Biochemical and structural analyses of a bacterial endo-β-1,2-glucanase reveal a new glycoside hydrolase family

β-1,2-Glucan is an extracellular cyclic or linear polysaccharide from Gram-negative bacteria, with important roles in infection and symbiosis. Despite β-1,2-glucan's importance in bacterial persistence and pathogenesis, only a few reports exist on enzymes acting on both cyclic and linear β-1,2-glucan. To this end, we purified an endo-β-1,2-glucanase to homogeneity from cell extracts of the environmental species Chitinophaga arvensicola, and an endo-β-1,2-glucanase candidate gene (Cpin_6279) was cloned from the related species Chitinophaga pinensis. The Cpin_6279 protein specifically hydrolyzed linear β-1,2-glucan with polymerization degrees of ≥5 and a cyclic counterpart, indicating that Cpin_6279 is an endo-β-1,2-glucananase. Stereochemical analysis demonstrated that the Cpin_6279-catalyzed reaction proceeds via an inverting mechanism. Cpin_6279 exhibited no significant sequence similarity with known glycoside hydrolases (GHs), and thus the enzyme defines a novel GH family, GH144. The crystal structures of the ligand-free and complex forms of Cpin_6279 with glucose (Glc) and sophorotriose (Glc-β-1,2-Glc-β-1,2-Glc) determined up to 1.7 Å revealed that it has a large cavity appropriate for polysaccharide degradation and adopts an (α/α)6-fold slightly similar to that of GH family 15 and 8 enzymes. Mutational analysis indicated that some of the highly conserved acidic residues in the active site are important for catalysis, and the Cpin_6279 active-site architecture provided insights into the substrate recognition by the enzyme. The biochemical characterization and crystal structure of this novel GH may enable discovery of other β-1,2-glucanases and represent a critical advance toward elucidating structure-function relationships of GH enzymes.


Edited by Ruma Banerjee
␤-1,2-Glucan is an extracellular cyclic or linear polysaccharide from Gram-negative bacteria, with important roles in infection and symbiosis. Despite ␤-1,2-glucan's importance in bacterial persistence and pathogenesis, only a few reports exist on enzymes acting on both cyclic and linear ␤-1,2-glucan. To this end, we purified an endo-␤-1,2-glucanase to homogeneity from cell extracts of the environmental species Chitinophaga arvensicola, and an endo-␤-1,2-glucanase candidate gene (Cpin_6279) was cloned from the related species Chitinophaga pinensis. The Cpin_6279 protein specifically hydrolyzed linear ␤-1,2-glucan with polymerization degrees of >5 and a cyclic counterpart, indicating that Cpin_6279 is an endo-␤-1,2-glucananase. Stereochemical analysis demonstrated that the Cpin_6279-catalyzed reaction proceeds via an inverting mechanism. Cpin_6279 exhibited no significant sequence similarity with known glycoside hydrolases (GHs), and thus the enzyme defines a novel GH family, GH144. The crystal structures of the ligand-free and complex forms of Cpin_6279 with glucose (Glc) and sophorotriose (Glc-␤-1,2-Glc-␤-1,2-Glc) determined up to 1.7 Å revealed that it has a large cavity appropriate for polysaccharide degradation and adopts an (␣/␣) 6 -fold slightly similar to that of GH family 15 and 8 enzymes. Mutational analysis indicated that some of the highly conserved acidic residues in the active site are important for catalysis, and the Cpin_6279 activesite architecture provided insights into the substrate recognition by the enzyme. The biochemical characterization and crystal structure of this novel GH may enable discovery of other ␤-1,2-glucanases and represent a critical advance toward elucidating structure-function relationships of GH enzymes.
Glycoside hydrolase (GH) 2 is a general term for enzymes that catalyze hydrolytic breakdown of the glycosidic linkages of gly-cosides, glycans, and glycoconjugates. GHs are distributed in almost all organisms and play important roles not only in various biological phenomena but also industrial processes. These enzymes are categorized into GH families based on their amino acid sequences in the Carbohydrate-Active enZyme (CAZymes) database, forming numerous classes in the database (1)(2)(3)(4)(5). The number of GH families has been increasing and currently ranges from GH1 to more than GH140. However, whereas GHs acting on ubiquitously available sugars have been well characterized, there have been few studies on GHs acting on sugars that are difficult to obtain.
Recently, we established a method of a large-scale enzymatic synthesis of linear ␤-1,2-glucan from inexpensive sugars, which makes it easier to study ␤-1,2-glucan-related enzymes and proteins (23). From now on, we denote linear ␤-1,2-glucan simply as "␤-1,2-glucan" unless otherwise noted. In this report, we propose a new GH family of bacterial endo-␤-1,2-glucanases on the basis of the structural and functional characteristics of the C. arvensicola and Chitinophaga pinensis enzymes. Furthermore, we determined the crystal structures of the C. pinensis enzyme in a ligand-free form and a complex form with Sop 3 and Glc.
43-kDa protein was purified to homogeneity from a cell extract of C. arvensicola in fraction 9 ( Fig. 2A). The protein showed endo-␤-1,2-glucanase activity but did not show ␤-glucosidase activity even in the absence of GDL (Fig. 2B).

