RANTES (CCL5) uses the proteoglycan CD44 as an auxiliary receptor to mediate cellular activation signals and HIV-1 enhancement

The chemokine RANTES (regulated on activation normal T-cell expressed and secreted; CCL5) is a member of the CC-chemokine family, a group of small proteins with a highly conserved tertiary.^1,2  Chemokines recruit and activate specific leukocyte populations. Multiple chemokine receptors with partially overlapping specificities for ligand binding have been described, suggesting a possible redundancy of the system. However, an increased selectivity of the action of chemokines is thought to be gained through their ability to discern and preferentially bind to certain glycosaminoglycan (GAG) subpopulations, a process considered important in mediating tissue-specific leukocyte recruitment.^3-6 

The chemokine RANTES has a complex influence on the biology of a variety of cell types including T lymphocytes, monocytes, natural killer cells, dendritic cells, basophils, and eosinophils.^7,8  At nanomolar concentrations, RANTES binds to and activates several 7-transmembrane G protein–coupled receptors (GPCRs), namely the chemokine receptors CCR1, CCR3, and CCR5. Ligation of these receptors activates a heterotrimeric Gαi protein–coupled signaling pathway, characterized by a transient Ca²⁺ influx,^7,9  and triggers activation of cell polarization and chemotaxis.¹⁰ 

In addition to these classic chemokine-activated responses, RANTES also induces several biochemical and biologic effects that are to date unique to this chemokine and that are triggered through a GPCR-independent pathway. Induction of this pathway is mediated by protein tyrosine kinases (PTKs), occurs at high micromolar concentrations of the chemokine, and leads to a sustained influx of Ca²⁺.⁹  We and others have previously demonstrated that the activation of this pathway requires an interaction of RANTES with cell surface GAGs and depends on the ability of the chemokine to form oligomers.^6,8,11-16  Stimulation of the GPCR-independent pathway by RANTES induces hyperphosphorylation and generalized cell activation.^7,8  A variety of effects is then triggered, such as induction of proliferation, apoptosis, and release of proinflammatory c ytokines.^{7,8,10,15,17-22} 

RANTES, like the other CCR5 ligands macrophage inflammatory protein 1α (MIP-1α; CCL3) and MIP-1β (CCL4), can inhibit the entry of HIV type 1 (HIV-1) strains that use CCR5 as an entry coreceptor with CD4 (R5 strains).^23,24  The chemokines inhibit entry of these R5 viruses by down-regulating CCR5 and by competing with the virus for binding sites on the same receptor.^23,25-28  However, RANTES has a bipartite effect, because it can also enhance HIV-1 replication in a dose-dependent manner.^{11,12,16,29-33}  We and others have shown that this effect is independent of known GPCRs for RANTES but is dependent on the ability of the chemokine to interact with cell surface GAGs and to form multimers.^6,11-16  Two mechanisms are involved in HIV-1 infectivity enhancement: the physical cross-linking of virions to target cells by oligomers of RANTES and the activation of different kinase-signaling pathways including PTK, focal adhesion kinase (p125 FAK), and the p44/p42 mitogen-activated protein kinase (MAPK).^{6,11,12,16,31}  Activated p44/p42 MAPKs have been suggested to phosphorylate the viral Vif protein and the matrix (p17) proteins, and also to influence reverse transcription and integration processes.^34-39  Activation of MAPK by RANTES and stromal-derived factor 1α (SDF-1α)/CXCL12 has also been suggested to enhance viral infectivity by promoting nuclear translocation and integration.^31,40 

Here we analyze the biochemical signals induced by the interaction of RANTES with GAGs on cells that lack GPCRs for RANTES and on primary CD4⁺ T cells, with emphasis on defining the receptor molecules involved in the GPCR-independent signaling pathway. We show that the proteoglycan CD44, via its GAG chains, can function as an alternate, signaling receptor for RANTES and is required for the activation of p44/p42 MAPK by RANTES. We provide evidence that both cellular activation and enhancement of HIV-1 infectivity by RANTES are linked to signals induced through this proteoglycan receptor. Our studies help to reveal the signaling functions of cell surface GAGs and the possible involvement of these molecules in chemokine function and the HIV-1 replication cycle.

