A Highly Conserved Arginine in gp120 Governs HIV-1 Binding to Both Syndecans and CCR5 via Sulfated Motifs*

HIV-1 has maximized its utilization of syndecans. It uses them as in cis receptors to infect macrophages and as in trans receptors to infect T-lymphocytes. In this study, we investigated at a molecular level the mechanisms that control HIV-1-syndecan interactions. We found that a single conserved arginine (Arg-298) in the V3 region of gp120 governs HIV-1 binding to syndecans. We found that an amine group on the side chain of this residue is necessary for syndecan utilization by HIV-1. Furthermore, we showed that HIV-1 binds syndecans via a 6-O sulfation, demonstrating that this binding is not the result of random interactions between basic residues and negative charges, but the result of specific contacts between gp120 and a well defined sulfation in syndecans. Surprisingly, we found that Arg-298, which mediates HIV-1 binding to syndecans, also mediates HIV-1 binding to CCR5. We postulated that HIV-1 recognizes similar motifs on syndecans and CCR5. Supporting this hypothesis, we obtained several lines of evidence that suggest that the 6-O sulfation recognized by HIV-1 on syndecans mimics the sulfated tyrosines recognized by HIV-1 in the N terminus of CCR5. Our finding that CCR5 and syndecans are exploited by HIV-1 via a single determinant echoes the mechanisms by which chemokines utilize these two disparate receptors and suggests that the gp120/chemokine mimicry may represent a common strategy in microbial pathogenesis.

The dominant cell surface heparan sulfate proteoglycans (HSPGs) 3 are the syndecans. Syndecans are transmembrane receptors highly expressed on adherent cells (i.e. epithelial cells, endothelial cells, and macrophages), but poorly expressed on suspension cells (i.e. T-lymphocytes) (1)(2)(3)(4). The syndecan family is composed of four members, syndecan-1 to -4. Their ectodomain bears linear heparan sulfate chains, which are composed of a repetition (30 -400 repeats) of a sulfated dis-accharide motif (5). The sulfation pattern of the heparan sulfates dictates the ligand specificity of syndecans. Syndecans via their heparan sulfates function as co-factors in cell-cell adhesion, in linking cells to ligands in the extracellular matrix, and in the binding and activation of cellular growth factors (5). Syndecans also function as receptors for HIV-1. We and others demonstrated that pretreatment of HIV-1 target cells (i.e. CD4ϩ T cell lines, CD4ϩ HeLa cells, and CD4ϩ CHO cells or macrophages) with heparinase, an enzyme that removes heparan sulfates from the ectodomain of syndecans, significantly reduces HIV-1 infectivity (1, 6 -11). Zhang et al. (11), using CHO cells that either express HSPGs (CHO-K1) or lack HSPGs (pgsA745), obtained evidence that HSPGs favor HIV-1 infection in a gp120-dependent manner. Previous work suggests that the requirement for syndecans and HSPGs in HIV-1 infection is particularly accentuated when target cells express low levels of entry receptors (CD4 and CCR5/CXCR4) such as CD4ϩ HeLa cells and macrophages (1). However, why and how HIV-1 uses sulfated syndecans to optimally infect these target cells remain to be understood. More interestingly, syndecans also serve as in trans receptors for HIV-1. Specifically, HIV-1 binds syndecans richly expressed on the endothelium (2). HIV-1 bound to syndecans remains infectious for a week, whereas cell-free virus loses its infectivity after a single day (2). Most importantly, HIV-1 attached onto the endothelium via syndecans represents an in trans source of infection for circulating T cells (2). Moreover, endothelial syndecans enhance HIV-1 replication in T cells via a yet uncharacterized mechanism (2). These findings suggest that syndecan-rich endothelium can provide a microenvironment that amplifies HIV-1 replication in T cells. Lastly, syndecans and HSPGs on brain microvascular endothelial cells play a significant role in HIV-1 transmigration through the blood-brain barrier (3,12). Altogether these observations suggest that syndecans, by acting as in cis and in trans receptors on different cells and tissues, may profoundly impact HIV-1 pathogenesis. In this study, we examined at a molecular level the mechanisms that control HIV-1-syndecan interactions to subsequently delineate the function and relevance of these interactions in vivo.

MATERIALS AND METHODS
Cells-Namalwa cells expressing human syndecan-1, -2, -3, and -4 and DC-SIGN were created as described previously (2). Two additional Namalwa cell lines were created by stably transfecting human CCR5 and CD4. Human CD4ϩ T cells and macrophages were isolated and infected as described previously (1).
