Probing the Acceptor Active Site Organization of the Human Recombinant β1,4-Galactosyltransferase 7 and Design of Xyloside-based Inhibitors

Background: Glycosyltransferase inhibitors have important applications in therapeutics and as chemical biology tools. Results: The human β1,4-galactosyltransferase 7 enzyme active site was mapped by modeling, mutagenesis, and in vitro/in cellulo assays, and novel inhibitors were synthesized. Conclusion: An efficient inhibitor of β1,4-galactosyltransferase 7 and glycosaminoglycan synthesis was obtained. Significance: This inhibitory molecule can be exploited to investigate glycosaminoglycan biology and modulate glycosaminoglycan synthesis in therapeutics.

Glycosaminoglycans (GAGs) 2 are linear heteropolysaccharide chains covalently attached to the core protein of a variety of proteoglycans (PGs). Because of their high structural diversity and their anionic characteristics, GAGs interact with a network of cellular and extracellular mediators including cytokines and chemokines, enzymes and enzyme inhibitors, matrix proteins, and membrane receptors (1). There is currently great emphasis on the crucial roles of GAGs in numerous physiological events including cell differentiation, proliferation and migration (2), and its pathological aspects, such as tumor formation, progression, and metastasis (3). Furthermore, because PGs are ubiquitously expressed in extracellular matrices and on cell surfaces of virtually every tissue, they are also involved in the normal and pathological functions of the cardiovascular and osteoarticular system (4), in amyloid disorders (5) and in axonal de-and regeneration (6). GAG biosynthesis is initiated by the formation of a tetrasaccharide linkage region (GlcA␤1-3Gal␤1-3Gal␤1-4Xyl␤1-O-) covalently linked to serine residues of the PG core protein (7). This tetrasaccharide acts as a primer for the elongation of major GAG chains, i.e. chondroitin/dermatan sulfate or heparin/heparan sulfate, which polymerization involves the coordinated activities of chondroitin-sulfate synthases and heparan-sulfate synthases (exostosins, EXT), respectively (8,9). Mature GAG chains are finally produced by the modifications of their constitutive disaccharide units catalyzed by epimerases and sulfotransferases, which considerably increase their structural and functional diversity (10,11).
The human xylosylprotein ␤1,4-galactosyltransferase (EC 2.4.1.1337, h␤4GalT7) catalyzes the transfer of the first Gal res-idue of the tetrasaccharide linkage from the activated sugar UDP-galactose (UDP-Gal) onto Xyl residues attached to the PG core protein (12). Because all GAGs share the same stem core tetrasaccharide, ␤4GalT7 is a central enzyme in GAG biosynthesis. Indeed, h␤4GalT7 mutations have been associated with a rare genetic condition, the progeroid form of Ehlers-Danlos syndrome (EDS), a group of connective tissue disorders characterized by a major deficiency in PG synthesis. As a consequence of GAG defect, EDS patients exhibit motor development delay, and musculoskeletal malformations, hypermobile joints, and wound healing defaults (13). Patients gene sequencing revealed the presence of missense mutations leading to L206P, A186D (14,15), and R270C substitutions (16) in the catalytic domain, resulting in a partially or totally inactive enzyme. Recently, we showed that R270C replacement reduced affinity toward the xyloside acceptor and strongly affected GAG chains formation in ␤4GalT7-deficient CHOpgsB-618 cells (17). There is currently no effective therapy for treating EDS patients.
Interestingly, the biosynthesis of GAGs can be manipulated by simple xylosides carrying a hydrophobic aglycone, which act as substrates and/or inhibitors of h␤4GalT7. Xyloside analogs have been shown to efficiently induce GAG synthesis bypassing the natural Xyl-substituted core protein of PGs for several decades (18,19). The xyloside-primed GAG chains are usually excreted and show interesting biological functions such as activation of fibroblast growth factor (FGF) signaling (20,21), antithrombotic (22), tissue regenerating (23), anti-angiogenic (24) and anti-proliferative properties (25,26). In addition, several groups have synthesized a series of xyloside analogs as potential inhibitors of GAG synthesis. Such compounds would represent highly valuable chemical biology tools to probe the functions of GAGs in cell systems and model organisms and as a starting point toward the development of pharmaceuticals, in particular anti-tumor agents. Recently, Garud et al. (27) and Tsuzuki et al. (28) used click chemistry to generate libraries of 4-deoxy-4fluorotriazole analogs comprising a set of hydrophobic molecules appended to the anomeric carbon of the xyloside. Siegbahn et al. (29,30) developed a collection of naphthyl and benzyl xylosides substituted on different positions of the Xyl moiety. These studies led to the discovery of promising xyloside-derived inhibitors of GAG synthesis when screened in cell models.
However, until recently, the development of substrates and inhibitors of ␤4GalT7 has been mostly limited to the synthesis of libraries of analog compounds and their testing in cell assays. Toward the rational design of h␤4GalT7 inhibitors, we have been involved in structure-activity relationship studies of the recombinant human enzyme for several years and identified critical active site amino acids implicated in catalysis and/or substrate binding (17,31,32). We previously investigated the importance of conserved 163 DVD 165 and 221 FWGWRGE-DDE 230 motifs in the organization of the catalytic domain. Our data have highlighted the crucial role of Trp 224 in substrate recognition and suggested a catalytic role for Asp 228 (31). These findings were in accordance with the structural data from the recently solved crystal structure of the catalytic domain of Dro-sophila melanogaster d␤4GalT7 (33) and the human enzyme (34).
In the current study, we developed a structure-guided approach for the design of xyloside inhibitors of h␤4GalT7 that were tested on its galactosyltransferase activity in vitro and on GAG biosynthesis in cell assays. We explored the organization of the acceptor binding pocket, specifically probing the functional and structural contribution of a set of residues located in the vicinity of the catalytic center, and highlighted the crucial role of three tyrosine residues, i.e. Tyr 194 , Tyr 196 , and Tyr 199 , in the architecture of the acceptor substrate binding site and the creation of a hydrophobic environment. Based on these and previous findings, we synthesized compounds that incorporate critical structural elements both on the xylopyranoside and on the aglycone moieties to tightly bind the acceptor site of h␤4GalT7. This work revealed that the 4-deoxy-4-fluoro-Xyl linked to 4-methylumbelliferone (4-MU) strongly inhibited h␤4GalT7 activity in vitro and efficiently impaired GAG synthesis in a cell context. Such a compound will be a valuable tool for the exploration of GAG and PG synthesis and opens avenues toward the development of bioactive oligosaccharide structures for GAG biosynthesis regulation in a number of diseases implicating disorders of GAG synthesis.

