The V2 domain of HIV gp120 mimics an interaction between CD4 and integrin ⍺4β7

The CD4 receptor, by stabilizing TCR-MHC II interactions, plays a central role in adaptive immunity. It also serves as the HIV docking receptor. The HIV gp120 envelope protein binds directly to CD4. This interaction is a prerequisite for viral entry. gp120 also binds to ⍺4β7, an integrin that is expressed on a subset of memory CD4+ T cells. HIV tropisms for CD4+ T cells and gut tissues are central features of HIV pathogenesis. We report that CD4 binds directly to ⍺4β7 in a dynamic way, consistent with a cis regulatory interaction. The molecular details of this interaction are related to the way in which gp120 interacts with both receptors. Like MAdCAM-1 and VCAM-1, two recognized ligands of ⍺4β7, the binding interface on CD4 includes 2 sites (1° and accessory), distributed across its two N-terminal IgSF domains (D1 and D2). The 1° site includes a sequence in the G β-strand of CD4 D2, KIDIV, that binds directly to ⍺4β7. This pentapeptide sequence occurs infrequently in eukaryotic proteins. However, a closely related and conserved sequence, KLDIV, appears in the V2 domain of gp120. KLDIV mediates gp120-⍺4β7 binding. The accessory ⍺4β7 binding site on CD4 includes Phe43. The Phe43 aromatic ring protrudes outward from one edge of a loop connecting the C’C” strands of CD4 D1. Phe43 is a principal contact for HIV gp120. It interacts with conserved residues in the recessed CD4 binding pocket. Substitution of Phe43 abrogates CD4 binding to both gp120 and ⍺4β7. As such, the interactions of gp120 with both CD4 and ⍺4β7 reflect elements of their interactions with each other. These findings indicate that gp120 specificities for CD4 and ⍺4β7 are interrelated and suggest that selective pressures which produced a CD4 tropic virus that replicates in gut tissues are linked to a dynamic interaction between these two receptors.


Introduction
The discovery that the HIV gains entry into CD4 + T cells by binding directly to the CD4 receptor [1][2][3], followed by the determination that neutralizing antibodies are directed exclusively to the gp120/41 envelope protein [4], provided a rationale for developing an HIV vaccine immunogen that would elicit CD4 binding site (CD4bs) antibodies.Features intrinsic to envelope-CD4 interactions suppress the elicitation of such antibodies [5,6].CD4 is comprised of 4 tandemly arrayed Ig -like ectodomains (D1-D4) followed by short transmembrane and cytoplasmic domains [7].As a first step, gp120 engages a phenylalanine at position 43 (Phe 43 ) in D1 [8,9].The Phe 43 aromatic sidechain protrudes from the edge of an upward facing loop near the apex of CD4 [10].The CD4bs on gp120 that accommodates Phe 43 appears as a spherical cavity.Insertion of Phe 43 into this cavity precipitates a significant conformational reorganization of the gp120/41 prefusion trimer [11][12][13][14].This includes the decoupling of the V2 and V3 domains, which in turn facilitates V3-coreceptor engagement and, ultimately, virus fusion with the cell membrane [14][15][16].Although we understand gp120-CD4 interactions in detail, the selective pressures that produced this interaction in the first instance remain unclear.
In considering the selective pressures that resulted in a CD4bs on gp120 that engages Phe 43 it is useful to consider CD4's normal role.Together with Lck, CD4 transiently migrates into the TCR where it binds to MHC II on APCs and stabilizes TCR interactions [17][18][19].Of note, gp120 and MHC II engage CD4 in distinct ways [20].Substitution of Phe 43 eliminates gp120 reactivity but does not disrupt CD4-MHC II interactions [21,22].Given its exposed presentation, it would appear that Phe 43 serves some function.Identifying a natural role for this aa might begin to reveal the selective pressures which resulted in the specialized interaction between gp120 and CD4.
In addition to CD4, the HIV envelope engages several ancillary receptors that may facilitate mucosal transmission.These include DC-SIGN [23], Siglec-1 [24], and the mannose receptor [25].Because it is difficult to model mucosal transmission in vitro, and even in non-human primates (NHP), the importance of these interactions is subject to debate.Integrin α 4 β 7 (α 4 β 7 ) is an additional ancillary receptor that we and others have shown binds gp120 [26][27][28][29].α 4 β 7 mediates T cell trafficking to gut associated lymphoid tissues (GALT) via interactions with two adhesion receptors, MAdCAM-1 and VCAM-1, that are expressed on high endothelial venules that line gut tissues [30,31].α 4 β 7 is not required for HIV entry or infection, and not all recombinant gp120s bind to it.We have speculated that α 4 β 7 provides a way for HIV to increase transmission efficiency by enhancing access to the plentiful supply of activated CD4 + CCR5 + T cells in GALT.In support of this proposition, higher frequencies of α 4 β 7 high CD4 + memory T cells correlate with increased susceptibility to HIV acquisition [32].Additionally, α 4 β 7 high CD4 + memory T cells collected from HIV infected subjects in the early stages of infection (FIEBIG II/III) are enriched in HIV proviral DNA, relative to other CD4 + T cell subsets [33], and this same subset is selectively infected and depleted from SIV infected macaques [34].
Finally, an α 4 β 7 -specific mAb was shown to block vaginal transmission of SIV in a repeated low-dose challenge model [35].
In a previous study we determined that α 4 β 7 clusters with CD4 and CCR5 on the surfaces of some α 4 β 7 high CD4 + effector memory T cells [36].This is puzzling.α 4 β 7 does not play a role MHC II-CD4/TCR interactions.Conversely, there is no direct role for CD4 in α 4 β 7 -mediated T cell trafficking.In this report we show that α 4 β 7 binds directly to CD4 in cis.In this way it adds to the list of integrins that engage counter receptors in both trans (opposing cell) and cis (same cell) [37] (S1A Fig) .We find that elements of this interaction are incorporated into the HIV envelope as a means of recognizing both receptors.These findings address the selective pressures that underlie the specific affinity of gp120 for CD4.