Sequence analysis
The 43-kDa protein was subjected to in-gel digestion to obtain its internal peptides. Based on a BLAST search limited to C. pinensis using the sequences constructed as described under "Experimental procedures," five sequences matched the same hypothetical protein (locus tag; Cpin_6279) ( Table 1). C. pinensis did not have paralogs of Cpin_6279. Cpin_6279 showed no sequence similarity to known GHs, although this protein is annotated as a glycoamylase (PF10091) in the Pfam database (25), implying that Cpin_6279 belongs to a novel GH family (described below).
The protein sequence of Cpin_6279 available in the RefSeq database (accession number; WP_044220121) consists of 457 amino acids. The N-terminal sequence of the 43-kDa protein purified from C. arvensicola (QVARATLAFDRT) corresponds to the sequence starting from the 32nd amino acid of Cpin_6279, whose N-terminal 18 amino acids were predicted to be a signal sequence by the SignalP 4.1 Server (26). This suggests that the predicted signal sequence plus 13 amino acids had been cleaved during the inner membrane translocation and protein purification. Indeed, the ␤-1,2-glucan-degrading activity of C. pinensis was not detected in culture filtrates (Fig. 1B). Thus, Cpin_6279 may be localized in the periplasmic space. The ␤-glucosidase activity in cell extracts seems to be derived from a GH3 LiBGL homolog (Cpin_1816), which contains a signal peptide (19). Cpin_6279 and Cpin_1816 seem to be involved in degradation of ␤-1,2-glucan in C. pinensis.

Action patterns of Cpin_6279 on ␤-1,2-glucan and sophorooligosaccharides
The action patterns on linear and cyclic ␤-1,2-glucans were analyzed by TLC. Both ␤-1,2-glucans were broken down into Sop 2 to Sop 5 at the early stage of the reaction (Fig. 5, A and B). These results confirmed that Cpin_6279 is an endo-type GH. Finally, both ␤-1,2-glucans were completely broken down into Sop 2 to Sop 5 in 60 min in contrast to the case of the partially A, fractionated proteins were subjected to SDS-PAGE, followed by silver staining (bottom), and then the enzymatic activity was measured (top). Lane M contains protein standard markers. Endo-␤-1,2-glucanase activity was evaluated as the ratio of ␤-1,2-glucan-hydrolyzing activity to ␤-glucosidase activity. The enzymatic reaction was conducted in a mixture (50 l) containing 80% (v/v) enzyme fraction, 0.2% (w/v) ␤-1,2-glucan (average DP 64), and 50 mM MOPS-NaOH buffer (pH 6.5) at 30°C for 16 h. An aliquot of the reaction mixture (40 l) was mixed with 160 l of a 1% (w/v) PAHBAH-HCl solution and then heated at 100°C for 5 min, followed by measurement of absorbance at 405 nm. The arrow indicates the protein band subjected to N-terminal amino acid sequence analysis. B, TLC analysis of endo-␤-1,2-glucanase in fraction 9. The enzymatic reaction was carried out in a mixture (10 l) containing 85% (v/v) fraction 9 concentrated using Amicon Ultra 30,000 molecular weight cutoff (Millipore), 0.2% (w/v) ␤-1,2-glucan (average DP 64), and 50 mM MOPS-NaOH buffer (pH 6.5) at 30°C for 16 h, and then an aliquot of the mixture was spotted onto a TLC plate. Lane M contains markers (1 l of 0.2% (w/v) each sugar). The asterisk indicates the origin on the TLC plate. Minus and plus indicate whether fraction 9 was added to the reaction mixture or not.
The action patterns on Sop n s were also analyzed by TLC. All tested Sop n s were decomposed into oligosaccharides with DP of 2-5 ( Fig. 5, C-E). The hydrolytic velocities of Cpin_6279rC for Sop 6 and Sop 7 were higher than that for Sop 5 . Notably, Sop 7 is predominantly cleaved into Sop 3 and Sop 4 , suggesting that the subsites of the enzyme extend from Ϫ3 to ϩ4 or Ϫ4 to ϩ3. Sop 3 and Sop 4 were not degraded (data not shown). Based on HPLC analysis, the specific activities as to Sop 5 , Sop 6 , and Sop 7

