shown that the number of α4β7+ CD4+ T cells at the site of rectal transmission is a risk factor for productive HIV infection in rhesus macaques [24]. Sexually transmitted infections such as HSV-2 have also been shown to increase expression levels of α4β7+ and enhance the risk for vaginal SHIV infection [25].

To further clarify the role for α4β7 in HIV infection, we made use of longitudinal viruses from the CAPRISA Acute Infection cohort based in Durban, South Africa, a region with one of the highest HIV incidence rates in the world [26]. We devised an α4β7-inhibition replication assay and tested dependence of the viruses on α4β7 for entry and replication using inhibitory mAbs. Here we show that variation in the α4β7 binding motif influences T/F virus α4β7-dependent replication. Furthermore, the immune environment in the genital mucosa at the time of HIV infection correlated with the transmission of particular binding motifs which are highly prevalent in South African subtype C viruses.

Go to:


α4β7 expressed on 293T cells binds infectious HIV

While monomeric gp120 has been shown to bind α4β7 [3], we sought to determine whether biologically relevant forms of HIV envelope expressed on the viral membrane, also bound the integrin. For this, plasmids encoding human α4 and β7 subunits were co-transfected and expressed in 293T cells (which do not express any HIV receptors) and stained with CD49d (α4)-PE (targeting an epitope on α4 overlapping the region of gp120 binding) and β7-FITC to confirm co-expression (Additional file 1, panel A). These cells were used in a direct binding assay, where HIV bound to α4β7 was detected by p24-FITC (Additional file 1, panels B and C). These cells were also used in a competition binding assay, where a reduction of CD49d (α4)-PE binding was measured relative to α4β7+ cells incubated without virus or the inhibitory mAbs, HP2/1 (anti-α4 mAb) and Act-1 (anti-α4β7 mAb). Two types of infectious viruses were used for these experiments; infectious molecular clones (IMCs) which comprised the entire proviral genome of isolates of interest and infectious envelope clones (IECs) which made use of a common pNL4-3 delta Env backbone co-transfected with different gp160 genes.

In agreement with published studies, SF162 was shown to bind to α4β7-expressing cells in a direct binding assay [3, 19] (Figure 1a). No significant differences were noted between the SF162 IMC or IEC in either of the binding assays or between another two IMC/IEC pairs used in this study (CAP210 and CAP239) (Additional file 2). Similar to what was shown previously (using monomeric gp120) [27], the CAP88 T/F IEC bound better to α4β7 expressed on 293T cells than the CAP88 12 month IEC (Figure 1a). These findings were confirmed in a competition assay where the CAP88 T/F was better able to compete for integrin binding with fluorescently labelled α4β7-directed mAbs than the 12 month IEC (*p = 0.018; paired t test), although not as efficiently as the inhibitory mAbs (Figure 1b). These data confirm the capacity of biologically relevant cell surface-expressed HIV-1 Env to bind α4β7.


An external file that holds a picture, illustration, etc.
Object name is 12977_2015_183_Fig1_HTML.jpg

Open in a separate window

Figure 1

HIV binding to cell-surface expressed α4β7. a Direct binding of SF162 (left panel) and CAP88 (right panel) viruses to α4β7 expressed on 293T cells measured by p24-FITC staining. For SF162, both the IEC and IMC (solid and dashed black lines respectively) bound α4β7 at similar levels which were higher than the untransfected α4β7 293T cells (grey line). The IEC of CAP88 T/F (solid black line) bound better than CAP88 12M (dashed black line). This is representative of 3 independent experiments. b Competition assay where CAP88 T/F (black dashed) bound to the integrin is measured as a decrease in α4-PE antibody binding relative to unbound α4β7 transfected 293T cells (black solid). HP2/1 and Act-1 binding in the absence of virus were used as positive controls (red and blue respectively) while the negative control was untransfected cells incubated with virus. Results are representative of four independent experiments. The CAP88 T/F virus bound significantly better to the α4β7 integrin compared to the 12 month IEC in the competition binding assay as shown in the bar graph (p = 0.018; paired t test). Bars represent the mean binding percentage of three independent experiments and the error bars represent the SEM.