Results
HIV gp120 enhances clustering of CD4 and α 4 β 7 CD4, CCR5 and α 4 β 7 colocalize on the surfaces of primary CD4 + T cells derived from both colon biopsies and blood derived CD4 + T cells cultured in retinoic acid (RA) [36].To determine whether CD4 and α 4 β 7 come into close-proximity, we carried out a flow-cytometry based proximity ligation assays (PLA) on freshly isolated CD4 + T cells using CD4 and β 7 mAbs as probes (S1B Fig naïve cell subsets.We observed positive PLA between CD4 and α 4 β 7 on both subsets, however the frequency of measurable PLA on α 4 β 7 high memory cells was ~2.5 greater than on naïve CD4 + T cells, indicating HIV gp120, like MAdCAM-1, can deliver costimulatory signals through α 4 β 7 in a way that promotes cellular proliferation and HIV infection [38,39].This finding prompted us to ask whether gp120 -mediated signaling might influence CD4-α 4 β 7 clustering.To this end, we employed fluorescence resonance energy transfer (FRET) measured by fluorescence lifetime imaging microscopy (FLIM).Primary CD4 + T cells were cultured in the presence of anti CD3, IL-2 and RA for 8 days and then placed on glass coverslips coated with HIV gp120 (stimulated) or without gp120 (unstimulated).Fluorescence lifetime measurements were performed on the CD4 donor mAb tagged with Alexa-488, in the presence or absence of an integrin β 7 acceptor mAb (tagged with Alexa-546).FLIM values were acquired at 20' and 60' time points.Cells treated with gp120 showed a substantial reduction in lifetime signals at 20' and a further reduction at 60' (S1C and S1D Fig) .These reductions in the lifetime of the donor indicate proximity of the CD4 and β 7 probes, with an approximate FRETing distances of <15nm, indicating a direct interaction.In considering the role such an interaction might play in CD4 + T cell function, we note numerous reports of integrins regulating the activity of other cell surface receptors through direct cis interactions [37] (S1 Fig) .A number of these cis interactions involve receptors that, like CD4, present tandem N-terminal Ig -like domains [40][41][42][43][44]. With these reports in mind, we sought to gather additional evidence for a direct physical interaction between CD4 and α 4 β 7 .

Specific binding of CD4 to α 4 β 7
The FRET/FLIM results described above indicate a direct interaction between CD4 and α 4 β 7 .
A determination that CD4 exhibits a specific affinity for α 4 β 7 would support this finding.To this end we employed a fluorescently labelled soluble dodecameric CD4 Ig fusion protein, termed D1D2-Igαtp, that includes the two N-terminal Ig-like domains of CD4 (D1D2, aa 1-180), followed by IgG1 CH2 and CH3 domains, followed by the 18 aa tail-piece domain of IgA (αtp) [45].In this oligomer, 12 D1D2-Igs are arranged in a disulphide linked daisy-chain (Fig 1A).Primary CD4 + T cells, some of which express α 4 β 7 , were reacted with D1D2-Igαtp   1D).We determined that D1D2-Igαtp bound, like β 7 mAb, to a minor subset of memory cells at relatively high levels, and to naïve cells at a lower level.Although we do not exclude the possibility of trans interactions, the direct interaction we show by FRET/FLIM, combined with the specific binding of CD4 to α 4 β 7 leads us to conclude that CD4 and α 4 β 7 interact in cis on the surface of CD4 + T cells (Figs 1 and S1).Molecular details of CD4-α 4 β 7 interactions, described below, support this conclusion.
We next compared the α 4 β 7 reactivity of dodecameric D1D2-Igαtp to smaller oligomers.While endecamers and decamers bound α 4 β 7 at levels comparable to dodecamers, we observed a marked reduction (~80%) in binding with smaller forms (Fig 1E).We note that for many integrins, including α 4 integrins, ligand recognition is modulated by clustering and avidity effects [47].Finding that only higher order CD4 oligomers bind α 4 β 7 is consistent with the role that avidity frequently plays in integrin-ligand interactions.Moreover, it helps explain how the specific affinity of CD4 for α 4 β 7 has gone unrecognized until now.
MAdCAM-1 and VCAM-1 bind to the N-terminal headpiece of α 4 β 7 only when it adopts an active conformation [48].α 4 β 7 conformation is regulated via "inside out signaling" [49], but can be artificially manipulated in vitro by altering relative concentrations of divalent cations.High Mn ++ concentrations promote an active state.We compared the conformational requirements of CD4 binding to α 4 β 7 with those of MAdCAM-1 and VCAM-1 by staining α 4 β 7 -expressing RPMI8866 cells with D1D2-Igαtp in 6 buffers supplemented with different combinations of divalent cations.Like MAdCAM-Ig and VCAM-Ig, D1D2-Igαtp failed to bind α 4 β 7 in the absence of divalent cations (Fig 1F).All three ligands bound to α 4 β 7 in the presence of Mn ++ .We observed low levels of binding of the two natural ligands in the presence of Mg ++ and no binding of D1D2-Igαtp.We conclude that D1D2-Igαtp selectively engages an activated form of α 4 β 7 normally associated with ligand binding.

