A Novel Structural Fold in Polysaccharide Lyases

Rhamnogalacturonan (RG) lyase produced by plant pathogenic and saprophytic microbes plays an important role in degrading plant cell walls. An extracellular RG lyase YesW from saprophytic Bacillus subtilis is a member of polysaccharide lyase family 11 and cleaves glycoside bonds in polygalacturonan as well as RG type-I through a β-elimination reaction. Crystal structures of YesW and its complex with galacturonan disaccharide, a reaction product analogue, were determined at 1.4 and 2.5Å resolutions with final R-factors of 16.4% and 16.6%, respectively. The enzyme is composed of an eight-bladed β-propeller with a deep cleft in the center as a basic scaffold, and its structural fold has not been seen in polysaccharide lyases analyzed thus far. Structural analysis of the disaccharide-bound YesW and a site-directed mutagenesis study suggested that Arg-452 and Lys-535 stabilize the carboxyl group of the acidic polysaccharide molecule and Tyr-595 makes a stack interaction with the sugar pyranose ring. In addition to amino acid residues binding to the disaccharide, one calcium ion, which is coordinated by Asp-401, Glu-422, His-363, and His-399, may mediate the enzyme activity. This is, to our knowledge, the first report of a new structural category with a β-propeller fold in polysaccharide lyases and provides structural insights into substrate binding by RG lyase.

Plant cell wall degradation is essential for plant pathogenic and saprophytic microbes to invade plants. The plant cell wall mainly consists of polysaccharides and proteins, with polysaccharides being more abundant. These polysaccharides are divided into three groups, cellulose, hemicellulose, and pectin (1). Pectin is more soluble in water than cellulose and hemicel-lulose, suggesting that this polysaccharide is the preferred target for plant-associated microbes. Pectin is further divided to three regions, i.e. polygalacturonan, rhamnogalacturonan type-I (RG-I), 2 and rhamnogalacturonan type-II (RG-II). In pectin molecules, polygalacturonan is present as a linear backbone, and RG-I and RG-II are attached to the backbone as branched chains (2)(3)(4). RG-I is a polymer with a disacchariderepeating unit consisting of L-rhamnopyranose and D-galactopyranouronic acid (GalA) as a main chain, and arabinans and galactans are attached to the main chain (5). RG-II has a backbone of polygalacturonan, and its side chains consist of a complex of ϳ30 monosaccharides, including rare sugars such as apiose and aceric acid (6). These plant cell wall polysaccharides become substrates to be degraded through reactions of polysaccharide hydrolases and/or lyases produced by plant pathogenic and saprophytic microbes. Hydrolases cleave glycoside bonds in polysaccharides via hydrolysis, and lyases do so by ␤-elimination reactions (7,8).
The degradation pathway for the polygalacturonan region has been characterized extensively in microbes (9 -12). The structure and function relationships of polygalacturonan-degrading enzymes, including hydrolases and lyases have also been well documented (13)(14)(15)(16)(17)(18)(19)(20). However, little information is available on the degradation of RG regions by microbes except for Aspergillus species (21) and Erwinia chrysanthemi (22,23). We identified the degradation pathway for the RG-I region in saprophytic Bacillus subtilis (24) and genetically and enzymatically characterized RG lyases YesW and YesX and unsaturated RG hydrolases YesR and YteR (25). RG lyases are responsible for cleaving ␣-1,4 bonds of the RG-I main chain through a ␤-elimination reaction (Fig. 1A), and unsaturated RG hydrolases act on the unsaturated RG saccharides produced through the reactions of RG lyases.
Carbohydrate-active enzymes such as glycoside hydrolases, polysaccharide lyases, glycosyl transferases, and carbohydrate esterases are categorized into over 200 families based on amino acid sequences in the carbohydrate-active enzymes (CAZy) data base (26). Lyases are classified into 18 polysaccharide lyase (PL) families (Table 1). Polysaccharide lyases commonly recognize uronic acid residues in polysaccharides, catalyze ␤-elimination, and produce unsaturated saccharides with CϭC double bonds at nonreducing terminal uronate residues. These characteristics of lyases indicate that they share common structural features determining uronate recognition and reaction manner (␤-elimination). Crystal structures of polysaccharide lyases in 11 families have been determined thus far (Table 1), and structure and function relationships of enzymes such as lyases for pectate, pectin, alginate, chondroitin, hyaluronan, and xanthan are being analyzed (15,16,(27)(28)(29)(30)(31)(32). However, little knowledge has been accumulated on the mechanisms of substrate recognition and catalytic reaction of RG lyases.