Subsite characterization of Cpin_6279
To determine the detailed action pattern of Cpin_6279, we examined the incorporation of 18 O atoms into the newly produced reducing ends using electrospray ionization (ESI)-MS after enzymatic hydrolysis of Sop 7 in the presence or absence of 18 O-labeled water (Fig. 7). The observed hydrolytic patterns are schematically shown in Fig. 7A 18 O was calculated from the ratio of the peak area of each 18 Oincorporating oligosaccharide to the peak area of the total amount of the corresponding oligosaccharide produced. These results demonstrate that Cpin_6279 mainly binds Sop 7 at subsites Ϫ4 to ϩ3 as shown in pattern I and suggest that recognition at subsite ϩ4 is weaker than that at subsite Ϫ4. We also investigated the hydrolytic pattern when Sop 7 was decomposed into Sop 2 and Sop 5 (Fig. 7, F and G showed a sufficiently high ratio of 78%, indicating that Cpin_6279 can also bind to subsites Ϫ5 to ϩ2, as shown in pattern III but not Ϫ2 to ϩ5. Therefore, subsite Ϫ3 is essential for binding.

Kinetic analysis
The kinetic parameters of Cpin_6279rC were determined using ␤-1,2-glucan and Sop 5 as substrates ( Table 2). The k cat and K m values were sufficiently high and low compared with those of other GHs. The higher k cat value and the lower K m value for ␤-1,2-glucan than those for Sop 5 imply the preference of more than six subsites on the enzyme, consistent with the HPLC analysis where the specific activities increased in proportion to DP.

Stereochemistry of hydrolysis catalyzed by Cpin_6279
To determine the stereochemical course of the Cpin_6279catalyzed hydrolysis, we performed 1 H NMR using ␤-1,2-glucan as a substrate (Fig. 8). Standard spectra (Sop 2-5 and ␤-1,2glucan with average DP 25) are shown in Fig. 8B. A reference spectrum (t ϭ 0 min) was recorded prior to the addition of the enzyme to the reaction mixture. Because the signals of the anomeric axial protons of liberated Sop n s (H1␤), which appeared around ␦ 4.7, overlapped with those of water and anomeric axial protons at internal glucosides, we compared the signals of the anomeric equatorial protons of liberated Sop n s (H1␣, ␦ 5.4) with the signals of the C2 protons of the non-reducing end glucoside in the liberated Sop n s (H2 NR , ␦ 3.35) regardless of the anomeric configuration. After the addition of the enzyme, the H1␣ signals increased along with the H2 NR signals (t ϭ 4 and 27 min) (Fig. 8, C and D). As the reaction proceeded (t ϭ 55 and 105 min and 24 h), the ratio of the H1␣ signals to the internal standard decreased, which is attributed to the mutarotation of anomers. The ratio of the H2 NR signals to the internal standard gradually increased at the late stage of the reaction (Fig. 8, C and D), which seems to be due to slower hydrolysis of Sop 5 . This finding indicates that Cpin_6279 is an inverting GH.
We also carried out polarimetric analysis for easier identification of the reaction mechanism (Fig. 8E). The observed optical rotation increased to a positive value on addition of Cpin_6279rC, quickly decreased after addition of an ammonia solution, and then reached equilibrium. This indicates that optical rotation derived from the reducing end ␣-anomer in ␤-1,2-glucan is greater than that from the ␤-anomer, as is the , and glycine-NaOH (pH 8.6 -10, closed rhombi). The highest activity was defined as 100% (stability, pH 6.5; and optimum, pH 6.0). B, highest activity was defined as 100% (50°C). As for the enzymatic reaction at 60 and 70°C, the data from 0 to 1.5 min were used to calculate the initial velocity.