EXPERIMENTAL PROCEDURES
Chemicals and Reagents-4-Methylumbelliferyl-␤-D-xylopyranoside (4-MUX), UDP-␣-D-Gal (UDP-Gal), and anti-goat IgG (whole molecule) peroxidase-conjugated antibody were provided from Sigma. Anti-Myc antibodies were from Invitrogen and anti-mouse IgG-peroxidase antibodies were purchased from Cell Signaling, whereas anti-decorin antibodies were from R&D Systems. Na 2 [ 35 S]SO 4 was from PerkinElmer Life Sciences. Cell culture medium was purchased from Invitrogen and restriction enzymes, T4 DNA ligase, and peptide N-glycosidase F from New England Biolabs. The eukaryotic expression vector pcDNA3.1(ϩ) and competent One Shot Top 10 Escherichia coli cells were provided by Invitrogen and the bacterial expression vector pET-41a(ϩ) and E. coli BL21(DE3) cells were from Novagen-EMD Chemicals. The QuikChange site-directed mutagenesis kit was from Stratagene and the transfection agent ExGen 500 from Euromedex.
Molecular Modeling of the h␤4GalT7 Active Site in the Presence of 4-MUX and UDP-Gal-The crystal structure of h␤4GalT7 bound to UDP and to the manganese ion (PDB code 4IRQ) was used as template (34). The crystal structure of d␤4GalT7 (PDB code 4M4K), an inactive mutant (D211N) of d␤4GalT7 in complex with UDP-Gal, and xylobiose was superposed to the human enzyme structure, which was straightfor-ward considering their strong sequence similarity (58% overall identity). Due to crystallization conditions, a Tris molecule is bound within the active site of the h␤4GalT7. When retrieved, it frees space within the cavity that can thus accommodate the Gal moiety. The coordinates of the Gal molecule from the d␤4GalT7 complex were merged to the UDP moiety of h␤4GalT7. This did not generate any steric clash within the active site. The resulting complex was then prepared using the Protein Preparation Wizard tool of the Schrödinger Suite (Schrödinger LLC, New York), with default settings (36). All water molecules were retrieved, except the one that coordinates the manganese ion. The hydrogen atoms were added to the protein and the ligand, ascribing a pH of 7.0. The histidine residues were treated as neutral. The selection of histidine enantiomers and the orientation of the asparagine and glutamine side chains were performed so as to maximize the hydrogen bond network. The partial atomic charges derived from the OPLS-2005 force field (37) were assigned to all ligand and protein atoms. Finally, an all-atom energy minimization with a 0.3 Å heavy-atom root mean square deviation criteria for termination was performed using the Impref module of Impact and OPLS-2005 (38). The 4-MUX ligand was prepared using the ligprep module (Schrödinger Release 2014 -22014). The docking program Glide was used in Standard Precision mode, with OPLS-2005, to run rigid-receptor docking calculations (39,40). The shape and physicochemical properties of the binding site were mapped onto a cubic grid with dimensions of 20 Å 3 centered on the xylobiose. During the docking calculations, the parameters for van der Waals radii were scaled by 0.80 for receptor atoms with partial charges less than 0.15e. Ring conformational sampling was not allowed to maintain the 4 C 1 conformation of the Xyl ring, and no constraint was introduced. A maximum of 100 poses were retained and ranked according to the GlideScore scoring function. The best-docked pose of the 4-MUX ligand showed a root mean square deviation on the Xyl ring heavy atoms of 0.5 Å with the crystallographic xylobiose ligand, thus validating the docking protocol able to recover the position of this moiety.
Expression Vector Construction-The h␤4GalT7 sequence (GenBank nucleotide sequence accession number NM_007255) was cloned by PCR amplification from a placenta cDNA library (Clontech), as previously described (41). For bacterial expression, a truncated form of h␤4GalT7 was expressed as a fusion protein with glutathione S-transferase (GST). The sequence lacking the codons of the first 60 N-terminal amino acids was amplified from the full-length cDNA and subcloned into NcoI and NotI sites of pET-41a(ϩ) to produce plasmid pET-␤4GalT7 (31). For the heterologous expression of h␤4GalT7 in mammalian cell lines, the full-length cDNA sequence was modified by PCR at the 5Ј end to include a KpnI site and a Kozak consensus sequence, and at the 3Ј end to include a sequence encoding a Myc tag and an XbaI site to be then subcloned into the KpnI-XbaI sites of the eukaryotic expression vector pcDNA3.1(ϩ) to produce pcDNA-␤4GalT7 as previously described (31). Mutations were constructed using the QuikChange site-directed mutagenesis kit, employing pcDNA-␤4GalT7 or pET-␤4GalT7 as template. Mutants were systematically checked by double strand sequencing. The human decorin cDNA sequence (GenBank accession number NM_001920.3) was cloned by PCR amplification from a placenta cDNA library (Clontech). For heterologous expression in eukaryotic cells, the full-length cDNA sequence was modified by PCR to include an AflII site, a Kozak consensus sequence at the 5Ј end, a sequence encoding a His 5 tag, and an XhoI site at the 3Ј end. This sequence was subcloned into pcDNA3.1(ϩ) to produce pcDNA-decorinHis as previously described (31).
