Crystal Structure of Family 14 Polysaccharide Lyase with pH-dependent Modes of Action*

The Chlorella virus enzyme vAL-1 (38 kDa), a member of polysaccharide lyase family 14, degrades the Chlorella cell wall by cleaving the glycoside bond of the glucuronate residue (GlcA) through a β-elimination reaction. The enzyme consists of an N-terminal cell wall-attaching domain (11 kDa) and a C-terminal catalytic module (27 kDa). Here, we show the enzyme characteristics of vAL-1, especially its pH-dependent modes of action, and determine the structure of the catalytic module. vAL-1 also exhibited alginate lyase activity at alkaline pH, and truncation of the N-terminal domain increased the lyase activity by 50-fold at pH 7.0. The truncated form vAL-1(S) released di- to hexasaccharides from alginate at pH 7.0, whereas disaccharides were preferentially generated at pH 10.0. This indicates that vAL-1(S) shows two pH-dependent modes of action: endo- and exotypes. The x-ray crystal structure of vAL-1(S) at 1.2 Å resolution showed two antiparallel β-sheets with a deep cleft showing a β-jelly roll fold. The structure of GlcA-bound vAL-1(S) at pH 7.0 and 10.0 was determined: GlcA was found to be bound outside and inside the cleft at pH 7.0 and 10.0, respectively. This suggests that the electric charges at the active site greatly influence the binding mode of substrates and regulate endo/exo activity. Site-directed mutagenesis demonstrated that vAL-1(S) has a specific amino acid arrangement distinct from other alginate lyases crucial for catalysis. This is, to our knowledge, the first study in which the structure of a family 14 polysaccharide lyase with two different modes of action has been determined.

the best of our knowledge, no report on a polysaccharide-degrading enzyme with both endo-and exotype activity.
This study deals with the identification of the protein module crucial for CWM/alginate degradation, characterization of peculiar pH-dependent modes of action, and determination of the x-ray crystal structure of family PL-14 enzymes. The structural determinants for enzyme catalysis and modes of action are also discussed.

EXPERIMENTAL PROCEDURES
Materials-Synthetic oligonucleotides utilized for PCR were purchased from Hokkaido System Science. Restriction enzymes were purchased from New England Biolabs. TALON metal affinity resin was purchased from Clontech. Mono S 10/100 GL and Superdex peptide 10/300 GL were purchased from GE Healthcare. Factor Xa was purchased from Novagen. Sodium alginate (average molecular mass, 110 kDa; M/G ratio, 56.5%/43.5%) from Eisenia bicyclis and Triton X-100 were purchased from Nacalai Tesque. GlcA sodium salt monohydrate was purchased from Sigma-Aldrich.
Plasmid Construction-To construct an Escherichia coli expression system for a fusion protein comprising glutathione S-transferase, vAL-1, and histidine-tagged sequences, the vAL-1 gene (1,047 bp) was inserted into the expression vector pET42b (Novagen) encoding glutathione S-transferase and two histidine-tagged sequences. For expression of vAL-1(S), the vAL-1(S) gene (732 bp) was inserted into pET21b (Novagen) encoding a histidine-tagged sequence. Detailed procedures for the plasmid constructions are described in the supplemental methods. To substitute Lys-197, Ser-209, His-213, Ser-219, Arg-221, Tyr-233, or Tyr-235 in vAL-1(S) with alanine or phenylalanine, site-directed mutagenesis was conducted using pET21b-vAL-1(S) as the template and pairs of synthetic oligonucleotides (supplemental Table S1) as primers with a QuikChange site-directed mutagenesis kit (Stratagene). The mutants were designed as K197A, S209A, H213A, S219A, R221A, Y233F, and Y235F. DNA manipulations were carried out according to the standard method (21).
Overexpression and Purification-E. coli strain BL21(DE3) was transformed with the constructed plasmids pET42b-vAL-1 and pET21b-vAL-1(S). E. coli cells harboring pET42b-vAL-1 were grown in Luria-Bertani medium (21), collected, suspended in 0.5 M NaCl and 20 mM Tris-HCl (pH 7.5) (buffer A) including 0.5% Triton X-100, and ultrasonically disrupted. The fusion protein consists of glutathione S-transferase, a linker including a Factor Xa site, vAL-1, and two histidine-tagged sequences in the linker and C terminus. After purification of the fusion protein on TALON metal affinity resin, vAL-1 was obtained through Factor Xa digestion and purification on a Mono S 10/100 GL column. The protein was dialyzed against 20 mM Gly-NaOH (Gly) (pH 8.4). After centrifugation at 20,000 ϫ g and 4°C for 20 min, the dialysate was used as purified vAL-1.
