Heparan Sulfate Mimicry

Cell-associated heparan sulfate (HS) is endowed with the remarkable ability to bind numerous proteins. As such, it represents a unique system that integrates signaling from circulating ligands with cellular receptors. This polysaccharide is extraordinary complex, and examples that define the structure-function relationship of HS are limited. In particular, it remains difficult to understand the structures by which HS interact with proteins. Among them, interferon-γ (IFNγ), a dimeric cytokine, binds to a complex oligosaccharide motif encompassing a N-acetylated glucosamine-rich domain and two highly sulfated sequences, each of which binds to one IFNγ monomer. Based on this template, we have synthesized a set of glycoconjugate mimetics and evaluated their ability to interact with IFNγ. One of these molecules, composed of two authentic N-sulfated octasaccharides linked to each other through a 50-Å-long spacer termed 2O10, displays high affinity for the cytokine and inhibits IFNγ-HS binding with an IC50 of 35–40 nm. Interestingly, this molecule also inhibits the binding of IFNγ to its cellular receptor. Thus, in addition to its ability to delocalize the cytokine from cell surface-associated HS, this compound has direct anti-IFNγ activity. Altogether, our results represent the first synthetic HS-like molecule that targets a cytokine, strongly validating the HS structural determinants for IFNγ recognition, providing a new strategy to inhibit IFNγ in a number of diseases in which the cytokine has been identified as a target, and reinforcing the view that it is possible to create”tailor-made“sequences based on the HS template to isolate therapeutic activities.

Current research increasingly implicates heparan sulfate (HS), 2 a highly sulfated glycosaminoglycan present in the extracellular matrix and at the cell surface, in a plethora of phenomena that include cell proliferation, cell adhesion, matrix assembly, chemoattraction, inflammation, immune response, development, lipid metabolism, angiogenesis, wound healing, and viral attachment (1,2). Mechanistically, this extensive functional repertoire often relies on the ability of HS to rec-ognize diverse proteins, the conformation, stability, local concentration, or biological activities of which are modified by the interaction (3)(4)(5)(6)(7).
Consistent with its wide protein binding activity, HS is structurally complex. It is composed of strongly anionic domains enriched in N-sulfated glucosamines and iduronic acids, typically 3-8 disaccharides long (referred to as NS or heparin-like domains), that bear a variable number of O-sulfate moieties and are hypervariable in sequence. These domains are separated by relatively regular regions encompassing a larger area that contain predominantly N-acetylated glucosamine and glucuronic acid domains (NA domains) and mixed NA/NS regions that make the transition between NA and NS domains (8). It has been thought that specific information for protein recognition resided within the NS domains in HS and, indeed, a large number of "heparin-binding proteins" interact with such structures. These include, for example, fibroblast growth factors (9), stromal cell-derived factor-1 (10), herpes simplex virus type 1 glycoprotein D (11), or antithrombin III (12). However, of the large number of heparin-protein complexes that can be experimentally demonstrated, the latter (heparin-antithrombin III) is the only one to date for which a specific sequence has been formally defined (13) and reproduced by chemical synthesis (14). Importantly, this achievement led to the development of an approved drug against deep venous thrombosis.
Interferon-␥ (IFN␥), a multifunctional T cell-secreted cytokine (15,16), has been identified as a heparin-binding protein some years ago (17). IFN␥ and growth factors (such as fibroblast growth factors) belong to distinct groups in regard to their regulation by HS. In particular, HS does not promote IFN␥ association to its cell surface receptor (IFN␥R), and IFN␥R does not bind to HS. 3 IFN␥ binding to HS was found to control the blood clearance, subsequent tissue targeting, and local accumulation of the cytokine. It also regulates IFN␥ activity by a unique mechanism involving a controlled processing of the carboxyl-terminal peptide (18,19).
