Crystal Structure of a Bacterial Unsaturated Glucuronyl Hydrolase with Specificity for Heparin*

Background: Bacterial unsaturated glucuronyl hydrolase (UGL) is essential for complete degradation of host glycosaminoglycans. Results: Crystal structure of Pedobacter heparinus UGL, Phep_2830 specific for heparin degradation, was determined. Conclusion: The pocket-like structure and lid loop of Phep_2830 are involved in heparin disaccharide recognition. Significance: This work contributes to understanding the bacterial degradation of host extracellular matrix components. Extracellular matrix molecules such as glycosaminoglycans (GAGs) are typical targets for some pathogenic bacteria, which allow adherence to host cells. Bacterial polysaccharide lyases depolymerize GAGs in β-elimination reactions, and the resulting unsaturated disaccharides are subsequently degraded to constituent monosaccharides by unsaturated glucuronyl hydrolases (UGLs). UGL substrates are classified as 1,3- and 1,4-types based on the glycoside bonds. Unsaturated chondroitin and heparin disaccharides are typical members of 1,3- and 1,4-types, respectively. Here we show the reaction modes of bacterial UGLs with unsaturated heparin disaccharides by x-ray crystallography, docking simulation, and site-directed mutagenesis. Although streptococcal and Bacillus UGLs were active on unsaturated heparin disaccharides, those preferred 1,3- rather than 1,4-type substrates. The genome of GAG-degrading Pedobacter heparinus encodes 13 UGLs. Of these, Phep_2830 is known to be specific for unsaturated heparin disaccharides. The crystal structure of Phep_2830 was determined at 1.35-Å resolution. In comparison with structures of streptococcal and Bacillus UGLs, a pocket-like structure and lid loop at subsite +1 are characteristic of Phep_2830. Docking simulations of Phep_2830 with unsaturated heparin disaccharides demonstrated that the direction of substrate pyranose rings differs from that in unsaturated chondroitin disaccharides. Acetyl groups of unsaturated heparin disaccharides are well accommodated in the pocket at subsite +1, and aromatic residues of the lid loop are required for stacking interactions with substrates. Thus, site-directed mutations of the pocket and lid loop led to significantly reduced enzyme activity, suggesting that the pocket-like structure and lid loop are involved in the recognition of 1,4-type substrates by UGLs.

Unsaturated GAG disaccharides are degraded to monosaccharides by unsaturated glucuronyl hydrolase (UGL), which is classified as a member of the glycoside hydrolase family 88 in the CAZy database (Fig. 1B) (7,13). In contrast to other hydro-lases, UGL recognizes a double bond in ⌬GlcUA and triggers the hydration of C-5 (14,15). Because ⌬GlcUA is a component of all unsaturated GAG oligosaccharides, UGLs are essential for the complete degradation of GAGs. To date, all bacterial UGLs are classified as members of the GH-88 family. However, different substrate specificities accommodate structural diversities of unsaturated GAG oligosaccharides with different sugar residues, glycosidic bonds, and degrees of sulfation. For example, streptococcal UGLs and Bacillus sp. GL1 enzyme (BacillusUGL) prefer sulfated and unsulfated unsaturated disaccharides with 1,3-glycosidic bonds, respectively (16,17). In contrast, Phep_ 2830, the UGL of Pedobacter heparinus (formerly known as Flavobacterium heparinum) degrades only unsaturated heparin disaccharides with 1,4-glycosidic bonds (18).
