The Heparin/Heparan Sulfate Sequence That Interacts with Cyclophilin B Contains a 3-O-Sulfated N-Unsubstituted Glucosamine Residue*

Many of the biological functions of heparan sulfate (HS) proteoglycans can be attributed to specialized structures within HS moieties, which are thought to modulate binding and function of various effector proteins. Cyclophilin B (CyPB), which was initially identified as a cyclosporin A-binding protein, triggers migration and integrin-mediated adhesion of peripheral blood T lymphocytes by a mechanism dependent on interaction with cell surface HS. Here we determined the structural features of HS that are responsible for the specific binding of CyPB. In addition to the involvement of 2-O,6-O, and N-sulfate groups, we also demonstrated that binding of CyPB was dependent on the presence of N-unsubstituted glucosamine residues (GlcNH2), which have been reported to be precursors for sulfation by 3-O-sulfotransferases-3 (3-OST-3). Interestingly, 3-OST-3B isoform was found to be the main 3-OST isoenzyme expressed in peripheral blood T lymphocytes and Jurkat T cells. Moreover, down-regulation of the expression of 3-OST-3 by RNA interference potently reduced CyPB binding and consequent activation of p44/42 mitogen-activated protein kinases. Altogether, our results strongly support the hypothesis that 3-O-sulfation of GlcNH2 residues could be a key modification that provides specialized HS structures for CyPB binding to responsive cells. Given that 3-O-sulfation of GlcNH2-containing HS by 3-OST-3 also provides binding sites for glycoprotein gD of herpes simplex virus type I, these findings suggest an intriguing structural linkage between the HS sequences involved in CyPB binding and viral infection.

thought to control the activities of HS-binding proteins, presumably through the interactions with structurally distinct HS species from different cell types and tissue loci (3,4). In support of this idea, immunohistochemical studies of tissues with antibodies to HS epitopes have demonstrated differences within HS in distinct cellular environments (5). Structural characterization of the binding heparin/HS domain of fibroblast growth factor-2 (FGF-2) has illustrated the importance of contiguous stretches of the disulfated disaccharide IdoUA2S-GlcNS Ϯ 6S for efficient interaction. Surprisingly, experimental data have demonstrated that 6-O-sulfation of GlcNS, although not directly involved in the binding of the protein to heparin/HS, is an absolute requirement for proliferative activity of FGF-2 (6,7). Therefore, the structural heterogeneity of HS could explain the remarkable tissue-specific activities of FGF-2, with regard to the ability of HS to form active complexes with the protein and its cognate signaling receptor on responsive cells (8). Another example of binding heparin/HS domain specifically involved in protein interaction is the well defined pentasaccharide required for the binding and activity of antithrombin-III. The most distinguishing feature of this motif is the unusual 3-O-sulfated GlcNS unit, which is essential for high affinity binding of antithrombin and anti-coagulant activity (9,10).
With regard to the ability of HS domains to form active complexes with extracellular soluble mediators, it could be hypothesized that many of the regulatory functions attributed to proteoglycans may actually be related to the interactions with cell-specific HS species. A class of recently identified inflammatory mediators is secreted cyclophilins (11)(12)(13)(14), which are mostly known as cyclosporin A-binding proteins (15,16). We demonstrated that cyclophilin B (CyPB) triggers chemotaxis and integrin-mediated adhesion of T lymphocytes mainly of the CD4 ϩ /CD45RO ϩ phenotype (11). These activities are dependent on the interactions with two types of binding sites, CD147 and cell surface HS. By using site-directed mutated ligands, we demonstrated that the peptides 3 KKK 5 and 14 YFD 16 , which are located in the N-terminal extension of CyPB, are required for efficient binding of CyPB to heparin and cell surface HS expressed on T cells (17). CyPB KKK-, for which the sequence 3 KKK 5 was replaced by AAA, and CyPB ⌬YFD , which corresponds to the protein depleted of the sequence 14 YFD 16 , were both deficient in their capability to induce cell adhesion to fibronectin (18). Most recently, we found that octasaccharides are the minimal length unit required for efficient binding of CyPB to heparin and cell surface HS of T lymphocytes. Molecular modeling confirmed that 3 KKK 5 and 14 YFD 16 tripeptides were spatially arranged in the three-dimensional structure of the protein, so that they may act synergistically to form a binding site for an octasaccharide (19). Therefore, an intriguing possibility is that the pro-migratory activity of CyPB may be dependent on interaction with specific binding HS domain present at the cell surface of responsive T cells.
The focus of this study was to characterize structural features within cell surface HS, which could explain the specific interaction with CyPB. Heparin is structurally related to the sulfated regions of HS, and we demonstrated that heparin and cell surface HS of T lymphocytes share the same capability to interact with CyPB (20). Therefore, heparin and chemically modified derivatives were used to determine the structural features within HS for optimal CyPB binding. In addition to the requirement of N-and O-sulfate groups, which is a common feature for binding of numerous heparin-binding proteins, we demonstrated that interaction of CyPB with heparin/HS was strictly dependent on the presence of GlcNH 2 residues. 3-OST are represented by seven distinct isoenzymes that are expressed at distinct levels in various human tissues, suggesting their involvement in making tissue-specific HS with different biological functions (9,21,22). Interestingly, certain isoforms of 3-OST have been reported to sulfate GlcNH 2 -containing disaccharides within HS (23,24). Therefore, we explored the relationships between the expression of multiple 3-OST isoforms and the responses triggered by CyPB binding to T cells. Our results suggest an intriguing implication of the GlcNH 2 residue and 3-O-sulfation by 3-OST3B isoform in efficient binding of CyPB to responsive T cells.
Biotinylation of Heparin and Oligosaccharides-N-Hydroxysuccinimide-LC-biotin (Pierce) was used to biotinylate heparin through internal free amino groups, as described (30). According to the manufacturer (Sigma), biotinylated heparin-BSA contains 50% (w/w) heparin and 3-5 biotin residues per molecule of conjugate. Heparin chains were linked via reductive amination of the terminal reducing sugar to BSA with a heterobifunctional cross-linking reagent, and conjugation of biotin to the albumin moiety was performed via biotinamidocaproic acid (31). To check whether biotinylation did not lead to partial derivatization of free amino groups of heparin, we quantified the amount of biotin residues that are distributed between glycanic and protein moieties of heparin-BSA conjugate, by using 4Ј-hydroxyazobenzene-2-carboxylic acid. Briefly, biotinylated heparin-BSA conjugate was treated with either 0.75 unit/ml heparinase I or 1% Pronase (w/v) overnight at 37°C, in order to liberate biotin residues that would be covalently linked to heparin or albumin moieties, respectively. Products of enzymatic digestion were then separated from not degraded material by gel filtration, and the amount of biotin in solution was determined by competition with known amounts of 4Ј-hydroxyazobenzene-2-carboxylic acid bound to avidin, according to the protocol provided by Pierce. As expected, biotin residues were only recovered in the Pronase-sensitive fraction, indicating that the procedure of biotinylation of heparin-BSA conjugate has only modified primary amino groups of some lysine residues in the albumin moiety. Therefore, this type of chemical linkage would allow access of ligands to the full length of the glycanic polymer and provide heparin biosensor surfaces with high binding capacity. Heparin-derived oligosaccharides were biotinylated at their reducing ends, as described in Ref. 32. Briefly, oligosaccharides and biotin-X-X-hydrazide (Calbiochem) were allowed to react for 2 h at room temperature. Unreacted biotin reagent was removed by ion exchange chromatography, and successful biotinylation of oligosaccharides was confirmed by a dot-plot procedure (32).