Peptide scanning of the α 4 β 7 binding site(s) on CD4 D1D2
To measure the specific affinity of CD4-α 4 β 7 interactions we developed a surface plasmon resonance (SPR) -based assay in which recombinant α 4 β 7 was passed over a surface coated with a 23 kD D1D2 protein (aa 1-183) [8].For comparison we prepared surfaces with recombinant MAdCAM-Ig, VCAM-Ig, monomeric gp120, and a SOSIP env trimer (S1 Table ).Assays were carried out in the presence of MnCl 2 to present α 4 β 7 in an active conformation.We obtained a dissociation constant (KD (M)) between α 4 β 7 and CD4 of 3.57e -9 .This rate of dissociation was We next modified our SPR -based assay in order to map the site(s) on CD4 that interacts with α 4 β 7 .In this modified assay α 4 β 7 was passed over a surface coated with CD4 D1D2 in the absence or presence of 42 overlapping 15 amino acid linear peptides spanning CD4 residues 1-178 (S2 Table ).Two peptides located in CD4 D1, #7 and #9, inhibited α 4 β 7 binding by ~17% (Fig 4A).The final peptide in CD4 D2, p42, inhibited binding by ~66%.Linear peptides often lack the true conformation of the protein from which they are derived.Additionally, the kinetic format of our assay allows for ligand-analyte rebinding during the association phase.These features render this inhibition assay somewhat insensitive.Nevertheless, the dynamic range of this assay is sufficient to identify interactions that are worthy of follow-up investigation (S4 Fig) .As such, our preliminary findings raised the possibility that both D1 and D2 domains of CD4 interact with α 4 β 7 .Of note, MAdCAM-1 and VCAM-1, like CD4, present two tandemly arrayed N-terminal IgSF domains (Fig 4B).In both cases, binding to α 4 β 7 is mediated by a 1˚binding site in their D1s and an accessory site, located in their D2s.The term accessory is a misnomer insofar as these sites plays an essential role in α 4 β 7 binding.Follow-up experiments focused on p9 and p42, which we herein refer to as pCD4 33-47 and pCD4 165-178 .pCD4 33-47 partially overlaps the D1 C'C" β-strands and a connecting loop.The primary gp120 contact site (Phe 43 ) lies at the N-terminal edge of this loop (Fig 4C).pCD4 165-178 includes Q 165 and K 166 .These residues lie primarily within the G β-strand of D2, close to the boundary with D3.Because peptide inhibition assays sometime mediate nonspecific effects, further analysis was performed to better establish the role of these two regions of CD4 in binding to α 4 β 7 .
CD4 165-178 binds directly to α 4 β 7 To further investigate the interaction between pCD4 165-178 and α 4 β 7 we attached a fluorescein dye to its amino terminus via a trioxatridecan-succinamic acid (Ttds) linker (Fig 5A ).A scrambled pCD4 165-178 was employed as a control.Both peptides were reacted with RA -treated primary CD8 + T cells that express high levels of α 4 β 7 .pCD4 165-178 reacted strongly in the presence, but not in the absence of Mn ++ (Fig 5B).The scrambled pCD4 165-178 showed limited binding (Fig 5C).α 4 mAbs 2B4 and HP2/1 reduced binding by ~80%, as did FG.FIB504 and vedolizumab inhibited binding by ~50%.This inhibition pattern resembles the one we obtained when inhibiting D1D2-Igαtp with the same antagonists (Fig 3).We conclude that the CD4 D2 G β-strand binds directly to the headpiece region of α 4 β 7 in a cation dependent manner.Support for this conclusion, in the form of site directed mutagenesis within the D2 G βstrand is provided below.
We and others have previously reported that some recombinant gp120s bind to α 4 β 7 in a cation -dependent manner [26][27][28]55,56] and determined that this interaction is mediated by conserved residues in the V2 domain.In support of this finding, a subtype B V2 consensus peptide spanning residues 170-184 (pV2 170-184 ) inhibits gp120 binding to α 4 β 7 [56].pV2 170-184  includes a conserved aliphatic aa-Asp motif (L 179 D 180 ).Substitution of L 179 D 180 reduces gp120 binding to α 4 β 7 substantially [26].Cardozo and colleagues showed, using a similar V2 peptide, that aa 170-172 (consensus QK[VE]) also plays a key role in gp120 binding to α 4 β 7 [28].With these observations in mind, we aligned pCD4 165-178 with pV2 170-184 (Fig 6B).Sequence similarity was evident.Most striking, CD4 residues 171-175 (KIDIV) are similar to V2 residues 178-182 (KLDVV).In V2, K 178 , L 179 and D 180 are the most frequently encoded amino acids at their respective positions (Fig 6C).Although V 181 in our subtype B V2 peptide does not align with CD4 I 174 , the presence of valine at this position is atypical. The majrity of gp120s (61%) in the Los Alamos National Laboratory (LANL) database encode either an isoleucine or leucine at position 181, such that the CD4 D2 sequence KIDIV is nearly identical to KLDIV, the most common V2 sequence spanning aa 178-182, and both include a metal ion coordination motif (ID and LD).Three additional conserved residues in this region of V2, Q 170 , K 171 , and F 176 could also be aligned with residues in pCD4 165-178 .In sum, a sequence overlapping the G βstrand of CD4 D2, that binds directly to α 4 β 7 , is nearly identical to a conserved sequence in gp120 V2 that also binds to α 4 β 7 .