Recent studies have focused on the physiological functions and structures of UGLs and have revealed peculiar mechanisms of UGL catalysis using artificial substrates (15) and that UGL gene disruption leads to reduced upper respiratory tract colonization by Streptococcus pneumoniae (19). Specific inhibitors of UGL are therefore expected to provide anti-bacterial drugs with no side effects. In addition, we demonstrated inducible mRNA expression of the streptococcal enzyme in the presence of GAG (16) and identified structural determinants of the preference of streptococcal UGL for sulfated substrates with 1,3glycosidic bonds (12). However, the enzyme recognition mechanism for 1,4-glycosidic bond-type substrates from heparin and heparan sulfate remains unknown. Degradation of heparins and heparan sulfates with 1,4-glycosidic bonds is also considered important for bacterial adherence and invasion to host cells because heparin, heparan sulfate, hyaluronan, and chondroitin sulfate are constituents of mammalian extracellular matrices (20). In this study, we examined the crystal structure of a P. heparinus UGL that has specific activity for unsaturated disaccharides with 1,4-glycosidic bonds and demonstrated the binding modes of these substrates with bacterial UGLs by x-ray crystallography, docking simulations, site-directed mutagenesis, and measurements of mutant enzyme kinetics.

EXPERIMENTAL PROCEDURES
Materials-Unsaturated chondroitin disaccharides were purchased from Seikagaku Biobusiness (Tokyo, Japan). Unsaturated heparin disaccharides were purchased from Iduron (Manchester, UK) and Sigma. Restriction endonucleases and DNAmodifying enzymes were purchased from Toyobo (Osaka, Japan). All other analytical grade chemicals were obtained from commercial sources.
Microorganisms and Culture Conditions-To express Phep_2238 or Phep_2649, E. coli BL21(DE3) cells harboring pET21b-Phep_2238 or pET21b-Phep_2649 plasmid were cultured at 37°C in Luria broth (Sigma) supplemented with sodium ampicillin (100 g ml Ϫ1 ). When turbidity at 600 nm reached 0.3-0.7, isopropyl ␤-D-thiogalactopyranoside was added to the culture to a final concentration of 0.1 mM, and the cells were further cultured at 16°C for 44 h. To overexpress Phep_2830, E. coli Rosetta gami B cells harboring pCold IV-Phep_2830 were cultured at 37°C in Luria broth supplemented with sodium ampicillin (100 g ml Ϫ1 ) and chloroamphenicol (33 g ml Ϫ1 ). When turbidity at 600 nm reached 1-1.2, the cells were cooled to 15°C with ice water, isopropyl ␤-D-thiogalactopyranoside was added to the culture to a final concentration of 0.4 mM, and the cells were further cultured at 15°C for 44 h.
Enzyme Assay-Reactions of SagUGL, SpnUGL, SpyUGL, and BacillusUGL were conducted at 30°C in 500-l solutions of 20 mM Tris-HCl (pH 7.5), 0.2 mM substrate, and enzyme. Reactions of Phep_2238, Phep_2649, and Phep_2830 were conducted at 30°C in 500-l solutions containing 100 mM KPB (pH 6.0), 0.2 mM substrate, and enzyme. Concentrations and pH of each buffer were adopted in accordance with previous reports (12,18). Enzyme activity was measured by monitoring decreases in absorbance at 235 nm, which corresponded to the loss of substrate C-C double bonds.
Docking Simulations-To perform docking analyses of UGLs and substrates, our coordinates of SagUGL were used from the Protein Data Bank (PDB). The structure of ligand-free SagUGL (PDB code 2ZZR) was selected as a receptor model. Docking analyses were performed using the AutoDock 4.2 program (23). The coordinates of C⌬6S were obtained from coordinates of SagUGL complexed with C⌬6S (PDB code 3ANK), and coordinates of unsaturated heparin disaccharides were obtained using ACD/ChemSketch Freeware, version 5.12 (Advanced Chemistry Development, Inc., Toronto, ON, Canada) and the OpenBabel program (24). Asp-175 and Lys-370 of SagUGL were treated as flexible residues because Asp-175 of SagUGL was critical for catalysis and Lys-370 showed movement during interactions with C⌬6S (12,16). The number of genetic algorithm runs was set to 20. Figures for protein structures and docking forms were prepared using the PyMOL program (25).