Immobilization of Heparin and Oligosaccharides-Streptavidin was immobilized on planar aminosilane surfaces according to the recommendations of the manufacturer (IAsys, Neosensors, Durham, UK) and then allowed to react with biotinylated derivatives. Glycosaminoglycans give a variable and poor response in optical biosensors (32)(33)(34). Therefore, the amount of biotinylated molecules immobilized onto the cuvettes was titrated indirectly by measuring the response to a concentration of HGF above the K d value (35,36). Sufficient amounts of each size of oligosaccharides or heparin were immobilized to yield a response of no more than 100 arc seconds (600 arc seconds corresponds to 1 ng of protein/mm 2 of the cuvette surface) in the presence of 25 nM HGF. The distribution of bound HGF, and by inference immobilized heparin or oligosaccharides on the surface of the biosensor cuvette, was inspected by examination of the resonance scan in the course of the association phase of binding reactions. This allowed showing that at all times HGF was distributed uniformly on the sensor surface and therefore was not aggregated. In addition, controls demonstrated that in phosphate-buffered saline supplemented with 0.2% Tween 20 (PBS-T, phosphate buffered saline/Tween 20), CyPB, FGF-2, or HGF did not bind to aminosilane surface or to a streptavidin surface derivatized with biotin hydrazide (results not shown).
Optical Biosensor Binding Assays-A binding assay consisted of adding the ligands at a known concentration in 5 l of PBS-T into a cuvette containing either heparin-or oligosaccharidederivatized aminosilane surface equilibrated in 45 l of PBS-T. The association reaction was followed over 250 s. The cuvette was then washed three times with 50 l of PBS-T, and the dissociation of bound ligand into the bulk PBS-T was followed over time. The surface was regenerated by washing twice with 50 l of 2 M NaCl, 10 mM phosphate, pH 7.2. All biosensor experiments were carried out at least three times on at least two different surfaces. Binding parameters were calculated from the association and dissociation phases of the binding reactions as described previously (30,(32)(33)(34)(35). A single binding assay yielded four binding parameters as follows: the initial rate of association, the on-rate constant (k on ), and the extent of binding, all calculated from the association phase, and the off-rate constant (k off ), calculated from the dissociation phase. To avoid artifactual second phase binding due either to rates of diffusion of soluble ligands approaching or exceeding the rate of association or to steric hindrance at the binding surface, k on was only determined at low concentrations of ligands, whereas k off was preferentially measured in separate experiments, using higher concentrations of CyPB and competing heparin (100 g/ml) in the dissociation buffer to avoid any rebinding artifacts (34). The equilibrium constant (K d ) was calculated both from the ratio of the dissociation and association rate constants and, independently, from the extent of binding at the equilibrium in order to provide an estimate of the self-consistency of the results. A single binding site model fitted the data at least as well as a two-site binding model in both the competitive binding assays and the kinetic experiments. Therefore, the binding reaction between the ligands and immobilized heparin was deemed to be monophasic, and a single site model was used to calculate all binding parameters using the nonlinear curve fitting FastFit software (Neosensors). In competitive binding assays, an aminosilane cuvette derivatized with biotinylated BSA-heparin was used with a customized program for the Iasys Autoϩ instrument (Neosensors). The cuvette was equilibrated at 20°C in 40 l of PBS-T, and 5 l of the relevant dilutions of chemically modified heparins in PBS-T was added. Once the base line was stable, 5 l of CyPB was added to initiate the association phase. The binding reaction was continued for 5 min. The surface was washed three times with 50 l of PBS-T and was regenerated with 2 M NaCl. No change was detected in the amount of CyPB bound to immobilized heparin in the absence of competing heparin derivatives at the start and the end of the experiments. The extent of binding was calculated by fitting the association curve to a single site binding model using the nonlinear curve fitting FastFit software.
Deaminative Cleavage of Heparin/HS with Nitrous Acid-To affect N-unsubstituted or N-sulfated GlcN residues, heparin, cell surface HS, and heparin-derived octasaccharides were cleaved with nitrous acid at high pH (pH 4.0) or at low pH (pH 1.5), respectively (37). In both reactions, freshly prepared reagent was added to the dry oligosaccharide sample and incubated for the indicated times at room temperature. Reactions were stopped by the addition of saturated Na 2 CO 3 solution, and the cleavage products were desalted on a Sephadex G-10 column (Amersham Biosciences). To separate CyPB-specific and nonspecific octasaccharides, heparin-derived octasaccharides were incubated with CyPB in 20 mM phosphate, 400 mM NaCl buffer, and the complexes formed by the association of the protein with specific octasaccharides were purified by gel filtration on a Sephadex G-25 column. Following desalting, specifically bound octasaccharides were eluted from the complex by incubation overnight with 1 M NaCl, desalted, and dried under vacuum. In some experiments, octasaccharides were labeled with the fluorescent probe 8-aminonaphthalene-1,3,6trisulfonic acid (ANTS) at the reducing end and then treated with nitrous acid. Fluorescent derivatization was performed by reductive amination as described previously (19). Briefly, dried sample (5-50 nmol) was dissolved in 5 l of acetic acid/water (3:17, v/v) containing 0.2 M ANTS (Molecular Probes, Leiden, Netherlands) and 5 l of dimethyl sulfoxide containing 1 M sodium cyanoborohydride (Fluka, Buchs, Switzerland). The mixture was incubated at 37°C for 16 h, desalted on a 10-ml desalting column of Sephadex G-15 column (Amersham Biosciences) equilibrated in water, and thereafter dried under vacuum by using a centrifugal vacuum evaporator. In parallel experiments, octasaccharides were first cleaved by nitrous acid and the degradation products were then derivatized with ANTS. In both cases, the cleavage products were resolved by electrophoresis.