Crystal structure of Cpin_6279
We first determined the ligand-free structure of Cpin_ 6279rC at 1.8 Å resolution (Table 3). Although these data were collected for a crystal soaked in a solution supplemented with Sop 3 (a main reaction product) as a cryoprotectant, the sugar ligand was not bound to the protein, probably due to crystal packing hindrance by the C-terminal His 6 tag (described below). The complex structure of N-terminally His 6 -tagged Cpin_6279 (Cpin_6279rN) with Sop 3 and Glc was next determined at 1.7 Å resolution by a co-crystallization method (Fig.  9A). Although Glc was not added to the crystallization sample at any preparation step, weak electron density of a Glc monosaccharide, which seems to be derived from Sop 3 , was observed in the binding groove ( Fig. 9, B and C). These ligand-free (Cpin_6279rC) and Sop 3 -Glc complex (Cpin_6279rN) crystals contained two molecules per asymmetric unit and belonged to space groups P3 1 and C2, respectively. The primary structure of the ligand-free form (Cpin_6279rC) consists of 449 residues, and the amino acid region in the determined structure extends from 24 to 449 in both chains. The C-terminal His 6 tags on both chains (residues 444 -449) were visible in the electron density map. The primary structure of the complex form (Cpin_6279rN) consists of 461 residues, and the amino acid region in the determined structure extends from 43 to 460 (corresponding to 23-440 in Cpin_6279rC) in both chains. The root mean square deviations (r.m.s.d.) for the C␣ atoms between all pairs of four molecules (two chains in the two structures) are within 0.1 Å. Hereafter, we focus on chain A unless otherwise noted. The overall structure exhibits an (␣/␣) 6 -barrel (Fig. 9A). A surface model depicts a cleft structure suitable for accommodating a large substrate (Fig. 9B). A structural homology search using the DALI server revealed that the structure of Cpin_6279 is very similar to those of homologous hypothetical proteins BF9343_0330 from Bacteroides fragilis (PDB code 3EU8), BACCAC_03554 from Bacteroides caccae (PDB code 4QT9), and BACUNI_03963 from Bacteroides uniformis (PDB code 4GL3) determined by means of structural genomics. The r.m.s.d. values for C␣ atoms are less than 1.3 Å, and the Z-scores are more than 63, reflecting their high sequence identity of ϳ67% with Cpin_6279. The structures of these homologous proteins also show the large cleft observed in Cpin_ 6279. In addition, the structure of Cpin_6279 exhibits modest similarity (r.m.s.d. values of 3.5-4.0 Å and Z-scores of Ͻ 20) but low sequence identity (Ͻ12%) with GH15 glucoamylase, cellobiose 2-epimerase, GH8 chitosanase, N-acetylglucosamine 2-epimerase, GH8 endo-␤-1,4-glucanase, and GH126 ␣-amylase.

Active-site architecture
In the complex structure, Sop 3 and Glc were observed at the center of the barrel (Fig. 9C). All Glc residues adopt a standard 4 C 1 conformation, and the C6 hydroxyl group is in a gauchegauche orientation. The hydroxyl groups of Sop 3 form hydro-   18 O-labeled water, respectively. The values above the peaks represent the molecular weights of the reaction products when using H 2 18 O. The asterisks denote the peaks derived from natural 13 C-containing oligosaccharides. Schematic diagrams in parentheses represent the minor products.
central Glc unit of Sop 3 . The O3 and O4 hydroxyl groups of the Glc form bifurcated hydrogen bonds with the ⑀-amino group of Arg-55. O4 and O6 of Glc form direct and water-mediated hydrogen bonds with Glu-54. In addition, O1 of Glc is hydrogen bonded to a water molecule. In the ligand-free form structure, the side chain of Glu-142 slightly turns downward and that of Arg-55 turns to a slightly distal position compared with the complex structure (Fig. 9C). The displacements in the ligandfree form structure are partly due to the crystal packing of the P3 1 space group crystal, in which the active site is completely blocked with the C-terminal His 6 tag from a symmetry mate molecule. In the binding pocket, six acidic amino acid residues (Glu-54, Asp-135, Asp-139, Glu-142, Glu-211, and Asp-400) are highly conserved in Cpin_6279 homologs (Figs. 9B and 10), indicating that these residues are candidates for the catalytic residues. Only Asp-400 does not form any hydrogen bonds with the Sop 3 or Glc molecule. The aromatic rings of Phe-131, Trp-192, Phe-204, Trp-269, and Tyr-330 are involved in hydrophobic interactions.