Intrinsic CD Spectroscopy-CD measurements were made with Aviv 62DS/720 CD spectrophotometer. Far UV spectra were recorded from 200 to 260 nm at 25°C with a 0.1-cm path length cell. The protein concentration for all samples was adjusted to 5 M. Spectra were collected at a scan speed of 3 nm/s and with a response time of 1s. Each spectrum was derived from an average of 3 scans. For calculation of mean residue ellipticity () the molecular weight of the proteins was determined as glycosylated species (76187.50 for wt, 73496.69 for V3R2, and 73581.80 for R298A). The data were expressed as mean residue ellipticity, , in deg cm 2 dmol Ϫ1 , which is defined as: ϭ obs (mdeg)/ 10nlc. Where obs is the CD in mill degrees, n is the number of amino acid residues; l is the path length of the cell, and c is the concentration of the protein in moles.
Virus Binding Assay-All viruses were derived and prepared from electroporated Jurkat T cells as previously described (2). Cells (0.25 ϫ 10 6 cells) were exposed to virus (1 ng of p24) in complete medium for 2 h at 37°C, washed twice, and lysed, and the p24 content was measured by ELISA (PerkinElmer Life Sciences). Viruses were preincubated or not for 1 h at 37°C with 10 g/ml anti-gp120 antibodies. 257-D, 268-D, 447-52D, B4a1, B4e8, and 4.8d were obtained from the National Institutes of Health AIDS Research and Reference Reagent Program, whereas 50.1, 58.2, and 83.1 (15) were a generous gift from R. Stanfield. The 412d FF and YY scFv were produced as described previously (16). For some experiments, viruses were incubated with peptides encompassing the V3 of consensus B (AIDS Research and Reference Reagent Program), sulfated and unsulfated CCR5 DYQVSSPIYDINYYTSE peptides (synthesized by American Peptide), or with heparin derivatives generously provided by J. Esko (17). Specific desulfation of heparan derivatives was verified by reverse phase ion pair and high performance liquid chromatography analyses (17).
gp120 ELISA-96-well plates were coated 16 h at 4°C with 10 g/ml Fc, wild-type, or R298A Fc-gp120 proteins. Wells were washed and blocked for 1 h at 25°C with 1% fetal calf serum and 0.5% Tween-20. Wells were then incubated with anti-gp120 antibodies (10 g/ml) for 1 h at 25°C and washed. Bound antibodies were detected by horseradish peroxidase antibodies (Amersham Biosciences) for 1 h at 25°C with 1% fetal calf serum and 0.5% Tween-20. Wells were washed and o-phenylenediamine (Sigma) substrate was added for 15 min at 25°C, and absorbance was read at 490 nm. Antibodies used for this ELISA are antibodies directed against N-glycan of gp120 such as 2G12 (21,22), against the CD4-binding site of gp120 such as b12 (23,24), or against the V3 region such as 447-52D (25)

RESULTS
The V3 Region of gp120 Contains the Syndecan-binding Site for HIV-1-We asked if gp120 serves as the main ligand for HIV-1 binding to syndecans. We created a panel of cell lines expressing each syndecan member (syndecan-1 to -4) (2). Human Namalwa B cells were chosen as parental cells, because they natively express no candidate gp120 receptors such as CD4, CCR5, HSPGs, and DC-SIGN. As positive controls, we created two additional cell lines expressing receptors known to bind gp120, CD4, and DC-SIGN. We verified by quantitative FACS analysis that these cell lines express similar amounts of gp120 receptors as described previously (2). Recombinant Fc-gp120 JR-CSF (gp120 derived from the primary R5 isolate JR-CSF fused to human Fc IgG1) binds both CD4 and DC-SIGN, but fails to bind parental cells (Fig. 1A). Importantly, Fc-gp120 binds all four syndecans, whereas Fc alone does not (Fig. 1A). Heparinase, which removes all heparan sulfates but preserves the integrity of the core protein (2), abolishes gp120 binding to syndecans, but not to CD4 and DC-SIGN (Fig. 1A). Thus, gp120 alone possesses the capacity to bind syndecans, and this interaction is mediated by the heparan sulfates of syndecans.
After demonstrating that gp120 alone binds syndecans, we searched for domains in gp120 necessary for this interaction. We generated gp120 proteins deleted for specific domains and tested their capacities to bind syndecans. gp120 deleted for its two variable regions V1 and V2 still binds syndecans (Fig. 1B). In contrast, the V3-deleted as well as the V1/V2/V3-deleted gp120 fail to bind syndecans but still bind CD4 and DC-SIGN, suggesting that V3 is critical for gp120-syndecan interactions.