In contrast to the above mentioned heparin-binding proteins, IFN␥ does not interact with isolated NS domains. The binding requires a larger sequence that encompasses an internal NA domain flanked at both sides by two NS domains (20). In such a structure the two external NS regions are believed to interact with the two carboxyl-terminal sequences of an IFN␥ dimer and bridge the two IFN␥ monomers by virtue of the internal domain (Fig. 1). Because of the structural heterogeneity of HS, such complex binding motif is obviously impossible to obtain in pure form and in a large quantity from natural sources. We thus used a chemical approach that has been recently developed (21) to obtain compounds that would mimic the IFN␥ binding site with the goal of validating the model of the IFN␥-HS complex initially proposed and obtaining material in homogeneous form and a large amount. * This work has been supported by the Commissariat à l'Energie Atomique, the CNRS (Contrat Physique et Chimie du Vivant), the Université Paris-Sud, and la Ré gion Rhô ne-Alpes (Programme Emergence). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. □ S The on-line version of this article (available at www.jbc.org) contains supplemental Fig. 1  A set of compounds in which authentic NS domains, ranging from tetrasaccharides to octasaccharides, linked to poly(ethylene glycol)based spacers of different lengths that would mimic the internal NA domain was produced by chemical synthesis, and the ability of the compounds to interact with IFN␥ was analyzed. One of these molecules inhibited the IFN␥/heparin interaction with an IC 50 of 35-40 nM. Importantly, this compound also inhibited the binding of IFN␥ to IFN␥R and, consequently, the biological activity of the cytokine. These results provide a new strategy to inhibit IFN␥ in a number of diseases in which this cytokine has been identified as a target and strongly validate the model proposed for the IFN␥-HS complex.

EXPERIMENTAL PROCEDURES
IFN␥ Production and Characterization-Human IFN␥ cDNA was cloned into a pET11a expression vector (Novagen) and used to transform Escherichia coli strain BL21 Star DE3 (Invitrogen). Cells were grown at 37°C in Luria broth medium containing 100 g/ml ampicillin and induced with 0.5 mM isopropyl 1-thio-␤-D-galactopyranoside for 5 h. Purification from inclusion bodies was performed as described (22) with slight modifications. Briefly, inclusion bodies were solubilized in 6 M guanidine HCl, and the protein was refolded by dilution to 0.2 mg/ml into 50 mM phosphate buffer and 0.5 M guanidine HCl, pH 7. IFN␥ was purified by ion exchange (Mono S HR 5/5 column) and gel filtration (Superdex 75 column) chromatography (Amersham Biosciences) and stored frozen in 10 mM Tris and 10 mg/ml mannitol, pH 6.8. Purified material was characterized by mass spectrometry and amino-terminal sequencing and quantified by amino acid analysis. Biological activities of IFN␥ (antiviral activity and up-regulation of HLA-DR antigen; see below) were found to be identical to that of commercially available IFN␥ (Promega), and both samples were used in this study.