RG lyases belong to family PL-4 and family PL-11, which mainly contain fungal RG lyases and bacterial RG lyases, respectively. The crystal structure of a family PL-4 RG lyase, RhgB, has been determined from Aspergillus aculeatus KSM 510 (33), although its structure and function relationship remains to be clarified. Family PL-11 RG lyases Rgl11A from Cellvibrio japonicus (34), Rgl11Y from Clostridium cellulolyticum (35), and YesW and YesX from B. subtilis (24) have been characterized enzymatically, but no structural analysis of these enzymes has, to our knowledge, been reported. Because family PL-11 enzymes show no significant homology to other proteins, including polysaccharide lyases, their structural fold exhibits a novel type that has not been observed in polysaccharide lyases analyzed so far. Structural analysis of family PL-11 lyases is expected to clarify their structure and function relationships, i.e. mechanisms for substrate recognition, catalytic reaction, and their physiological functions in bacterial infection and/or saprophytic processes in relation to plants.
Herein we report the determination of the three-dimensional structure of family PL-11 YesW, and its complex with a reaction product analogue by x-ray crystallography, and identification of the active site of the enzyme and amino acid residues responsible for substrate recognition.

EXPERIMENTAL PROCEDURES
Materials-RG-I (from potatoes) and polygalacturonan (from citrus pectin) were purchased from Megazyme. Pectin was purchased from Wako Pure Chemical and pectic acid from Lancaster Synthesis. Galacturonan disaccharide (GalA-GalA) was purchased from Sigma. The disaccharide is similar to a reaction product (⌬GalA-GalA) released from polygalacturo-  Assays for Enzymes and Proteins-RG lyase YesW was incubated at 30°C for 5 min in a reaction mixture (1 ml) consisting of 0.5 mg/ml RG-I, 50 mM Tris-HCl (pH 7.5), and 2 mM CaCl 2 . Activity was determined by monitoring the increase in absorbance at 235 nm arising from the double bond formed in the reaction products. One unit of enzyme activity was defined as the amount of enzyme required to produce an increase of 1.0 in absorbance at 235 nm per minute by using a cuvette with a light path 1 cm long. Protein content was determined by the method of Bradford (36), with bovine serum albumin as the standard.
Protein Expression and Purification-Protein expression and purification were conducted as described previously (24,37). Briefly, YesW expressed in Escherichia coli cells was purified by two-step column chromatography, i.e. Ni 2ϩ -Chelating Sepharose TM Fast Flow and HiLoad TM 16/60 Superdex TM (200 pg). The N-terminal 37 amino acid residues published in the DNA/ protein data base (YesW, 620 amino acid residues, GenPept accession no. CAB12524) are excised as a signal peptide in E. coli cells (37). The purified enzyme (64 kDa) with a specific activity of 200 units/mg consists of 591 amino acid residues, including the C-terminal histidine tag sequence (8 amino acid residues, LEHHHHHH) derived from the expression vector, pET21b (Novagen).