Mutational analysis
To predict the catalytic residues, we carried out mutational analysis by replacing the highly conserved acidic amino acid residues in the active site with isosteric residues (Asn or Gln) ( Table 4). Mutations at Asp-139, Glu-142, and Glu-211, which are located around the non-reducing end of Sop 3 (Fig. 9C), reduced the specific activity to Ͻ0.2% compared with the wild type. As for E54Q and D135N, the specific activity also decreased, but it was not as critical as the above three mutants. D400N was less effective on the specific activity, consistent with the structural observation that Asp-400 does not participate in any ligand recognition (Fig. 9C).

Discussion
Although several studies on bacterial and fungal endo-␤-1,2glucanases were reported over 30 years ago (20 -22), their genes have not been identified. In this study, we could purify a bacterial endo-␤-1,2-glucanase and clone its gene from a Chitinophaga species by using enzymatically synthesized linear ␤-1,2glucan. Biochemical and structural characterizations were carried out using the recombinant Cpin_6279 protein, revealing that the enzyme has at least seven subsites ranging from Ϫ4 to ϩ3 in a long cleft of the (␣/␣) 6 barrel molecule. Because Cpin_6279 showed no significant amino acid sequence similarity with any known enzymes, the enzyme and its homologs will be classified as a novel GH family, GH144. 3 Among the uncharacterized proteins homologous to Cpin_6279, only the ligandfree crystal structures of three hypothetical proteins from Bac-teroides have become available by means of structural genomics. These hypothetical proteins were incorrectly annotated as glucoamylases because they were classified into the glycoamylase family (PF10091) by Pfam. Cpin_6279 showed no glucoamylase activity (data not shown). Therefore, the structure of the complex with Sop 3 is important evidence that shows the correct function of Cpin_6279.
Bioinformatic analysis of Cpin_6279 homologs showed that the homologous proteins are mainly distributed in Gram-negative bacteria and not in eukaryotes. A phylogenetic tree illustrates that the proteins in the new GH family are mainly from the Bacteroidetes and Proteobacteria and can be divided into distinct clades (Fig. 11). Notably, not only soil bacteria but also gut bacteria such as Bacteroides thetaiotaomicron, Bacteroides fragilis, and Bacteroides ovatus have a homologous protein.
C. pinensis does not have a gene cluster for dissimilation of ␤-1,2-glucan, although the genomes of some other bacteria encode such gene clusters. Many members of the Bacteroidetes harboring Cpin_6279 homologs seem to use the SusC (TonBdependent receptor)/SusD (carbohydrate-binding protein)like system for the catabolism of ␤-1,2-glucan. These susCDlike genes are located around the Cpin_6279 homolog gene, which is a feature of the Bacteroidetes for degradation of complex polysaccharides known as polysaccharide utilization loci (29). Other bacteria appear to utilize ABC proteins or the other systems for dissimilation. In addition to these genes, GHs such as LiBGL homologs (GH3) and LiSOGP homologs (GH94) and LacI transcription factors tend to form gene clusters together with these uptake systems. Functional predictions suggest these proteins are involved in the synergetic decomposition of ␤-1,2-glucan.
Some cyclic ␤-1,2-glucan-producing bacteria such as Agrobacterium tumefaciens and Rhizobium meliloti possess a Cpin_6279 homolog. In a number of bacterial exopolysaccharide secretion systems, a GH or polysaccharide lyase is found in the operon of that system. These degrading enzymes are suggested to be important for efficient production of exopolysaccharides, clearing the residual exopolysaccharides in the periplasmic space and forming a scaffold for secretion of the polymers into the extracellular space (30 -33). The cellular concentrations of cyclic ␤-1,2-glucan in Rhizobium and Agrobacterium species range from 5 to 20% of the whole-cell dry weight, and the concentration depends on the species, medium composition, and growth phase (7). Although almost all the Cpin_6279 homologs from root nodule bacteria do not form a gene cluster with cyclic ␤-1,2-glucan synthase and related proteins, they may play a role in clearance of the residual cyclic ␤-1,2-glucan in the periplasm.   The crystal structure of Cpin_6279 bound with Sop 3 and Glc provided structural evidence for binding of Sop n s. Considering the preferable binding of Sop 7 to subsites Ϫ4 to ϩ3 suggested by ESI-MS analysis (Fig. 7A), the Sop 3 and Glc observed in the crystal structure may occupy subsites ϩ1 to ϩ3 and subsite Ϫ3, respectively (Fig. 9C). In the binding cleft, highly conserved Glu-54, Asp-135, Asp-139, Glu-142, Glu-211, and Asp-400 are located (Fig. 9B). These residues except for Asp-400 form direct or indirect hydrogen bonds with the hydroxyl groups of Sop 3 and Glc (Fig. 9C). Mutation at these residues, especially at Asp-139, Glu-142, and Glu-211, obviously decreases the activity ( Table 4), suggesting that these residues play an important role in catalysis. However, none of the three residues are in the direct hydrogen transfer distance range as to the O2 atoms of all the Glc units in Sop 3 (Fig. 9C) or are proximal to space between Sop 3 and Glc. Fig. 12 shows a comparison of the topological positions of the conserved acidic residues in Cpin_6279 with the catalytic residues in two representative inverting GHs (GH15 and GH8) having an (␣/␣) 6 -barrel fold. It has been known that the majority of inverting GHs with an (␣/␣) 6 -barrel, including GH15 (Fig. 10B), possess catalytic residues at loops between ␣5 and ␣6 (catalytic acid) and between ␣11 and ␣12 (catalytic base), whereas several families possess them at circularly permutated positions (34). GH8 (Fig. 10C) and GH48 have catalytic residues in loops of ␣1-␣2 (catalytic acid) and ␣7-␣8 (catalytic base), and GH9 has them in ␣11-␣12 (catalytic acid) and ␣5-␣6 (catalytic base). Glu-211 (␣5-␣6) of Cpin_6279 is topologically equivalent with the general acid residue in GH15 (Fig. 10A). However, Glu-211 does not appear to be involved in proton transfer to the glycosidic bond but anchors the central Glc unit of Sop 3 (Fig. 9C). The positions of Asp-139 and Glu-142 (␣3-␣4) are unprecedented in known inverting GHs. These observations may imply a non-canonical reaction mechanism for the novel GH enzymes, as in the case of several exceptional inverting GHs such as GH6 Cel6A (35), GH45 PcCel45A (36), GH55 Lam55A (37), GH95 ␣-1,2-L-fucosidase (38), and GH124 Cel124A (39). Presumption of the positions of subsite Ϫ1 and a nucleophilic water is difficult from the complex structure in this study.
C. pinensis is a Gram-negative bacterium that exhibits potential as to the discovery of novel CAZymes (24,40). This bacterium possesses diverse carbohydrate-active enzymes including 191 GHs that can be classified into 56 different families. In addition, C. pinensis shows poor growth on cellulose and starch, which are the most abundant sugar resources on earth. Therefore, it is expected that novel GHs will be further found in this organism.
Studies on endo-␤-1,2-glucanase did not greatly advance from 1987 until recently because it had been difficult to prepare large amounts of ␤-1,2-glucan. The identification, characterization and crystal structure of this enzyme will facilitate further discovery of ␤-1,2-glucan-related enzymes and help us better understand the structure-function relationships of GH enzymes. Moreover, our study will provide solid foundations for a wide range of research concerning ␤-1,2glucan and Sop n s.