Synthesis of Oligosaccharides and Glycoconjugates-The general approach used to synthesize HS oligosaccharides and glycoconjugate mimetics is illustrated in Fig. 2 and briefly described here (a fully detailed description of each synthetic step can be found in Ref. 21). A disaccharide building block with orthogonal allyl and para-methoxybenzyl protecting groups on the anomeric and 4Ј position, respectively, was first prepared as described (23,24). The allyl group was cleaved (25) followed by the activation of the anomeric position in the form of a trichloroacetimidate derivative to obtain a disaccharide donor (88% yield). Treatment of the same building block with 2,3-dichloro-5,6-dicyano-1,4-benzoquinone gave rise to a disaccharide acceptor (83% yield). Donor and acceptor were then condensed in dichloromethane at Ϫ40°C, using tert-butyldimethylsilyl triflate as promotor, to yield the desired tetrasaccharide with total ␣ stereoselectivity in 90% isolated yield. Using the same procedure, the tetrasaccharide was converted into tetrasaccharide acceptors and donors with respective yields of 81 and 87%. Hexasaccharides and octasaccharides were then obtained, using a 2 ϩ 4 or 4 ϩ 4 oligomerization strategy with total ␣ stereoselectivity. Oligosaccharides were then deacetylated, the azido groups were reduced, the resulting amino groups were sulfated along with the free hydroxyl groups, and the methyl esters were saponified. Hydrogenolysis of the oligosaccharides benzyl groups gave the non-conjugated tetrasaccharides, hexasaccharides, and octasaccharides in 85% to quantitative yields. To obtain mimetics of the IFN␥ binding site, oligosaccharides were conjugated, before hydrogenolysis, with bis-thio-poly(ethylene glycol) under UV light irradiation (360 nm) as described (26). The benzyl protecting groups were then removed by hydrogenolysis after oxidation of the thioether linkages using pH 7 buffered oxone (27), giving rise to the glycoconjugates 2T n , 2H n , and 2O n (where T, H, and O stand for tetrasaccharide, hexasaccharide, and octasaccharide, respectively, and 2 indicates that two oligosaccharides are linked through a spacer having n equal to 5, 10, or 32 ethylene glycol repeats). All samples were sterilized by filtration through 0.22-m filters, quantified as described (28), and analyzed by polyacrylamide gel electrophoresis analysis (29), 1 H NMR, and, for 2T 5 , 2H 32 , and 2O 10 , heteronuclear single quantum coherence experiments. Size-defined, heparin-derived oligosaccharides were also obtained by enzymatic depolymerization of pig intestinal mucosa heparin as described (10).
Biotinylation of Heparin and Biacore-based Binding Assay-Heparin, 1 mM in phosphate-buffered saline, was reacted for 24 h at room temperature with 10 mM biotin-LC-hydrazide. The mixture was extensively dialyzed against water to remove unreacted biotin and freeze-dried (10). Two flow cells of CM3 or CM4 sensor chips were activated with 50 l of 0.2 M N-ethyl-NЈ-(diethylaminopropyl)-carbodiimide and 0.05 M N-hydroxysuccinimide, after which 50 l of streptavidin (Sigma-Aldrich) at 0.2 mg/ml in 10 mM sodium acetate buffer, pH 4.2, was injected. Remaining activated groups were blocked with 50 l of ethanolamine (1 M), pH 8.5. Typically, this procedure allowed coupling 2000 -2500 resonance units (RU) of streptavidin on both flow cells. Biotinylated heparin (5 g/ml) was captured to a level of 50 RU on one surface, and the other one was left untreated and served as negative control. Before use, the chip was submitted to several injections of Hepes-buffered saline containing 1.25 M NaCl and then washed by continuous flow of HBS-P buffer (10 mM Hepes, 0.15 M NaCl, and 0.05% P20 detergent, pH 7.4). For binding assays, IFN␥, either alone or in combination with oligosaccharides, was simultaneously injected over both negative control and heparin surfaces for 5 min at 25°C and 50 l/min. The heparin surface was regenerated with a 250 l pulse of 1.25 M NaCl.
Biotinylation of Soluble IFN␥ Receptor and Biacore Base Binding Assay-Soluble IFN␥R (R&D Systems) at 250 g/ml in 20 mM phosphate buffer, pH 6, was reacted for 20 min in the dark at 4°C, with 10 mM sodium periodate to oxidize glycans of the molecule. The reaction was quenched with 15 mM glycerol, and the sample was dialyzed against the 20 mM phosphate buffer. Biotin-LC-hydrazide was then added to a concentration of 5 mM, and the mixture was incubated for 4 h at 4°C. Ethanolamine (100 mM final) was then added to the sample, which was then extensively dialyzed against phosphate-buffered saline, pH 7.2. Biotinylated soluble IFN␥R (40 g/ml) was captured on top of a streptavidin sensor chip (prepared as described above) to a level of 150 -200 RU. Before use, the surface was conditioned with 10 2-min pulses of 10 mM HCl. For binding assay, samples (IFN␥ either alone or in combination with oligosaccharides) diluted in HBS-P buffer maintained at 25°C were injected over the IFN␥R surface at a flow rate of 50 l/min for 5 min. The IFN␥R surface was regenerated with 10 mM HCl.