X-ray Crystallography-To determine the three-dimensional structure of Phep_2830, the purified enzyme was crystallized using sitting drop vapor diffusion. Solutions containing 1 l of proteins (10 mg ml Ϫ1 ) in 20 mM Tris-HCl (pH 7.5), 1 mM dithiothreitol, and 1 mM EDTA were mixed with equal volumes of reservoir solution containing 25% polyethylene glycol 6000, 0.1 M HEPES (pH 7.5), and 0.1 M lithium chloride. Protein solutions were then incubated at 20°C, and single crystals were grown for about 2 months. The crystal was flash-cooled under a cold nitrogen stream. Diffraction data were collected at ϭ 1 Å using an ADSC Q315 detector at the BL38B1 station of SPring-8, Harima, Japan. The data were processed using the HKL2000 program (27). Molecular replacements for structure determinations were conducted using the Molrep program supplied in the CCP4 program package with coordinates of SagUGL (PDB code 2ZZR) as an initial model (28,29). Structure refinement was conducted with the Phenix.refine program supplied in the PHENIX program package (30,31). Randomly selected 5% reflections were excluded from refinement and were used to calculate R free . After each cycle of refinement, the model was manually adjusted using the Coot program (32). The final model quality was checked using the PROCHECK program (33). Superposi-tioning of protein models and calculation of their root mean square deviations were conducted using the LSQKAB program supplied with the CCP4 program package (34).

RESULTS AND DISCUSSION
Enzyme Activity of Streptococcal UGLs toward Unsaturated Heparin Disaccharides-Three streptococcal UGLs (SagUGL, SpnUGL, and SpyUGL) were purified to homogeneity, and their enzyme activities (units mg Ϫ1 ) toward a variety of unsat-urated heparin disaccharides were measured (Table 1). All three streptococcal UGLs degraded H⌬NS with greater efficiency than the other unsaturated heparin disaccharides. Kinetic parameters of SagUGL toward H⌬NS were determined as follows: K m , 1.9 mM; k cat , 3.9 s Ϫ1 . SagUGL is known to exhibit the highest enzymatic activity toward C⌬6S with 1,3-glycosidic bond (12) as follows: K m , 0.10 mM; k cat , 10 s Ϫ1 , indicating that the enzyme activity of SagUGL toward unsaturated heparin disaccharides was lower than that toward unsaturated chon-   (12). The enzyme activities of Bacillu-sUGL with each unsaturated heparin disaccharide were also measured. BacillusUGL exhibited the highest enzyme activity toward H⌬NAc0S distinct from streptococcal UGLs (Table 1). BacillusUGL also demonstrated specificity for unsulfated unsaturated chondroitin disaccharides (17). These results and our previous reports (16,17) demonstrate that streptococcal UGLs exhibit high enzyme activity toward disaccharides with specific sulfate groups, whereas BacillusUGL preferentially degrades unsulfated substrates.