Inspection of nonhuman primate orthologs indicates that CD4 residues Q 165 , K 166 , F 170 , along with KIDIV (aa 171-175) are conserved in 19 of 21 Old World monkeys, including those of Asian origin that are employed in NHP models of HIV infection, but are absent in New World monkeys and other mammals (Figs 6D, 6E and S3).Next, we considered the possibility that similar sequences might appear frequently in other cell surface proteins, particularly proteins with Ig -like domains, which would reduce the significance of the sequence similarity we observed.To this end, we queried the ~560,000 proteins included in the SwissProt database.We initially analyzed a motif, QKx(3)Fx[LI]D[IV]V, that includes all of the residues in CD4 D2 that align with gp120 V2 (x is any amino acid and brackets allow for either of two residues at a single position).Using the ScanProsite tool we determined that the probability this motif would appear randomly was 6.8 per 100,000 (p = 6.8e -6 ) (Fig 6F).We obtained 7 hits, all of which were primate CD4 genes (S3 Table ).We then restricted our search to include only KIDIV.The probability that this sequence appears randomly is 30 per 100,000 (p = 3e -5 ).We observed 41hits among 560,000 proteins.8 were primate CD4 genes.The only other cell surface receptor encoding KIDIV was IL-1R, which does not react with α 4 β 7 .This suggests that other elements of CD4 D2 are necessary for α 4 β 7 reactivity.These findings indicate that a CD4 D2 sequence that binds directly to α 4 β 7 is not common in the eukaryotic proteome, indicating that the presence of a closely related sequence in the gp120 V2 domain was worthy of further investigation.
CD4 165-178 and V2 170-184 cross-inhibit α 4 β 7 ligands The sequence similarity between pCD4 165-178 and pV2 170-184 suggested that they engage the α 4 β 7 headpiece in a similar way.To determine if this is the case, we employed an SPR -based assay.Biosensor surfaces were coated with either CD4 D1D2 or a 42 aa cyclic V2 peptide derived from the subtype E 92TH023 gp120 (cV2 92TH023).A third surface coated with MAdCAM-Ig was employed as a positive control.Soluble α 4 β 7 was passed over these surfaces.We then evaluated the capacity of staggered 15 aa peptides that span either CD4 D2 or gp120 V2 to inhibit this interaction (Fig 7A ).
Before proceeding, we evaluated CD4 peptides p43, p44 and p45 that extend our initial screen presented in Fig 4A to include residues in D2 c-terminal to A 178 , the final residue in pCD4 165-178 .p44 and p45 were unable to inhibit α 4 β 7 binding to D1D2 (Fig 7B ), indicating that CD4 residues c-terminal to A 178 do not likely contribute to α 4 β 7 reactivity.p43 showed reduced inhibition relative to pCD4 165-178 , suggesting that Q 165 and K 166 , which are absent in p43 but present in pCD4 165-178 , contribute to the α 4 β 7 reactivity.This is notable because these two residues align with gp120 V2 aa Q 170 K 171 , which were singled out by Cardozo and colleagues as key determinants of V2-α 4 β 7 reactivity [28].
As presented in S3 Fig the G β-strand of CD4 D2 is well conserved in Old World monkeys, but not in New World monkeys.Because divergent sequences can adopt similar structures, we asked whether a D2 G β-strand peptide derived from a New World monkey CD4 could retain a structure required to bind α 4 β 7 .Aotus nancymaae is a South American night monkey whose D2 G β-strand sequence diverges from human CD4 in a significant way.An Aotus peptide analogous to pCD4 165-178 was employed as an inhibitor of α 4 β 7 binding to surface immobilized D1D2, as described above.We included a Gorilla pCD4 165-178 , which differs slightly from human CD4, as a reference.The Aotus and Gorilla peptides both inhibited α 4 β 7 binding in a similar way (S5 Fig) .This suggests that the cis interaction between α 4 β 7 and CD4 is retained in Aotus, and perhaps other New World monkeys and demonstrates that α 4 β 7 can retain CD4 binding even with a sequence that diverges significantly from the conserved KIDIV sequence.This observation is consistent with gp120 V2 mimicking CD4 when one considers that HIV evolved out of Old World monkeys.