E. coli cells harboring the plasmid pET21b-vAL-1(S) were grown in LB medium, collected, suspended in buffer A including 0.5% Triton X-100, and ultrasonically disrupted. These cells produced a protein consisting of vAL-1(S) and a histidinetagged sequence at the C terminus. The protein was purified on TALON metal affinity resin. The enzyme solution was dialyzed against 20 mM Gly (pH 8.4). After centrifugation at 20,000 ϫ g and 4°C for 20 min, the dialysate was used as purified vAL-1(S). Detailed procedures for purification of vAL-1 and vAL-1(S) were described in the supplemental methods.
The vAL-1(S) mutants K197A, S209A, H213A, S219A, R221A, Y233F, and Y235F were also expressed in E. coli cells and purified similarly. Circular dichroism spectra of these mutants were measured by a J-720 spectropolarimeter (Jasco) in the range of 250 -190 nm, confirming that there was no significant difference in secondary structures among vAL-1 (S) and mutants. The selenomethionine derivative of vAL-1(S) was expressed in E. coli B-834(DE3) cells (22). The derivative protein was purified as mentioned above for vAL-1(S).
Assay of Protein Concentration and Enzyme Activity-The protein concentration was determined by the bicinchoninic acid method (23) with bovine serum albumin as the standard or by measuring the absorbance at 280 nm assuming that E 280 ϭ 1.5 corresponds to 1 mg/ml protein.
Alginate lyase activity was determined by the thiobarbituric acid method (24). Considering that the absorbance at 549 nm of 0.01 mol for ␤-formylpyruvic acid corresponds to 0.29 (25), 1 unit of enzyme activity was defined as the amount of enzyme required to liberate 1 mol of ␤-formylpyruvic acid/min at 37°C. The specific activity of vAL-1, vAL-1(S), and the mutants was assayed with a reaction mixture containing 1.0 mg/ml sodium alginate, 200 mM Hepes (pH 7.0) or Gly (pH 10.0), and 0.1 mg/ml enzyme. The kinetic parameters k cat and K m for vAL-1(S) were determined with several concentrations of sodium alginate (0.01, 0.02, 0.05, 0.1, 0.2, 0.5, 1.0, 2.0, 3.0, 4.0, and 5.0 mg/ml), 200 mM Hepes (pH 7.0), and 0.1 mg/ml enzyme. The sizes of products released from sodium alginate were measured by size exclusion chromatography. The products were prepared  DECEMBER 18, 2009 • VOLUME 284 • NUMBER 51 in a reaction mixture consisting of 10.0 mg/ml sodium alginate, 200 mM Hepes (pH 7.0) or Gly (pH 10.0), and 1.0 mg/ml vAL-1(S) incubated for different periods (0, 1, 2, 3, 4, 10, and 20 h). After the sample solutions were boiled for 10 min, and the supernatants were collected by centrifugation at 15,000 ϫ g and 25°C for 20 min to eliminate denatured vAL-1(S). Then 100 l of supernatant was applied onto a Superdex peptide 10/300 GL column equilibrated with 20 mM potassium phosphate (pH 7.0). The exact molecular mass of each product was determined by electrospray ionization mass spectrometry at a cone voltage of Ϫ50 V by using LCT Premier XE (Waters).

Structure and Function of Family 14 Polysaccharide Lyase
N-terminal Amino Acid Sequence-Protein solution including the truncated vAL-1 was subjected to SDS-PAGE (26). A polyvinylidene difluoride membrane wet with 100% methanol was soaked in 10% methanol and 10 mM CAPS-NaOH (pH 11.0). The proteins in the gel were transferred to the membrane by electroblotting at a current of 2 mA/cm 2 for 40 min and then detected with a dye solution of Ponceau S. The N-terminal amino acid sequence of the protein corresponding to vAL-1(S) (27 kDa) was determined by Edman degradation with a Procise 492 protein sequencing system (Applied Biosystems).