Recognition of Unsaturated Heparin Disaccharides by Streptococcal UGL-Structure determinations of SagUGL in com-
plex with H⌬NS were difficult, possibly due to its lower affinity for this disaccharide. Thus, to elucidate the binding modes of unsaturated heparin disaccharide with SagUGL, the structure of SagUGL in complex with H⌬NS was predicted using the docking simulation program AutoDock. To evaluate the suitability of the AutoDock program, docking simulations were performed with our coordinates of ligand-free SagUGL (PDB code 2ZZR) and C⌬6S. The simulated complex structure with the lowest binding energy was almost identical to the crystal structure of SagUGL/C⌬6S except that Lys-370 forms a hydrogen bond with a sulfate group (PDB code 3ANK; Fig. 2, A and  B), indicating that the AutoDock docking simulation accurately determined the substrate-bound UGL structure. Thus, the binding mode of H⌬NS to SagUGL was calculated using an AutoDock program, indicating the accommodation of H⌬NS in the active site of SagUGL (Fig. 2C). The direction of the pyranose ring of GlcN toward ⌬GlcUA of H⌬NS opposed that of GalNAc in C⌬6S (Fig. 1C). Due to these differences in direction, the amino group of GlcN in H⌬NS is located at a similar position to the C-6 of GalNAc in C⌬6S, and the position of a sulfate group in H⌬NS corresponds to that of the sulfate group in C⌬6S. The directions of H⌬NS pyranose rings indicated by docking simulations were similar to those of heparin in solution, which were determined by x-ray scattering (35). This docking simulation suggests that Ser-365, Ser-368, and Lys-370 residues of SagUGL form hydrogen bonds with the sulfate group of H⌬NS. To confirm the functions of Ser-365, Ser-368, and Lys-370 of SagUGL, the SagUGL mutants S365A, S365G, S368A, S368G, and K370A were purified and assayed. Kinetic parameters of each mutant were determined and compared with those of wild-type SagUGL ( Table 2). K m values of S365A, S365G, S368A, and S368G for H⌬NS were higher than that of the wild-type enzyme, whereas that of K370A was lower. The k cat values of S365G, S368G, and in particular K370A for H⌬NS were also decreased (k cat of K370A was ϳ50-fold lower than that of WT). This mutant analysis shows that Ser-365 and Ser-368 of SagUGL are involved in binding to H⌬NS, whereas Lys-370 enhances the reaction efficiency of SagUGL rather than the binding affinity for H⌬NS. In contrast, although the binding modes of the other unsaturated heparin disaccharides to SagUGL were calculated, no structures of substrate-bound SagUGL were obtained (data not shown). These docking simulations and enzyme assays of SagUGL (Table 1) suggest that, with the exception of H⌬NS, its affinity toward unsaturated heparin disaccharides is also remarkably low.
Docking simulations of H⌬NS with SagUGL, site-directed mutagenesis, and kinetic analyses demonstrate that Ser-365 and Ser-368 residues of SagUGL contribute to recognition of the sulfate group of H⌬NS and that the Lys-370 residue is involved in degradation of the substrate. These amino acid residues constitute the motif SXXSXK, which plays an important role in the recognition of the sulfate group of C⌬6S (12). Streptococcal UGLs acted on unsaturated heparin disaccharides, although their enzyme activities toward these substrates were low. In addition, putative heparan sulfate lyase is encoded in the vicinity of the streptococcal UGL gene, suggesting that streptococcal UGLs contribute to the degradation of heparan sulfate.
Substrate Specificity of P. heparinus UGLs-The structural determinants of SagUGL recognition of H⌬NS sulfate groups were calculated using docking simulations. However, the intrinsic UGL-binding mechanism of substrates containing 1,4-glycosidic bonds remains to be clarified. Therefore, the mechanisms behind UGL recognition of unsaturated heparin disaccharides were analyzed using high affinity UGL interactions with unsaturated heparin disaccharides. One (Phep_2830) of the UGLs of P. hepa-

Complete Bacterial Degradation of Heparin
rinus, which assimilates heparin as a carbon source, is known to specifically degrade substrates containing 1,4-glycosidic bonds. Thirteen UGL genes of P. heparinus were assigned by complete genome sequences (36) and the three UGLs Phep_2238, Phep_2649, and Phep_2830 were overexpressed in E. coli cells and purified to homogeneity and assayed for activity (Table 1).