A gp120 binding residue in CD4 D1 mediates α 4 β 7 -reactivity
Both MAdCAM-1 and VCAM-1 utilize two binding sites (1˚and accessory) to engage α 4 β 7 .Our preliminary peptide scan (Fig 4) indicated that pCD4 33-47 , which overlaps a loop connecting the C'-C" strands in D1, inhibited α 4 β 7 binding in a modest way (~17%).This observation, by itself, is insufficient to conclude that CD4 D1 is involved in binding to α 4 β 7 .However, with MAdCAM-1 and VCAM-1 in mind, we explored this possibility further.MAd-CAM-1 and VCAM-1 accessory sites, which are both loops, were initially identified by site directed mutagenesis [53,54].Mutating these accessory sites reduced their α 4 β 7 reactivity dramatically.We followed a similar strategy for CD4 by constructing S42G/F43V-D1D2 -Igαtp, which incorporates mouse CD4 substitutions (Fig 8A).Phe 43 is of interest because it is the primary contact for gp120.In previous studies an S42G/F43V CD4 D1D2 mutant was shown to fully abrogate CD4-gp120 interactions without disrupting the overall conformation of D1D2 [8,57].We constructed an additional CD4 D2 mutant, K171D, that incorporates a substitution within the KIDIV sequence in the D2 G β-strand, expecting that it would disrupt α 4 β 7 binding and serve as a control.Wild type and mutant dodecamers were reacted with α 4 β 7 -expressing CD8 + T cells.D1D2-Igαtp bound to α 4 β 7 in a dose dependent manner (Fig 8B).K171D D1D2-Igαtp showed reduced binding to α 4 β 7 .Of note, S42G/F43V-D1D2 -Igαtp showed an even greater reduction in reactivity, suggesting that a region of CD4 D1 that includes the C'C" strands and a connecting loop mediates the CD4-α 4 β 7 interaction.We next asked whether the CD4 D1 mAb Leu3A could block α 4 β 7 binding to CD4.Phe 43 is a key residue in the Leu3A epitope, and Leu3A is a potent neutralizing antibody that blocks gp120 binding to CD4 [57].To this end, we employed an SPR -based kinetic binding assay in which soluble α 4 β 7 was passed over surface immobilized CD4 D1D2, in the presence vs absence of Leu3A.Leu3A reduced α 4 β 7 binding to the same degree as preloading (blocking) the surface with soluble α 4 β 7 (Fig 8C).Ibalizumab, a CD4 D2 mAb, which we employed as a reagent control, failed to inhibit α 4 β 7 binding.Taken together these results indicate that, in addition to the G β-strand of D2, a region of D1 including Phe 43 comprises a 2 nd region of CD4 that mediates α 4 β 7 binding.
Finally, we asked whether gp120, CD4 and α 4 β 7 can form a ternary complex.Labelled D1D2-Igαtp was bound to cell surface expressed α 4 β 7 in the presence vs absence of unlabeled gp120.With a 5X molar excess of gp120, we observed a >70% reduction in binding (Fig 8D).This finding is consistent with results presented in Fig 8B and suggests that binding sites on CD4, α 4 β 7 and gp120 utilize shared epitopes in a way that reduces the likelihood that all three proteins interact simultaneously.
We conclude that CD4, like MAdCAM-1 and VCAM-1 utilizes two sites to mediate α 4 β 7 interactions.Relative to MAdCAM-1 and VCAM-1, the two sites in CD4 are inverted, such that the 1˚CD4 binding site, as defined by the I/LD cation binding motif, lies in D2, while the accessory site lies in D1 (Fig 4B).This inversion may reflect a cis interaction in the case of CD4 vs. trans interactions for MAdCAM-1 and VCAM-1.The way that HIV has incorporated elements of this interaction into its envelope, summarized in S2B Fig, raises several questions, that are discussed below.