IFN␥ Bioassays-Antiviral activity was determined in triplicate with a standard microtiter inhibition of cytopathic effect assay against the vesicular stomatitis virus on monolayers of WISH cells essentially as described (30). An IFN␥-induced HLA-DR antigen was measured with an enzyme-linked bio-immunoassay (31). For that purpose, Colo 205 cells seeded in a 96-well microtiter plate were treated with IFN␥ during a 4-h period. After 72 h of additional culture, cells were fixed with ice-cold ethanol, and HLA-DR antigen was detected with an anti-HLA-DR monoclonal antibody (1 g/ml; BD Biosciences) followed by a goat anti-mouse IgG-horseradish peroxidase conjugate (1:2500; Amersham Biosciences), both in Hanks' buffer sodium salt containing 0.02% Tween 20 and 1% bovine serum albumin.

RESULTS
Design, Synthesis, and Characterization of the Glycoconjugates-Our previous studies (20) suggested that in the IFN␥-HS complex two NS NOVEMBER 11, 2005 • VOLUME 280 • NUMBER 45

JOURNAL OF BIOLOGICAL CHEMISTRY 37559
domains directly interact with two IFN␥ carboxyl termini and bridge the two cytokine monomers through an internal NA domain (Fig. 1). Although the atomic structure of IFN␥ has been solved by x-ray crystallography, the organization of the carboxyl termini, which are believed to be intrinsically disordered, is unknown downstream of the residue Ala-123 (32). Because the exact distance that spans the two heparin binding domains is thus not defined, we prepared synthetic glycoconjugates in which poly(ethylene glycol)-based spacers of different lengths were used to link chemically obtained NS domains. For that purpose, a disaccharide building block was first prepared and converted into either a disaccharide donor or a disaccharide acceptor, which were condensed together (2 ϩ 2) to give a tetrasaccharide (T). A similar strategy (2 ϩ 4 or 4 ϩ 4) was used to prepare hexasaccharides and octasaccharides ( Fig.  2A). The glycoconjugates 2T 5 , 2T 10 , 2T 32 , 2H 5 , 2H 10 , 2H 32 , 2O 5 , 2O 10 , and 2O 32 were obtained through conjugation of the oligosaccharides onto ␣,-bis-thio-poly(ethylene glycol) spacers of different lengths (Fig.  2B). 1 H NMR and heteronuclear single quantum coherence spectra were in accordance with the expected structures (data not shown). PAGE analysis (Fig. 3) showed that synthetic tetrasaccharides, hexasaccharides, and octasaccharides (lanes 2, 4, and 6) have the same migration pattern as the corresponding heparin-derived oligosaccharides (lanes 1, 3, and 5) but, in contrast to the natural size-purified mixtures, are homogenous. Conjugation to poly(ethylene glycol) linkers was clearly evidenced by the shift in migration observed for 2O 5 (Fig. 3, lane  7), 2O 10 (lane 8), and 2O 32 (lane 9), each of which migrated at the position of natural octadecasaccharides and larger. Similar migration changes were observed for tetrasaccharide and hexasaccharide conjugation (data not shown). In some case, conjugated oligosaccharides migrated as apparent doublets on the gel. However, because NMR analyses of these molecules were consistent with single structures, this result should not be caused by sample heterogeneity (see supplemental data, available in the on-line version of this article).