Phep_2238 degraded H⌬NAc0S and H⌬NS more efficiently than other unsaturated heparin disaccharides. Moreover, Phep_2238 exhibited comparably lower enzyme activity toward substrates with 1,3-glycosidic bonds, such as C⌬0S, C⌬4S, and C⌬6S, than those with 1,4-glycosidic bonds. The kinetic parameters K m and k cat of Phep_2238 toward H⌬NS were 0.073 mM and 2.3 s Ϫ1 , respectively. Phep_2649 preferred C⌬6S but also  FEBRUARY 21, 2014 • VOLUME 289 • NUMBER 8 degraded C⌬0S, H⌬NAc0S, and H⌬NS, although its specific activity was lower than that of the streptococcal UGLs, Bacillu-sUGL, Phep_2238, and Phep_2830. K m and k cat values of Phep_2649 toward H⌬NS were 0.052 mM and 0.17 s Ϫ1 , respectively. Although a UGL that degrades unsaturated chondroitin disaccharide (C⌬6S) was previously isolated as a 1,3-glycuronidase from P. heparinus (37), this enzyme differs from Phep_2238 and Phep_2649 in amino acid composition, isoelectric point, and molecular weight. In contrast, Phep_2830 exhibited enzyme activity toward only unsaturated heparin disaccharides, as described previously (18). K m and k cat values of Phep_2830 toward H⌬NAc6S were 0.029 mM and 16 s Ϫ1 , respectively.

Complete Bacterial Degradation of Heparin
Crystal Structure of Phep_2830-Because Phep_2238 and Phep_2830 showed high enzyme activity toward various unsaturated heparin disaccharides, these enzymes were crystallized to clarify the mechanisms by which UGL recognizes substrates with 1,4-glycosidic bonds based on tertiary structures. Phep_2830 was successfully crystallized, and x-ray diffraction data were collected. The crystal structure of Phep_2830 was determined at a resolution of 1.35 Å using molecular replacement with the SagUGL structure (PDB code 2ZZR) as an initial model. Data collection and model refinement statistics are summarized in Table 3. The final model contains one monomer enzyme from Gly-28 to Thr-398, a molecule of HEPES, and 422 water molecules. The N-terminal region from Met-1 to Asn-27 could not be assigned because the electron density map was too thin. Similar to SagUGL and BacillusUGL, the overall structure of Phep_2830 has a ␣ 6 /␣ 6 -barrel architecture, and Phep_2830 contains 12 ␣ helices and 5 ␤ strands (Fig. 3A). The root mean square deviation for all of the 336 C ␣ atoms between Phep_2830 and SagUGL was 1.7 Å, indicating that both have a common basic scaffold structure. In contrast, ϳ10 C-terminal amino acid residues of Phep_2830 protrude forward from the outside of the protein. Previous studies show that Asp-149 of BacillusUGL and Asp-175 of SagUGL act as critical catalysts (14,16). Thus, to investigate the catalytic mechanisms of P. heparinus UGLs, corresponding residues of Phep_2238 (Asp-182) and Phep_ 2830 (Asp-174) were substituted with Asn. Both mutant enzymes were inactive, indicating that the catalytic mechanisms of Phep_2238 and Phep_2830 are similar to those of SagUGL and BacillusUGL.
Active sites were structurally compared by superimposing coordinates of Phep_2830 on those of SagUGL (PDB code 2ZZR) and BacillusUGL/C⌬0S (PDB code 2AHG; Fig. 3

, B-D).