Expression and Purification of the Soluble Form of h␤4GalT7-A single colony of E. coli BL21(DE3) cells transformed with the pET-␤4GalT7 plasmid was cultured overnight at 37°C in Luria broth (LB) medium containing 50 g/ml of kanamycin. The overnight culture was transferred into fresh LB medium (1:100 dilution), supplemented with 50 g/ml of kanamycin, and incubated at 37°C until the A 600 value reached 0.6 -0.8. Expression of h␤4GalT7 was induced by addition of 1 mM isopropyl ␤-D-thiogalactopyranoside to the cell suspension, which was then incubated overnight at 20°C under continuous shaking (200 rpm). The bacterial cells were then harvested by centrifugation at 7,000 ϫ g for 10 min at 4°C. The pellet was resuspended in Lysis buffer (50 mM sodium phosphate, 1 mM phenymethylsulfonyl fluoride, 1 mM EDTA, and 5% (v/v) glycerol, pH 7.4) supplemented with protease inhibitor mixture tablets (1 tablet/12 ml; Roche Diagnostics) and Benzonase Nuclease (250 units/10 ml, Sigma). The suspended cells were then sonicated for 8 cycles of 30 s, at 30% power (Badelin Sonoplus GM70) with a 20-s interval on ice between each cycle. Soluble proteins were collected from the supernatant after centrifugation for 25 min at 12,000 ϫ g and clarification by filtration (0.2 M Supor Membrane; PALL-Life Science). 10 ml of clarified extracts were applied onto a 1-ml glutathione-Sepharose High Performance column (GSTrap HP; GE Healthcare) connected to an AKTA prime plus instrument (GE Healthcare). Protein was eluted as 1-ml fractions using 50 mM Tris-HCl, pH 8.0, containing 10 mM reduced glutathione buffer. Protein purity of the eluted fractions was evaluated by 12% (w/v) SDS-PAGE analysis, followed by staining with Coomassie Brilliant Blue. Fractions containing the pure protein were used to determine the kinetic parameters of the enzyme. The same procedure was used for purification of the mutants. Protein concentration was measured using Quant-iT TM assay kit and Qubit TM spectrofluorimeter.
Determination of the in Vitro Kinetic Parameters of h␤4GalT7-The kinetic parameters k cat and K m toward 4-MUX and UDP-Gal were determined as described (31). Briefly, 0.2 g of purified wild-type or mutated GST-h␤4GalT7 were incubated for 30 min at 37°C in a 100 mM sodium cacodylate buffer, pH 7.0, 10 mM MnCl 2 , with concentrations from 0 to 5 mM 4-MUX in the presence of fixed 1 mM UDP-Gal to determine the apparent K m toward 4-MUX, and with concentrations from 0 to 5 mM UDP-Gal in the presence of fixed 5 mM 4-MUX to determine the apparent K m toward UDP-Gal. The incubation mixture was then centrifuged at 10,000 ϫ g for 10 min at 4°C. The supernatant was analyzed by high performance liquid chromatography (HPLC) with a reverse phase C18 column (xBridge, 4.6 ϫ 150 mm, 5 m, Waters) using Waters equipment (Alliance Waters e2695) coupled to a UV detector (Shimadzu SPD-10A). Kinetic parameters were determined by nonlinear least squares regression analysis of the data fitted to the Michaelis-Menten rate equation using the curve-fitter program of Sigmaplot 9.0 (Erkraft, Germany).
In Vitro Competition Assays of h␤4GalT7 Activity by C4-modified Xylosides-The in vitro inhibition assays of the wild-type GST-h␤4GalT7 were carried out using 0.2 g of purified protein incubated for 30 min at 37°C in a 100 mM sodium cacodylate buffer, pH 7.0, 10 mM MnCl 2 , with 0.5 mM 4-MUX and 1 mM UDP-Gal, in the presence of concentrations from 0 to 5 mM 4H-Xyl-NP, 4H-Xyl-MU, or 4F-Xyl-MU. Quantification of the reaction product was carried out by HPLC, as described above. The enzyme activities were reported as a function of the logarithmic values of inhibitor concentration. IC 50 values were determined by fitting the experimental dose-response curves using the curve-fitter program of Sigmaplot 9.0 (Erkraft, Germany). K i values were calculated from IC 50 values according to the Cheng-Prusoff's equation (42,43).
In Cellulo Analysis of GAG Chains Biosynthesis by Na 2 [ 35 SO 4 ] Incorporation-GAG chains biosynthesis using 4-MUX as primer substrate was determined with CHOpgsB-618 cells (American Type Culture Collection). Cells were cultured in Dulbecco's modified Eagle's medium/F-12 (DMEM/F-12) (1:1), supplemented with 10% fetal bovine serum (Dutscher), penicillin (100 units/ml)/streptomycin (100 mg/ml), and 1 mM glutamine, then transfected with the wildtype or mutant pcDNA-␤4GalT7-Myc plasmid or with the empty pcDNA3.1 vector at 70% cell confluence. Transfected cells were then incubated in low sulfate medium (Fisher) supplemented with 10 Ci/ml of Na 2 [ 35 SO 4 ] in the presence of 0.5 or 10 M 4-MUX for 16 h. For GAG chain isolation 1 ml of culture medium was applied to a G-50 column (GE Healthcare) to separate radiolabeled GAG chains from the non-incorporated Na 2 [ 35 SO 4 ] and radiolabeling was quantified by scintillation counting. In parallel, the h␤4GalT7 expression level was checked by Western blotting using a primary anti-Myc (1/5,000) and a secondary anti-mouse antibody (1/10,000). To test the inhibitory potency of C4-modified xylosides, the molecules were added at 0 to 100 M concentration together with 4-MUX (5 M) for 16 h prior to isolation and quantification of radiolabeled GAGs. To test the cytotoxicity of xyloside inhibitors in CHOpgsB-618 cells expressing the wild-type h␤4GalT7, cells were seeded at 150,000 cells/well in 12-well plates, and incubated for 48 h at 37°C in the presence of 0 to 400 M inhibitor, or 4-MUX as a control. The ratio of viable cells upon the total number of cells was determined using the cell counter TC20 (Bio-Rad) in the presence of a vital marker (trypan blue).
In Cellulo Analysis of Decorin Core Protein Glycosylation-CHOpgs-B618 cells stably transfected with pcDNA-decorinHis encoding the human decorin core protein (31,44) were transiently transfected with pcDNA3.1 or with recombinant vector encoding either the wild-type or mutated h␤4GalT7-Myc as described above. 48 Hours following transfection, the cell medium was collected, concentrated by centrifugation at 4°C for 15 min at 3000 ϫ g, using the Amicon Ultracell 30 MWCO concentrating system (Merck Millipore, Germany), and submitted to SDS-PAGE (25 g of protein/well). The glycosylation level of the decorin core protein was monitored by immunoblot using a 1/5,000 dilution of primary polyclonal anti-human decorin antibody (VWR) and a 1/10,000 dilution of secondary anti-goat antibody coupled to horseradish peroxidase (Sigma), then quantified using ImageJ software. Briefly, the level of decorin glycosylation was expressed as relative band intensity by normalizing the band intensity value for the glycosylated form upon the total intensity value for the bands corresponding to the glycosylated and non-glycosylated forms of decorin core protein. The expression level of the decorin core protein in pcDNA3.1-transfected cells was used as the negative control and served as a loading control. On the other hand, the level of glycosylated decorin in cells expressing wild-type h␤4GalT7 was used as positive control for decorin glycosylation.