Crystallization and X-ray Diffraction-After concentration by ultrafiltration using Centriprep (Millipore), purified YesW (24 mg/ml) in buffer containing 20 mM Tris-HCl (pH 7.5), 2 mM CaCl 2 , and 0.2 M NaCl was crystallized at 20°C by sitting-drop vapor diffusion using Linbro tissue-culture plates and microbridges (Hampton Research) (37). The reservoir solution volume in each well was 1 ml, and the droplet was prepared by mixing 5 l of the protein solution with 5 l of the reservoir solution. A crystal suitable for x-ray analysis was obtained after microseeding by using a reservoir solution consisting of 55% 2-methyl-2,4-pentanediol and 0.1 M Tris-HCl (pH 8.4). To obtain a derivative for phasing, the YesW crystal was soaked at 20°C for 3 h in a reservoir solution containing 20 mM KAuCl 4 . The YesW crystal was also soaked at 20°C for 20 h in a sugar solution containing 1.5 M GalA-GalA, 0.1 M Tris-HCl (pH 8.4), and 2 mM CaCl 2 to prepare the enzyme-sugar complex. The native crystal of YesW on a nylon loop (Hampton Research) was placed directly in a cold nitrogen-gas stream at Ϫ173°C, and x-ray diffraction images of the crystal were collected at Ϫ173°C under the nitrogen-gas stream with a Jupiter 210 charge-coupled device detector and synchrotron radiation of wavelength 0.8000 Å at the BL-38B1 station of SPring-8 (Hyogo, Japan).   During data collection for singlewavelength anomalous diffraction (SAD) analysis using the crystal derived with Au 3ϩ , the synchrotron radiation wavelength was adjusted to 1.0357 Å. The wavelength for SAD data collection was defined from x-ray absorption fine-structure (XAFS) measurement of the derivative crystal. The 720 diffraction images (total, 720°) from the derivative crystal with 1.0°oscillation were collected as a consecutive series of datasets. After 360 images were collected, the crystal was vertically translated by 0.5 mm along the long axis to reduce radiation damage. Diffraction data for native and derivative crystals were processed using the HKL2000 program package (38). X-ray diffraction images from the YesW-GalA-GalA complex crystal were collected at 4°C with a Bruker Hi-Star multiwire area detector using CuK␣ radiation generated by a MacScience M18XHF rotating-anode generator, and diffraction data were processed with the SADIE and SAINT programs based on manuals provided by Bruker. Data collection statistics from the native, derivative, and complex crystals are summarized in Table 2.

Phasing and model building
Structure Determination and Refinement-The crystal structure of YesW was solved by SAD using the Au 3ϩ derivative crystal. Au 3ϩ sites and initial phasing were determined by using the SOLVE program (39). Density modification (solvent flattening and histogram matching) and automatic model building were done by the DM (40) and ARP/ wARP programs (41), respectively, in the CCP4 program package. The Coot program (42) was used for manual modification of the initial model. The 50.0-to 1.40-Å resolution dataset was truncated with the CCP4 program package and used for subsequent refinement. Initial rigid body refinement and several rounds of restrained refinement against the dataset were done using the Refmac5 program (43). Water molecules were incorporated where the difference in density exceeded 3.0 above the mean and the 2F o Ϫ F c map showed a density of over 1.0 . At this stage, 2-methyl-2,4-pentanediol molecules and ions were included in the calculation and refinement continued until convergence at 1.4-Å resolution (R ϭ 16.4, R free ϭ 18.2%). A crystal soaked with galacturonan disaccharide was isomorphous with the crystal used for the native set. An F o Ϫ F c map (contoured at 3.0 ) at 2.5-Å resolution was obtained using reflection data for the sugar-soaked crystal, and the phase was calculated from the final sugar-free YesW model. The structure of the enzyme-sugar complex was refined using the Coot program with the parameter file for GalA at the PRODRG site. Protein models were superimposed and their root-mean-square-deviations (r.m.s.d.) were determined with the LSQKAB program (44), a part of CCP4 (45). Final model quality was checked with PRO-CHECK (46). Ribbon plots were prepared using the MOL-SCRIPT (47) and RASTER3D (48), or PyMOL 3 programs. Coordinates used in this report were taken from the RCSB Protein Data Bank (PDB) (50).
Enzyme Kinetics-For determining kinetic parameters (V max and K m ) with various substrates, appropriate concentrations of RG-I, pectin, pectic acid, and polygalacturonan were used. V max and K m were calculated using the Michaelis-Menten equation with KaleidaGraph software (Synergy Software).