Anti-α4β7 antibodies partially inhibit virus replication

We next determined whether binding to α4β7 enhanced HIV infection of primary CD4+ T cells induced to express α4β7 using all-trans retinoic acid (ATRA), the effect of which was measured by flow cytometry (Additional file 3). For this, we developed an α4β7 inhibition replication assay which measured viral replication (levels of p24) in ATRA-treated CD4+ T cells in the presence of HP2/1 or Act-1 mAbs over a 10 days period. Accurate titration of the inhibitory mAbs was crucial as saturating concentrations were shown to enhance viral replication in this assay (Additional file 4). This effect is likely to be the result of signalling processes and homotypic clustering of cells enhanced by ligands to and antibodies against α4β7 [12, 19, 28]. This titration is unique to this study and may have allowed us to develop an assay in which we see consistent partial inhibition of replication by the inhibitory mAbs, in contrast to other studies [19, 21]. The optimal mAb concentration for viral inhibition by HP2/1 was 0.275 nM and for Act-1 was 2.2 pM which were used in all subsequent experiments.

A total of 60 IECs from 11 CAPRISA participants were tested in this assay to determine if the α4β7 receptor was commonly used by HIV. While replicative capacity differed among IECs, all showed lower levels of replication (range 15–88%) in the presence of antibodies to α4β7, confirming usage of the integrin (Additional file 5). There was no difference in replication inhibition levels between HP2/1 and Act-1, suggesting that the inhibition was a result of blocking the heterodimeric α4β7 surface molecule (Figure 2a). Treatment with anti-α4β7 mAbs resulted in only partial inhibition of replication, compared to an anti-CD4 mAb that resulted in complete abrogation of replication of all viruses consistent with its essential role in viral entry. These data re-enforce previous findings that α4β7 serves as an attachment factor and unlike CD4 and CCR5, is not essential for HIV infection [3, 19, 21, 29].

An external file that holds a picture, illustration, etc.
Object name is 12977_2015_183_Fig2_HTML.jpg

Figure 2

HIV dependence on α4β7 for replication over the course of infection. a Kinetic growth curves of IECs from CAP88 T/F, 1 month, 12 and 39 months post-infection (green lines) in the presence of α4β7 inhibitory mAbs Act-1 (blue) and HP2/1 (red) or anti-CD4 mAb (purple) measured as p24 (ng/ml) over 10 days. Inhibition by Act-1 and HP2/1 was partial at the point of exponential growth of the virus control and did not differ between the two mAbs. Curves are representative of three independent experiments, each one in triplicate and error bars representative of SEM. b Longitudinal IECs (from 3 individuals) including the T/F virus and multiple 2–39 months post-infection viruses were tested for their dependence on α4β7 for viral replication. The percentage inhibition by HP2/1 (red) and Act-1 (blue) is expressed as the difference in p24 concentration between HP2/1 or Act-1 treated and untreated cells at the point of exponential viral growth. Differences in dependence across all time points in each of these three individuals were significant by a repeated measures ANOVA (p < 0.0001) as well as between CAP88 T/F and 1 month p.i, (**p < 0.001) CAP200 T/F and 6 months p.i. (*p < 0.01) and CAP206 T/F and 2 months p.i (**p < 0.001). Pairwise comparisons were adjusted by the Tukey method. Bars represent means of between two and three independent experiments with the error bars indicating SEM.

The dependence on α4β7 for viral replication changes over time

In order to assess whether the dependence on α4β7 varied over the course of HIV infection, we selected three participants who had a complete longitudinal set of IECs, including the T/F virus and viruses from shortly after infection (1 or 2 months post-infection) as well as later time points up to 3 years post-infection. All 3 individuals showed a similar pattern, with the T/F virus having a greater dependence on α4β7 for replication than viruses from 1 to 2 months (CAP88, CAP200 and CAP206; p < 0.001, p < 0.01 and p < 0.001 respectively using a repeated measures ANOVA) (Figure 2b). However, α4β7 dependence of viruses from later time points was not significantly different from T/F viruses. Longitudinal samples from additional CAPRISA participants showed a similar trend, but this analysis was limited because clones from either the T/F or later time-points were not available (Additional file 5). We found no association between dependence on α4β7 and markers of disease progression such as CD4 counts or viral loads when corrected for duration of infection (data not shown).