Molecular Modeling of the h␤4GalT7 Acceptor Binding Site-
In the present study, we aimed to identify amino acids important for structural organization of the h␤4GalT7 acceptor substrate binding site. We took advantage of the recent crystal structure of h␤4GalT7 in complex with UDP (34) to build a molecular model of this enzyme in complex with both the sugar donor UDP-Gal and the acceptor 4-MUX (Fig. 1A). The modeled structure is in a closed conformation, considered to be the catalytically competent form, and the hydrogen bond network around the UDP moiety is fully conserved. We first examined the position of a series of tyrosine residues, i.e. Tyr 194 , Tyr 196 , and Tyr 199 , that were suggested to be involved in binding of the xylobiose in the d␤4GalT7 structure (33). Our computational analysis indicates that Tyr 194 stabilizes both the donor and acceptor substrates location by establishing a hydrogen bond with a ␤-phosphate oxygen of UDP-Gal and a -stacking interaction with the 4-methylumbelliferyl moiety, respectively (Fig.  1A). Residue Tyr 196 is not hydrogen bonded to the substrates but to the side chain of residue Asp 229 , allowing its second carboxylic oxygen to be suitably oriented to form a hydrogen bond with the O2 atom of Xyl. The spatial orientation of Tyr 199 inside the substrate binding pocket allows the formation of a hydrogen bond between its side chain hydroxyl and the O2 atom of the Gal moiety of UDP-Gal. Altogether, residues Tyr 194 , Tyr 196 , and Tyr 199 form a strongly hydrophobic cluster that is required for correct binding of the substrates.
Analysis of the His 195 position, a conserved amino acid located between the two active site Tyr 194 and Tyr 196 residues, shows no hydrogen bond involving its side chain. However, the backbone nitrogen atom of this residue is hydrogen bonded with the CO group of 4-MUX (Fig. 1A). As illustrated in Fig. 1B, Arg 226 is located on the surface of the acceptor binding site contributing to an amphipathic entry door with the aromatic residues. In our model, there is no hydrogen bond involving the side chain of Arg 226 . Instead, its backbone nitrogen atom is hydrogen bonded with the O3 atom of the Xyl moiety of 4-MUX (Fig. 1A).
The structural impact of Arg 270 on enzyme activity, in the context of EDS was also addressed. The model structure of h␤4GalT7 reveals that Arg 270 belongs to the flexible loop (261-284) that moves upon donor substrate binding, thus creating the acceptor substrate binding site (Fig. 1A). This conformational change leads to the closed and catalytically competent conformation of the active site. However, the crystal structure of the human enzyme (34), as well as our own model in complex with both the donor and acceptor substrates do not highlight specific interactions established by this residue, although its close location to the surface of the active site has to be underlined (Fig. 1B).
Kinetic Properties of the Human Recombinant h␤4GalT7 Mutants Expressed in E. coli-To assess the functional importance of the residues of the acceptor binding site highlighted by our model, we carried out point mutagenesis and analyzed the consequences of conservative and non-conservative mutations on the kinetic parameters of h␤4GalT7 expressed and purified from recombinant E. coli cells. The wild-type enzyme and engineered mutants were produced as truncated fusion proteins lacking the 60 N-terminal amino acids (including the trans-membrane domain and part of the stem region) linked to GST and purified by affinity chromatography (data not shown). This led to 1.0 to 2.5 mg of pure protein per liter of culture for wildtype and mutant h␤4GalT7. Kinetic assays were performed using 4-MUX as acceptor substrate, which allowed quantification of the transfer reaction product by UV detection coupled to HPLC. The k cat and K m values of the wild-type enzyme toward UDP-Gal and 4-MUX shown in Table 1 were in agreement with previous work (17,31).
Substitution of Tyr 194 by alanine led to an inactive enzyme, and its conservative substitution by phenylalanine did not restore the galactosyltransferase activity of h␤4GalT7 (Table  1), indicating a critical role of this residue and, importantly, of the hydroxyl group of the tyrosine side chain. The mutation of Tyr 196 to alanine totally abolished enzyme activity, whereas replacement of this residue by phenylalanine led to a slightly active enzyme. The Y196F mutation did not impair enzyme affinity toward the donor substrate to a major extent but this mutant presented a lower affinity toward 4-MUX with a K m value about 3-fold that of the wild-type enzyme ( Table 1). As observed in the case of Tyr 194 and Tyr 196 , the non-conservative mutation Y199A led to a total loss of enzyme activity. However, similarly with what was observed for Tyr 196 , substitution of Tyr 199 by phenylalanine led to an active h␤4GalT7 enzyme with K m values toward UDP-Gal and 4-MUX and a k cat value only weakly affected compared with the wild-type enzyme. These data suggest that the aromatic ring of phenylalanine at position 199 is sufficient to support xyloside binding and activity.
Substitution of His 195 by alanine, glutamine, or arginine was carried out. The K m values of all three mutants toward UDP-Gal and 4-MUX were mostly comparable with those of the wild-type enzyme, indicating that these mutations had no major effect upon h␤4GalT7 affinity toward its substrates. Moreover, the substitutions at position 195 did not affect the rate of reaction transfer, as indicated by the k cat values that were essentially unchanged (Table 1). Altogether, these results indicate that the side chain of His 195 does not play a critical role in xyloside binding and h␤4GalT7 catalytic activity. Substitution of Arg 226 by alanine or lysine did not impair the affinity toward the substrates with K m values for UDP-Gal and 4-MUX, which were in the same range to that of the wild-type enzyme, and produced a moderate decrease (about 2-fold) of the catalytic constant value ( Table 1).
The Arg 270 residue is mutated to cysteine in the progeroid form of the EDS syndrome, and we previously showed that this mutation led to a significant decrease in h␤4GalT7 activity, mainly due to a reduced affinity toward 4-MUX (about 10-fold, see Ref. 17). To ascertain the contribution of this residue in h␤4GalT7 activity and xyloside binding, we performed kinetic assays following the conservative R270K and non-conservative R270A mutations. The k cat values for both mutants were about two times lower than that found for the wild-type enzyme and the K m value toward UDP-Gal was almost unaffected ( Table 1).