The following polyclonal antibodies (pAbs) and monoclonal antibodies (mAbs) were used for immunoprecipitation (IP) and Western blot analysis: mouse mAb anti–heparan sulfate (anti-HS) 10E4 and anti–chondroitin sulfate (anti-CS; MC21C) from Seikagaku (Tokyo, Japan); goat pAbs against RANTES from PeproTech (London, United Kingdom), mouse mAb anti-phosphotyrosine (P-tyr) 4G10 and mouse mAb anti-phosphoserine/threonine (P-ser/thr) MPM-2 from Upstate Biotechnology (Lake Placid, NY), rabbit pAbs against p44/p42 MAPKs from Upstate Biotechnology, mAb against phospho-p44/p42 MAPKs from New England Biolabs (Beverly, MA), rabbit anti-FAK[pY397] pAbs from Biosource International (Camarillo, CA), and rabbit anti-FAK pAbs from Pharmingen; anti-Src rabbit pAbs from Biosource International; rabbit anti-Fyn pAb (FYN3) from Santa Cruz Biotechnology (Santa Cruz, CA); mouse anti-Hck (clone 18) mAb from Transduction Laboratories (Lexington, KY); anti-Shc rabbit pAbs from Upstate Biotechnology; mouse anti-CD44 mAb 365 (used for IP) and clone F10-44-2 (for Western blot) from Serotec (Oxford, United Kingdom); mouse anti-CD138 mAb B-B4 (for IP) from Immunotech (Marseille, France) and mouse anti-CD138 mAb (for Western blot) from Serotec.

The following antibodies conjugated with phycoerythrin (PE) or fluorescein isothiocyanate (FITC) were used for fluorescence-activated cell sorting (FACS) analysis: murine mAbs anti-CD44 (IgG2b), anti-CD138 (IgG1), anti-CD4 (IgG2a), and anti-RANTES (IgG2b) as well as all murine isotype control antibodies were from Caltag Laboratories (Burlingame, CA).

Recombinant human RANTES and its derivatives Met-RANTES, Met-RANTES E66S, produced by polymerase chain reaction (PCR) with the oligonucleotide corresponding to the Glu66Ser mutation, and [⁴⁴AANA⁴⁷] RANTES were generated in the host Escherichia coli as described previously.^41,42 

All RANTES molecules reacted with the polyclonal anti-RANTES antibodies used for analyses (Figure 3C and data not shown). The inhibitor pertussis toxin was from Biomol Research Laboratories (Plymouth Meeting, PA). All other reagents, when not specified differently, were from Sigma-Aldrich (Poole, United Kingdom).

HeLa-CD4 cells were provided by David Kabat (University of Oregon, Portland). The cells were maintained in Dulbecco minimal essential medium (DMEM; BioWittaker, Walkersville, MD) containing 10% fetal calf serum (FCS) and 2 mM l-glutamine, and split twice a week. For use in signal cascade analyses, HeLa-CD4 cells were washed with phosphate-buffered saline (PBS) twice, then maintained in DMEM with 0.1% FCS for 48 hours prior to RANTES addition.

Peripheral blood mononuclear cells (PBMCs) were isolated from healthy blood donors by Ficoll-Hypaque centrifugation and then stimulated for 3 days with 5 μg/mL phytohemagglutinin (PHA) in RPMI 1640 medium (BioWhittaker) containing 10% FCS, 10 U/mL interleukin 2 (IL-2; National Institutes of Health Research Laboratories, Bethesda, MD), 2 mM l-glutamine, and antibiotics. After stimulation, CD4⁺ T cells were purified from these PHA-stimulated PBMCs by positive selection with anti-CD4 immunomagnetic beads (Dynal Biotech, Oslo, Norway) according to the manufacturer's instructions. The purified CD4⁺ T cells were cultured in RPMI 1640 supplemented with 10% FCS, 100 U/mL IL-2, 2 mM l-glutamine, and antibiotics, for 3 to 5 days before being used in analyses. For analysis of protein phosphorylation state, CD4⁺ T cells were cultured for 24 hours in medium without IL-2, then serum-starved for 5 hours in RPMI 1640 containing 0.5% FCS before being treated with RANTES.

Enzymatic digestion of cells surface GAGs was done by treating cells with heparinase I, II, and III (500, 250, and 50 mU/mL, respectively; all from Sigma-Aldrich) and chondroitinase-ABC (50 mU/mL; Sigma-Aldrich) in RPMI 1640 medium at 37°C for 3 hours. When indicated, the enzymatic digestion was performed after the IP step with heparinase I, II, and III (200, 100, and 20 mU/mL, respectively) and chondroitinase-ABC (50 mU/mL) in RPMI 1640 medium at 37°C for 2 hours. The efficiency of enzymatic digestion was monitored by Western blot analysis using anti-HS mAb 10E4 or anti-CS mAb MC21C.