Carbohydrate Electrophoresis and Mobility Shift Assay-Heparin, cell surface HS, and products from either deaminative cleavage or heparinase-I digestion (4 g per sample) were incubated in the absence or presence of CyPB (4 g) in 40 l of electrophoresis binding buffer, containing 20 mM Tris-HCl, 400 mM NaCl, 1 mM EDTA, 1 mM dithiothreitol, pH 7.9, for 30 min at room temperature. The samples were then supplemented with 10 l of 60% (v/v) glycerol and subjected to electrophoresis in a 10% (w/v) native polyacrylamide gel in 10 mM Tris, 1 mM EDTA, pH 7.4. Electrophoresis was carried at 100 V for 2-3 h at 4°C, in an SE 600 Standard gel system (18 ϫ 16 cm) (Hoefer Scientific Instruments, San Francisco). The electrophoretic buffer used was 40 mM Tris, 40 mM acetic acid, 1 mM EDTA, pH 8.0. A mixture of bromphenol blue and phenol red was used as electrophoresis markers. The position of free and CyPB-bound oligosaccharides was then visualized by staining the gel with 0.08% (w/v) aqueous azure A (38). To check the position of CyPB, the gel was transferred onto nitrocellulose (Sartorius, Göttingen, Germany) and immunostained with specific rabbit polyclonal anti-CyPB antibodies as described (19). ANTS-labeled oligosaccharides were mixed in electrophoresis buffer, supplemented with 60% glycerol (8:2, v/v), and subjected to electrophoresis in a 20% (w/v) native polyacrylamide gel (1 g per lane, 5 l). Electrophoresis was carried at 100 V for 1-2 h at 4°C in an SE 250 Mighty Small II mini-gel system (10 ϫ 8 cm) (Hoefer). At the end of the electrophoresis, images were acquired with the gel Doc 2000 Image analysis apparatus from Bio-Rad, equipped with a 365 nm UV-transilluminator. Analysis was performed with the supplied software Quantity One.
Flow Cytofluorimetry Analysis-Phenotype of peripheral blood CD4 ϩ T lymphocytes was determined by incubating cells (5 ϫ 10 5 cells per sample) in Dulbecco's phosphate-buffered saline containing 0.5% BSA (DPBS/BSA) supplemented with the appropriate monoclonal anti-CD antibodies or the respective isotype-matched control IgG for 1 h at 4°C (28). After washing, cells were labeled for 1 h at 4°C with fluoresceinconjugated goat anti-mouse IgG (1/64). For the detection of cell surface HS, T cells were first washed rapidly in DPBS containing 0.6 M NaCl, in order to remove cationic molecules that could mask HS epitopes, and thereafter incubated with mouse monoclonal anti-HS antibody (clone MAB2040, Chemicon Int., Temecula, CA; 1/100) or irrelevant control antibodies (1/2000) for 1 h at 4°C in DPBS/BSA. After washing, fluorescein-conjugated anti-mouse IgM antibodies were added for another 1-h incubation. Cells were washed twice and immediately used for analysis. Data were monitored on a flow cytofluorimeter (FACSCalibur, BD Biosciences) and analyzed with Cellquest software. Results are expressed as variations of fluorescence mean values (⌬FMV %) (28).
Cell Binding Assays-For competitive binding assays, interaction between CyPB and cell surface HS was performed in 96-well microtiter plates (Nunc-Polylabo, Strasbourg, France), essentially as described (18). Briefly, peripheral blood CD4 ϩ T lymphocytes (10 ϫ 10 6 cells per ml in DPBS/BSA) were allowed to adhere to immobilized CyPB (1 g per well) in the presence of increasing concentrations of heparin derivatives for 1 h at 20°C. After washing with PBS, adherent cells were fixed with 3% (v/v) formaldehyde, pH 7.8, stained with 1% (w/v) methylene blue in 100 mM borate buffer, pH 8.2, and then lysed with 0.1 M HCl. The absorbance, which is proportional to the number of adhered cells, was measured at 590 nm with a microplate Bio-Rad reader model 550. Cell adhesion was estimated by using standard curves in which absorbance was related to cell numbers. Results were obtained from at least three separate experiments performed with CD4 ϩ T lymphocytes from different donors and are expressed as percentage of initially added cells that remained fixed to the adhesive substrate. Statistical significance was determined using the Student's t test, and p values Ͻ0.05 were considered as significant.
RNA Interference-Synthetic siRNA duplexes with symmetric 3Ј-deoxythymidine overhangs (Eurogentec, Seraing, Belgium) were used to carry out RNA interference. A synthetic siRNA duplex (si3-OST-3), corresponding to the 3-OST-3 mRNA sequence 5Ј-CGG ACA AGC ACU UCU ACU UTT-3Ј (nucleotides 1808 -1826 and 1295-1313 of human 3-OST-3A and 3-OST-3B sequences, respectively), was used to inhibit the expression of mRNAs encoding 3-OST-3 isoforms. The sequence was subjected to a BLAST search analysis, and no significant identity to others sequences could be detected. To check the specificity of the RNA interference, a siRNA duplex (siGFP) corresponding to GFP was used as a negative control, 5Ј-GAA CGG CAU CAA GGU GAA CTT-3Ј. Cells were transfected with 2 g of siRNA duplexes in serum-free medium using DreamFect reagent (OZ-Biosciences, Marseilles, France), according to the manufacturer's instructions. To monitor the transfection efficacy, a fluorescein-tagged siRNA duplex was used, and the transfection rate was evaluated by fluorescenceactivated cell sorter and found to beϾ85%. The cell responsiveness to CyPB was analyzed 72 h post-transfection.
Analysis of p44/42 MAPK Activation-Jurkat T cells were cultured overnight in the absence of serum and then stimulated by 50 nM CyPB at 37°C. At various times, cells (2 ϫ 10 6 per sample) were washed in cold PBS and lysed for 3 h at 4°C in 100 l of lysis buffer (20 mM phosphate buffer, pH 7.4, 350 mM NaCl, 10 mM KCl, 1 mM EDTA, 1% (v/v) Triton X-100, 20% (v/v) glycerol) supplemented with 1 mM sodium orthovanadate, 10 mM sodium fluoride, and protease inhibitor mixture (Roche Applied Science). The lysates were clarified by centrifugation at 10,000 ϫ g for 30 min at 4°C and used for immunoblotting. Proteins were separated on 10% (w/v) SDS-PAGE and transferred onto nitrocellulose membranes. The blots were blocked for 1 h at room temperature in 150 mM NaCl, 20 mM Tris-HCl, 0.1% (v/v) Tween 20, pH 7.6 (TBS-T TBS-T, Tris-buffered saline/Tween 20), supplemented with 3% (w/v) BSA. Membranes were then incubated for 2 h at room temperature with the appropriate primary antibodies in TBS-T supplemented with 1% (w/v) BSA. After washing, immunoreactive proteins were visualized with horseradish peroxidase-conjugated secondary antibodies (1-h incubation, 1/5000), by using a chemiluminescence detection kit (ECL) (Amersham Biosciences).