Inhibition of the IFN␥-Heparin Interaction by the Glycoconjugates-To investigate the ability of the 12 different synthetic compounds to interact with IFN␥, an inhibition assay was set up in which the cytokine, either alone or coincubated with each of the 12 molecules (at 37.5, 75, or 150 nM) to be analyzed, was injected over both a heparin-functionalized sensor chip and a streptavidin sensor chip used as a control surface. Injection of IFN␥ over the heparin surface produced at equilibrium a binding response of 180 RU, whereas a response of 5 RU was observed over the streptavidin surface (data not shown). Results showed (Fig. 4) that whereas the tetrasaccharides and hexasaccharides were completely inactive, the octasaccharide slightly prevented the IFN␥-heparin binding (no more than 20% inhibition was observed with the highest dose of octasaccharide), indicating that by themselves these oligosaccharides did not display significant affinity for the cytokine. Conjugation of tetrasaccharides or hexasaccharides to the different poly(ethylene glycol) linkers did not lead to significant inhibition activity (25-30% of inhibition at 150 nM; Fig. 4C). In contrast, octasaccharides conjugated to spacers with 5, 10, or 32 poly(ethylene glycol) repeats (2O 5 , 2O 10 , or 2O 32 ) clearly inhibited the binding of IFN␥ to heparin, the most active molecule (2O 10 ) displaying a 50% inhibition at ϳ35-40 nM (Fig. 4A). These results strongly suggest that the glycoconjugates function by bridging the two IFN␥ monomers and that proper spacing of the oligosaccharides are important for the binding process. Similar experiments FIGURE 1. A model of the IFN␥-HS complex. IFN␥ is a homodimer that contains four clusters of basic residues (boldfaced), two of which (Lys-Thr-Gly-Lys-Arg-Lys-Arg and Arg-Gly-Arg-Arg) are located in the unstructured carboxyl-terminal end of the protein, that function as HS binding sites. The motif that IFN␥ recognizes along the HS chain consists of two NS domains that directly interact with the two carboxyl termini of the cytokine and are linked to each other by an internal NA domain. Among the nine glycoconjugates that have been synthesized to mimic such structure, the most active compound (n ϭ 10) is shown.
performed with 15-and 5-kDa heparin indicated that 20 and 130 nM were required, respectively, to produce 50% inhibition (data not shown).
The Glycoconjugates Inhibit the IFN␥-IFN␥R Interaction-A number of studies demonstrated that the IFN␥ carboxyl termini are critical for bioactivity (33). In particular, it has been suggested that residues 128 -132 (Lys-Arg-Lys-Arg-Ser) within this domain of the cytokine are required for receptor binding, and a model has been proposed where this basic cluster is located near an acidic patch of the IFN␥R (32). We thus investigated whether the synthetic oligosaccharides would inhibit the binding of the cytokine to its receptor. For that purpose, we immo-bilized the ectodomain of the IFN␥R on a Biacore sensor chip and analyzed the ability of the synthetic oligosaccharides to inhibit the IFN␥-IFN␥R interaction. IFN␥ was preincubated with increasing concentrations of the different molecules to be analyzed and then injected over the IFN␥R surface (Fig. 5). The first two sets of sensorgrams show that unconjugated oligosaccharides (Fig. 5A) and conjugated tetrasaccharides (Fig. 5B) had no or minimal effect on the IFN␥-IFN␥R interaction. Conjugated hexasaccharides (2H 5 , 2H 10 , and 2H 32 ) yielded a modest inhibition (Fig. 5C) as did the conjugated octasaccharides 2O 5 and 2O 32 (Fig. 5, D and E), whereas the 2O 10 had a marked effect on the binding process (Fig. 5F). The 2O 10 compound activity is similar to that of a 6-kDa heparin fragment.