Discussion
In this study we identified a specific interaction between CD4, the HIV entry receptor, and α 4 β 7 an integrin that mediates lymphocyte trafficking to GALT.We found that the way these two receptors interact with each other is reflected in their individual interactions with HIV gp120.This finding supports the proposition that memory CD4 + T cells that co-express α 4 β 7 play a specific and important role in HIV pathogenesis [58,59].
We previously reported that, on CD4 + T cells, CD4 and α 4 β 7 can appear sufficiently close to each other to interact [36].Here we report the results of two assays, one based on SPR measurements of α 4 β 7 binding to CD4 D1D2, and a second based on binding of oligomerized sCD4 to cell surface α 4 β 7 , that provide clear evidence of a specific interaction between these two receptors.We do not exclude the possibility that CD4 interacts with α 4 β 7 in trans, however the combination of proximity and affinity argues for a cis interaction.It is well established that cis interactions between integrins and leukocyte receptors can regulate immune responses [37].Several of these interactions involve receptors that, like CD4, present amino-terminal Ig -like domains [40,42].The potential for α 4 β 7 to regulate CD4, or vice versa, is a subject worthy of future investigation, and may be relevant to efficacy of several α 4 integrin antagonists widely used in the treatment of IBDs and MS.
The selective pressures that underlie HIV evolving the capacity to bind to and signal through two receptors that interact with each other remain to be determined.Because α 4 β 7 adopts an active conformation only under certain conditions it is possible that α 4 β 7 provides a means for HIV to target cells that are susceptible to productive infection.This possibility is consistent with studies reporting the preferential infection of α 4 β 7 high CD4 + T cells in vitro [36] and in vivo [33,60].We know that α 4 β 7 is associated with migration to gut tissues, and thus gp120 affinity for α 4 β 7 could help explain the gut -tropic nature of HIV.Because CD4 migrates into the TCR in a transient way, where it plays an essential role in antigen recognition and the stabilization of immune synapses, we can speculate that interactions between CD4 and α 4 β 7 might regulate TCR activity.Perhaps HIV is exploiting this function to its own advantage.Finally, the close-proximity of α 4 β 7 to CD4 may impact viral entry and fusion.α 4 β 7 is not required for viral entry and cells lacking α 4 β 7 are clearly infected.However, productive infection is favored in activated cells, and gp120 delivers signals through α 4 β 7 [26,39].Thus, engagement and signaling through α 4 β 7 either before or following CD4 binding may enhance infection in cells that are suboptimally activated.In this regard, integrins including α 4 β 7 mediate rearrangement of the cytoskeleton in the context of cellular activation [61].Both MAdCAM-Ig and VCAM-Ig inhibit CD4 binding to α 4 β 7 , indicating that CD4 engages the ligand binding groove in the α 4 β 7 headpiece.CD4-α 4 β 7 interactions could only be detected in the presence of Mn ++ .Mn ++ drives the extension of the α 4 β 7 leg domains in a way that places the headpiece >20nm above the cell surface.The N-terminus of CD4 appears no more than 12 nm above the cell surface.How then could CD4 engage the α 4 β 7 headpiece in cis?We recognize two possibilities.Upon cellular activation, both the TCR and CD4 selectively migrate toward the raised tips of microvilli [62].As such, the contours of a T cell could provide a way for an extended form of α 4 β 7 to engage CD4 when CD4 appears above the plane of the cell membrane.If in fact CD4-α 4 β 7 complexes are associated with specialized membrane structures, this could explain the selective pressures that resulted in a gp120 that mimics CD4α 4 β 7 interactions.Alternatively, some integrins headpieces can adopt an active conformation while their leg domains remain bent [49].Current evidence indicates that the activated form of α 4 β 7 is extended [63], but perhaps the possibility of an additional form of α 4 β 7 that is activated but bent requires further investigation.
The structural similarities between CD4 D1D2, and the D1D2s of MAdCAM-1 and VCAM-1 suggest that CD4 might similarly utilize two sites (1˚and accessory site) distributed across its two N-terminal domains to engage α 4 β 7 .Peptides spanning residues 165-178 in D2, and 33-47 in D1 partially inhibited α 4 β 7 -CD4 interactions.Mutating residues in either of these two regions abrogated α 4 β 7 binding in the same way that substitutions in either the 1˚or accessory sites of MAdCAM-1 and VCAM-1 also eliminate α 4 β 7 reactivity.Their 1˚binding sites reside in their respective D1 domains, and each includes an essential Asp-aliphatic aa motif (MAdCAM-1: LD, VCAM-1: ID) that coordinates a divalent cation.An analogous motif (I 172 D 173 ) appears in CD4 D2, and binding of a D2 peptide that includes these two residues is Mn ++ -dependent.We therefore assigned CD4 residues 165-178 as the 1˚α 4 β 7 binding site.
MAdCAM-1 and VCAM-1 encode distinct accessory α 4 β 7 binding sites.Our findings indicate that the CD4 D1 C'C" strands and a connecting loop, also functions as an accessory binding site.Substituting residues Ser 42 and Phe 43 with corresponding mouse residues eliminated α 4 β 7 reactivity.These substitutions do not disrupt the overall antigenic structure of D1 [57], and a high-resolution crystal structure of D1D2 with an F 43 /V substitution found that this protein and w.t D1D2 were essentially isomorphous [64].We conclude that the loss of α 4 β 7 reactivity mediated by these aa substitutions reflects a specific effect.Supporting this finding, mAb Leu3A, whose epitope includes Phe 43  [57], inhibits D1D2 binding to α 4 β 7 .Whether CD4 D1 mediates CD4-α 4 β 7 interactions directly or indirectly is a subject of current investigation.
The sequence similarity between the α 4 β 7 binding sites in CD4 D2 and gp120 V2 is relevant to HIV vaccine development insofar as this region of V2 played a central role in RV144, the only HIV vaccine trial to achieve efficacy.Reduced risk in RV144 correlated with weakly-neutralizing antibodies specific to a region of V2 encompassed by aa 169-181 [65][66][67].Moreover, V2 residues, K 169 and I 181 were subject to vaccine -mediated immune pressure (sieving) in vaccinees who became infected [68].I 181 is included in the KIDIV sequence and K 169 is adjacent to Q 170 K 171 .These observations indicate that the immune responses associated with reduced risk in RV144 target a region of gp120 V2 that mimics the 1˚α 4 β 7 binding site in CD4 D2.Additionally, the primary contact sites of two RV144 -derived mAbs, CH58 and CH59, that are considered representative of the relevant RV144 Ab response lie within V2 aa 170-184 [69][70][71].These mAbs block V2 adhesion to α 4 β 7 [29,56,71].The prevailing view holds that reduced risk in RV144 was associated with V2 Ab effector functions [72].We wonder whether the mechanistic explanations of RV144 reduced risk require further refinement.Such information may contribute to our understanding of the failure of HVTN702, a vaccine regimen intended to elicit V1V2 antibodies [73].Overall, our findings provide an additional rationale for continuing to investigate the potential utility of V2 antibodies as part of an efficacious HIV vaccine.
The limitations of this study include a recognition that we do not know how the direct interaction between α 4 β 7 and CD4 impacts HIV infection.However, the presence of a sequence in V2 that mimics CD4 D2 and shares with it the capacity to directly bind α 4 β 7 provides strong evidence that HIV is sensitive, in some way, to α 4 β 7 -CD4 interactions.This conclusion is consistent with the preferential infection in vivo of memory CD4 + T cells that co-express CD4 and α 4 β 7 [33].Depending on whether CD4-α 4 β 7 interactions enhance or inhibit infection, it follows that circumstances that promote this interaction define a condition in which CD4 + T cells are either more or less susceptible to productive infection.Because productive infection is strongly impacted by the activation state of a cell, determining if and how CD4-α 4 β 7 interactions regulate CD4 participation in TCR-MHC II interactions may help resolve this question.Additionally, although not the subject of this study, our understanding of how gp120 interacts with α 4 β 7 is also incomplete, but relevant to the findings presented here.In contrast to the 1˚α 4 β 7 binding site in CD4 D2, which is well exposed [10], the analogous α 4 β 7 binding site in gp120 V2 is partially hidden in the predominant ground state represented in prefusion trimers [13,[74][75][76][77][78].We do not know at which step in the transition of gp120 from the closed trimer to the open coreceptor bound state that V2 interacts with α 4 β 7 , nor whether this interaction precedes or follows CD4 interactions with gp120.Perhaps a more detailed understanding of V2 conformation as the trimer moves away from its ground state will resolve the context in which gp120 engages α 4 β 7 [14,78].
In conclusion, in this report we identify two new aspects of the CD4 receptor and the way that it interacts with gp120 that are relevant to HIV pathogenesis.First, CD4 binds directly to α 4 β 7 , the integrin most closely associated with T cell trafficking to GALT.Second, gp120 interactions with CD4 and α 4 β 7 reflect their interactions with each other.These findings underscore the view that α 4 β 7 -expressing CD4 + T cells play a specific and important role in HIV pathogenesis.