Kinetics of CyPB Binding to Heparin-derived Oligosaccharides-
We recently reported that an octasaccharide is the minimal structural unit that interacts with CyPB in solution (19). To extend these results, the kinetics of the interactions between CyPB and reducing-end biotinylated oligosaccharides immobilized on a streptavidin-derivatized cuvette were determined in an optical biosensor. As expected, CyPB failed to bind to a biosensor surface where hexasaccharide has been immobilized. Conversely, the binding of CyPB to an octasaccharide was typical (Fig. 1A). The binding reaction was fast because it reached a maximum extent within 2-3 min. Importantly, binding was always monophasic, and there was no evidence for secondary binding sites. Thus, the distribution of the data points around a one-site model was random and within the limits of instrument noise (Fig. 1B). Moreover, the plot of k on , which represents the observed on-rate, determined using a one-site binding model, against concentrations of CyPB was linear, indicating that binding was not limited by diffusion (Fig. 1C). The interaction of CyPB with the longer oligosaccharide was similarly monophasic (data not shown), and thus a one-site binding model was used for all data analyses ( Table 1). The association rate constant (k a ) of CyPB for octasaccharides was relatively fast. Increasing the length from dp8 to dp14 did not change significantly the value of k a . The dissociation rate constant (k d ) was measured independently using high concentration of CyPB and including heparin (100 g/ml) in the dissociation buffer to prevent re-binding (34). Overall, k d did not vary with the length of the oligosaccharides. Therefore, the dissociation constant (K d ) for the interaction between CyPB and oligosaccharides, determined from the kinetic parameters, was similar and in the range of 10 -50 nM ( Table 1). The K d of CyPB for the oligosaccharides was also calculated from the amount of CyPB binding observed near equilibrium. Overall, the values correspond to those determined from the kinetic parameters. Therefore, binding data are internally self-consistent and are likely to reflect the intrinsic values for these interactions. Moreover, these values are close to the one previously described for CyPB binding to cell surface HS (K d ϳ 10 nM) (20), indicating that the use of optical biosensor to examine the interactions between CyPB and immobilized oligosaccharides reports accurately on the binding of CyPB to heparin/HS.
Competitive Binding Assays with Chemically Modified Heparins-To examine the structural features of heparin that are required for CyPB binding, we first used a streptavidincoated cuvette for which immobilization of heparin was performed through biotinylation of internal free amino groups ( Fig. 2A, panel a). Indeed, we have routinely used this biosensor surface to examine the interactions with FGF-2 and HGF (32,33,36). As reported previously, we found that FGF-2, for which N-and 2-O sulfate groups are the main structural requirements for interaction with heparin (6,7,30), efficiently bound to this biosensor surface. In contrast, CyPB failed to interact with heparin immobilized via its internal free amino groups. Binding was indeed not distinguishable from nonspecific interaction, even at a concentration of 12.5 M equivalent to 1000 ϫ K d . Therefore, the procedure of heparin immobilization, which led to unavailability of GlcNH 2 residues and/or surrounding environment, was probably related to the inability of CyPB to interact with this biosensor surface.
The use of biotinylated albumin-heparin conjugate has been already reported to provide heparin biosensor surface with high binding capacity (39). As expected, both FGF-2 and CyPB were found to bind to the heparin-BSA conjugate immobilized onto a streptavidin-derivatized cuvette ( Fig. 2A, panel b). Surprisingly, a 3-fold lower response was observed, however, in the binding reaction with CyPB compared with FGF-2. CyPB and FGF-2 have similar molecular masses (21 and 18 kDa, respectively), and both the ligands were used at a concentration equivalent to 5ϫ K d . Moreover, we checked that biotinylation of heparin-BSA conjugate did not lead to partial derivatization of free amino groups of the heparin moiety, indicating that the conjugate allows access of ligands to the full length of heparin. Nevertheless, the overall content of unsubstituted amino groups is very low in heparin by comparison with N-and 2-Osulfated groups. Thus, the difference in the binding of CyPB and FGF-2 to immobilized heparin-BSA conjugate is likely to be due to limiting structural features in heparin that enables interaction with CyPB.
The kinetics of the interaction of CyPB with immobilized heparin-BSA conjugate was similar to the ones observed with oligosaccharides immobilized via reducing-end biotinylation. Binding was always monophasic, and no evidence was found for secondary binding sites (Fig. 2B). The dissociation constant calculated from the maximum extent of binding (K d equilibrium) was similar to the value calculated from the kinetic parameters, indicating that these binding data are internally self-consistent (Table 1). Interestingly, CyPB possessed a relatively fast k a for the heparin-BSA conjugate compared with immobilized oligosaccharides. Nevertheless, the K d values were in the same range  S.E. of the k a was derived from the deviation of the data from a one-site binding model, calculated by matrix inversion using the FastFit software provided with the instrument. Three independent sets of k on were measured, and the three resulting values for k a and their errors were combined. b The correlation coefficient of the linear regression through the k on values is represented by r. c The k d is the mean Ϯ S.E. of five values, obtained at high concentrations of CyPB, in the presence of competing heparin. d The K d (kinetic) was calculated from the ratio of k d /k a , and its S.E. is the combined S.E. of the two kinetics parameters. e The K d (equilibrium) was calculated from the extent of binding at equilibrium, and its S.E. is the combined S.E. of three independent determinations of K d . as those obtained for interactions with oligosaccharides immobilized via reducing-end biotinylation, because of a higher k d ( Table 1). The differences in the rate constants probably reflect the fact that the probability of binding events was higher for heparin compared with the one for oligosaccharides, per-haps because access to the full length of the polysaccharide may facilitate encounter with structural information that enables interaction with CyPB.
To identify these structural features, the capability of chemically modified heparins to compete for CyPB binding to immobilized heparin was then analyzed (Fig. 2C). Unmodified heparin was used as a control to ensure that soluble heparin concentrations were high enough to get a full inhibition of CyPB binding to the cuvette. The IC 50 of soluble unmodified heparin was found at around 2 g/ml, and more than 90% of CyPB binding could be competed with 25 g/ml heparin. In contrast, N-desulfated and N-desulfated/N-reacetylated heparins were far less effective competitors of CyPB binding to immobilized heparin, because high concentrations of these two chemically modified heparins only reduced CyPB binding by 20%. In the same way, 2-O-and 6-O-desulfated heparins were poor competitors. These data indicate that 6-O-, 2-O-, and N-sulfated groups are crucial for the interaction between CyPB and heparin. These results are, however, not surprising because involvement of sulfated groups for efficient binding is a common feature of numerous heparin-binding proteins. Interestingly, N-acetylated heparin was not as efficient as unmodified heparin to inhibit CyPB binding onto the cuvette. No more than 50% of inhibition could be obtained in the presence of high concentrations of this chemically modified heparin. The modification of this heparin derivative was acetylation of unsubstituted amino groups, further indicating that unsubstituted amino groups are critically required for CyPB binding to heparin. Taken together, these data suggest that CyPB does require unusual and rare modification in addition to sulfation to interact with heparin.