Crystallization and X-ray Data Collection-vAL-1(S) was crystallized by the sitting drop method and sparse matrix screening by using commercial crystallization kits (Hampton Research). vAL-1(S) crystals were obtained under two kinds of crystallization conditions; the first consists of 100 mM trisodium citrate dihydrate, 20% polyethyleneglycol (PEG)-4000, and 20% 2-propanol with 10 mg/ml vAL-1(S) (solution A), and the other consists of 100 mM Hepes (pH 7.0), 200 mM ammonium formate, and 20% PEG-3350 with 14 mg/ml of vAL-1(S) (solution B). The crystals formed in solution A were soaked in the corresponding mother liquid containing 20% glycerol for cryoprotection. Diffraction data were collected by synchrotron radiation at a wavelength of 0.7 Å in a resolution range of 50 -1.2 Å at the SPring-8 Beamline (Japan). The selenomethionine derivative of the protein was also crystallized in solution A. Diffraction data were collected by synchrotron radiation at a wavelength of 0.98 Å in the resolution range of 50 -1.7 Å. The crystals formed in the solution B were soaked at 20°C for 10 min in a solution containing 100 mM Hepes (pH 7.0) or Gly (pH 10.0), 10% PEG-3350, and 200 mM GlcA. The crystals formed at pH 7.0 were then soaked in a solution of 50 mM Hepes (pH 7.0), 200 mM GlcA, and 35% PEG-3350 for cryoprotection. The data obtained were processed, merged, and scaled using the HKL2000 program package (DENZO and SCALEPACK) (27). The data collection statistics are shown in Table 1.
Structure Determination-The single anomalous dispersion method was used to determine the structure of vAL-1(S). The initial phase was determined with the SOLVE program using the anomalous data of the selenomethionine derivative (28). After density improvement by solvent flattening and histogram matching in the DM program (29), an initial model was built using the ARP/wARP program (30). The crystal structure of vAL-1(S) formed in solution A was determined by manual modeling with the Coot program (31) and anisotropic B-factor refinement with the SHELXL program (32). The crystal structures of GlcA-bound vAL-1(S) molecules at pH 7.0 and 10.0 were determined by molecular replacement with the ligandfree vAL-1(S) structure mentioned above as a search model, manual modeling with the Coot program, and refinement with the SHELXL program. The refinement statistics are shown in Table 1.
To clarify whether or not vAL-1 exhibits alginate lyase activity, the protein was overexpressed in E. coli cells, purified, and characterized. The thiobarbituric acid assay for alginate lyase revealed that vAL-1 functions as an alginate lyase with an optimum pH of around 10 ( Fig. 2A, left panel). Although vAL-1 optimally degrades CWM at pH 7.0 (11), its alginate lyase activity was low at neutral pH.
Activation of Alginate Lyase through Truncation of N-terminal Domain-vAL-1 (38 kDa) purified from E. coli cells gradually converted to the 27 kDa protein when stored in buffer with high protein and/or low salt concentrations. The 27 kDa protein was considered to be a truncated form of vAL-1(S). The N-terminal amino acid sequence of vAL-1(S) was determined to be NH 2 -TNVI, corresponding to 106 TNVI 109 of vAL-1. The molecular mass and N-terminal amino acid sequence revealed that vAL-1(S) comprises 252 amino acid residues including Thr-106 to the C-terminal histidine-tagged sequence. Because vAL-1(S) shows sequence identity with other family PL-14 lyases, the recombinant vAL-1(S) expressed in E. coli cells was purified and characterized.