Strain and culture conditions
C. arvensicola (JCM2839) and C. pinensis (DSM2588) were purchased from the Japan Collection of Microorganisms (Ibaraki, Japan), and the Leibniz Institute DSMZ-German Collection of Microorganisms and Cell Cultures (Braunschweig, Germany), respectively.
C. arvensicola was pre-cultured in 5 ml of YP medium (1% polypeptone, 0.5% yeast extract, and 0.5% NaCl, adjusted to pH 7.2) at 30°C for 24 h. C. pinensis was pre-cultured in 5 ml of CY medium (0.3% casitone, 0.14% CaCl 2 , 2H 2 O, and 0.1% yeast extract, adjusted to pH 7.2) at 25°C for 48 h. For investigation of the ␤-1,2-glucan-hydrolyzing activity, these pre-cultured bacteria were inoculated into 5 ml of chemically defined medium (22) containing 0.5% ␤-1,2-glucan (average DP 64) or 0.5% Glc as a sole carbon source. C. arvensicola was grown at 30°C for 24 h. C. pinensis was grown at 30°C for 2 days (the Glc-containing medium) or 12 days (the ␤-1,2-glucan-containing medium). These bacteria were collected by centrifugation at 13,000 ϫ g for 4 min, and the collected cells were suspended in 500 l of 20 mM sodium phosphate (pH 7.0). The cells were then disrupted by sonication and centrifuged at 20,000 ϫ g for 10 min to obtain cell extracts. The culture filtrates and cell extracts were concentrated and buffered with 20 mM MOPS-NaOH (pH 7.0) using Amicon Ultra 10,000 molecular weight cutoff (Millipore, Billerica, MA) to remove low molecular weight compounds such as Glc in the medium. For purification of the endo-␤-1,2-glucanase, the pre-cultured C. arvensicola was inoculated into 1 liter of chemically defined medium containing 0.5% ␤-1,2-glucan (average DP 25) and then grown at 30°C for 24 h.