The 2O 10 Glycoconjugate Displays Anti-IFN␥ Activity-In view of the above data, we investigated whether 2O 10 would inhibit IFN␥ activity. We thus measured the ability of 2O 10 to prevent the induction of the HLA-DR antigen by IFN␥, using Colo 205 cells. We found that optimum HLA-DR antigen was observed 72 h after the initial stimulation with 5 ng/ml of IFN␥, and these conditions were used in the following experiments. Cells were treated with IFN␥ that was preincubated with a range of concentration of 2O 10 (0 -25 g/ml), and HLA-DR expression was quantified 72 h later. Our data (Fig. 6A) showed that 2O 10 clearly displayed anti-IFN␥ activity, with an IC 50 close to 1.5 g/ml (250 nM). Analysis of all the other glycoconjugates demonstrated that their anti-FIGURE 2. Oligosaccharides and glycoconjugates synthesis. A, synthesis of the protected tetrasaccharides, hexasaccharides, and octasaccharides. A single disaccharide building block was converted into a disaccharide acceptor (arrow a) or a donor (arrow b) and then condensed together (2 ϩ 2) to give a tetrasaccharide (arrow c). This tetrasaccharide was similarly transformed into either a tetrasaccharide donor (arrow a) or acceptor (arrow b) that were then converted, using a 2 ϩ 4 or 4 ϩ 4 strategy (arrow c), into the expected hexasaccharide and octasaccharide. These compounds were treated as described under "Experimental Procedures" to give the non-conjugated tetrasaccharides, hexasaccharides, and octasaccharides. B, general strategy for the preparation of the glycoconjugates. The sulfated and benzylated tetrasaccharides, hexasaccharides, or octasaccharides were conjugated to bis-thio-poly(ethylene glycol) linkers of different lengths (n ϭ 5, 10, or 32) under UV light irradiation (arrow d). The nine benzylated glycoconjugates thus obtained were debenzylated after oxidation of the thioether linkages to give the awaited nine deprotected glycoconjugates (arrow e). IFN␥ activity closely match their ability to inhibit the IFN␥-IFN␥R interaction (Fig. 6B).

DISCUSSION
Structural heterogeneity of HS is the basis of the many functions these molecules fulfill, in particular through their unique ability to interact with a large array of proteins (34). Despite increased interest in the field, progress has been hampered by the extraordinary complexity of HS. In particular, the characterization of the protein-HS interface and the isolation of the corresponding binding domains are very difficult, an aspect particularly prominent in the case of IFN␥, the binding site of which is not simply contained within a single structural domain of the polysaccharide (20).
In this context, we chose a chemical strategy to investigate the general motif organization that IFN␥ recognizes on HS. According to the model of the IFN␥-HS complex (see Fig. 1) that was used as a working hypothesis, different heparin-like oligosaccharides were synthesized and linked to each other using molecular spacers of distinct lengths. Binding studies first pointed out that properly spaced interacting sequences acted in a concerted manner to form a functional unit, because none of the unconjugated oligosaccharides displayed efficient binding activity. Optimal activity was observed when a spacer of 10 poly(ethylene glycol) repeats (ϳ50-Å-long) was introduced between the two NS domains. Interestingly, the last two defined residues (Ala-123) in the IFN␥ crystallographic structure are at a distance of 23 Å. The 50-Å spacer should thus function by optimally presenting the two binding NS domains to the IFN␥ basic clusters (see Fig. 1), in contrast to the 5 and 32 poly(ethylene glycol) repeats that are 33-and 114-Å-long, respectively, and presumably might impose accommodations to meet an appropriate binding conformation. Our results also clearly demonstrated the importance of the oligosaccharide size for the binding, because the glycoconjugates containing tetra-NS or hexa-NS domains, although linked with the optimal spacer, were unable to significantly interact with IFN␥. The 2O 10 molecule (two octasaccharides linked with the 10 poly(ethylene glycol) repeat spacer) thus appears to be the best mimic of the IFN␥ binding domain. Because its central domain is not HS derived, the 2O 10 molecule is much less charged than the natural equivalent molecule. However, its binding activity and its ability to inhibit IFN␥ are identical to that of heparin, demonstrating that it represents a fully functional binding unit. The C 2 symmetry of this compound may also contribute to its binding activity by fitting the symmetry of the IFN␥ dimer better than a natural single HS chain.