To more definitely demonstrate that V3, in the context of whole virus, is necessary for HIV-1 binding to syndecans, we examined the capacity of a virus deleted for V3 to bind syndecans. The virus deleted for gp160 (NL4.3 ⌬gp160) binds syndecans poorly compared with wildtype (NL4.3) virus, suggesting that gp120/gp41 mediates HIV-1 binding to syndecans. Furthermore, the V3-deleted virus (⌬V3 HXB) lost its capacity to bind to syndecans compared with wild-type (HXB) or V1/V2-deleted (⌬V1/V2 HXB) viruses (Fig. 1D, top panel). However, the V3-deleted virus kept its capacity to bind DC-SIGN and CD4 (Fig.  1D, middle and bottom panels), suggesting that the syndecan-binding site in gp120 does not overlap with CD4-and DC-SIGN-binding sites. Moreover, heparinase treatment of the cells impaired the ability of HIV-1 to bind syndecans (Fig. 1D), indicating that the heparan sulfates contain the motifs necessary for HIV-1 capture. Thus, our data demonstrate that V3 serves as the major locus for HIV-1 binding to syndecans and that this binding occurs via the sulfated heparans of syndecans.
A Single Highly Conserved Arginine Located at the Base of the V3 Region Is Necessary for gp120 Binding to Syndecans-We then examined if specific residues in V3 govern syndecan recognition by HIV-1. We tested a panel of HIV-1 for their capacities to bind syndecans as described previously (2), hoping to identify specific motifs in V3 that would correlate with their capacities to bind syndecans. However, we did not observe an obvious correlation between the capacities of viruses to bind syndecans and their co-receptor usage or gp120 subtype ( Fig.  2A). Syndecans capture chemokines and growth factors via contacts between their heparan sulfates and basic consensus sequences within the ligand (5). Thus, we also examined if a correlation exists between the number of basic residues within V3 and the capacity of the virus to bind syndecans. Surprisingly, again, no obvious correlation was observed. For example, viruses containing only five basic residues in V3 (i.e. JR-CSF and 93IN101) bind syndecans as well as viruses containing eight or nine basic residues (i.e. HXB2 and MN) ( Fig. 2A). However, our observation that all viruses bind more efficiently to syndecan-expressing cells than to parental cells suggests that all HIV-1 contain intrinsic capacities to bind syndecans. We thus postulated that a few residues in V3 are necessary and sufficient to mediate HIV-1-syndecan interactions. First, we asked if the basic residues of V3 are important for HIV-1 binding to syndecans. We substituted the five basic residues of V3 of JR-CSF in Fc-gp120, and tested the resulting mutant protein for its capacity to bind syndecans. The five basic residues Arg-298, Arg-304, Lys305, Arg-308, and Arg-326 were replaced by alanines (called RRKRR/AAAAA).
Fc-gp120 RRKRR/AAAAA binds syndecans poorly, but binds to CD4 and DC-SIGN well (Fig. 2B). To identify which of these basic residues are critical for HIV-1-syndecan interactions, we tested V3 peptides, which encompass V3, for their capacities to prevent HIV-1 binding to syndecans. Peptides encompassing the N terminus of V3 are more potent inhibitors than peptides encompassing the tip or the C terminus of V3 (Fig. 2C). This finding is in accordance with our observation that the anti-V3 257-D antibody directed against the base of V3 interferes with HIV-1 binding to syndecans (Fig. 1C).