Purification of endo-␤-1,2-glucanase from C. arvensicola
C. arvensicola cultured in 1 liter of chemically defined medium was harvested by centrifugation at 16,000 ϫ g for 15 min, and the harvested cells (3.8 g) were suspended in 20 ml of 5 mM MOPS-NaOH buffer (pH 7.5) (buffer A). The cells were disrupted by sonication and then centrifuged at 27,000 ϫ g for 10 min to obtain a cell extract. The cell extract was applied onto an anion-exchange column (DEAE-Sepharose CL-6B (10 ml); GE Healthcare, Buckinghamshire, UK) pre-equilibrated with buffer A. After the column had been washed with buffer A, bound proteins were eluted with a 100-ml linear gradient of 0 -500 mM NaCl in buffer A at a flow rate of 2.0 ml/min and collected in 2.5-ml aliquots. The fractions containing ␤-1,2glucanase activity were pooled and dialyzed against 5 mM sodium phosphate buffer (pH 6.5) (buffer B). The solution was applied onto a hydroxyapatite column (Bio-Scale MiniCHT Ceramic Hydroxyapatite Cartridges (5 ml); Bio-Rad) pre-equil- Figure 11. Phylogenetic tree of a novel GH family. Sequences for phylogenetic analysis were retrieved for each genus in the KEGG database and confined to one paralog per species. After multiple sequence alignment using MUSCLE had been performed (57), a phylogenetic tree was constructed using MEGA7 (58) based on the neighbor-joining method. Species harboring the gene of a Cpin_6279 homolog are categorized into phyla and surrounded by solid lines. The organism possessing the gene cloned in this study is shown with a black background and white letters. Cpin_6279 is denoted by a closed circle. ibrated with buffer B. After the column had been washed with buffer B, bound proteins were eluted with a 50-ml linear gradient of 5-500 mM sodium phosphate buffer (pH 6.5) at a flow rate of 2.0 ml/min and collected in 2.5-ml aliquots. The fractions containing ␤-1,2-glucanase activity were concentrated using Amicon Ultra 30,000 molecular weight cutoff (Millipore). The concentrated fraction was applied onto a Superdex 200 column (10/300 GL; GE Healthcare) pre-equilibrated with 50 mM sodium phosphate buffer (pH 6.5) containing 150 mM NaCl at a flow rate of 1.0 ml/min and collected in 0.5-ml aliquots. The fractions containing ␤-1,2-glucanase activity were applied onto a hydroxyapatite column again as described above except that the elution was carried out with a 30-ml linear gradient, and the eluate was collected in 1.0-ml aliquots. Ä KTA (GE Healthcare) was used for all fractionations. The fractionated proteins were subjected to SDS-PAGE, followed by staining with 2D-Silver Stain II (COSMO BIO Co., Ltd., Tokyo, Japan) or Coomassie Brilliant Blue R-250 (Nacalai Tesque, Kyoto, Japan). Precision Plus Protein TM unstained standards (Bio-Rad) were used as protein markers.