Ethics statement
All primary T cells utilized in these studies were isolated from PBMCs, collected from healthy donors through a NIH Department of Transfusion Medicine protocol that was approved by the Institutional Review Board of the National Institute of Allergy and Infectious Diseases (NIAID), National Institutes of Health.Informed consent was written and was provided to study participants.

Cell line and reagents
RPMI8866 cells, a human B lymphoma cell line that constitutively expresses α 4 β 7 was purchased from Sigma-Aldrich.Cells were maintained in RPMI-1640 (Gibco) containing 10% heat inactivated fetal bovine serum (Gibco), 2% penicillin/streptomycin/glutamine (Gemini Bio-Products) and 0.1% all trans RA (Sigma).Cells were cultured for a minimum of 7 days prior to use.Primary CD4 + and CD8 + T cells were purified from buffy coats by negative bead selection.Cells were cultured in RPMI-1640 plus anti CD3, IL-2, and RA for a minimum of 8 days prior to use.CD4 D1D2 protein was a gift from Dr. E. Berger.Linear CD4 peptides and FAM conjugated peptides with > 90% purity were provided by Biopeptide.Cyclic V2 peptides were supplied by JPT Peptide Technologies.α 4 and β 7 expression vectors were purchased from Origene Technologies.Synthetic CD4 D1D2 and D1D2-Igαtp genes were purchased from ATUM.Site directed mutagenesis was performed by Genscript.Transient transfections were carried out with Polyfect (Invitrogen), using standard protocols.CHO cell derived A244 gp120 (Lot 26539-1) was provided by Global Solutions for Infectious Diseases (San Francisco, CA).Purity was estimated at 97.1%.Vedolizumab was provided by the NIH Clinical Center Pharmacy Department.FG was purchased from MedChem Express.Human integrin α 4 mAb 2B4, MAdCAM-Ig, VCAM-Ig and soluble α 4 β 7 were purchased from R&D Systems.Ibalizumab was generously provide by Dr. David Ho.Other mAbs were purchased from BD Biosciences.Dye labelling of D1D2-Igαtp was carried out using a LYNX HRP conjugation kit (Bio-Rad), following the manufacturer's instructions.

Flow cytometry binding assays
Cells were stained with fluoresceine conjugated mAbs or recombinant proteins using standard procedures preceded by Fc receptor blocking with human IgG.Buffers used were 10 mM HEPES 150 mM NaCl (HBS Buffer) with 100 μM CaCl 2 . 1 mM MnCl 2 was also included, unless otherwise specified.Buffer without divalent cations included 0.1 mM EDTA.Unlabeled mAb or protein inhibitors were preincubated with cells or proteins at a 5X molar excess for 15 min on ice.FAM labeled peptides were used at 1.25 μg/ 200,000 cells in a 100 μl volume.Data were acquired using a FACSCanto II (BD Biosciences) and analyzed with FlowJo software (BD Biosciences).Plotted MFI values represent absolute values.

Surface plasmon resonance
Experiments were performed using a Biacore 3000 instrument (Cytiva Life Sciences) using CM4 or CM5 sensor chips.The data were evaluated using BIAevaluation 4.1.1software (Cytiva Life Sciences).The chip surface was activated by injecting 35 μl of a 1:1 mixture of 0.05 M Nhydroxysuccinimide and 0.2 M N-ethyl-N-(dimethylaminopropyl) carbodiimide at 5 μl/min.NeutrAvidin, HIV gp120, sCD4 (D1D2) or Hu-MAdCAM-Ig (R&D Systems) at concentrations of 5 μg/ml in 10mM NaOAc, pH 4.5, were immobilized to 250-750 resonance units (RU).After the proteins were immobilized to the desired densities, unreacted sites were blocked with 35 μl of 1 M Tris-HCl (pH 8.0).Biotinylated cyclic V2 peptides (1 μg/ml in 20 mM Tris-HCl, pH 8.0) were bound to the NeutrAvidin surfaces to densities of approximately 250-300 RU.One surface was activated and blocked without ligand as a control surface for non-specific binding.Signal from the control surface was subsequently subtracted.Running buffer was HBS (pH 7.4), 1 mM MnCl 2 , 0.005% Tween-20, 0.05% soluble carboxymethyl-dextran.Binding experiments were carried out at a flow rate of 25 μl/min at 25˚C.After a 2 min injection, the surface was washed for an additional 2 min in running buffer allowing dissociation of the bound ligand from the surface.The surfaces were regenerated by multiple injections of 4.5 M MgCl 2 at a flow rate of 100 μl/min.Inhibition of α 4 β 7 binding by linear V2-loop and CD4 derived peptides was carried out by pre-incubating α 4 β 7 with the peptides in running buffer at the indicated concentrations for 2 hours at room temperature prior to passing them over the prepared surfaces as described above.Cross-inhibition experiments were performed by passing either buffer or the listed analytes over a surface to which sCD4 (D1D2) had been immobilized, followed by a secondary sequential injection of α 4 β 7 .The ability of each analyte to inhibit either itself or α 4 β 7 was determined by comparing the binding of α 4 β 7 in the absence or presence of the listed pre-bound analytes as previously described [79].Illustration of the dynamic range of range of linear peptide inhibition assays is provided in S4 Fig.