Given that all viruses tested so far possess the capacity to bind syndecans ( Fig. 2A), we searched for conserved basic residues at the base of V3 that may be responsible for the contact between the virus and syndecans. We observed that one arginine at the base of V3, Arg-298, is highly conserved among all HIV-1 subtypes (Fig. 2D). We mutated this arginine in the context of Fc-gp120 JR-CSF and tested the resulting mutant protein to bind syndecans. Importantly, the R298A lost its capacity to bind syndecans (Fig. 2B), suggesting that the highly conserved arginine in position 298 of V3 of JR-CSF gp120 is necessary to mediate gp120 binding to syndecans. Although the R298A gp120 protein still binds CD4 and DC-SIGN (Fig. 2B), we cannot exclude the possibility that the introduction of mutations at this position induces conformational changes in V3 that would explain the inability of the  ). B, same as A. C, syndecan-2ϩ and DC-SIGNϩ cells were exposed to JR-CSF (1 ng of p24) for 2 h at 37°C, washed, and lysed, and p24 content was measured by ELISA. JR-CSF was preincubated or not for 1 h at 37°C with anti-gp120 antibodies. Results are expressed in percentage of attachment by fixing at 100% the attachment of JR-CSF in the absence of antibody. D, syndecan-2ϩ, DC-SIGNϩ, and CD4ϩ cells treated or not with heparinase were exposed to wild-type or mutant NL4.3 or HXBc2 viruses. Results are expressed in percentage of p24 captured relative to the initial inoculum. Results are representative of two independent experiments. FIGURE 2. A single highly conserved arginine at the base of the V3 region is necessary for gp120 binding to syndecans. A, syndecan-2ϩ cells were exposed to a panel of HIV-1 isolates. Results are expressed in percentage of p24 captured relative to the initial inoculum. B, syndecan-2ϩ, CD4ϩ, and DC-SIGNϩ cells were incubated with wild-type (Fc-gp120), V3-deleted (Fc-gp120⌬V3), all V3 basic residues mutated (Fc-gp120 RRKRR/AAAAA: R298A, R304A, K305A, R308A, and R326A), two V3 basic residues mutated (Fc-gp120 RR/AA: R298A and R326A), and a unique V3 basic residue mutated (Fc-gp120 R298A) Fc-gp120 proteins. C, syndecan-2ϩ cells were exposed to JR-CSF in the presence of peptides that encompass the consensus B V3 (AKTIIVQLNESVEINCTRPNNNTRKSIHIGPGRAFYTTGEIIGDIRQAHCN). Results are expressed in percentage of attachment by fixing at 100% the attachment of JR-CSF in the absence of peptide. D, consensuses for the V3 region. E, 96-well plates coated with Fc, wild-type, or R298A Fc-gp120 proteins were exposed to anti-gp120 antibodies. F, gel electrophoresis of Fc-gp120 proteins (Coomassie staining). G, far-UV CD spectra of the wild-type gp120 (WT) (filled squares), the R298A mutant (298) (filled triangles), and the double R298A/R326A mutant (v3r2) (filled circles) are shown. The spectra were measured at 25C with a 0.1-cm path length cell at a protein concentration of 5 M. gp120 mutant protein to bind syndecans. To rule out this possibility, we tested by ELISA anti-V3 antibodies for their capacities to recognize the R298A gp120 protein. As expected, anti-V3 antibodies do not bind the ⌬V3 gp120 protein, whereas antibodies directed against the CD4-binding site (b12) or N-glycans (2G12) do bind the ⌬V3 gp120 protein (Fig.  2E). Importantly, a majority of anti-V3 antibodies, including those known to recognize conformational epitopes such as 447-52D (30), bind wild-type and R298A gp120 proteins equally well (Fig. 2E). Note that the anti-V3 4.8d antibody recognizes the R298A mutant poorly. Although this antibody recognizes wild-type gp120 poorly too, we cannot exclude the possibility that the R298A substitution directly or indirectly perturbs the epitope recognized by 4.8d. Nevertheless, gel electrophoresis (Fig. 2F) and circular dichroism (Fig. 2G) analyses strongly suggest that the secondary structure of the R298A gp120 mutant is not altered significantly compared with wild-type gp120. Although we cannot exclude the possibility that subtle changes result from the mutation, our data suggest that mutating the arginine 298 does not profoundly affect the V3 conformation and that this arginine truly represents a key residue in gp120-syndecan contact.

HIV-1 Binding to Syndecans via 6-O-Sulfation
A Highly Conserved Arginine in gp120 Mediates HIV-1 Binding to Syndecans-We next asked if this highly conserved arginine also plays a critical role in the context of whole virus. We replaced the arginine 298 by an alanine within the proviral clone encoding HXB2 ConsB virus, which contains the consensus B sequence of the V3 of R5 HIV-1 isolates, creating the HXB2 R298A mutant virus. As a control, we substituted the other highly conserved arginine located at the base of the V3 (arginine 326), creating the HXB2 R326A mutant virus. In contrast to wild-type and R326A viruses, the R298A virus binds syndecans poorly (Fig. 3A). All three viruses bind DC-SIGN efficiently (Fig. 3A). This demonstrates that a highly conserved arginine (Arg-298) at the base of the V3 is critical for gp120-and HIV-syndecan interactions. We verified by gp120 ELISA that similar amounts of gp120 were incorporated among viruses as described previously (4) (data not shown).