vAL-1(S) showed no CWM lyase activity, possibly because of the lack of binding affinity between its N-terminal domain and CWM. In contrast, vAL-1(S) degraded alginate at pH 10.0 with a specific activity of 0.10 milliunits/mg, and its activity was comparable with that of vAL-1 (0.086 milliunits/mg). This indicates that the N-terminal domain of vAL-1 is unessential for its alginate lyase activity. Surprisingly, vAL-1(S) showed 50-fold higher alginate lyase activity at pH 7.0 (1.1 milliunits/mg) than vAL-1 (0.024 milliunits/mg). The N-terminal domain is thought to hinder the expression of alginate lyase activity at neutral pH. Unlike vAL-1, vAL-1(S) showed the highest activity at pH 7.0 ( Fig. 2A, right panel). A Michaelis-Menten plot showed that the kinetic parameters k cat and K m of vAL-1(S) at pH 7.0 were 0.039 Ϯ 0.002 min Ϫ1 and 0.10 Ϯ 0.03 mg/ml, respectively. Similar truncation of the N-terminal repeated region was conducted in the family PL-7 alginate lyase A1-II', although the enzyme activity was comparable between the truncated and full-length forms (33). Although it remains unclear how the N-terminal domain hinders alginate degradation at neutral pH, two possibilities are as follows: (i) In the case of the family PL-5 alginate lyase A1-III (34), the noncatalytic lid loop over the active cleft binds to the catalytic residue in the absence of substrate. The N-terminal domain of vAL-1 may also interact at neutral pH with the active site in the absence of CWM and thereby inhibit the approach of other polysaccharides. At alkaline pH, the interaction is lowered, and the alginate molecules can be accommodated at the active site.
(ii) The isoelectric point of the N-terminal domain is estimated to be considerably high (pI 10). This positive charge at neutral pH may induce the tight interaction of the N-terminal domain with highly acidic alginate rather than CWM. Because alginate is a high molecular mass polysaccharide (110 kDa), the interaction between the polysaccharide and N-terminal domain prevents another alginate molecules from binding to the active site. At alkaline pH, the interaction is reduced, and the polymer is thereby degraded by the catalytic module. Anyway, further experiments are required to clarify the reaction mechanism by full-length vAL-1.
Mode of Action-To clarify the mode of action of vAL-1(S) by which it degrades alginate, the alginate degradation profile was determined by size exclusion chromatography. The degrada-tion products were detected by monitoring the absorbance at 235 nm derived from C4ϭC5 double bonds of unsaturated saccharides. In the absence of vAL-1(S), alginate degradation was not observed at pH 7.0 and 10.0. In the presence of vAL-1(S), several peaks corresponding to unsaturated di-, tetra-, penta-, and hexasaccharides released from alginate were observed at pH 7.0 after reaction for 2.0 h (Fig. 2B, left panel). The amount of di-, tri-, and tetrasaccharides increased with time, and they seemed to be final products. At pH 10.0, unsaturated disaccharides were detected in the early stage (3.0 h) of the reaction. As the reaction proceeded, the amount of only disaccharides increased, and other products were scarcely detected, except a small amount of trisaccharides (Fig. 2B, right panel). The trisaccharides together with disaccharides seem to be generated from alginate pentasaccharides as the minimum substrate. A small peak corresponding to products smaller than disaccharides was observed. These products are not considered to be monosaccharides because no monosaccharides were detected by thin layer chromatography (supplemental Fig. S1).
The reaction profiles at pH 7.0 and 10.0 suggest that vAL-1(S) shows pH-dependent modes of action, i.e. the enzyme tends to act endolytically on alginate at neutral pH but exolytically at alkaline pH. The modes of action were confirmed to be influenced strictly by pH and not by the reaction components, i.e. Hepes and glycine, because similar results were obtained in 200 mM Tris-HCl (pH 7.0) and ammonia-ammonium chloride (pH 10.0). In fact, the reaction profile at pH 8.5 was a combination of those at pH 7.0 and 10.0. The reaction products at pH 8.5 comprised alginate disaccharides as well as tri-and tetrasaccharides, whereas few disaccharides were included in the reaction products at pH 7.0 under the same conditions (supplemental Fig. S1). This is, to our knowledge, the first endo and exo polysaccharide lyase; these enzymes are usually specifically endo or exo, although molecular conversion of endo/exo modes was achieved in polysaccharide-degrading enzymes by protein engineering through the addition/removal of the loop recognizing terminus of the substrate (18). Thus, x-ray crystallogra-phy of vAL-1(S) was conducted to determine the three-dimensional structure of family PL-14 enzymes and clarify the structural determinants for the modes of action.