Effect of Tyr 194 , Tyr 196 , Tyr 199 , His 195 , Arg 226 , and Arg 270
Mutations on the Galactosyltransferase Activity of h␤4GalT7 Toward 4-MUX in Cellulo-To check the importance of the selected residues on the h␤4GalT7 function in a cellular context, we designed an experimental procedure involving CHOpgsB-618 cells transfected with h␤4GalT7 cDNA encoding either the wild-type or mutant enzymes fused to a Myc tag sequence at their C-terminal end, as described under "Experimental Procedures." CHOpgsB-618 cells expressing the recombinant enzymes were tested for in cellulo galactosyltransferase activity in the absence or presence of 4-MUX as the exogenous acceptor substrate. The expression level of enzymes was checked by immunoblot analysis using ImageJ software. As shown in Fig. 2, an additional upper band at ϳ39 kDa was observed (Fig. 2, inset). This ϳ39-kDa band can be attributed to the N-glycosylated form of the enzyme as it disappears upon peptide N-glycosidase F digestion (Fig. 2, inset, left panel). Both bands were taken into account to quantify the total enzyme expression level. The results indicate that all mutants considered in these experiments were expressed at a similar level to that of the wild-type protein (Fig. 2, inset, right panel). As shown in Fig. 2, the GAG synthesis level in cells expressing wild-type h␤4GalT7 was about 7-and 9.5-fold higher in the presence of 5 and 10 M 4-MUX, respectively, than in the absence of acceptor substrate, indicating that CHOpgsB-618 cells expressing h␤4GalT7 are able to prime efficiently GAG chains synthesis from 4-MUX, in agreement with previous studies (17,31).
As expected, cells expressing either Y194A or Y194F mutant in the presence of 4-MUX showed GAG synthesis levels comparable with those obtained with cells transfected with empty vector (Fig. 2). These in cellulo assays confirm that any mutation affecting the Tyr 194 position leads to a total loss of h␤4GalT7 activity. Substitution of Tyr 196 to alanine dramatically reduced the GAG synthesis rate whose level in the presence of 4-MUX was comparable with that obtained with cells transfected with empty vector. This is consistent with the loss of enzymatic activity observed in the in vitro assays (Table 1). By contrast, the conservative mutation Y196F allowed GAG chain

Kinetic parameters of wild-type and mutant GST-␤4GalT7
Kinetic parameters towards donor (UDP-Gal) and acceptor (4-MUX) substrates were determined as described under "Experimental Procedures." The results are the mean values of three independent determinations Ϯ S.D. on assays performed in duplicate.

JOURNAL OF BIOLOGICAL CHEMISTRY 7663
priming from 4-MUX. However, the GAG expression level reached with this mutant was about 2-3-fold lower than that of the wild-type enzyme, at 4-MUX at 5 and 10 M concentrations, respectively (Fig. 2). This result is consistent with the drastic decrease of the k cat /K m value toward 4-MUX found for the purified Y196F mutant. Comparable results were obtained with cells expressing h␤4GalT7 whose sequence is mutated on the Tyr 199 position. Indeed, cells expressing the Y199A mutant were unable to synthesize GAG chains from 4-MUX, whereas cells expressing Y199F showed GAG chain synthesis at a level about half of that observed with cells expressing wild-type enzyme (Fig. 2). These results are in line with the reduced efficiency exhibited in vitro by the enzyme substituted on the Tyr 199 position (Table 1). Altogether, these results demonstrate that both conservative and non-conservative mutations affecting Tyr 194 , Tyr 196 , or Tyr 199 significantly impaired GAG chains biosynthesis in a cellular context, in line with in vitro data (Table 1).
In the presence of 4-MUX, the GAG synthesis rate in cells expressing H195A, H195Q, and H195R mutants was moderately reduced, i.e. 10 to 15% lower than that of cells expressing the wild-type enzyme (Fig. 2). These results indicate that the side chain of this residue does not influence galactosyltransferase activity of h␤4GalT7 in the context of 4-MUX-primed GAG chains in eukaryotic cells, corroborating the findings that none of the mutations of His 195 significantly affect in vitro activity ( Table 1). The level of [ 35 SO 4 2Ϫ ] incorporation in the presence of 4-MUX in cells expressing R226A was about 2 times lower than that of the wild-type enzyme (Fig. 2). In addition, replacement of Arg 226 by lysine slightly increased the GAG expression level compared with alanine substitution, reaching about 60% that obtained with the wild-type, at 5 and 10 M 4-MUX. Corroborating in vitro kinetic parameters, these cellular assays indicate that modification of the side chain of Arg 226 produces minor effects on galactosyltransferase activity. We finally examined the impact of mutations of the Arg 270 residue upon GAG synthesis in eukaryotic cells. We observed that the GAG synthesis rate in cells expressing the R270A mutant was about 55% lower than that of the wild-type enzyme, at 5 and 10 M 4-MUX concentration (Fig. 2). The GAG synthesis level of the conservative mutant R270K was also about 2-fold reduced compared with the wild-type (Fig. 2). These results confirm that mutations of Arg 270 significantly affect the capacity of h␤4GalT7 to synthesize GAG chains from 4-MUX in a cellular context.
Effect of Tyr 194 , Tyr 196 , Tyr 199 , His 195 , Arg 226 , and Arg 270 Mutations on the Ability of h␤4GalT7 to Initiate the Glycosylation of the Decorin PG in Cellulo-We next determined whether the mutations would affect GAG chain formation on the core protein of decorin, used as a model PG (31). To this aim, CHOpgsB-618 cells were engineered to stably express the recombinant human decorin, and were transiently transfected with a pcDNA3.1 vector encoding the wild-type or mutant forms of h␤4GalT7. This allowed monitoring of the GAG sub- stitution of the secreted PG by Western blot analysis (Fig. 3,  inset, left and right panels). The rates of decorin PG glycosylation in cells expressing wild-type or mutant h␤4GalT7 were determined as described under "Experimental Procedures," then reported onto the histogram (Fig. 3). Results showed that non-conservative mutations of the tyrosine residues at positions 194, 196, and 199 as well as mutations of Tyr 194 and Tyr 196 to phenylalanine fully abolished the glycosylation of decorin (Fig. 3). These data were consistent with the drop of GAG chains primed from 4-MUX in cells expressing h␤4GalT7 mutated at these positions (Fig. 2). However, the substitution of Tyr 196 to phenylalanine induced a more dramatic effect on the glycosylation of decorin than on the in vitro or in cellulo activity toward 4-MUX. Furthermore, results shown in Fig. 3 indicate that conservative mutation Y199F allowed recovery up to 70% of the decorin glycosylation level compared with cells expressing the wild-type h␤4GalT7.