Arginine is a positively charged residue, which is involved in proteinprotein or protein-RNA interactions and that contains a long side chain with two terminal amino (NH 2 ) groups at the position and one secondary amine (NH) at the ⑀ position. To examine if a positively charged residue at position 298 is essential for syndecan binding, the arginine  NOVEMBER 25, 2005 • VOLUME 280 • NUMBER 47 was replaced with different classes of amino acids: a positively charged (lysine), a negatively charged (glutamic acid), a neutral (glutamine), and a hydrophobic residue (leucine). R298A, R298K, R298E, R298Q, and R298L viruses were tested for their capacities to bind syndecans. In contrast to the R298A virus, the R298K virus binds syndecans at wildtype levels (Fig. 3B), indicating that a positively charged residue at position 298 of V3 is important for syndecan utilization. Furthermore, R298E and R298L viruses, like the R298A virus, bind syndecans poorly (Fig. 3B), suggesting that negatively charged or neutral residues at position 298 cannot support HIV-1 binding to syndecans. Interestingly, the R298Q virus binds syndecans at wild-type levels (Fig. 3B), suggesting that the presence of an additional amine group on the side chain of the residue at position 298 (arginine, lysine, or glutamine) is critical for HIV-1-syndecan contact. These findings provide the first evidence that an amino group on the side chain of the residue located at position 298 of the V3 of gp120 is required for syndecan utilization by HIV-1.

HIV-1 Binding to Syndecans via 6-O-Sulfation
The Highly Conserved Arginine 298 Is Necessary for HIV-1 Binding to Both Syndecans and CCR5-To delineate the contribution of HIV-1syndecan interactions to HIV-1 infectivity, we examined the replication of the R298A virus in CD4ϩ T-lymphocytes and macrophages. Given that T-lymphocytes do not express syndecans (1), we did not anticipate that HIV-1-syndecan interactions modulate HIV-1 infectivity. In contrast, because macrophages express high syndecan levels upon cellular differentiation (1, 31), we anticipate that syndecan interactions significantly enhance HIV-1 infectivity in this cell type as previously reported (1). Surprisingly, in contrast to wild-type virus, the R298A virus fails to replicate in both T-lymphocytes and macrophages (Fig. 4A). One possibility to explain the failure of the R298A virus to infect syndecannegative T-lymphocytes is that the mutation disrupts the binding of gp120 with HIV-1 entry receptors, CD4 and/or CCR5. R298A mutation does not affect the recognition of gp120 by CD4 (Fig. 2B), suggesting that R298A gp120-CD4 interactions are preserved. We thus asked if the R298A mutation alters the contact between gp120 and CCR5. We intro-duced CCR5 into Namalwa cells and tested these CCR5-positive cells for gp120 binding. Given that CD4 facilitates the binding of gp120 to CCR5 (32), Fc-gp120 binding to CCR5-positive cells was performed in the presence of soluble CD4. Wild-type Fc-gp120 binds poorly to parental CCR5-negative cells (data not shown) or to CCR5-positive cells in the absence of soluble CD4 (Fig. 4B). In contrast, wild-type gp120 protein binds efficiently to CCR5-positive cells in the presence of soluble CD4 (Fig. 4B). Importantly, the R298A gp120 mutant protein fails to bind to CCR5-positive cells even in the presence of soluble CD4, suggesting that this arginine is critical for gp120-CCR5 interactions. This observation is in accordance with a previous study, which showed that CCR5 co-receptor utilization involves this highly conserved arginine in V3 of gp120 (20). Thus, our data demonstrate that HIV-1 exploits a single highly conserved residue in V3 of gp120 to utilize two entirely distinct HIV-1 receptors: syndecans and CCR5.