RESULTS AND DISCUSSION
Phasing and Structure Determination-As is the case with the native crystal, the Au 3ϩ derivative YesW crystal belongs to space group P2 1 with unit cell parameters of a ϭ 56.9 Å, b ϭ 104.9 Å, c ϭ 99.6 Å, and ␤ ϭ 96.5°. Two molecules are present in an asymmetric unit. For phasing by SAD analysis, 720 diffraction images were collected at up to 1.9-Å resolution from this derivative crystal and processed at a high redundancy of 14.0 with separate Friedel pairs. Data collection statistics are shown in Table 2. The initial phase was determined at 0.333 figureof-merit by the SOLVE program. The phase was improved greatly, and the figure-of-merit increased to 0.836 after density modifications such as solvent flattening and histogram matching with the DM program. The initial model was built to consist of 1,137 amino acid residues with side chains by the ARP/wARP program. Subsequent modification of the initial model through manual modeling and refinement contributed to the protein model with 1,166 amino acid residues except for the C-terminal histidine tag sequence. After initial rigid body refinement, the model was refined against the 50.0-to 1.40-Å resolution dataset. Data collection statistics are summarized in Table 2.
Refined Model Quality-The refined model includes 1,166 amino acid residues, 1,329 water molecules, 11 2-methyl-2,4pentanediol molecules, and 20 calcium ions for two YesW molecules, designated molecules A and B, in an asymmetric unit. All amino acid residues except for the C-terminal regions corresponding to the histidine tag sequence (LEHHHHHH) could be assigned well in the 2F o Ϫ F c map. The final overall R-factor for the refined model was 16.4% with 217,773 unique reflections within a 28.4-to 1.40-Å resolution range. The final overall free R-factor calculated with randomly selected 5% reflection  was 18.2%. Based on theoretical curves in the plot calculated according to Luzzati (51), the absolute positional error was estimated to be 0.14 Å at a resolution of 1.40 Å. Ramachandran plot analysis (52), in which the stereochemical correctness of the backbone structure is indicated by (, ) torsion angles (53), shows that 89.2% of non-glycine residues lie within the most favored and 9.8% in additionally allowed regions. Five residues, Asn-152, Ala-327, Asn-490, Ser-506, and Ala-594, in each molecule (A and B) fell into generously allowed regions. One cis-peptide was observed between Glu-248 and Pro-249 residues. Refinement statistics are summarized in Table 2.
Overall Structure-The overall structure (Fig. 2, A and B) and topology of the secondary structure elements (Fig. 2C)   Structural homologues of YesW were searched for in the PDB using the DALI program (54). Several proteins with a ␤-propeller as a basic scaffold were found to exhibit significant structural homology to YesW (Table 3). Almost all show an eight-bladed ␤-propeller structure. The overall structure of YesW is most similar to that of a ligase, centromere DNA-binding protein (PDB entry 1NEX, Z ϭ 14.6) from Saccharomyces cerevisiae with an r.m.s.d. of 4.0 Å for 297 C ␣ .
Calcium-binding Sites-Ten calcium ions are bound to a YesW molecule. This abundance of calcium ions was supported by the calcium content determined by inductively coupled plas-ma-atomic emission spectrometry (Shimadzu ICPS-8000, Kyoto, Japan). Each blade except for blade D contains one or two calcium ions. Seven of these ions are located in the loops between strands 1 and 2 in each blade. Loops, including calcium ions, are preceded by a ␤-turn. Strand-loop-strand structures in blades A, B, C, F, G, and H, especially aspartic acid residues in the first position of the loop, are well conserved (Fig. 3A; the structure in blade B is shown in the upper inset) and similar to the EF-hand-like calcium-binding motif (70) in alkaline protease from Pseudomonas aeruginosa (71) and integrin ␣ V ␤ 3 from Homo sapiens (72). Blade E contains two calcium ions surrounded by two loops. The loop between strands 1 and 2 is designated as loop 1, and that between strands 3 and 4 is designated loop 2. Asp-369, Asp-371, Asp-373, Lys-375, and Glu-377 residues from loop 1 coordinate with one calcium ion, whereas Asp-371, Asp-373, and Glu-377 from loop 1 and Asp-386 and His-387 from loop 2 coordinate with another calcium ion ( Fig.  3A; lower inset). Loop 1 forms a typical EF-hand-like calciumbinding motif as is seen in other calcium-binding sites (Fig. 3A). In the calcium-binding site in blade E, Asp-371, Asp-373, and Glu-377 from loop 1 coordinate with both calcium ions, enclosing two calcium ions. Although a large number of EF-hand and EF-hand-like calcium-binding motifs have been found, this coordination involved in binding to two calcium ions in blade E is a novel type of calcium-binding motif. No calcium binding is observed in blade D, possibly due to the presence of the first tyrosine residue in the loop in place of an aspartic residue.