We also assessed the role of the His 195 residue in the glycosylation process of decorin. The results shown in Fig. 3 indicate that none of the mutations, H195A, H195Q, or H195R, significantly affected the level of decorin glycosylation. These data are consistent with the results obtained on the in vitro and in cellulo activity of the enzyme toward 4-MUX. Together, mutagenesis experiments indicate that modification of the amino acid side chain at position 195 did not greatly affect xyloside binding and galactosyltransferase activity of h␤4GalT7. Investigation of the effect of the Arg 226 mutation upon the ability of CHOpgsB-618 cells to glycosylate decorin showed that the glycosylation level reached with cells expressing R226A or R226K was about 65% of that obtained with cells expressing the wild-type enzyme.
These results were consistent with the in vitro and in cellulo GAG chain synthesis assays (Fig. 3).
Because we aimed to better understand the molecular basis of the EDS syndrome, it was important to further investigate the impact of mutations affecting the Arg 270 position upon the decorin glycosylation. The decorin glycosylation level reached with cells expressing either R270A or R270K mutant was about half of that obtained with cells expressing the wild-type h␤4GalT7, consistent with in vitro and in cellulo galactosyltransferase assays (Fig. 3).
Xyloside Inhibitors Design and in Vitro and in Cellulo Competition Assays-We next took advantage of the knowledge gained from our investigation of the organization of the acceptor substrate binding site to synthesize and test xyloside analogs as potential inhibitors of h␤4GalT7. To this end, in vitro competition assays were performed as described under "Experimental Procedures." The specific activity as a function of the logarithm values of the inhibitor concentrations are reported in Fig. 4B and data fitted to the logistic equation provided IC 50 values as reported in Table 2.
Because the C4-position is critical for both binding and transfer of the Gal residue from UDP-Gal onto the xyloside acceptor, we first synthesized a 4-deoxy derivative of 4-MUX (4H-Xyl-MU, Fig. 4A) and tested this compound as inhibitor of h␤4GalT7 in vitro. 4H-Xyl-MU was able to inhibit up to 50% of the initial activity at a 2 mM concentration (Fig. 4B), with an IC 50 value of about 1 mM and a K i value of about 0.5 mM (Table  2). To test whether hydrogen bond formation between 4-MUX and the protein via His 195 is important for the inhibitory potency, we synthesized 4H-Xyl-NP, which the aglycone struc- . Effect of wild-type and mutated h␤4GalT7 expression on decorin core protein glycosylation in CHOpgsB-618 cells. Cells stably expressing the human recombinant decorin core protein were transfected with the recombinant vector encoding either the wild-type (WT) or mutated h␤4GalT7. The decorin glycosylation level was monitored by immunoblot then quantified using ImageJ software, as described under "Experimental Procedures." Immunoblot analyses of the decorin core protein glycanation level are shown as inserts for CHOpgsB-618 cells transfected with empty (pcDNA) or with the recombinant vector coding for WT, Y194A, Y194F, Y196A, Y196F, Y199A, or Y199F h␤4GalT7 (left panel), and for WT, R270A, R270K, R226A, R226K, H195A, H195Q, or H195R h␤4GalT7 (right panel). The band observed at a molecular mass of ϳ35 kDa can be attributed to the decorin core protein, whereas the wide upper band corresponding to a molecular mass Ն75 kDa corresponds to the glycosylated decorin. Data are mean Ϯ S.E. from three independent experiments performed in triplicate. Statistical analysis was carried out by the Student's t test with **, p Ͻ 0.01 and ***, p Ͻ 0.001 versus decorin glycosylation in cells expressing the wild-type h␤4GalT7.

Structure-guided Inhibitors of h␤4GalT7
ture is unable to establish such an interaction, and compared its inhibitory effect to 4H-Xyl-MU. This compound produced a decrease of h␤4GalT7 activity toward 4-MUX less than 25% at the highest concentration (Fig. 4B) that did not allow determining IC 50 and K i values. These results clearly indicated that 4H-Xyl-NP is a weak inhibitor of h␤4GalT7. We next substituted the equatorial hydrogen of the C4 atom of the Xyl moiety by a fluorine atom, closer to oxygen in terms of electronegativity and predicted to fit the active site in terms of steric hindrance. 4F-Xyl-MU led up to 60% inhibition of the h␤4GalT7 activity at the highest concentration (Fig. 4B), with an IC 50 of 0.06 mM and a K i of 0.03 mM. The inhibition constant for this compound is more than 10 times lower than that reached for the deoxy analog (Table 2).
To complement the in vitro assay, we assessed the ability of the synthesized xyloside derivatives to inhibit GAG chain biosynthesis in cellulo. Addition of 4H-Xyl-NP produced a moderate but significant 20% decrease of GAG chain synthesis in CHOpgsB-618 cells, when used at 50 and 100 M (Fig. 5A). This correlated with the weak in vitro inhibition level obtained with this compound. The compound 4H-Xyl-MU allowed a larger inhibition of GAG chain synthesis, with up to 30% reduction of the GAG synthesis rate at similar concentrations (Fig. 5B). The best inhibitory effect was observed when performed in the pres-   ence of 4F-Xyl-MU leading to up to 50% inhibition of the initial GAG chain synthesis rate at 100 M concentration (Fig. 5C). Interestingly, preliminary results indicated that 400 M 4F-Xyl-MU inhibited the initial decorin glycosylation rate by about 50%, without affecting the viability of CHOpgsB-618 cells (data not shown). Altogether, the latter data confirmed that this fluorinated compound should be considered as a promising non-cytotoxic xyloside-based inhibitor of h␤4GalT7.