Analysis of longitudinal sequences revealed no changes in the α4β7 tri-peptide motif in CAP88 and CAP206 while in CAP200 there was a change from SDV to PDI by 6 months post infection, but this was due to dual infection (Sheward, unpublished) (Additional file 6). For CAP88, there were only three amino acid differences between the T/F and the 2 month clone; L568R, a highly conserved residue in the N-heptad repeat of gp41 and two changes in the cytoplasmic tail. For the CAP206 pair there was an introduction of a predicted N-linked glycan (PNG) at position 462 in V5 and a D474N mutation in the C5 region (Additional file 7). CAP200 showed a total of 39 non-synonymous changes in envelope at 6 months Overall, no common sequence signature was associated with changes in α4β7 reactivity among these three participants.

Since decreased loop length and PNG density of the gp120 V1/V2 and C3/V4 regions have previously been shown to correlate with increased α4β7 binding [27], we analysed these features among all 60 clones. We found that α4β7 dependence positively correlated with the length and predicted glycan density of V1/V2 which includes the α4β7 binding site and negatively correlated with C3/V4 length, however these associations were only weakly supported (Additional file 8, panel A). We also compared individual predicted glycan sites (excluding those in variable regions which could not be accurately aligned) across all 60 sequences (Additional file 8, panel B). Viruses with high α4β7 dependence had significantly higher frequencies of PNG234 (p = 0.009) and PNG334 (p = 0.006), while PNG332 (p = 0.026) was present less frequently as compared by the Fisher exact test. These data suggest that specific glycans may play a role in α4β7 dependence but it is likely that additional factors influence interactions with the integrin.

Impact of host factors at transmission on α4β7 dependence for replication

Since the role of α4β7 is likely to be most relevant at transmission, we focused on the 8 T/F viruses included in this study. These exhibited a wide range of Act-1 inhibition (22–69%) indicating that high α4β7 dependence is not a typical feature of T/F viruses (Figure 3a). We investigated if this variation could be explained by host factors present at the time of transmission. STIs have been identified as a major cause of inflammatory cytokine upregulation and immune cell recruitment to the genital mucosa, in some cases typified by the homing function of α4β7 [7, 30]. Given this and the integrin’s role in supporting HIV replication, we considered whether these factors could create an environment conducive for α4β7 interaction. Only Trichomonas vaginalisChlamydia trachomatis and bacterial vaginosis (BV) were detected in some individuals at the time of transmission. Strikingly, individuals with BV (Nugent scores ≥7) had viruses with significantly higher dependence on α4β7 than those who did not (61.58 vs. 29.64% α4β7 dependence; p = 0.029 Mann–Whitney test) (Figure 3b).

An external file that holds a picture, illustration, etc.
Object name is 12977_2015_183_Fig3_HTML.jpg

Figure 3

Bacterial vaginosis and genital cytokines associated with α4β7-dependent T/F viruses. a T/F viruses of CAP225, CAP88, CAP200 and CAP206 showed high (>50%, indicated by the dotted line) α4β7 dependence (green) while the remaining four showed lower α4β7 dependence (blue). Dependence on α4β7 was determined using Act-1 mAb inhibition. Bars represent the mean of four independent experiments, with error bars indicating SEM. b Individuals infected with T/F viruses that had higher dependence on α4β7 were significantly more likely to be BV positive at the time of infection (p = 0.029; Mann–Whitney test). c Concentrations of cytokines in CVL (n = 31) and plasma (n = 30) were determined and correlated with T/F virus dependence on α4β7 for replication. IL-7, IL-8 and IL-1α showed significant univariate correlations in the CVL. In contrast only eotaxin was significantly associated in plasma (shown in the grey boxes) while IL-8, IL-1α and IL-7 showed no correlation in plasma. No significance was maintained at a multivariate level. Relevant p values and Spearman’s coefficients are shown where *p < 0.05.