Whole cell extracts (WCEs) were prepared as described previously.¹⁶  Briefly, the cells were lysed in 20 mM HEPES (N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid) buffer (pH 7.9) containing 1.5% Nonidet P-40 (NP-40), 25% glycerol, 140 mM NaCl, 1 mM EDTA (ethylenediaminetetraacetic acid), 1 mM MgCl₂, 1 mM dithiothreitol (DTT), 1 mM phenylmethylsulfonyl fluoride (PMSF), and a protease inhibitor cocktail (Boehringer Mannheim, Mannheim, Germany). For analysis of protein phosphorylation state, the lysate buffer was supplemented with 1 mM sodium orthovanadate and 10 mM NaF. Equal amounts of WCEs were incubated with appropriate antibodies (1 μg) and protein G–Sepharose 4 Fast Flow beads (Amersham Pharmacia Biotech, Uppsala, Sweden) at 4°C for 15 hours. Where indicated, enzymatic digestion of HS and CS was performed following IP. Immunoprecipitated protein complexes were washed 3 times with lysis buffer prior to separation by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE). Depending on the proteins analyzed, 10%, 12%, or 4% to 20% gels (Bio-Rad, Hemel Hempstead, United Kingdom) were used as specified in the figure legends. Following electrophoresis, proteins were transferred to polyvinylidene fluoride (PVDF) membranes (Millipore, Bedford, MA), blocked, then incubated with the appropriate primary antibodies at 4°C overnight. The immunoblotted proteins were visualized with horseradish peroxidase–linked secondary antibodies (Dako, Glostrup, Denmark) using enhanced chemiluminescent (ECL) substrate according to the manufacturer's specifications (Amersham Pharmacia Biotech). To reprobe blots, membranes were incubated in stripping buffer (100 mM β-mercaptoethanol, 2% SDS, 62.5 mM Tris (tris(hydroxymethyl) aminomethane)–HCl, pH 6.7) at 55°C for 1 hour and rinsed with PBS several times prior to a second Western blot analysis.

FACS analysis was carried out as described previously using a FACScalibur flow cytometer (Becton Dickinson, Franklin Lakes, NJ).²⁸  Statistical analysis was done after background correction. Mean fluorescence intensity (MFI) values are presented.

CD44 gene–specific sense and antisense 21-nucleotide (nt) single-strand (ss) RNAs, with symmetrical 2-nt 3′(2′-deoxy) thymidine overhangs were chemically synthesized and high-performance liquid chromatography (HPLC) purified by Xeragon (Huntsville, AL). The CD44 gene–specific sense sequence with the GenBank accession number CD44exon3 (CD44e3), accession number L05408, nt 126-154 was used. RNA sequences corresponding to CD44e3 were: sense 5′-GUAUGACACAUAUUGCUUCTT; antisense 5′-GAAGCAAUAUGUGUCAUACTT. To monitor the sequence specificity of CD44 RNA interference (RNAi) and to control for adverse effects additional double-strand (ds) RNAs with the inverted (inv) CD44e3 sequence was synthesized in parallel: inverted sense 5′-CUUCGUUAUACACAGUAUGTT; inverted antisense 5′-CAUACUGUGUAUAACGAAGTT. For RNAi experiments, dsRNAs were generated by mixing equimolar amounts (20 μM) of sense and antisense ssRNAs in annealing buffer (100 mM potassium acetate, 30 mM HEPES-KOH, 2 mM magnesium acetate, at pH 7.4) for 1 minute at 94°C, followed by 60 minutes incubation at 37°C.

HeLa-CD4 cells (1.5 × 10⁵ cells) were transfected with 300 nM (4.5 μg) dsRNA in serum-free OptiMEM medium using Oligofectamine reagent (Gibco BRL/Life Technologies, Grand Island, NY) following the manufacturer's instructions. Mock cells were cultured in parallel and transfected with the transfection mixture lacking dsRNA. Two consecutive transfections were performed at 24-hour intervals.⁴³  Cells transfected with CD44-specific RNAi were used 3 days after transfection for further analysis.

The efficiency of CD44-specific RNAi was assayed by analyzing the expression of CD44-specific mRNA and CD44 protein expression by FACS analysis.