The achievement of such a molecule has considerable interests for different purposes. First, in contrast to natural derived compounds, this molecule is homogeneous and large amounts of it can be prepared. A possibility that arises from this quality is the development of structural studies of the cytokine in complex with the glycoconjugate, a point that Complexes were injected at 50 l/min over a heparin-activated sensor chip, and the binding response was recorded at the end of the injection phase. The response obtained in the absence of oligosaccharides (100%) was equal to 180 RU. Each experiment was repeated 2 or 3 times using two batches of independently synthesized molecules. The S.E. never exceeded 10% of the mean value. was not possible before. Like many disordered segments, the carboxylterminal peptide of IFN␥ may fold on binding to its target (35) and thus might be characterized by x-ray crystallography. Because this part of the protein is a critical regulatory element of the cytokine activity (33), resolving its structure is an important issue. Secondly, the 2O 10 molecule is clearly a prototype with considerable potential for refinement. In particular, we have not yet addressed the importance of the sulfation profile within the NS octasaccharides. Combinatorial synthesis is a possible tool to generate distinct sulfo forms and/or an iduronic/glucuronic pattern (23) and should give rise to molecules with better defined and specific oligosaccharide sequences, a point that can now be investigated in the context of the 2O 10 scaffold.
Another important aspect of the present work is the observation that, once bound to the 2O 10 molecule, IFN␥ did not interact anymore with its cell surface receptor and, consequently, is biologically inactive. The importance of the IFN␥ carboxyl-terminal domain for receptor binding has been investigated by carboxyl-terminal deletions and site-directed mutagenesis (33). These studies all point out that preservation of the Lys-Arg-Lys-Arg-Ser sequence is critical for activity. Our data are thus consistent with the view that 2O 10 directly interacts with this sequence and blocks binding to both HS and IFN␥R. The fact that a carboxylterminal region of the cytokine can bind to two such unrelated molecules as 2O 10 and IFN␥R could be somewhat surprising. However, this part of the molecule is disordered, a feature that has been recently recognized as frequently involved in regulatory functions and that confers functional advantages to the protein, including the ability to bind in different conformations to different targets (35).
IFN␥ is a key cytokine involved in the development and maintenance of the Th1 arm of the immune system (16). However, aside from normal functions in host defense, IFN␥ production may also contribute to a number of diseases characterized by pro-inflammatory and autoaggressive Th1 response (36). These include autoimmune pathologies such as rheumatoid arthritis, systemic lupus erythematosus, multiple sclerosis, insulin-dependent diabetes mellitus, psoriasis, and alopecia areata (37) or chronic inflammatory bowel diseases such as Crohn disease and ulcerative colitis (38). IFN␥ is also a key contributor to hyperoxia-induced lung injury (39) and, because of its ability to induce a major histocompatibility complex in many tissues, is associated with graft rejection (40). This cytokine is thus considered as an attractive target for the treatment of such diseases, and strategies have been developed to antagonize its biological activity. So far these strategies have focused on the use of soluble IFN␥R (41,42) or neutralizing antibodies, and in some cases these were found to offer substantial clinical benefits (37,43). Humanized anti-IFN␥ is also currently evaluated in Crohn disease (44). On the other side, administration of a HS binding cationic peptide (derived from the IFN␥ carboxyl terminus) has delayed the time of rejection in a mouse model of allogenic skin flap transplantation (45). Most interestingly, the 2O 10 molecule described here inhibits the binding of IFN␥ to both HS and IFN␥R, the two known ligands of the cytokine. Such a molecule should thus inhibit both bioactivity and local concentration within tissue and therefore should compare very favorably with neutralizing antibodies that have been selected for their ability to inhibit IFN␥ bioactivity but not binding to HS.
There is enormous potential for the development of heparin-like structures as drugs for a range of diseases in addition to the current antithrombotic target (46,47). Based on the 2O 10 scaffold, our future work will investigate whether such type of molecules could be of some interest in pathologies for which IFN␥ has been identified as a target.