HIV-1 Does Not Bind Randomly to Cell Surface Negative Charges But Rather Specifically to the 6-O-Sulfate Group of Glucosamine Units in
Syndecans-If it is true that a single residue in V3 is critical for the recognition of two disparate receptors, syndecan and CCR5, one can envision that these receptors contain similar gp120-binding motifs. Because heparinase abolishes HIV-1-syndecan interaction and gp120deleted HIV-1 fails to bind syndecans, this suggests that the sulfated heparans of syndecans represent the major binding sites for gp120 (2). If specific sulfated heparan motifs in syndecans serve as binding sites for gp120, what are the motifs in CCR5 that eventually mimic these motifs? CCR5 contains an N-terminal region that is acidic and tyrosine-rich (33,34). CCR5 is modified by sulfation of its N-terminal tyrosines (35). The sulfated tyrosines contribute to the binding of CCR5 to MIP-1␣ and MIP-1␤ as well as to gp120 (35)(36)(37)(38). Substitution of these tyrosines impairs HIV-1 entry without altering CCR5 expression (35). Thus, we hypothesized that the sulfated motifs of syndecans mimic the tyrosinesulfated motifs of CCR5 (Fig. 5A). To test this hypothesis, we examined   NOVEMBER 25, 2005 • VOLUME 280 • NUMBER 47 whether sulfated heparans could prevent gp120-CCR5 interactions, and conversely, whether tyrosine-sulfated CCR5 peptides could prevent gp120-syndecan interactions (Fig. 5B). Soluble heparan sulfate, wildtype heparin, and oversulfated heparin efficiently prevent HIV-1 binding to syndecans, whereas 2-O-and 3-O-desulfated heparin partially blocks HIV-1 binding to syndecans (Fig. 6A), suggesting that these compounds compete with the cell surface-sulfated syndecans for HIV-1 recognition. Oversulfated heparin inhibits HIV-1 binding more effi-ciently than wild-type heparin, indicating that the degree of sulfation of syndecans is important for HIV-1 recognition. Note that the presence of soluble CD4 does not affect HIV-1 binding to syndecans. More interestingly, 6-O-desulfated heparin, which is still highly sulfated due to intact 2-O and 3-O sulfations (17), fails to inhibit HIV-1 binding to syndecans (Fig. 6A). This finding suggests that the interaction between HIV-1 and syndecans is not simply the result of random interactions between basic residues in gp120 and negative charges in syndecans, but

HIV-1 Binding to Syndecans via 6-O-Sulfation
the result of specific interactions between gp120 and a well defined sulfation, the 6-O sulfation.
More importantly, the sulfated, but not the unsulfated CCR5 peptide, prevents HIV-1-syndecan interactions (Fig. 6C). This further suggests that the sulfated N terminus of CCR5 contains motifs that resemble those formed by the 6-O sulfation in syndecans. The inhibitory effect of sulfated CCR5 peptide is only moderate in the absence of soluble CD4 further confirming that HIV-1 binds to CCR5 in a CD4-independent manner.
A few anti-gp120 antibodies inhibit HIV-1-CCR5 interactions, because they contain sulfated tyrosines, which mimic the tyrosine-sul-FIGURE 6. HIV-1 recognizes analogous sulfated motifs in syndecans and CCR5. A, syndecan-2ϩ cells were exposed to JR-CSF in the presence of heparin derivatives, washed, and lysed, and p24 content was measured by ELISA. Results are expressed in percentage of p24 captured relative to the initial inoculum. B, CCR5ϩ cells were incubated with Fc-gp120 in the presence of heparin derivatives or sulfated CCR5 peptides. Bound Fc-gp120 was detected by FACS using an anti-human Fc fluorescein isothiocyanate antibody. Results are expressed in fluorescence units. C, syndecan-2ϩ cells were exposed to JR-CSF virus in the presence of sulfated CCR5 peptides and sulfated anti-gp120 antibodies, washed, and lysed, and p24 content was measured by ELISA. 447-52D and b12 antibodies were used as positive and negative control, respectively. Results are expressed in percentage of p24 captured relative to the initial inoculum. Results are representative of two independent experiments. All experiments were conducted in the presence (with sCD4) or absence (without sCD4) of soluble CD4. NOVEMBER 25, 2005 • VOLUME 280 • NUMBER 47 fated N terminus of CCR5 (Fig. 5) (16). We thus tested the effect of these antibodies on HIV-1-syndecan interactions. We used the tyrosine-sulfated 412d antibody (412d YY) and its nontyrosine-sulfated version (412d FF) that contains phenylalanines instead of tyrosines (16). 412d YY, but not 412d FF, blocks HIV-1-CCR5 interactions (Fig. 6C) and decreases R5 virus infection (i.e. ADA and YU2) (16). The sulfated 412d YY antibody blocks HIV-1 binding to syndecans more efficiently than the unsulfated 412d FF antibody (Fig. 6C). This is expected, given that the unsulfated 412d FF antibody binds poorly gp120 (10 -20% of sulfated 412d YY antibody) (16). The sulfated 412d YY antibody-mediated inhibitory effect was comparable to that of the anti-V3 447-52D antibody, which efficiently prevents HIV-1 binding to syndecans (Fig. 1C). Soluble CD4 amplifies the inhibitory effect of the sulfated 412d antibody (Fig. 1C), because 412d recognizes a CD4-induced gp120 epitope (16). 447-52D (anti-V3 region) and b12 (anti-CD4 binding site) antibodies were used as positive and negative control, respectively. These data remarkably suggest that the heparan-sulfated motifs of syndecans, the tyrosine-sulfated motifs of CCR5, and tyrosine-sulfated anti-gp120 antibodies contain analogous sulfated motifs that permit gp120 contact.