CD44 mRNA⁴⁴  and, to normalize for input of total RNA, glyceraldehyde 3-phosphodehydrogenase (GAPDH)–mRNA⁴⁵  were quantified by reverse transcription (RT) real-time PCR in separate reactions. CD44 mRNA was measured using primer mf92 (5′-ACTGGGAGGTGTTGTATGTGAGGATGTA-3′) for cDNA synthesis, primers mf90 (5′-CACGTGGAGAAAAATGGTCGCTACA-3′), mf91 (5′-GTTTGCTGCACAGATGGAGTTGG-3′) and mf93 (5′FAM-ATCGGATTTGAGACCTGCAGGTATGGGTT3′-TAMRA) for amplification and detection. GAPDH mRNA was measured, using primer mf46 (5′-GGCAACAATATCCACTTTACCAG-3′) for cDNA synthesis, the primers mf45 (5′-TCGACAGTCAGCCGCATCTT-3′) and mf46 and mf70 (5′FAM-AAGGTCGGAGTCAACGGATTTGGTCGT-3′TAMRA) for amplification and detection, respectively.

PCR was performed using a single-tube system (Qiagen 1-step RT-PCR kit; Qiagen, Hilden, Germany) and an additional “hot-start” using Ampliwax (Applied Biosystems, Rotkreuz, Switzerland) to separate cDNA synthesis and PCR as follows. Aliquots of a lower phase mix were prepared containing 10 μL 1 × reaction buffer (including 2.5 mM MgCl₂), 2.3 μM amplification primers (CD44: mf90, mf91; GAPDH: mf45, mf46), 0.46 μM of the fluorescent probe (CD44: mf93; GAPDH: mf70), 1.15 mM MgCl₂, and 23 ng/μL poly-A carrier RNA. Ampliwax was added to each reaction and lower phases were sealed by incubation at 80°C for 5 minutes and cooling to room temperature. Upper phase mix (10 μL 1 × reaction buffer [including 2.5 mM MgCl₂], 0.26 μM of the primer for first-strand cDNA synthesis [CD44: mf92; GAPDH: mf46], 0.92 mM dNTP and 9% [vol/vol] enzyme mix), and template RNA (3 μL) were added. cDNA synthesis and subsequent amplification were performed in a real-time thermocycler (i-cycler; Bio-Rad) as follows: 50°C 30 minutes, 95°C 15 minutes, and 60 cycles (95°C 10 seconds, 60°C 1 minute) with monitoring of fluorescence at 60°C. Quantification was performed using an external standard curve by monitoring 10-fold serial dilutions of total human PBMCs. CD44 mRNA levels were standardized to GAPDH mRNA content and expressed relative to those in mock-transfected control cells.

The extent of HIV-1 entry was determined using a single-cycle infection assay as described previously.^11,12,16  Amphotropic murine leukemia virus (MuLV) envelope-pseudotyped, luciferase (luc)–expressing HIV-1 reporter virus stocks (HIV-1_MuLV-luc) were produced using the calcium phosphate technique as described previously.^46-48  HIV-1 p24 antigen concentrations in the virus stocks were measured by an in-house p24 antigen enzyme-linked immunosorbent assay^49,50  to standardize the amount of input virus.

One day before infection, 1 × 10⁴ HeLa-CD4 cells were plated into wells of a 96-well tissue culture plate. The cells were allowed to adhere for 5 hours and then treated with 640 nM (5 μg/mL) RANTES for 24 hours. Cells were then washed, supplemented with fresh medium, and infected with HIV-1_MuLV-luc in the absence of RANTES. The infection step was carried out in a total volume of 200 μL for 2 hours at 37°C, before removal of the virus and replenishment of the cells with fresh medium. Seventy-two hours after infection, the cells were lysed in 50 μL1 × reporter lysis buffer (Promega, Madison, WI). The luciferase activity of a mixture of 100 μL luciferase substrate (Promega) and 30 μL cell lysate was measured in relative light units (RLU) using a DYNEX MLX microplate luminometer (Chantilly, VA). Eight replicate wells per treatment condition were tested.

We previously described that RANTES interaction with target cell GAGs activates a PTK-signaling cascade.^12,16  The purpose of our present study was to define the molecules associated with binding and delivering these signals. RANTES, as other chemokines, binds selectively to specific GAG subpopulations and was found to have a broad range of affinities with 3 orders of magnitude higher binding affinity for heparin than for chondroitin sulfate.^5,51  Here we probed if RANTES interacts with both heparan sulfate (HS) and chondroitin sulfate (CS) moieties on HeLa-CD4 cells. To this end cells were treated for 24 hours with 640 nM RANTES followed by IP with anti-HS or anti-CS antibodies. RANTES was found to coimmunoprecipitate with both HS and CS on HeLa-CD4 cells (Figure 1A). These interactions were almost completely abolished when the GAGs were digested with heparinase and chondroitinase-ABC enzymes after immunoprecipitation, proving the specificity of the RANTES-GAG association (Figure 1A). Furthermore, enzymatic digestion of GAGs on RANTES-treated cells, prior to cell lysis and IP with RANTES-specific antibodies, reduced the amount of detectable chemokine to background levels (Figure 1B) indicating that RANTES does not directly bind to the protein core of proteoglycans.