Involvement of GlcNH 2 in CyPB Binding to Heparin and Cell Surface HS-In the next experiments, we analyzed the requirement of free amino group of GlcNH 2 in the binding of CyPB to heparin and cell surface HS derived from peripheral blood CD4 ϩ T cells. To this end, we used an approach based on the ability of HNO 2 at pH 4.0 to specifically target GlcNH 2 residues within heparin/HS and convert them to terminal 2,5-anhydromannose residues (37). Thereafter, the binding of CyPB to chemically treated heparin or HS was visualized by electrophoretic mobility shift assay (EMSA). HNO 2 treatment of heparin at pH 4.0 produced oligosaccharides whose size decreased for the first 45 min and remained unchanged thereafter (Fig.  3A, panel a). We previously reported that the minimal length unit for CyPB binding was an octasaccharide (19). Interestingly, most of the oligosaccharides produced by HNO 2 treatment at pH 4.0 were of dp Ն 14, ruling out the possibility that a loss in CyPB binding was because of the presence of short heparinderived products. In the absence of any treatment, CyPB formed a complex with heparin, which was retained at the top of the gel. The intensity of the complex decreased over the time course of degradation by HNO 2 , suggesting that the cleavage of heparin at GlcNH 2 residues paralleled the loss of binding sites for CyPB. Indeed, 20 min of HNO 2 treatment reduced the intensity of the complex by 50%, whereas no more than 20% of the protein was retained after 45 min of treatment (Fig. 3A,  panel b). To confirm these findings, the gel was transferred onto nitrocellulose and immunostained with anti-CyPB anti- and FGF-2 (60 nM) to heparin were analyzed by using streptavidin-coated cuvettes where heparin was immobilized either via direct biotinylation of internal free amino groups of the polysaccharide (panel a) or via interaction with biotinylated BSA-heparin in which heparin was conjugated to the protein moiety through its reducing end (panel b). Nonspecific binding (n.s.) was obtained by adding CyPB to a streptavidin-coated cuvette in the absence of immobilized heparin. Data are representative of three independently performed experiments. B, binding kinetics of CyPB to biotinylated heparin-BSA conjugate was measured as described under "Experimental Procedures." CyPB at different concentrations was added to a cuvette where heparin was immobilized via interaction with biotinylated BSA-heparin conjugate, and the association reaction was followed until at least 90% of the fitted one-site binding curve was covered. The concentrations of CyPB in nM are indicated on the representative set of binding interactions of CyPB with heparin. Plot of the k on , calculated by nonlinear regression using a one-site binding model, against concentrations of CyPB, is shown. C, for competitive binding experiments, the extent of binding of CyPB (75 nM) to heparin immobilized on an aminosilane surface was measured in the presence of increasing concentra- bodies. As expected, the protein was only visualized at the top of the gel, indicating that interaction of CyPB with oligosaccharides did not lead to the formation of small complexes that could migrate within the gel (Fig. 3A, panel c). To obtain further evidence on the crucial role of GlcNH 2 residues, we then tested the interaction of CyPB with oligosaccharides obtained from digestion with heparinase I, which preferentially cleaves heparin and HS at GlcNS Ϯ 6S-IdoA2S linkages but has no activity on GlcNH 2 residues (27,40). As expected, partial digestion of heparin, which generated large oligosaccharides higher than dp14, did not significantly reduced the intensity of the complex. Conversely, long term depolymerization of heparin by heparinase I (24-h incubation) led to a dramatic loss in CyPB binding, which was because of the generation of short oligosaccharides (Fig. 3B). We then reproduced the same experiment with cell surface HS. As already described (19), the interaction between CyPB and HS derived from CD4 ϩ T lymphocytes could be visualized as a shifted band at the top of the gel. A 45-min treatment by HNO 2 at pH 4.0 reduced by 80% the intensity of the complex, indicating that cleavage of HS at GlcNH 2 residues dramatically reduced the binding of CyPB. Conversely, partial digestion of HS with heparinase I (2-h incubation) did not significantly reduce the intensity of the complex (Fig. 3C). Therefore, these data indicate that only part of the GlcNS content would be necessary for interaction, whereas GlcNH 2 residues and/or the neighboring environment represent the limiting structural features that enable specific binding of CyPB to heparin/HS.
To further demonstrate the requirement of GlcNH 2 in the binding of CyPB to cell surface HS, we analyzed the ability of chemically N-modified heparins to inhibit the binding of peripheral blood CD4 ϩ T lymphocytes to immobilized CyPB (Fig. 4). As reported previously, unmodified heparin potently inhibited the interaction between T cells and immobilized CyPB (20), with a IC 50 of 0.3 g/ml. Interestingly, a similar 50% inhibition required a 10-fold higher concentration of either N-acetylated heparin or oligosaccharides obtained from degradation of heparin by HNO 2 at pH 4.0. Taking into account that both competitors did not contain unsubstituted amino groups, these data further support a role for GlcNH 2 in the binding to CyPB. Finally, we found that N-desulfated heparin was a poor competitor, whereas N-desulfated/N-reacylated heparin was devoid of any inhibitory properties. Altogether, these results clearly demonstrate that the efficient binding of CyPB to cell surface HS of peripheral CD4 ϩ T lymphocytes requires interaction with both GlcNS and GlcNH 2 residues. The profile of migration of free and CyPB-bound oligosaccharides was then visualized by staining the gel with azure A. Panel c, gel was transferred onto nitrocellulose, and the protein was immunostained with rabbit polyclonal anti-CyPB antibodies. B, heparin was treated with heparinase I (0.75 unit/mg of heparin) for the indicated times and next subjected to EMSA (4 g per lane) in the presence of 4 g of CyPB. The profile of migration of free and CyPB-bound oligosaccharides was visualized by staining the gel with azure A. C, cell surface HS chains purified from peripheral blood T lymphocytes were either not treated (lanes 1 and 3) or treated (lanes 2 and 4) with nitrous acid, pH 4.0, for 45 min or heparinase I for 2 h. Thereafter, samples were subjected to electrophoresis (4 g per lane) in the absence (lanes 1 and 2) or the presence of 4 g of CyPB (lanes 3 and 4). The profile of migration of free and CyPB-bound oligosaccharides was visualized with azure A. Representative gels of at least three separate experiments are shown.