DISCUSSION
HIV-1 has maximized its utilization of syndecans. It uses them as in cis receptors to facilitate infection of macrophages (1). It also uses syndecans as in trans receptors to infect T cells (2). By serving as in trans receptors, syndecans expressed on the high endothelial venules can create hot spots for high efficiency HIV-1 transmission to T cells, which traffic in and out the lymph nodes (2). Because the surface area of the endothelial lining of the vasculature is estimated to be 600 m 2 , the syndecan-rich endothelium may represent a significant in trans viral reservoir. Moreover, syndecans on brain microvascular endothelial cells modulate cell-free HIV-1 transmigration through the blood-brain barrier (3,12). Given that endothelial and macrophage syndecans are potent receptors for HIV-1 (2), they are likely to represent key determinants in HIV-1 pathogenesis.
To obtain a more profound understanding of the mechanistic interplay between HIV-1 and syndecans, we sought mutant viruses that are unable to bind syndecans. We demonstrated that the V3 of gp120, but not V1 and V2, contains the determinants necessary for HIV-1 binding to syndecans. This is in accordance with the observation that polyanions mask epitopes on gp120 recognized by anti-V3 antibodies (39). We demonstrated that V3 interacts with the heparan sulfates of syndecans to mediate contact between the virus and the cell surface receptor. Moreover, we found that a single highly conserved arginine at the N terminus of V3 (Arg-298) is critical for gp120-and HIV-1-syndecan contact. This is in accordance with our findings that V3 peptides encompassing the N terminus of V3 as well as antibodies directed to the N-terminal ␤-strand of V3, interfere with the binding of HIV-1 to syndecans. Interestingly, an additional amine group on the side chain of the residue at position 298 (arginine, lysine, or glutamine) is required for syndecan utilization by HIV-1. Importantly, we found that a specific sulfation in syndecans is critical for HIV-1-syndecan interaction, the 6-O sulfation. This demonstrates that the interaction between HIV-1 and syndecans is not simply the result of random interactions between basic residues in gp120 and negative charges on human cells, but the result of specific interactions between gp120 and a well defined sulfation, the 6-O sulfation. This also suggests contacts between by the amine group on the side chain of Arg-298 in V3 and the 6-O sulfation on syndecans. Supporting the notion that specific sulfations govern interactions between a ligand and its receptor, Wang et al. showed that the 2-O sulfation is necessary for fibroblast growth factor] binding to syn-decans, whereas the 6-O sulfation is critical for the formation of a fibroblast growth factor]-syndecan-fibroblast growth factor] receptor ternary complex (40). Thus, distinct O-sulfations differentially contribute to contact between a ligand and its receptor. Our finding that HIV-1 binds syndecan via a specific sulfation is reminiscent of HSV-1, which binds cells through interactions of gB and gC with heparan sulfates (41). However, HSV-1 requires the 3-O but not the 6-O sulfation (41), suggesting that viruses can share and exploit syndecans using distinct sulfations.
We found that the residue responsible for HIV-1 binding to syndecans, Arg-298, is also responsible for HIV-1 binding to CCR5. The introduction of mutations at this position does not alter the structure of gp120. The V3 can be removed from gp120 without affecting its binding to CD4 and DC-SIGN. Anti-V3 antibodies recognize wild-type and Arg-298 mutant gp120 equally well, even antibodies that recognize conformational epitopes. Mutating Arg-298 does not influence gp120 incorporation (20). Conservative mutations in Arg-298 preserve the ability of gp120 to bind CCR5 (20). Thus, apparently the global structure of the V3 of Arg-298 mutants is preserved and thus cannot explain their inability to utilize syndecans and CCR5. Our surprising observation that gp120, via the highly conserved Arg-298, interacts with two entirely different receptors, led us to postulate that it recognizes similar motifs on syndecans and CCR5. Supporting this hypothesis, we obtained several lines of evidence that suggest that the sulfated heparan motif recognized by HIV-1 on syndecans mimics the sulfated tyrosine motif recognized by HIV-1 in CCR5. Specifically, sulfated heparan derivatives inhibit both gp120-syndecan and -CCR5 interactions. Conversely, tyrosine-sulfated CCR5 peptides prevent gp120-syndecan and -CCR5 interactions. Importantly, 6-O-desulfated heparans, although still highly 2-O-and 3-O-sulfated (17), do not interfere with gp120-syndecan and -CCR5 interactions. This suggests that Arg-298 can bind equally well the sulfated tyrosines in CCR5 and the 6-O sulfation of syndecans. Moreover, sulfated, but not unsulfated anti-gp120 antibodies, block HIV-1 binding to both CCR5 and syndecans. Altogether these data suggest that the 6-O sulfation motifs of syndecans, the tyrosine-sulfated motifs of CCR5 and tyrosine-sulfated anti-gp120 antibodies contain analogous sulfated motifs that are similarly recognized by HIV-1, likely via Arg-298 of gp120. The fact that sulfated anti-gp120 antibodies that mimic these sulfations spontaneously arise underscores the importance of HIV-1 sulfation contacts for HIV-1 pathogenesis. Given that the N terminus of CXCR4 is also tyrosine-sulfated (35), it will be interesting to analyze HIV-1-syndecan and -CXCR4 interactions.