The other two calcium ions are located at the opposite side of the ␤-propeller with EF-hand-like motifs. One is coordinated with two side chains and two main chains of amino acid residues, Asp-153, Asn-592, Ala-594, and Asn-596 (Fig. 3B, upper  inset). The other calcium ion is coordinated with four side chains of amino acid residues, His-363, His-399, Asp-401, and Glu-422 (Fig. 3B, lower inset) and located close to the deep cleft.  Active Site Structure-YesW acts on polygalacturonan as well as RG-I (24) (Fig. 1). We determined kinetic parameters (V max and K m ) of YesW with RG-I, pectin, pectic acid, and polygalacturonan (Table 4). This kinetic analysis indicated that YesW acts on RG-I as a preferred substrate, but pectin-derived polysaccharides are also degraded by this enzyme. To identify the active site of YesW, a crystal structure of the enzyme in complex with galacturonan disaccharide (GalA-GalA), a reaction product analogue, was determined.
The refined model in an asymmetric unit consisted of 1,166 amino acid residues and two GalA-GalA molecules. Two identical YesW monomers present in the asymmetric unit were designated as molecules A and B. Each protein molecule consisted of 583 amino acid residues and accommodated a molecule of GalA-GalA in the cavity in the deep cleft, indicating that the active site of YesW is located in the deep cleft at the center of the ␤-propeller (Fig. 4, A and B). These disaccharides in molecules A and B are well fitted in the electron density map ( Fig. 4C) with B-factors of 27.0 and 27.3 Å 2 . Constituent galacturonic acids are designated G1 and G2 from the nonreducing end. The r.m.s.d. between molecules A and B was calculated as 0.524 Å for all residues (583 C ␣ atoms). Because no significant difference exists in the interaction of the enzyme with the disaccharide between molecules A and B, the interaction between YesW molecule A and GalA-GalA is detailed hereafter.
The r.m.s.d. between the apo and GalA-GalA complex structures of YesW was calculated as 0.555 Å for all residues (583 C ␣ atoms), indicating no significant conformational change occurs between protein structures with and without GalA-GalA. Several amino acid residues and water molecules are responsible for binding to GalA-GalA (Fig. 4D). The data in Table 5 presents the interaction of the enzyme with the disaccharide in the complex. Of the seven direct hydrogen bonds between the enzyme and disaccharide atoms, carboxyl groups in G1 and G2 are directly bound to Arg-452, Thr-534, Lys-535, and Tyr-595 through the formation of five hydrogen bonds. The direct interaction with G1 by the three residues Arg-452, Thr-534, and Lys-535 is stricter than that with G2, suggesting that the L-rhamnopyranose residue is accommodated in the G2-binding site in the case of RG-I as a substrate (these chemical formulae are shown in Fig. 1, A and B). Positively charged residues such as Arg-452 and Lys-535 are crucial for binding to acidic polysaccharides and/or neutralizing the negative charge of the carboxyl group. Lys-535 also binds to O5 of G1 and Gln-181 to O1 of G2. In addition to direct hydrogen bonds, six water-mediated hydrogen bonds exist between the enzyme and G2 atoms: O2 ϭ Wat-707 ϭ Thr-534 (3. In addition to hydrogen bond interactions, carbon-carbon (C-C) contacts were also observed between the enzyme and disaccharide atoms (Table 5). Residues such as Arg-452, Thr-534, Lys-535, and Tyr-595 show interactions with the disaccharide through the formation of C-C contacts as well as hydrogen bonds. Tyr-595 is partially parallel to the pyranose ring of G2, indicating that the residue undergoes a partially stacked interaction with the sugar ring. Near the substrate-binding site, a calcium ion is coordinated by four residues, His-363, His-399, Asp-401, and Glu-422, and shows water-mediated interaction with the O6B atom of G2 (Fig. 4D).