Subsites were defined such that Ϫn represents the nonreducing terminus, ϩn represents the reducing terminus, and cleavage occurs between these sites (38). The amino acid residues and their positions at subsite Ϫ1, which is the binding site of ⌬GlcUA, were common to all three UGLs (Fig. 3B). However, the structures at subsite ϩ1, which is the binding site of an amino sugar, differed significantly (Fig. 3, C and D). In particular, Ser-365, Ser-368, and Lys-370 of SagUGL comprise the motif SXXSXK, which contributes to the recognition of a sulfate group (12). These three residues are not conserved in Phep_2830 and correspond to the Ala-363, Tyr-366, and Ser-368 residues of Phep_2830, respectively. These observations indicate that Phep_2830 recognition of substrate sulfate groups differs from that of SagUGL. Regarding hydrophobic amino acid residues forming a stacking interaction with an amino sugar at the subsite ϩ1, the Tyr-338 residue of BacillusUGL (or the Tyr-364 residue of SagUGL), which is located around the C-6 position of GalNAc in BacillusUGL/C⌬0S is substituted with the Gly-362 residue in Phep_2830, whereas the Trp-134 residue of BacillusUGL (or the Trp-161 residue of SagUGL) is conserved in Phep_2830. Due to this difference, this large space in Phep_2830 occurs around Gly-362, and is distinct from that of SagUGL and BacillusUGL. As described above, the Phe-164 residue of Phep_2830 that corresponds to Glu-163 of SagUGL or Pro-136 of BacillusUGL also lies at the binding site of an amino sugar (Fig. 3, C and D). The loop comprising amino acid residues 163-169 of Phep_2830, which is designated loop A and corresponds to residues 162-170 of SagUGL and 135-144 of BacillusUGL, covers the active site. However, these loops of both SagUGL and BacillusUGL are distal from those at the active site. Accordingly, the Phe-164 residue of Phep_2830 is proximal to the active site (Fig. 3D). These Phep_2830-specific amino acid residues may be involved in stacking interactions with substrates, especially with amino sugars. Indeed, whereas Arg-57, Arg-66, and Glu-369 residues of Phep_2830 are arranged at subsite ϩ1, these are not conserved in the active sites of either SagUGL or BacillusUGL.
Binding Mode of Unsaturated Heparin Disaccharides to Phep_2830-To clarify the mechanism by which Phep_2830 recognizes unsaturated heparin disaccharides, we attempted to prepare complexes of Phep_2830 with these substrates but failed. Thus, the binding modes of unsaturated heparin disaccharides (H⌬NAc0S, H⌬NS, H⌬6S, and H⌬NAc6S) to Phep_ 2830 were estimated in a similar way to those of SagUGL using the AutoDock program. In these simulations, structures of H⌬NAc0S, H⌬NS, and H⌬NAc6S-bound Phep_2830 were successfully calculated and that of the Phep_2830-H⌬NAc0S complex is shown in Fig. 2D. Two common features were observed in the three docking structures that exhibited low binding energies. Similar to the docking simulation of SagUGL with H⌬NS, the direction of the pyranose ring of GlcNAc (or GlcN sulfated at the N position) toward ⌬GlcUA was opposed to that of Gal-NAc in complex with SagUGL/C⌬6S and BacillusUGL/C⌬0S (Fig. 1C). In addition, the acetyl groups of H⌬NAc0S and H⌬NAc6S (or the sulfate group of H⌬NS) were predicted to be accommodated in the pocket-like structure comprising Arg-57, Arg-66, Trp-73, Gly-362, Ala-363, Tyr-366, Ser-368, and Glu-369 residues, which was designated the acetyl/sulfate groupbinding pocket (Fig. 2, D and E). No similar pocket-like structures were observed in SagUGL and BacillusUGL, which prefer substrates with 1,3-glycosidic bonds (Fig. 3C). The Trp-162 and Phe-164 residues in loop A of Phep_2830 (Fig. 3D) are located close to the C-6 position of an amino sugar of unsaturated heparin disaccharides. As described above, Trp-161, Trp-134, and Trp-162 residues of SagUGL, BacillusUGL, and Phep_2830, respectively, are situated at almost identical positions, whereas the residues of SagUGL and BacillusUGL that correspond to Phe-164 of Phep_2830 are distal from subsite ϩ1.