Next, we investigated the cytokine milieu in the genital compartment at the time point when the T/F viruses were isolated. The concentrations of interleukin (IL)-7 (r2 = 0.89), IL-8 (r2 = 0.78) and IL-1α (r2 = 0.86) in cervicovaginal lavages (CVLs) correlated significantly with α4β7 dependence in a univariate analysis (p = 0.007, p = 0.038, p = 0.013 respectively by Spearman’s correlation) but this was lost when the p-values were adjusted for multiple comparisons, likely a result of the small sample size (Figure 3c; Additional file 9). These associations were not mirrored in plasma where only eotaxin, an eosinophil chemoattractant whose function is mediated by α4β7 [31], showed a significant univariate correlation (p = 0.031; r2 = 0.80) but a non-significant adjusted p value of 0.899.

The sequence of the α4β7-binding motif influences virus binding and replication

While the aspartic acid in the tri-peptide binding motif in the V2 domain is highly conserved, there is variation at the first and third amino acid residues. Interestingly, the sequence of the tri-peptide α4β7 binding motifs could be used to stratify the 8 T/F viruses based on replication dependence (Figure 4a). The 4 T/F viruses with high α4β7 reactivity had P/SDI/V motifs while the 4 T/F viruses with low α4β7 reactivity had LDI/L motifs (p = 0.029, Mann–Whitney test).

Open in a separate window

Figure 4

Viruses with P/SDI/V α4β7 binding motifs are more reactive with the integrin. a T/F viruses with P/SDI/V showed a higher dependence on the integrin compared to those with LDI/L motifs (n = 4, blue) (*p = 0.029; Mann–Whitney test). b Mutation of CAP8 T/F from LDI to PDI and CAP256 T/F from LDL to SDI resulted in a significant increase in α4β7 dependence (*p = 0.02 for both, paired t test). Bars represent three independent experiments and error bars indicate the SEM. c CAP8 and CAP256 mutants also showed increased α4β7 binding compared to the wild-type in the direct assay using p24-FITC MFI as a read-out, representative of three independent experiments.

To determine if the sequence of the motif impacted on binding to α4β7, we mutated the tri-peptide motif from P/SDI/V to LDI/L (or vice versa) in 5 of the 8 T/F viruses. The α4β7 binding motif of the CAP8 T/F virus (which showed low α4β7 dependence) was mutated from LDI to PDI. Similarly the CAP256 T/F was mutated from LDL to SDI. In both cases, mutated viruses had significantly increased α4β7 dependence (p = 0.02, paired t test, Figure 4b). Furthermore, binding to the integrin as measured by the direct p24 binding assay was enhanced when motifs were mutated from LDI/L to P/SDI (Figure 4c). Notably, when viral motifs were mutated from P/SDI to LDI/L, none of the mutated viruses were able to replicate.

Analysis of the α4β7 binding motif, presence of BV and the genital and plasma cytokine milieu in the CAPRISA 002 cohort

To further examine the host and viral factors associated with α4β7 dependence, we analysed additional women in the CAPRISA 002 cohort where STI, cytokine and sequence data were available. Of the 30 women who had STI clinical information, 18 were infected with viruses with P/SDI/V and 12 with LDI/L motifs. In this larger analysis, BV diagnosis (Nugent score ≥7) was significantly associated with viruses having P/SDI/V motifs (17/18 P/SDI/V motifs vs 7/12 LDI/L motifs; p = 0.026, Fisher exact test) (Figure 5a), mirroring what was seen in the smaller sub-group of 8 women. Using available cytokine concentrations for CVL (n = 25 women) and plasma (n = 28 women), we investigated the relationship with different viral motifs in each compartment. CVL exhibited a dramatically different association profile to plasma (Figure 5b). In CVL, IL-17 showed the strongest positive association with the P/SDI/V motif and IL-10 the strongest inverse correlation, although these did not correlate with α4β7 dependence (Additional file 9). Of the three cytokines in CVL (IL-7, IL-8 and IL-1α) that were previously shown to correlate with α4β7 dependence for replication (Figure 3c), IL-7 and IL-8 concentrations were 2- to 5-fold higher in CVL from women who were infected with viruses containing P/SDI/V motifs compared to women without these motifs in agreement with the original observation.


An external file that holds a picture, illustration, etc.
Object name is 12977_2015_183_Fig5_HTML.jpg

Leave a comment

All comments are moderated before being published