DISCUSSION
␤4GalT7 is a unique enzyme in the GAG biosynthetic pathway with regard to its capacity to use exogenous xyloside molecules as substrates and/or inhibitors that can efficiently modulate GAG synthesis in vitro and in vivo (19,20,45). This glycosyltransferase is also central in the GAG synthesis process because the formation of the tetrasaccharide linker is a prerequisite to the polymerization of both heparan sulfate and chondroitin/dermatan sulfate chains. The human enzyme thus represents a prime target for the design of effectors of GAG synthesis as drugs to correct GAG disorders associated with numerous malignant conditions such as genetic diseases and cancer. To meet this challenge, we pioneered structure-function studies of the recombinant h␤4GalT7. We previously carried out structural, thermodynamic, and phylogenetic investigations that identified key amino acid residues mainly implicated in the recognition and binding of the donor substrate (31,46). We also provided insight into the molecular basis of the GAG defects characterizing rare forms of EDS syndrome (17,47). In the present work, to develop xyloside compounds that will specifically target the h␤4GalT7 activity for a therapeutic purpose, we explored the architecture of the acceptor substrate binding site. To this aim, we combined functional investigations including site-directed mutagenesis, kinetic analyses, in vitro and in cellulo evaluation of galactosyltransferase activity and GAG synthesis, and a computational approach. This allowed mapping the acceptor binding site and to design and synthesize a potent xyloside-based inhibitor of GAG synthesis.
We first targeted a set of three tyrosine residues, Tyr 194 , Tyr 196 , and Tyr 199 , as well as His 195 belonging to the same conserved motif, that occupy a strategic position surrounding the xyloside acceptor substrate (Ref. 34 and the present data). Our mutational analysis led to a remarkable observation because alanine substitution of each of these tyrosine completely abolished h␤4GalT7 activity. The tyrosine-alanine mutants (i) were devoid of galactosyltransferase activity in vitro, (ii) were unable to prime GAG synthesis from 4-MUX in cellulo, and (iii) did not promote decorin glycosylation, thus supporting a prominent function of this set of aromatic residues. Mutating Tyr 194 , Tyr 196 , and Tyr 199 with phenylalanine revealed that the substitution differently affected h␤4GalT7 activity depending on the position. Noteworthy, the presence of the hydroxyl group of Tyr 194 was indispensible because the conservative Y194F mutant completely lacked in vitro or in cellulo galactosyltransferase activity toward 4-MUX and was unable to sustain glycosylation of decorin. This observation is likely to be explained by a functionally important interaction between the hydroxyl group of this tyrosine and the ␤-phosphoryl group of UDP-Gal that was observed in all structures and models of ␤4GalT7 (Ref. 34 and this report, see Fig. 1). Most importantly, our computational model and experimental data suggested that the critical role of Tyr 194 also arises from a stacking interaction between the aromatic ring of this residue and the 4-methylumbelliferyl moiety of 4-MUX. Altogether, these data indicate that Tyr 194 occupies a strategic location in the catalytic center and interacts with both the donor nucleotide and the aglycone group of 4-MUX. In the case of Tyr 196 , the presence of the hydroxyl group of the tyrosine residue was also a major structural element because its replacement by phenylalanine only slightly restored the activity toward 4-MUX in vitro and in cellulo. The Y196F mutant did not sustain decorin glycosylation, also supporting an important role of this residue in the glycosylation of endogenous proteoglycans. Our model provides a molecular explanation to these results, because it shows that Tyr 196 is not directly involved in the binding of the acceptor substrate but rather forms a hydrogen bond between its hydroxyl group and Asp 229 , this latter residue establishing an important interaction with O2 of the Xyl moiety. This supports the idea that interactions with the hydroxyl in the C2 position control a strict geometry via both Asp 229 and Tyr 196 , in agreement with the physiologically important regulatory role of Xyl-phosphate substitution in position 2 on GAG synthesis (41,48). Furthermore, our model suggests that this residue is part of the cavity floor in agreement with structural data indicating that the acceptor substrate xylobiose is located in a hydrophobic binding pocket formed by Tyr 177 , Tyr 179 , and Trp 207 in the Drosophila structure, and by Tyr 194 , Tyr 196 , and Trp 224 in h␤4GalT7. The present data complements our previous findings demonstrating that Trp 224 is a crucial active site residue (31). Differently to the preceding studied tyrosines, h␤4GalT7 tolerated well the substitution of tyrosine to phenylalanine at position 199 leading to a mutant that was active toward 4-MUX in vitro and was able to prime GAG synthesis from 4-MUX and onto the decorin core protein. Consistently, Tyr 199 is substituted by a phenylalanine in the Drosophila enzyme, suggesting that the presence of a hydrophobic aromatic ring is sufficient at this position. Further analysis of our molecular model predicts hydrogen bond formation between the tyrosine hydroxyl group of Tyr 199 and O2 of the Gal moiety of UDP-Gal. However, no significant change in the K m value toward UDP-Gal was observed for the Y199F mutant, indicating that this interaction may not play a critical functional role in nucleotide binding. The location of Tyr 199 favors a role as contributor to the hydrophobic surrounding of the acceptor substrate binding pocket together with Tyr 194 , Tyr 196 , and Trp 224 residues when h␤4GalT7 is in its closed conformation (34).
Investigation of the structural role of His 195 led to the most interesting findings for the design of efficient substrates and inhibitors of h␤4GalT7. We predicted that the nitrogen atom of the peptide backbone of this residue forms a hydrogen bond with the carbonyl group of the coumarin moiety of 4-MUX. This is in full agreement with our mutational study that showed no major effect upon changing the side chain of the histidine residue at this position by alanine, glutamine, or arginine substitution on the h␤4GalT7 activity monitored in vitro and in cellulo. However, the functional importance of an interaction between the His 195 backbone and 4-MUX was clearly emphasized by the stronger inhibitory effect of 4-deoxy-Xyl-MU compared with 4-deoxy-Xyl-NP. We also found that 4-MUX was used as a substrate with a better affinity than Xyl-NP (data not shown). Noteworthy, among a series of naphthyl xylosides, Siegbahn et al. (29) showed that 6-hydroxy-naphthyl-Xyl was able to prime GAG chains more efficiently than any other unsubstituted derivative in breast cancer cell lines. Altogether, this gives strong evidence that His 195 through its backbone provides an important structural element for efficient binding of 4-MUX derivatives, and offers a molecular explanation for the superiority of 4-MUX synthesized in this study over naphthyl and benzyl-substituted xylosides previously reported in the literature (25,30).