Analysis of the N-Substituent Pattern of the CyPB-binding
Heparin Unit-We recently used an approach based on carbohydrate electrophoresis with ANTS-labeled oligosaccharides to demonstrate that an octasaccharide is the minimal length unit required for efficient binding of CyPB to heparin and cell surface HS (19). To further confirm the involvement of a GlcNH 2 residue in the interactions with CyPB, we then evaluated the susceptibility of this minimal heparin unit to nitrous acid degradation. To this end, heparin-derived octasaccharides were first separated according to their capability to interact with the protein. Following gel filtration to remove unbound octasaccharides (fraction A), the complex formed by the association of CyPB with specific octasaccharides was incubated in the presence of 1 M NaCl, and initially bound octasaccharides were separated from the protein by gel filtration (fractions B). Then both fractions were individually treated with HNO 2 , pH 4.0, and the products of degradation were derivatized with ANTS and subjected to electrophoresis. As shown in Fig. 5A, most of the octasaccharides that did not bind CyPB were resistant to nitrous acid cleavage at pH 4.0, indicating that they did not contain any detectable amount of GlcNH 2 residues. In contrast, octasaccharides that bound CyPB were susceptible to HNO 2 cleavage at pH 4.0. Indeed, the profile of elution of treated fraction B was characterized by the presence of two degradation products, which corresponded to disaccharides and hexasacharides. Moreover, an estimation of relative amounts of each product suggested the presence of a single GlcNH 2 unit per octasaccharide. In parallel experiments, the same fractions were subjected to low pH nitrous acid treatment, which results in specific scission at N-sulfated disaccharides with a concomitant loss of the N-sulfate groups. Treatment of fraction A essentially resulted in the liberation of disaccharides, indicating that unbound octasaccharides mainly contained GlcNS resi-dues. This result was expected because the major disaccharide unit in heparin is IdoUA2S-GlcNS6S. Surprisingly, treatment of fraction B also resulted in the liberation of disaccharides. We could expect that these oligosaccharides, which contain GlcNH 2 residues, were partially resistant to attack by low pH nitrous acid and generated degradation products larger than disaccharides. Such an unexpected cleavage was however reported by Liu et al. (23,24), who demonstrated that low pH nitrous acid also targets 3-O-sulfated GlcNH 2 residues within heparin sequence. Therefore, these data indicate that the heparin-derived octasaccharide that binds CyPB is likely to contain one 3-O-sulfated GlcNH 2 and neighboring GlcNS Ϯ 6S.
We therefore used the same approach to obtain information on the position of the GlcNH 2 residue within the binding heparin unit. To this end, octasaccharides of both the fractions A and B were first labeled with ANTS at the reducing end of the molecule and then treated with HNO 2 , pH 4.0. The products of hydrolysis were subsequently separated by electrophoresis (Fig.  5B). The ANTS moiety served as a tag to differentiate oligosaccharides derived from the reducing end of the parent compounds. Treatment of fraction B resulted in the liberation of a fluorescent product that migrated as dp6 standard. No fluorescent disaccharides could be visualized, indicating that the heparin unit that interacts with CyPB contains a GlcNH 2 at position 2 from the nonreducing end of the octasaccharide. As expected, the degradation products obtained from the fraction A migrated as dp8 standard. Finally, treatment of both fractions at pH 1.5 resulted in the liberation of disaccharides, thus confirming the preceding findings. Altogether, these results indicate that CyPB does not bind to octasaccharides for which GlcNH 2 residues are lacking or not correctly positioned.

Role of 3-OST-3 in the Generation of Binding Sites for CyPB-
The different 3-OST isoforms exhibit distinct substrate specificities and thereby generate specialized HS sequences with diversified biological functions. For instance, 3-OST-3 isoenzymes, which are represented by two closely related isoforms with similar substrate specificity, i.e. 3-OST-3A and 3-OST-3B, introduce a sulfate substituent to the 3-OH position of GlcNH 2 residues within the disaccharide unit IdoUA2S-GlcNH 2 Ϯ 6S and thereby generate a 3-O-sulfated GlcNH 2 residue within the HS (23,24). To test whether the presence of GlcNH 2 residue is correlated with the ability of 3-OST to generate a binding unit for CyPB, we analyzed by RT-PCR the relative expression of 3-OST isoenzymes in responsive T cells. As shown in Fig. 6, purified CD4 ϩ T lymphocytes and Jurkat T cells predominantly expressed mRNA encoding 3-OST-3B, whereas other 3-OST isoforms were only slightly detected or absent. As control, we then analyzed the expression pattern of mRNAs encoding 3-OST isoenzymes in epithelial HeLa cells and promonocytic leukemia THP-1 cells. Indeed, we recently reported that THP-1 cells are highly responsive to CyPB, by a mechanism dependent on the interactions with cell surface HS (29). In contrast, we have observed that although HeLa cells produce high levels of HS by comparison with T lymphocytes, they are devoid of high affinity binding sites for CyPB (data not shown). Interestingly, we found that THP-1 cells mainly express mRNAs encoding 3-OST-3A, 3-OST3-3B, and 3-OST-5. Conversely, 3-OST-5 was the main isoenzyme expressed in HeLa cells, whereas mRNAs that encode 3-OST-1, 3-OST-2, 3-OST-3B, and 3-OST-6 were barely detected. Therefore, the findings that high expression of mRNA encoding 3-OST-3 isoenzymes correlates with cellular binding of CyPB supports the hypothesis that the GlcNH 2 residue present in the HS motif that interacts with CyPB could be 3-O-sulfated.
To gain insights into the relationships between 3-OST-3 expression and generation of binding sites for CyPB, we then developed a model based on the use of RNA interference. The 3-OST-3 siRNA was first tested for its ability to suppress mRNAs encoding 3-OST-3 isoforms specifically (Fig. 7A). As expected, transfection of Jurkat T cells and HeLa cells resulted in a significant 3-OST-3B mRNA down-regulation. Inhibition was found to reach 90% at day 2 in Jurkat T cells, by comparison with control cells. A significant decrease in the expression of mRNAs encoding 3-OST-3A and 3-OST-3B was also observed in THP-1 cells, indicating that the siRNA was efficient to reduce the expression of both the 3-OST-3 isoforms. Moreover, the expression of other mRNAs encoding 3-OST isoforms was not affected in Jurkat T cells, HeLa cells, and THP-1 cells, indicating that the siRNA specifically targets 3-OST-3 isoforms. Finally, no significant decrease in the expression of mRNAs encoding 3-OST-3 or other 3-OST isoforms could be detected in the presence of irrelevant siRNA, confirming the specificity of the method. We then examined the effect of 3-OST-3 mRNA down-regulation on the expression of cell surface HS by flow cytofluorimetry (Fig. 7B). As expected, transfection of Jurkat T cells with 3-OST-3 siRNA did not significantly modify the cell surface immunostaining with anti-HS antibodies, by comparison with nontransfected cells or cells transfected with irrelevant siRNA (⌬FMV ϭ 198 Ϯ 21% versus 219 Ϯ 18 and 191 Ϯ 25%, respectively). As a control, we found a 4-fold decrease in cell surface immunostaining following treatment with heparinase I (⌬FMV ϭ 56 Ϯ 13%), confirming that suppression of mRNA encoding 3-OST-3 had no detectable inhibitory effect on the expression of HS at the cell surface. We then analyzed the interaction between CyPB and cell-surface HS in transfected cells. As a control, we first checked that CyPB binding was potently inhibited (Ͼ75% of initial binding) by heparinase I treatment. Interestingly, knockdown of 3-OST-3 reduced CyPB binding by more than 60%, indicating that 3-O-sulfation of GlcN residues is required for the interactions between the protein and cell surface HS (Fig. 8A). We recently demonstrated that the main intracellular event initiated by CyPB binding to T lymphocytes is the activation of the p44/42 MAPK pathway (29). When analyzing the phosphorylated status of ERK1/2, we found that transfection of Jurkat T cells with 3-OST-3 siRNA dramatically reduced the ability of CyPB to activate p44/42 MAPK, indicating that expression of 3-OST-3B is also required for the  initiation of signaling events (Fig. 8B). Taken together, these findings suggest that efficient binding of CyPB and consequent induction of cell responses are dependent on the presence of a specific 3-O-sulfated HS motif.