We next studied the interaction between RANTES and GAGs expressed on activated primary human CD4⁺ T cells (Figure 1C) using the following RANTES mutants: the RANTES antagonist Met-RANTES, which has impaired ability to signal through CCRs⁵² ; the mutant Met-RANTES E66S that in addition does not oligomerize and cannot enhance HIV-1 infectivity^11,13 ; the mutant [⁴⁴AANA⁴⁷] RANTES, which has intact CCR-signaling capacity but whose primary GAG-binding motif BBXB at residues 44-47 has been altered to reduce its ability to bind GAGs and enhance HIV-1 infectivity.⁶  Wild-type RANTES, the signaling-impaired Met-RANTES, and the signaling- and oligomerization-defective Met-E66S RANTES bound equally well to primary CD4⁺ T cells (Figure 1C). However, no binding above background levels was observed with [⁴⁴AANA⁴⁷] RANTES. This indicates that on primary CD4⁺ T cells, just as with HeLa-CD4 cells, most RANTES associated with the cell surface binds to GAGs. Moreover, when studying RANTES binding to HeLa-CD4 cells and primary CD4⁺ T cells by FACS analysis we found that a dose-dependent deposition of wild-type RANTES protein but not of [⁴⁴AANA⁴⁷] RANTES occurred with both cell types (Figure 1D-E).

In agreement with our observations on HeLa-CD4 cells,¹⁶  we found that RANTES activates p44/p42 MAPK in primary CD4⁺ T cells in a dose-dependent manner (Figure 2A). Even a low concentration of RANTES (50 nM) was sufficient to increase MAPK activation, although only marginal deposition of RANTES on target cells occurred at this concentration (Figure 1E). To rule out that the observed MAPK activation was due to signaling through the RANTES chemokine receptors CCR1, CCR3, or CCR5, we studied the effect of RANTES on MAPK activation in the presence of the potent inhibitor of Gαi proteins, Bordatella pertussis toxin. Pertussis toxin did not inhibit MAPK activation induced by RANTES (640 nM; Figure 2B). Thus, RANTES activates a Gαi-independent signaling pathway in CD4⁺ T cells even though these cells possess specific GPCRs for RANTES that are linked through Gαi proteins.

In concordance with our binding studies (Figure 1C-E), we found that [⁴⁴AANA⁴⁷] RANTES failed to induce p44/p42 MAPK activation on HeLa-CD4 cells (Figure 2C) or primary CD4⁺ T cells (Figure 2E) nor did it induce FAK activation (Figure 2D) on HeLa-CD4 cells. The CCR5 signaling-impaired Met-RANTES mutant activated MAPK to the same extent as the wild-type RANTES protein in primary CD4⁺ T cells (Figure 2E). However, in agreement with our previous observation on HeLa-CD4 cells, the oligomerization-defective Met-E66S mutant failed to induce MAPK activation in these cells. These data once more indicate that MAPK activation through an oligomerization-competent RANTES molecule is induced by GAG interaction but not through classical GPCR involvement.

Based on our observations 2 possible roles for proteoglycans in mediation of the signals by RANTES can be envisioned: The GAG side chains of proteoglycans could serve as an anchor for the chemokine, to enable it to interact efficiently with one or more yet unidentified signaling receptors. Alternatively, the proteoglycan itself could be both the binding anchor and signaling receptor. Our first approach was to investigate the latter possibility. Because tyrosine kinases are crucially involved in transmitting the signal induced by RANTES,^11,16  we focused our investigation on 2 well-described proteoglycans, CD44 and CD138/syndecan-1, both of which associate with activated Src tyrosine kinases⁵³  and are expressed on the target cells investigated.^54-56  The HeLa cell line we used for our analyses expresses high levels of CD44 and relatively low but significant levels of CD138 (Freissler et al⁵⁷  and data not shown). Activated human CD4⁺ T cells are known to express high levels of CD44, but lack CD138/syndecan-1 expression.^58-60 

We found that RANTES can be coimmunoprecipitated with CD44 and CD138 from HeLa-CD4 cells treated for 24 hours with RANTES (640 nM; Figure 3A-B). IPs with anti-RANTES antibodies followed by Western blot analysis revealed that multiple isoforms of CD44 (p42, p75, p95) and CD138/syndecan-1 (p35, p75) are expressed on HeLa cells and were coprecipitated with RANTES (data not shown). The interactions of RANTES with CD44 and CD138/syndecan-1 were totally abolished when cell surface GAGs were digested with heparinase and chondroitinase-ABC enzymes prior to IP with anti-CD44 or anti-CD138 antibodies (Figure 3B). Thus RANTES interacts with CD44 and CD138/syndecan-1 through GAG side chains, but not via binding to the protein core of the proteoglycans. These observations agree with our findings that RANTES binds predominantly to GAG moieties on the cell surface and is not primarily associated with protein structures (Figure 1).