Given that HIV-1 binds syndecans without the need for cell-surface CD4 and that HIV-1 binds syndecans even in the presence of soluble CD4, this suggests that Arg-298 is accessible to syndecans with or without CD4. First, it is important to note that the V3 region must be (at least partially) exposed, given that several anti-V3 antibodies block HIV binding to syndecans. One can envision that the long linear heparan sulfate chains of syndecans can reach Arg-298 at the base of the V3 loop without the need for a major structural change (i.e. CD4-induced V3 region exposure). In other words, the "linear" nature of the heparan sulfate chains may obviate the need for gp120 structural change. In contrast, Arg-298 as well as additional gp120 residues necessary for CCR5 binding are not sufficiently exposed for adequate CCR5 contact. In this scenario, the initial CD4 binding is absolutely required for subsequent CCR5 interaction. Moreover, one can envision that the number of residues within gp120 necessary for gp120-CCR5 contact is superior to that of the residues necessary for gp120-syndecan contact. Indeed, it is possible that only a few residues within gp120 are required for 6-O sulfation contact with syndecans, whereas many residues are required for CCR5 contact (i.e. larger contact interface).
Why does HIV-1 require syndecans for optimal infection, only when target cells express low levels of CCR5 and CD4? It is thought that gp120 initially binds the sulfated N terminus and subsequently to an extracellular loop of CCR5. One can envision that syndecans substitute for the sulfated N terminus of CCR5 in the initial binding of HIV-1 and thus facilitate subsequent interactions with CCR5. Supporting this hypothesis, overexpression of an N-terminal tyrosine CCR5 mutant is required for HIV-1 infection in CD4-positive HeLa cells (42,43). Another possibility is that sulfated syndecans predispose CCR5 for optimal HIV-1 entry. When target cells express minimal amounts of CCR5, the cooperation of multiple CCR5 molecules represents a precondition for successful infectivity (42)(43)(44). Thus, syndecans may facilitate HIV-1 entry by concentrating CCR5 molecules poorly expressed on target cells. Syndecans and CCR5 or CXCR4 exist as preformed complexes on macrophages and HeLa cells (45)(46)(47). This supports the possibility that syndecans may contribute to the cooperation of multiple CCR5 molecules in the HIV-1 infection pathway. It is also interesting to note that, although syndecans and CCR5 share Arg-298 in gp120, syndecans do not interfere in HIV-1 entry. One can speculate that, although some gp120 trimers interact with syndecans others interact with CD4 and CCR5.
Our study demonstrates that HIV-1 via a single residue in V3 of gp120 can interact with two disparate receptors via analogous sulfated motifs. This is reminiscent of chemokines. Like gp120, RANTES, MIP-1, and MCP-2 bind CCR5 (48,49) as well as syndecans (HSPGs) (45,46,50). Both gp120 and chemokines have syndecan binding sites, which could overlap with chemokine receptor binding sites (51). Both gp120 and chemokines bind CCR5 in a two-step binding model: first to the sulfated N terminus of CCR5 and secondly to an extracellular loop of CCR5 (48,49). Both gp120 (at least V3 peptides) and chemokines (MIP-1 and RANTES) share structural similarities at a ␤ hairpin comprised of two anti-parallel ␤ strands (52). Both HIV-1 and chemokines bind endothelial syndecans to enhance their endurance in vivo (2,49). Chemokine mutants unable to bind syndecans retain their chemotactic activities in vitro, but not in vivo (53), suggesting that HIV-1-syndecan interactions may play an even more critical role in vivo than in vitro. Altogether these analogies suggest that HIV-1 encodes a chemokinelike protein, gp120. This is reminiscent of another pathogen, Toxoplasma gondii that encodes the cyclophilin-18 protein, which also binds both syndecans and CCR5 via its sulfated N terminus and acts as a potent CCR5 antagonist for R5 viruses (54). Further work is required to determine if this chemokine mimicry is a common strategy in microbial pathogenesis to evade the immune system.