DISCUSSION
The fine structure of HS varies considerably depending on the tissue of origin, developmental stage, and pathophysiological conditions, and it is thought that rare structural sequences within HS species contribute to selective protein binding. Such a diversity of structure is imposed by tightly regulated patterns of sulfation and epimerization upon the basic polysaccharide backbone during synthesis within the Golgi (1)(2)(3)(4). We have reported that HS are absolutely required for CyPB to trigger integrin-mediated adhesion of peripheral blood T cells to the extracellular matrix (11). In this study, we attempted to determine the structural features of HS that are responsible for the specific binding of CyPB to responsive cells.
To gain insights into the relationship of structural information and biological activity, we currently used heparin and its derivatives in our binding experiments, because we previously demonstrated that binding of CyPB to heparin and cell surface HS from T lymphocytes was quite similar (19,20). In a first set of experiments, we used reducing-end immobilized oligosaccharides from dp6 to dp14 to examine the kinetics of the interactions. Indeed, the observation that different methods of immobilizing heparin onto biosensor surfaces can affect the affinity of interactions (41) has made the use of pure heparin oligosaccharides immobilized via reducing-end biotinylation more appropriate to determine the binding parameters (32,36). Previous studies have demonstrated that the minimal binding unit within heparin that could interact with CyPB in solution was an octasaccharide (19). As expected, we confirmed here that CyPB failed to bind to immobilized hexasaccharides. Conversely, kinetic studies showed that oligosaccharides of dp Ն 8 all bind CyPB and that the affinity for octasaccharides is similar to that for longer oligosaccharides. We then examined the structural features of heparin that are required for CyPB binding. In previous works, we routinely used a streptavidin-coated cuvette for which immobilization of heparin or HS was performed via biotinylation of internal free amino groups (30,32,33,35,36). Although this method of immobilizing heparin was reported to affect the binding capacity of some proteins (41), we had not observed any significant changes in the binding affinities of FGF-2 and HGF by comparison with immobilization via reducing end of oligosaccharides (32,36). Unfortunately, we found here that CyPB failed to bind onto this biosensor cuvette. This observation confirms that immobilizing heparin via free intrachain amino groups can dramatically affect the binding of some particular heparin-binding proteins, probably by blocking interactions with GlcNH 2 residues and/or surrounding environment. We then used a biotinylated albumin-BSA conjugate, in which conjugation of biotin to the protein moiety allowed access of CyPB to the full length of the polymer (39). As expected, the binding affinity K d of CyPB for this conjugate was similar to those obtained with oligosaccharides immobilized via reducing-end biotinylation. Moreover, these values are close to the one previously described for CyPB binding to cell surface HS (K d ϳ 10 nM) (20), indicating that the use of this cuvette reports accurately on the interactions between CyPB and heparin/HS. Interestingly, we observed that binding of FGF-2 to immobilized heparin-BSA conjugate was higher compared with that of CyPB. This discrepancy could not be attributed to differences in the binding affinities, because we previously reported that the K d of FGF-2 for heparin or oligosaccharides was in the same range to that of CyPB (32,33). Actually, N-and 2-O sulfate groups, which are the main structural requirements for interaction with FGF-2 (6,7,30), are not limiting structural features in heparin. In contrast, the overall content of free amino groups is very low in heparin/HS (27,42). Thus, requirement of GlcNH 2 residues and/or surrounding environment for interaction is likely to represent the limiting structural information that explains the lower binding of CyPB to immobilized heparin-BSA. Competitive binding experiments performed with chemically modified heparins further support this hypothesis. We found that 2-O, 6-O, and N-sulfate groups were required for the binding of CyPB. These results were expected because involvement of sulfate groups is a common feature in the interaction of numerous proteins to heparin/HS. Most importantly, we also demonstrated that N-acetylated heparin was a poor competitor of CyPB to binding to immobilized heparin, highlighting the crucial role of free amino group of GlcNH 2 in the interactions.
To validate the findings obtained with the optical biosensor approach, we then analyzed the interaction of CyPB with heparin and cell surface HS by EMSA. Cleavage of heparin/HS with nitrous acid at pH 4.0, which targets GlcNH 2 residues, dramatically reduced the interactions with CyPB in solution. Conversely, partial digestion of heparin/HS with heparinase I did not significantly modify the interaction with CyPB in solution. Finally, we further demonstrated that N-acetylated heparin or fragments of heparin treated with nitrous acid at pH 4.0 were poor competitors of CyPB to binding to the cell membrane of T lymphocytes. Altogether, these findings demonstrate that binding of CyPB is not simply the consequence of electrostatic interactions between basic residues in the protein and negative charges on the polysaccharide, but also the result of specific contacts between the protein and a rare modification in the heparin/HS-binding unit.