We further observed that high concentrations (640 nM) of wild-type but not mutant [⁴⁴AANA⁴⁷] RANTES bound to CD44 (Figure 3C middle panel). At low concentrations, RANTES (50 nM) bound weakly to CD44 but not to CD138 on HeLa-CD4 cells (Figure 3C). RANTES also bound to CD44 on primary CD4⁺ T cells but only at high, not low concentrations tested (Figure 3D). This could in part be due to the relative limited sensitivity of the IP procedure.

The highly conserved cytoplasmic⁶¹  tail of CD44 is known to undergo specific phosphorylation or dephosphorylation of serine residues on ligand binding.⁵⁴  Here we show that different isoforms of CD44 expressed on HeLa-CD4 cells were phosphorylated at serine/threonine residues after a 30-minute exposure to RANTES (Figure 4A). Changes in the phosphorylation status of 5 protein bands were detected in response to RANTES: Serine/threonine phosphorylation was increased on bands I, II, IV, and V, whereas phosphorylation of band III decreased. In addition, there was a rapid induction of tyrosine phosphorylation on 3 major protein bands (approximate molecular mass 120, 60, and 45 kDa) that coimmunoprecipitated with CD44 (Figure 4B).

We confirmed that the 60-kDa protein bands represented members of the Src family of tyrosine kinases. Specifically, we observed association of CD44 with the kinases Fyn, Src, and Hck (Figure 4C and data not shown). The 2 other protein bands were identified as p125 FAK (Figure 4D) and the p46 isoform of Shc, an adapter molecule found in different intracellular pathways (Figure 4E and Ravichandran⁶² ). The protein complex between CD44, Src-kinases, p125 FAK, and Shc existed also in the absence of RANTES stimulation, but association of CD44 with these intracellular proteins increased significantly on chemokine treatment (Figure 4C-E). A dose-dependent induction of tyrosine phosphorylation of several proteins associated with CD44 was also observed on RANTES treatment of primary CD4⁺ T cells (Figure 4F, A-D).

In contrast to the activation of CD44 by RANTES, we did not observe specific phosphorylation of CD138 or any association of this receptor with other signaling molecules (data not shown).

We next investigated the CD44 dependency of the RANTES-induced signaling pathway, by using RNAi^63,64  to deplete endogenous CD44 protein expression from our target cells. Transfection of HeLa-CD4 cells with CD44e3 dsRNAs resulted in a pronounced CD44 mRNA down-regulation reaching 89% reduction on day 3 (Figure 5A). Similarly, when measuring protein expression by FACS we found that the percentage of cells expressing high levels of CD44 was reduced from 87.9% (MFI 420) in mock-transfected cells to 11% (MFI 399) in cells transfected with CD44e3 dsRNAs (Figure 5B). To monitor the sequence specificity for CD44 RNAi, invCD44e3 dsRNA was used as a control in all experiments. The invCD44e3 dsRNA construct caused a minor reduction of CD44 mRNA and protein expression in some but not all experiments (Figures 5, 6) concordant with previous reports on RNAi methodology.⁶⁵  There was no effect of either dsRNA CD44e3 or invCD44e3 transfection on expression of the unrelated protein CD4 (data not shown).

The extent of RANTES deposition on HeLa-CD4 cells after 24 hours of incubation with the chemokine was not influenced by the reduction in CD44 expression (Figure 5C). Residual CD44, CD138, and other proteoglycans expressed on these cells may provide sufficient GAG-binding sites for high levels of RANTES binding to still occur. In contrast, p44/p42 MAPK activation by RANTES was markedly reduced after knock-down of CD44 when compared to activity in mock-transfected cells and cells transfected with invCD44e3 dsRNA (Figure 5D), thus indicating that p44/p42 MAPK activation by RANTES is predominantly mediated by CD44 and requires a high level of expression of this proteoglycan.