Structure Determination-Crystals of vAL-1(S) were obtained in solution A containing PEG-4000, 2-propanol, and trisodium citrate. The crystal structure of vAL-1(S) was determined at 1.2 Å resolution by the single anomalous dispersion method with an R-factor of 12.9% and a free R-factor of 15.9%.
The crystal structure of vAL-1(S) shows that it comprises 485 amino acid residues, 783 water molecules, two citrate molecules, and six glycerol molecules. Two protein molecules, A and B, are included in an asymmetrical unit. Except for the N-terminal residue in molecule A, two residues in molecule B, and the 16 C-terminal histidine-tagged residues in the two molecules, all of the coordinates were built according to the 2F o Ϫ F c map.
Structure of vAL-1(S) in Complex with GlcA-Because the alginate monosaccharides, D-mannuronic acid, and L-guluronic acid were not available, the structure of GlcA-bound vAL-1(S) was determined to clarify the enzyme catalysis and modes of action. Although the vAL-1(S) crystals formed in solution A were soaked in several solutions containing GlcA, GlcA was not found bound to vAL-1(S) in the crystal structure. vAL-1(S) crystals were also obtained in solution B containing PEG-3350, ammonium formate, and Hepes (pH 7.0). The crystals, belonging to the P2 1 2 1 2 1 space group, were soaked in GlcA-containing solution at pH 7.0 or 10.0 and used for x-ray data collection. The crystal structures of GlcA-bound vAL-1(S) at pH 7.0 and 10.0 were determined by molecular replacement with the coordinates of ligand-free vAL-1(S) mentioned above as a search model ( Fig. 4 and Table 1). No significant conformational change was found in the GlcA-bound vAL-1(S) structures at the two pH levels compared with the ligand-free model; the rootmean-square deviation is 0.25 Å for all C ␣ atoms between the GlcAbound and ligand-free structures.
pH has a significant effect on the electric charge of protein molecules. At pH 10.0, the molecule surface displays the formation of a highly positively charged region composed of Lys-197, Arg-221, Tyr-233, and  Structure and Function of Family 14 Polysaccharide Lyase DECEMBER 18, 2009 • VOLUME 284 • NUMBER 51 Tyr-235 (Fig. 4C). From the interaction between vAL-1(S) and GlcA at pH 10.0, this highly positively charged region is thought to be important for substrate binding at alkaline pH. In fact, the results of site-directed mutagenesis described in the next section support this hypothesis regarding the structural determinant for substrate binding at pH 10.0. Many polysaccharide-degrading enzymes possess a catalytic pocket and loops that restrict the mode of action, whereas vAL-1(S) with an open cleft forms no such pocket and loops. The binding affinity at each subsite is important for determining the mode of action of open cleft enzymes. For example, in a report on Aspergillus niger endopolygalacturonases I and II with an open substrate-binding cleft, substitution of a single amino acid residue at subsite Ϫ5 changes the binding affinity and thereby influences the endo/exo mode of action (36). In the case of vAL-1(S), the substrate binding affinity at each subsite is dependent on pH and thereby determines the mode of action.
Mutation in the Active Cleft-Polysaccharide lyases are characteristic in that they commonly cleave glycoside bonds through ␤-elimination reaction (37), although there is great diversity in the arrangement of catalytic residues crucial for the enzyme reaction. Because the family PL-14 enzymes are homologous in their primary structures, some amino acid residues conserved in this family were considered to play an important role in the catalytic reaction. Most conserved residues were found to be positioned at strands A3-A6 and the loop L1, all of which constitute the cleft (Fig.  5A). The amino acid residues Lys-197, Ser-209, His-213, Ser-219, Arg-221, Met-223, Tyr-233, Tyr-235, and Phe-325 are located on the molecular surface of the cleft (Fig. 5B). In fact, GlcA molecules are bound to residues located in the cleft. Thus, vAL-1(S) mutants with a mutation in the cleft were constructed and enzymatically characterized. Site-directed mutagenesis was performed in Lys-197, Ser-209, His-213, Ser-219, Arg-221, Tyr-233, and Tyr-235 to evaluate their contributions to the catalytic reaction and/or substrate binding. Specific activity was measured at pH 7.0 and 10.0 for K197A, S209A, H213A, S219A, R221A, Y233F, and Y235F ( Table 3). All of the mutants showed lower activity than the wild-type enzyme, indicating that an active site of vAL-1(S) is present on the cleft. The mutations of Lys-197 and Ser-219 especially led to a great decrease in enzyme activity both at pH 7.0 and 10.0, suggesting that these two residues play key roles in the catalytic reaction and/or substrate binding. Although H213A, R221A, Y233F, and Y235F retained their activity to some extent at pH 7.0, these mutants showed extremely low activity at pH 10.0, suggesting that His-213, Arg-221, Tyr-233, and Tyr-235 are important for an enzyme reaction at alkaline pH.