We also discovered a unique active site basic residue, i.e. Arg 226 . Interestingly, this residue is located between the aromatic-rich 221 FWGWG 225 sequence, containing Trp 224 that interacts with the ␤-phosphate O-atom of the donor substrate, and the 227 EDDE 230 sequence containing acidic residues that are involved in Xyl binding and in the transfer reaction (31). Our functional analysis showed that modifying the side chain of Arg 226 by site-directed mutagenesis did not affect enzyme affinity toward acceptor or donor substrate. This is in line with the computational analysis indicating that the nitrogen atom of the peptide backbone of Arg 226 , but not its side chain, interacts with the O3 atom of the Xyl moiety. Fig. 1, A and B, clearly shows that this residue, together with Trp 224 , are brought close to the aromatic triad in the closed conformation of h␤4GalT7.
We also investigated the role of Arg 270 , which substitution by a cysteine residue is implicated in the progeroid form of EDS (49). Our previous studies revealed that this genetic mutation dropped in vitro h␤4GalT7 activity and impaired GAG chains synthesis in CHO-pgsB618 cells (17). These effects were suggested to be due to a loss of hydrogen bonding between the lateral chain of Arg 270 and the hydroxyl group of a serine residue from the PG core protein (17). This idea was supported by later molecular modeling indicating that Arg 270 borders the catalytic site of h␤4GalT7 in the closed conformation (Ref. 34 and this work, see Fig. 1A). However, the precise mechanism by which Arg 270 modulates in vitro and ex vivo h␤4GalT7 activities remains unclear. Indeed, current crystal structures and molecular models do not point to a specific role of this residue in catalysis or substrate binding, consistent with kinetic data showing that mutations of Arg 270 in alanine and lysine moderately affect h␤4GalT7 activity and affinity, and the observation that in Drosophila, the corresponding position is occupied by a lysine residue. The Arg 270 residue is located in the flexible loop (261-284) that moves upon donor substrate binding, thus creating the acceptor substrate binding site. This conformational change leads to the closed and catalytically competent conformation of the active site. It thus can be expected that any mutations affecting the loop movement would impair the transfer reaction. However, why the substitution of Arg 270 by a cysteine residue that causes the progeroid form of EDS patients, produces more deleterious consequences than alanine or lysine mutations requires further investigation of the conformational modifications operating during the catalytic cycle.
Our current and previous functional and computational approaches provide a detailed cartography of the h␤4GalT7 acceptor substrate binding pocket for the rational design of xyloside-based inhibitors (31). We show that the active site organization is governed, on one side, by a series of aromatic amino acids comprising three Tyr residues, i.e. Tyr 194 , Tyr 196 , and Tyr 199 , which together with Trp 224 create a hydrophobic environment and provide stacking interactions essential to the binding of both the xylosyl and aglycone parts of the acceptor substrate. On the opposite side of the site, it involves a network of hydrogen-bonding interactions between three charged amino acids, i.e. Asp 228 , Asp 229 , and Arg 226 , and the hydroxyl groups of the Xyl moiety and other active site residues. Until now, most studies aiming to inhibit cellular and extracellular GAG synthesis have been targeted to the synthesis and testing of xyloside derivatives acting as substrates of ␤4GalT7 thus reducing the glycosylation of endogenous proteoglycan core proteins (19). This approach successfully provided promising pharmacological agents, in particular anti-tumor compounds (45,50). However, because such molecules behave both as exogenous primers of GAG synthesis and inhibitors of endogenous GAG formation, deciphering their mechanism of action remains challenging. With the perspective of designing xyloside derivatives that selectively act as inhibitors of GAG formation, we opted for C4-modified analogs, whose modification at the C4 position prevents the catalytic transfer, and first synthesized deoxy derivatives. In addition, we took advantage of the information gained from our structural and mutational analyses. We considered two key elements of the aglycone binding, i.e. the strategic position of Tyr 194 that forms stacking interactions with the aglycone part of the acceptor substrate as well as the interaction between His 195 N-backbone and the carbonyl group of the coumarinyl moiety. In agreement with our prediction, our results clearly show that the 4-deoxy-xyloside appended to 4-MU was superior to the naphthyl-substituted molecule, as indicated by in vitro and in cellulo studies, supporting the idea that the hydrogen bond between His 195 and the carbonyl group of the coumaryl group is crucial. Interestingly, Tsuzuki et al. (28) found that among triazole xyloside derivatives generated by click chemistry bearing various aromatic and nonaromatic aglycones, the p-nitrophenyl analog was the best inhibitor of PG synthesis when screened in endothelial cells. Although detailed docking of these triazole derivatives should be performed, it is tempting to speculate that the formation of a hydrogen bond interaction between the aglycone nitro group and His 195 enhances the inhibitory potential compared with the other substituted benzyl derivatives. Furthermore, we show here that the 4-deoxy-4-fluorinated 4-MUX was superior to the unsubstituted 4-deoxy analog, indicating that addition of an electronegative atom at this position is an important element in the design of potent inhibitors. The 4-hydroxyl group is involved in two hydrogen bonds with the carboxyl group of Asp 228 and the 4-hydroxyl group of Gal, respectively. The replacement of the hydrogen donor hydroxyl group by a fluorine atom that is larger than hydrogen and which van der Waals radius and electronegativity are closer to oxygen, at the C4 position, enhanced binding interactions with h␤4GalT7. This corroborated previous studies showing that several fluorinated xylosides act as "good" substrates or inhibitors of GAG synthesis (18,29,30). In the same way, fluorinated thrombin inhibitors showed improved factor Xa binding (51). Further docking calculations of fluoro-substituted xylosides are underway to assess the mechanism underlying the improved interactive properties upon fluorine incorporation.
In summary, we developed a powerful approach for the design of xyloside inhibitors that specifically target h␤4GalT7. By integrating structural elements important for the binding of both the Xyl moiety and the hydrophobic aglycone, we synthesized a xyloside-based inhibitor of h␤4GalT7. We generated a compound that both impact in vitro galactosyltransferase activity and affect GAG synthesis in cells, opening promising pharmacological applications. This molecule also represents a valuable chemical biology tool to explore the biological effects of GAGs.