Our previous studies had demonstrated that blocking of the PTK-signaling pathway by appropriate inhibitors resulted in an approximately 50% decrease of the HIV-1 infectivity-enhancing effect induced by RANTES.^12,16  We therefore investigated here whether silencing of CD44 expression and the consequent reduction of p44/p42 MAPK activation by RANTES influences the HIV-1 infectivity-enhancing activity of the chemokine. We measured HIV-1_MuLV-luc infection of HeLa-CD4 cells with or without CD44 depletion. A significant (P < .0001; Mann-Whitney test) increase in HIV-1 infectivity after 24 hours of RANTES treatment was observed in mock-transfected cells and cells transfected with the inverted control RNAi probe (10.3- and 9.1-fold increase above entry in absence of RANTES, respectively, Figure 6). However, in cells expressing reduced levels of CD44 after RNAi transfection, the enhancement of HIV-1 infection by RANTES was 40.8% lower and reached only a 6.1-fold increase. This decrease in RANTES-induced infectivity enhancement was highly significant compared to both mock treated (P < .0001; Mann-Whitney test) and inverted control treated cells (P = .0008; Mann-Whitney test). The minor reduction in infectivity enhancement observed in the inverted control treated cells in comparison to mock-treated cells did not reach significance (P = .4437; Mann-Whitney test). Taken together, our data strongly suggest that RANTES via engagement of CD44 induces p44/p42 MAPK activation, which in turn participates in increasing HIV-1 infectivity.

Proteoglycans are known to be important cofactors of various specific ligand-receptor interactions such as cytokines, chemokines, and certain growth factors.^59,66  The GAG moiety of proteoglycans can be composed of a variety of related structures, with heparin, heparan sulfate, chondroitin sulfate, and dermatan sulfate being important constituents.

To date, the association of chemokines with GAGs is considered to play a pivotal role in directing inflammatory responses by localizing cytokines and chemokines at sites of inflammation and thereby forming immobilized gradients, which are believed essential for triggering cell migration.^{3,4,6,51,67-73}  Moreover, the differential affinity of chemokines to certain GAG subclasses has been considered as an important feature of the chemokine network to increase specificity and selctivity.^3-6  Binding of chemokines to cell surface GAGs is required for their interaction and signaling through their specific GPCRs in vivo but not in all in vitro systems analyzed.^{6,42,70,72,74}  Formal proof of the chemokine-GAG interaction in vivo has recently been demonstrated since chemokines deficient in GAG binding are unable to recruit cells when administered into the peritoneum.⁴² 

The versatile effects of RANTES have drawn particular attention over recent years. Besides its actions as typical proinflammatory chemokine through interaction with its cognate, G protein–coupled chemokine receptors RANTES was shown to be a potent leukocyte activator and to trigger a variety of cellular activation signals through interaction with GAGs via induction of a GPCR-independent pathway.^7,8  It has been previously suggested that an engagement of the T-cell receptor-CD3 complex by RANTES induces the GPCR-independent, PTK-dependent signal, but this had not been confirmed experimentally.¹⁹ 

We report here that binding of RANTES to proteoglycans on primary CD4⁺ T cells and HeLa-CD4 cells seems to occur solely through association with their GAG side chains. Interaction of the chemokine with both heparan and chondroitin sulfate was observed, but we were unable to find evidence for a direct interaction of RANTES with the protein core of proteoglycans. Binding of RANTES to target cell surfaces depended with both the transformed HeLa-CD4 cell line and primary CD4⁺ T cells on an interaction with GAGs as affirmed by the failure of the GAG-binding impaired mutant [⁴⁴AANA⁴⁷] RANTES to associate with the target cell surface.

It is of note that RANTES and derivatives (Met-RANTES), which were previously found to enhance HIV-1 infectivity, were strong activators of p44/p42 MAPK, whereas mutants that lack enhancing activity (Met-RANTES E66S and [⁴⁴AANA⁴⁷] RANTES) also failed to induce MAPK activation^11-13,16  (Figures 3 and 4). Thus it is essential for the ability of RANTES to activate the p44/p42 MAPK pathway in HeLa-CD4 cells and in primary CD4⁺ T cells, that the GAG-binding BBXB motif of the protein and the capacity of the chemokine to form multimers are preserved. The mutant [⁴⁴AANA⁴⁷] RANTES, in contrast to the mutants containing the N-terminal methionine (Met-RANTES, Met-RANTES E66S), has no deficiency in signaling through CCR1 or CCR5.^6,13,14,52  Therefore the general lack of MAPK activation we observed in response to treatment with this mutant in stimulated primary CD4⁺ T cells, which express RANTES chemokine receptors, strongly suggests that the MAPK pathway is activated solely by interaction of RANTES through GAGs and not the cognate GPCR chemokine receptors. Furthermore, because a low but significant activation of this GPCR-independent pathway in primary CD4⁺ T cells occurs at physiologic, low nanomolar concentrations, activation of this pathway might also bear biologic significance.