As is the case with the family PL-14 alginate lyase vAL-1(S), family PL-7, 18, and 20 enzymes also have a ␤-jelly roll structure as a basic scaffold (Table 2). Previous reports propose that the family PL-7 alginate lyase A1-IIЈ contains a catalytic site composed of the four residues Arg, Gln, His, and Tyr (38,39). This arrangement of four residues is also conserved in the PL-18 alginate lyase. Gln and Tyr in the structure of the family PL-20 FIGURE 5. Conserved amino acid residues in family PL-14 lyases. A, alignment of family PL-14 alginate lyases by using the Clustal W program with alginase from H. discus discus (HDD_alginase; GenPept accession number BAE45131), alginate lyase from H. discus hannai (HDH_Aly; GenPept accession number BAC87758), exo-oligoalginate lyase from H. discus hannai (HDH_Alex; GenPept accession number BAE81787), alginate lyase from Chlorella virus CVN1 (CVN_Aly; GenPept accession number BAA83789), and vAL-1(S) (GenPept accession number BAB19127). The amino acid residues conserved in family PL-14 enzymes are mainly located on strands A3-A6 and loop L1. The conserved non-glycine residues exposed to the molecule surface are shown in yellow boxes. vAL-1(S) was also aligned with family PL-7 alginate lyase A1-II' from Sphingomonas sp. 1 (GenPept accession number BAD16656) and PL-20 endo-␤-1,4-glucuronan lyase from Trichoderma reesei (GenPept accession number BAG80639). B, arrangement in the cleft of hydrophilic amino acid residues conserved in family PL-14 enzymes (stereo diagram). The figure was drawn using MolFeat (FiatLux). ␤-1,4-glucuronan lyase are well superimposed on those in family PL-7 and 18 lyases, whereas His and Arg are replaced with Ile and His, respectively (40). To examine the catalytic site in the family PL-14 alginate lyases, vAL-1(S) was superimposed on family PL-7 A1-IIЈ. Despite the common scaffold, the arrangement of the four residues is not conserved in family PL-14 alginate lyases; vAL-1(S) has Gly, Gly, Phe, and Phe at the corresponding positions, implying that family PL-14 enzymes have a different specific catalytic site (Fig. 5A). Site-directed mutants with a low activity indicated that Lys-197, Ser-209, His-213, Ser-219, Arg-221, Tyr-233, and/or Tyr-235 are conserved in the PL-14 family and are involved in enzyme catalysis. Lys-197 and Ser-219 are especially crucial for the enzyme reaction at both pH 7.0 and 10.0.
In conclusion, this is the first report on the molecular identification of alginate lyase, which was shown to have two pHdependent modes of action, and structural determination of family PL-14 enzymes. Distinct from family PL-7, 18, and 20 polysaccharide lyases, this family PL-14 lyase has a specific amino acid arrangement at its catalytic site. Further, it exhibits pH-dependent modes of action, which regulate the binding affinity level with the substrate at each subsite.
Seaweed alginate is widely utilized in the food, chemical, and pharmaceutical industries as a stabilizing, thickening, or emulsifying reagent (17), whereas the polymer produced by bacteria such as Pseudomonas aeruginosa functions as an important virulence factor during lung infections in cystic fibrosis patients (41). Thus, alginate lyases hold promise as biochemicals for processing edible seaweed alginate and removing bacterial biofilm alginate (42,43). More recently, alginate lyases have been expected to become potential enzymes in the bioenergy generation from alginate (44). The alginate lyase with two different modes of action is applicable to various fields, and the regulation of endo/exo modes by electric charges at subsites contributes to molecular design of useful polymer-degrading enzymes.