FRET/FLIM
Primary CD4 + T cells were cultured in IL-2 + anti CD3 + RA (10nM) for a minimum of 10 days.RA -mediated upregulation of α 4 β 7 was verified by flow cytometry prior to mAb staining and FRET\FLIM Imaging.FLIM sample staining was carried out 2 days after fresh media exchange.Samples were reacted with the following antibodies using standard immunofluorescence methods.Donor mAbs (AF488 conjugated) includes OKT4 (BioLegend 317420), FIB504 (lab conjugated), Leu3A (Biolegend 344618).Acceptor mAbs (AF546 conjugated) includes FIB504 (lab conjugated), OKT4 (lab conjugated).Mock and gp120 treatments were performed in the same volume of PBS (cells: 2x10 6 /mL), for 20 or 60 min at 37˚C in the dark, in the presence of 5% CO 2 .Treatments were stopped by 4 vigorous washes in high volumes (3x) of ice-cold complete media, followed by 1 wash in PBS + BSA, then fixed in 2% MeOHfree PFA at 4C.After the completion of the staining procedure (~16 hrs) samples were cytospun onto imaging slides at 400 rpm for 6 min.Slides were then mounted with imaging coverslips using prolong gold mounting media with antifade-no DAPI (Invitrogen) and cured thoroughly in the dark at room temperature before imaging.Images were acquired using a Leica DMI6000 confocal microscope coupled with Becker and Hickel Lifetime measurement module.Data were collected in excess of 1000 photons per pixel to fit the curve in a multimodal decay plot.Images were post-processed using ImageJ/NIHimage analysis software.

Proximity ligation assay
Flow cytometry based PLA assays (Sigma Aldrich) were carried out per the manufacturer's instructions.In brief, an oligonucleotide conjugated anti CD4 mAb (clone DB81) (plus strand) and an oligonucleotide conjugated anti integrin β 7 mAb (clone FIB27) (minus strand) were generated using a Duolink In Situ Probemaker kit.Freshly isolated purified CD4 + T cells were fixed in 1% PFA and blocked with buffer containing 2% donkey serum, 10% BSA and 0.2% gelatin.Cells were incubated with mAb probes for 30 min at 37˚C in buffer containing 0.2% donkey serum, 10% BSA and 0.2% gelatin.Additional fluorescein based mAbs (vedolizumab and CD45RA) were included.After washing, cells were treated with DNA ligase and Duolink ligation buffer for 30 min at 37˚C.After washing, cells were treated with Duolink amplification buffer and DNA polymerase O/N at 37˚C.Cells were washed and treated with Duolink detection buffer (Far Red) for 30 min at 37˚C.After washing cells were resuspended in PBS and analyzed by flow-cytometry.All washes were carried in Duolink wash buffer.

CD4 and gp120 V2 sequence alignments and analysis
All gp120 aa sequences were numbered based on HXB2.Frequencies of gp120 V2 aa were established by accessing the HIV ENV web alignment from Los Alamos database (https://www.hiv.lanl.gov/content/sequence/NEWALIGN/align.html(accession date 09/29/2022)).Sequences from different HIV subtypes and recombinants were utilized and represent the full spectrum of HIV envelope sequences from the LANL database.All non-group M sequences were excluded from this analysis.Geneious 11.1.5software was used to determine the frequencies of I, L and V at gp120 position 181.HIV ENV web alignment was used to generate the logo figure on WebLogo3 software (DOI: 10.1101/gr.849004).Protein motifs were scanned against the Swiss-Prot database using the Prosite tool (doi:10.1093/nar/gks1067,https://doi.org/10.1093/nar/gkl124).the private views of the author, and are not to be construed as official, or as reflecting true views of the Department of the Army or the Department of Defense.The investigators have adhered to the policies for protection of human subjects as prescribed in AR 70-25.

Fig 8 .
Fig 8. Mutating a gp120 contact site in CD4 D1 abrogates CD4-α 4 β 7 interactions.(A) Positions of amino acid substitutions introduced into D1D2-Igαtp are indicated on a space-filling model of CD4 D1D2.Substitutions at S42/ F43 in D1 (green) and K171 in D2 (purple) are shown.Approximate locations of mAb Leu3A and ibalizumab epitopes are indicated.(B) Flow-cytometric analysis of increasing amounts of D1D2-Igαtp and mutant derivatives binding to α 4 β 7 on CD8 + T cells.Y-axis indicates % cells bound.(C) SPR analysis of soluble α 4 β 7 binding to a D1D2 coated surface following preincubation of the surface with ibalizumab (red), mAb Leu3A (light green), or α 4 β 7 (dark green).Y-axis indicates the mass of soluble α 4 β 7 bound at the end of the association phase.Mass of prebound inhibitors is subtracted.(D) Flow-cytometric analysis of unlabeled gp120 inhibiting PE labeled D1D2-Igαtp binding to α 4 β 7 on CD8 + T cells.The integrin α 4 antagonist FG was employed as a control.Y-axis indicates MFI.https://doi.org/10.1371/journal.ppat.1011860.g008 flow-cytometry.A second oligomeric Ig fusion protein, MicA-Igαtp, which includes the two N-terminal domains of MicA, was employed as a specificity control.D1D2-Igαtp, but not MicA-Igαtp, bound to cells, and this binding was inhibited by an integrin α 4 mAb but not by a control mAb (Fig1B).To establish specificity, 293T cells were co-transfected with α 4 and β 7 expression vectors and stained with D1D2-Igαtp in the absence or pres-4 β 7 is expressed at intermediate levels on the majority of naïve CD4 + T cells in blood.As noted above, most memory CD4 + T cells in blood show either little or no α 4 β 7 expression.However a minor subset (~5-20%) express high levels of α 4 β 7 .Differential expression of α 4 β 7 on naïve vs memory cell can be visualized by co-staining cells with β 7 and CD45RO mAbs (Fig 2 , MgCl 2 and MnCl 2 .Y-axis indicates gMFI.https://doi.org/10.1371/journal.ppat.1011860.g001and analyzed by α