The occurrence of GlcNH 2 units was first believed to reflect artificial loss of N-sulfate groups during handling of HS samples. Moreover, the mechanism of GlcNH 2 formation during HS biosynthesis remains largely unclear. Nevertheless, accumulating data have demonstrated a functional relevance of such a modification in biological and pathological phenomena. A monoclonal antibody that recognizes GlcNH 2 units in HS was reported to stain basement membranes and epithelial and mesenchymal tissues (5). The presence of GlcNH 2 residues was found to correlate with the ability of L-selectin to bind to endothelial cells (43), although the interaction was not critically dependent on GlcNH 2 residues (44). It was proposed that GlcNH 2 residues provide a cleavage site for endogenous NO-derived nitrite and thus contribute to recycling of glypican-1 (45). The overall content of GlcNH 2 is generally low and varies between HS species from 1.2 to 7.5% (w/w) of total GlcN content (27,42). GlcNH 2 residues are preferentially located in the transition zones between N-sulfated and N-acetylated region within HS (46,47). Recently, Westling and Lindahl (42) have analyzed the sequences around GlcNH 2 residues in HS from diverse preparations. They confirmed that the GlcNH 2 -containing sequences are predominantly located within the N-acetylated domains or in transition sequences between N-acetylated and N-sulfated domains. Moreover, they found that the structures upstream and downstream of GlcNH 2 residues are highly diverse. Thus, 20 -30% of the adjacent disaccharide units could be N-sulfated. Although most of the uronic acid residues immediately upstream of GlcNH 2 are GlcUA, some of these units may be subjected to C 5 -epimerization and 2-O-sulfation. GlcNH 2 residues may be also 6-Osulfated and even 3-O-sulfated, and the highly sulfated but N-unsubstituted unit IdoUA2S-GlcNH 2 Ϯ 6S would be located to the highly sulfated domain of HS chains. Interestingly, we demonstrated here that the GlcNH 2 residue within the heparin-derived octasaccharide that binds CyPB could be cleaved by low pH nitrous acid. Such an unexpected cleavage was reported by Liu et al. (23,24), who demonstrated that nitrous acid at pH 1.5 also targets 3-O-sulfated GlcNH 2 residues within heparin sequence. Therefore, these data indicate that the heparin sequence that binds CyPB is likely to contain one 3-O-sulfated GlcNH 2 . We also demonstrated that the GlcNH 2 residue required for interaction of CyPB with heparin is located at position 2 from the nonreducing end of specific octasaccharides. In our experiments, heparinase I was used to generate heparin-derived octasaccharides. As this enzyme cleaves HS essentially where GlcNS Ϯ 6S-IdoUA2S linkage occurs (27,40), these data indicate that the uronic acid residue immediately upstream to GlcNH 2 could be IdoUA2S, suggesting that the heparin sequence that binds CyPB probably contains the disaccharide unit IdoUA2S-GlcNH 2 3S Ϯ 6S. This disaccharide unit was found to be the main GlcNH 2 -containing unit in intestinal HS, although it could not be detected in HS from the aorta. Moreover, GlcNH 2 -containing disaccharide units were also identified as targets for certain 3-OST isoforms that introduce a sulfate substituent at C 3 and thereby provide a binding site for HSV-1 glycoprotein gD (20,48,49). Altogether, these data strongly support the hypothesis that biosynthetic mechanisms, whereby GlcNH 2 residues are implicated, are dependent on the tissue of origin and may provide specialized and rare structural sequences that selectively interact with distinct effector proteins.
The different 3-OST isoforms are expressed at distinct levels in various human tissues, suggesting their involvement in making tissue-specific HS with different biological functions. For instance, 3-OST1 transfers sulfate groups to the 3-OH position of the GlcNS residue that is linked to a GlcUA residue at the nonreducing end (GlcUA-GlcNS Ϯ 6S) and provides a binding site for antithrombin III (9). HS modified by 3-OST-2, -3A, -3B, -4, and -6 bind to HSV-1 gD and assist viral entry (21,22,50,51). By contrast, 3-OST-5 exhibits broad substrate specificity and transfers a sulfate group on the hydroxyl group in C 3 of GlcNS or GlcNH 2 (52). To identify whether the HS sequence that binds CyPB could be 3-O-sulfated, we then analyzed the expression pattern of 3-OST isoenzymes in peripheral blood CD4 ϩ T lymphocytes, Jurkat T cells, and THP-1 cells. We found that these cells, which are responsive to CyPB, mainly express mRNA encoding 3-OST-3B alone or in combination with 3-OST-3A. Interestingly, specific down-regulation of the expression of the mRNA encoding these isoenzymes by siRNA potently reduced binding of CyPB to Jurkat T cells and consequent activation of p44/42 MAPK pathway, indicating that 3-OST-3B is a key enzyme in the biosynthesis of the HS sequence involved in the interaction with CyPB. 3-OST-3 isoforms transfer sulfate groups to the 3-OH position of GlcNH 2 residue that is linked to an adjacent upstream IdoUA2S residue (IdoUA2S-GlcNH 2 Ϯ 6S) (23,24,48). Therefore, the unique substrate specificity of 3-OST-3 further supports the hypothesis that the binding HS unit for CyPB contains the disaccharide unit IdoUA2S-GlcNH 2 3S Ϯ 6S.
Infection by HSV-1 requires a two-step process, i.e. attachment followed by entry into cells, and it is known that HS are involved in assisting viral binding as well as viral entry. An HS motif containing IdoUA2S and GlcNS(Ac)6S residues is sufficient to make the initial contact with HSV-1 via interaction with the virion envelope glycoprotein gC or in some cases glycoprotein gB. After attachment, interaction with glycoprotein gD triggers the fusion between the viral envelope and cell membrane leading to entry of the virus. Several cellular receptors for HSV-1 gD have been described, including a member of the tumor necrosis factor receptor family, named HVEM, and two members of the immunoglobulin superfamily, designated as nectin-1 and nectin-2 (53,54). In addition, a specific 3-O-sulfated HS was reported to give rise to another family of binding sites for gD and to assist viral entry in some target cells (21,22,50,51). Interestingly, 3-OST-3 was suggested to provide unique binding sites for the glycoprotein gD via the generation of the disaccharide IdoUA2S-GlcNH 2 3S Ϯ 6S (23,24,48,49). Actually, 3-OST-3-modified HS units are rarely found in HS, suggesting that the specific HS sequence required for binding of CyPB and glycoprotein gD could be the same. In this way, our findings that mRNAs encoding 3-OST-3 isoforms were barely expressed in HeLa cells correlate with the absence of high affinity binding sites for CyPB on epithelial cells and the requirement of nectin-1 as an entry receptor for HSV-1 within these cells (49). Finally, several lines of evidence indicate that HS proteoglycans do not only stabilize the interaction of CyPB with cognate signaling receptor but also act as co-stimulatory molecules to initiate signaling events, cytoskeleton rearrangement, and integrin activation (11,12,19,29). These observations suggest that HSV-1 might hijack 3-O-sulfated HS proteoglycans recognized by CyPB and take them out of their original functions to facilitate its entry within some target cells.
In conclusion, we demonstrated here that efficient binding of CyPB to HS is dependent on the presence of a GlcNH 2 residue and the expression of 3-OST-3 in responsive cells, supporting the notion that this interaction is the result of specific contacts between CyPB and well defined modifications in HS. Growing data suggest a role for secreted cyclophilins in the development and/or progression of physiological inflammatory responses and pathological disorders (13,14,55,56). Because the interaction of cyclophilins with cell surface HS is a key step for inducing cellular signaling events, our results provide new information for further determining the regulatory functions of HS proteoglycans in assisting the activity of extracellular cyclophilins, as well as in the development of synthetic HS-derived oligosaccharides, which may serve as therapeutic agents for treating inflammatory disorders.