To confirm that this pocket-like structure and loop A contribute to recognition of unsaturated heparin disaccharides, Arg-57, Phe-164, and Gly-362 residues of Phep_2830 were substituted with Ala, Ala, and Tyr, respectively. The resulting mutants R57A, F164A, and G362Y with N-terminal His tags were expressed in E. coli and were purified using affinity columns. Although their expression was confirmed by SDS-PAGE (Fig. 4), two mutants of the pocket-like structure (R57A and G362Y) exhibited no detectable enzyme activity with unsaturated heparin disaccharides as well as unsaturated chondroitin  FEBRUARY 21, 2014 • VOLUME 289 • NUMBER 8 disaccharides. The K m value (1.4 mM) of F164A toward H⌬NAc6S was about 50-fold higher than that (0.029 mM) of the wild-type enzyme, whereas there was no significant difference in k cat between wild-type enzyme (16 s Ϫ1 ) and F164A (22 s Ϫ1 ). These site-directed mutagenesis experiments indicated that the pocket-like structure and loop A play a significant role in recognizing substrates. In contrast, the absence of the acetyl group may have led to low specific activity of Phep_2830 toward H⌬6S and hampered docking simulations of the complex of Phep_2830 with H⌬6S. Hence, substrates containing 1,4-glycosidic bonds are readily inserted into the acetyl/sulfate groupbinding pocket of Phep_2830 through stacking interactions of loop A residues, Trp-162 and Phe-164, with an amino sugar. Simultaneously, Trp-162 and Phe-164 residues may inhibit binding of Phep_2830 with substrates containing 1,3-glycosidic bonds.

Complete Bacterial Degradation of Heparin
Comparisons of the active site structures and docking simulations suggest that the acetyl/sulfate group-binding pocket comprising Arg-57, Arg-66, Trp-73, Gly-362, Ala-363, Ser-368, and Glu-369, and the loop A including Trp-162 and Phe-164, are important for binding of Phep_2830 to unsaturated heparin disaccharides. Among these amino acid residues, Arg-57, Trp-73, Gly-362, Glu-369, and Trp-162 are conserved in Phep_2238 and Phep_2649 (Fig. 5). Moreover, Arg-66, Ala-363, Tyr-366, Ser-368, and Phe-164 of Phep_2830 correspond with Val-73, His-373, Gly-376, Ser-378, and Ser-171 of Phep_2238 and Arg-59, Ser-362, His-365, Asn-367, and Val-158 of Phep_2649, respectively. These differences in primary structure may lead to changes in pocket shapes but do not occupy the space. However, Ser-171 of Phep_2238 and Val-158 of Phep_2649, which correspond to Phe-164 of Phep_2830, do not form stacking interactions because they are smaller than Phe. Hence, these structural features may enable recognition of substrates with 1,3-glycosidic bonds by Phep_2238 and Phep_2649. Although the pocket-like structure of Phep_2649 was arranged with high probability, the activity of the enzyme for unsaturated heparin disaccharides was lower than that of Phep_2238 and Phep_2830. In addition, the relative length of amino acid residues of Phep_2649 that correspond to loop A may contribute to low enzyme activity.
Bacteroides species, human intestinal bacteria, are known to produce family PL-12, -13, and -21 heparinases depolymerizing heparin to unsaturated disaccharides through a ␤-elimination reaction (39 -43), whereas complete degradation of heparin to monosaccharides remains to be clarified. Four UGL-homologous proteins are encoded in the genomes of Bacteroides stercoris (BACSTE_02470, BACSTE_03202, BACSTE_03210, and BACSTE_03707) and Bacteroides thetaiotaomicron (BT_0146, BT_2913, BT_3348, and BT_4658). Arg-57 and Gly-362, key residues comprising the pocket-like structure of Phep_2830, are conserved in these bacteroides UGLs (supplemental Fig.  S1), suggesting that, similar to Phep_2238 and Phep_2830, bacteroides enzymes preferentially degrade unsaturated heparin disaccharides to constituent monosaccharides. Consequently, structural determinants of Phep_2830 for 1,4-specificity found in this study promotes a better understanding of the complete degradation of heparin by the Bacteroides species.
In conclusion, tertiary and active site structures of UGLs that are specific for unsaturated heparin disaccharides containing 1,4-glycosidic bonds were determined for the first time. Comparisons of the active site structures, docking simulations, and site-directed mutagenesis experiments demonstrate the significance of the acetyl/sulfate group-binding pocket and the lid loop at subsite ϩ1 in Phep_2830 for recognition of substrates with 1,4-glycosidic bonds.