Thus far, chemokine binding to GAGs had not been attributed to specific proteoglycans, but it has been previously shown that proteoglycan cores can be important determinants of ligand binding to GAG receptors.⁷⁵  Because our studies had underlined the importance of the interaction of RANTES with GAGs for the induction of signaling events, we sought to define whether specific proteoglycans preferentially interact with RANTES and function as both anchoring and signaling receptor for this chemokine. Based on their expression patterns and signaling capacities we selected 2 types of proteoglycans, CD44 and syndecans, for further examination of their interaction with RANTES. The primary target cells for HIV-1 in vivo, CD4⁺ T lymphocytes, macrophages, and dendritic cells, all express CD44, which is expressed as several isoforms that are generated by alternative splicing of transcripts and differ in size as well as glycosylation pattern.^58,60  In addition, macrophages and dendritic cells also express syndecans. Most cells express at least one or more of the 4 known syndecans (syndecan-1, -2, -3, and -4).^59,76  We focused our investigations on CD138/syndecan-1 due to the availability of reagents. Both CD44 and syndecans associate with activated Src-tyrosine kinases and interact with the actin cytoskeleton in cholesterol-rich lipid rafts.^53,54,56,77  Of note, binding of RANTES and MIP-1β (CCL4) to GAGs on CD44 has been reported.^78,79 

Our biochemical analyses demonstrated clearly that RANTES associated with both CD44 and CD138 via their GAG chains. However, only the RANTES interaction with GAG moieties on CD44 led to the formation of a signaling complex composed of CD44, src kinases, and adapter molecules, whereas no such activation or association with signaling molecules was detected for CD138. Thus far we have no evidence for the involvement of further receptors in the RANTES signaling process. Whether or not the failure to detect CD138 activation is due to the relatively low expression levels of this receptor on HeLa-CD4 cells and signal transduction through these molecules actually does occur remains to be determined. However, if cellular activation by RANTES through CD138 or other molecules on HeLa cells occurs, it must be of minor consequence because knock-down of CD44 alone obliterated p44/p42 MAPK activation on exposure to RANTES. Previous studies have demonstrated that RANTES is unique in its ability to stimulate cellular activation.^8,9,17  Whether the interaction of RANTES with CD44 reflects a specific property of this chemokine remains to be determined. Preliminary experiments with MIP-1α (CCL3) gave no indication of cellular activation via CD44-GAG association by this chemokine.

Stimulation of HIV-1 infectivity by RANTES and its derivatives has been previously described to depend in part on the activation of the p44/p42 MAPK.^{11,12,16,31,40}  Activation of this pathway could have an impact on the HIV-1 life cycle indirectly by activating transcription factors,³⁸  and directly by phosphorylating viral proteins and thereby regulating reverse transcription and integration events.^35-37,39  Accordingly, the induction of MAPK activation has been considered as a potential mechanism that directly increases HIV-1 infectivity.^31,40  Because silencing of CD44 expression in HeLa-CD4 cells prevented the activation of p44/p42 MAPK by RANTES, we were able to further probe the impact of this pathway on RANTES enhancement of HIV infection. The reduction in CD44 expression observed on RNAi treatment and the associated loss in MAPK signaling by RANTES did not interfere with RANTES binding to cells but was linked to a substantial reduction in HIV infectivity enhancement by RANTES.

In conclusion, our work demonstrates that the chemokine RANTES interacts specifically with the proteoglycan CD44 through association with GAG side chains on this molecule. We further provide evidence that this association occurs at both low, physiologically relevant concentrations and at higher, probably supraphysiologic concentrations of the chemokine and induces the formation of a signaling complex composed of CD44, src kinases, and adapter molecules. The ensuing MAPK activation depends on high expression levels of CD44. Therefore, CD44, via its GAG chains, functions as an alternate signaling receptor for RANTES. This is the first direct evidence that the interaction of a chemokine with GAGs goes beyond a mere sequestration and anchoring of the chemokine to cells.^3,4,6,70,80  Furthermore, our studies help reveal the signaling functions of cell surface GAGs and the possible involvement these molecules may play in the HIV-1 replication cycle.

We thank Esther Beerli, Peter Rusert, and Frédéric Borlat for technical assistance.