RG lyase cleaves ␣-1,4 glycoside bonds between L-rhamnopyranose and GalA in the RG-I molecule through the ␤-elimination reaction and releases unsaturated saccharides with unsaturated GalA at the nonreducing terminus. The catalytic reaction by polysaccharide lyases is divided into three steps as follows: (i) positively charged residues stabilize or neutralize the negative charge on the C-6 carboxylate anion, (ii) a general base catalyst abstracts the proton from C-5 of the uronic acid residue, and (iii) a general acid catalyst donates the proton to the glycoside bond to be cleaved. To identify the role of YesW residues in the active site, we constructed YesW mutants, K535A, R452A, Y595F, D401N, E422Q, H363A, and H399A, in which Lys-535, Arg-452, Tyr-595, Asp-401, Glu-422, His-363, and His-399 were replaced by Ala, Ala, Phe, Asn, Gln, Ala, and Ala, respectively. Kinetic parameters of purified mutants were determined using RG-I as a substrate (Table 6). All the mutants except for Y595F significantly reduced the enzymatic activity (V max /K m ). The increase in K m of K535A and R452A indicates FIGURE 5. Alignment of amino acid sequences of family PL-11 RG lyase. YesW and YesX, RG lyases from B. subtilis strain 168 (GenPept accession no. CAB12524 for YesW and CAB12525 for YesX); Rgl11Y, RG lyase from C. cellulolyticum (GenPept accession no. AAG45161); Rgl11A, RG lyase from C. japonicus (GenPept accession no. AAK20911). Amino acid sequences were aligned using the ClustalW program. Identical and similar amino acid residues in the four RG lyases are denoted by asterisks and dots, respectively. Secondary structure elements of YesW are as shown above. Amino acid residues interacting with GalA-GalA disaccharide are in blue, and those with calcium are in orange. Amino acid residues surrounded with a black square constitute an EF-hand like motif. that the direct interaction between these positively charged residues and the carboxyl group of the substrate is essential for binding to the acidic polysaccharide molecule. A significant decrease of V max was also observed in K535A and R452A. Both mutants K535A and R452A had difficulty in abstracting the proton from C-5 of GalA, possibly due to lack of the interaction with or stability of the substrate carboxyl group. The replacement of Tyr-595 by Phe caused a significant decrease of K m , suggesting that a hydrophobic stacking interaction between the aromatic and pyranose rings is important for substrate binding. The decrease of V max in Y595F suggests that the hydroxyl group of Tyr-595 is important for the catalysis. In D401N, E422Q, H363A, and H399A, a disruption of the coordinate bond between these residues and the calcium ion was thought to occur. The decrease in V max of these mutants probably indicates that YesW shows a calcium-mediated activity as is seen for pectate lyases such as PelC and Pel9A from E. chrysanthemi (20,49). Although no tertiary structure of family PL-4 RG lyase RhgB in complex with the substrate has been determined, a possible active site is proposed based on the sequence alignment of family PL-4 enzymes and pH optimum for the enzyme reaction (33) (Fig. 4E). Residues such as Arg-107, Arg-111, Lys-150, Tyr-203, Tyr-205, His-210, Arg-451, and Arg-455 in RhgB appear to be important for catalytic activity or substrate binding. His-210 is thought to function as a catalytic base. Because the architecture of this putative active site in RhgB differs from that in YesW (Fig. 4, D and E), amino acid residues as base and acid catalysts are probably specific to the PL-4 or -11 families.
Conclusions-This is, to our knowledge, the first report on the structural determination of a family PL-11 RG lyase and establishment of a new structural category in polysaccharide lyases involving an eight-bladed ␤-propeller fold. The substrate-binding site is located in the deep cleft at the center of the ␤-propeller, where positively charged (Lys-535 and Arg-452) and aromatic (Tyr-595) amino acid residues are crucial for binding to the substrate through hydrogen bond formation and stacking interaction. Additionally, a calcium ion coordinated by Asp-401, Glu-422, His-363, and His-399 may be required for the enzyme activity.