Measurement of hydrolytic activity during purification
␤-1,2-Glucan-hydrolyzing activity (the sum of the activities of ␤-1,2-glucanase and ␤-glucosidase) was determined by measuring the increase in reducing ends using the p-hydroxybenzoic acid hydrazide (PAHBAH) method (41). A reaction mixture (50 l) containing 40% (v/v) enzyme fraction, 0.2% (w/v) ␤-1,2-glucan (average DP 64), and 50 mM MOPS-NaOH buffer (pH 6.5) was incubated at 30°C for 24 h and then heated at 100°C for 5 min to stop the reaction. An aliquot of the reaction mixture (20 l) was mixed with 120 l of a 1% (w/v) PAHBAH-HCl solution and heated at 100°C for 5 min, followed by measurement of absorbance at 405 nm on a Spectramax 190 (Molecular Devices) with a 96-well microplate (EIA/RIA plate, 96-well half-area; Corning, NY). These instruments were used for all spectrophotometric measurements. During the purification from the cell extract of C. arvensicola, ␤-1,2-glucan-hydro-lyzing activity was defined as the amount of released reducing ends (Glc equivalence) in ϳ24 h based on a Glc standard curve.
Endo-␤-1,2-glucanase activity was analyzed by TLC in the presence of GDL, a known ␤-glucosidase inhibitor (42). The enzymatic reaction was carried out in the reaction mixture described above in the presence of 50 mM GDL, and an aliquot of the reaction mixture was analyzed by TLC.
␤-Glucosidase activity was determined by measuring the increase in pNP using pNP-␤-Glc as a substrate. A reaction mixture (100 l) containing 40% (v/v) enzyme fraction, 5 mM pNP-␤-Glc, and 50 mM MOPS-NaOH buffer (pH 6.5) was incubated at 30°C for 24 h, and an aliquot of the reaction mixture (20 l) was mixed with 180 l of 0.2 m Na 2 CO 3 . The increase in p-nitrophenol (pNP) was calculated from absorbance at 405 nm based on a pNP standard curve. ␤-Glucosidase activity was defined as the amount of released pNP in 24 h based on a pNP standard curve.

Sequence analysis
The protein band of the 43-kDa candidate for the target enzyme was transferred to a Sequi-Blot TM PVDF membrane (Bio-Rad) and then stained with Coomassie Brilliant Blue. The N-terminal sequence of the protein was analyzed by GenoStaff Co., Ltd. (Tokyo, Japan). Internal sequences of the protein were analyzed as described below. The protein detected on SDS-PAGE was routinely digested with trypsin (Promega, WI) or chymotrypsin (Promega). The obtained peptides were subjected to LC/MS/MS analysis as described in the previous report (43). The peptide sequences of the target enzyme were analyzed with an in-house MASCOT server as described in the previous report (43) except that the search parameters were as follows: taxonomy, other bacteria, and fragment mass tolerance Ϯ0.6 Da. The peptide sequences were also analyzed by de novo sequencing using SPIDER 5 (44) on PEAKS Online 6 (Bioinformatics Solutions Inc., Waterloo, Canada) (45). From the trypsinized peptides, we selected ones that possessed more than six sequential amino acids with confidence above 80% Glu-54 is located between ␣1 and ␣2. Asp-135, Asp-139, and Glu-142 are located between ␣3 and ␣4. Glu-211 is located between ␣5 and ␣6. Asp-400 is located between ␣11 and ␣12. B, overall structure of GH15 glucoamylase (PDB code 1AGM). Glu-179 (catalytic acid) is located between ␣5 and ␣6. Glu-400 (catalytic base) is located between ␣11 and ␣12. C, overall structure of GH8 endoglucanase (PDB code 1KWF). Glu-95 (mutated to Gln-95, catalytic acid) is located between ␣1 and ␣2. Glu-278 (catalytic base) is located between ␣7 and ␣8.
with SPIDER 5. Then we searched for peptides from chymotrypsinized peptides that have more than five sequential common amino acids with the above selected trypsinized peptides. The internal sequences of the purified enzyme were constructed by combining these peptides. All amino acids that differ between these two peptides were arranged based on either trypsinized or chymotrypsinized peptides.

Cloning, expression, and purification
The gene encoding the Cpin_6279 protein (GenBank TM accession number ACU63684.1) was amplified by PCR (94°C for 2 min and then 30 cycles of 94°C for 15 s, 40°C for 30 s, and 68°C for 2 min) with a forward primer 5Ј-CATACCACATAT-GATGTCATGTGGAGGTTC-3Ј (NdeI site underlined) and a reverse primer 5Ј-ATCAACTCGAGCTTCAGGTAAGGG-CTCTG-3Ј (XhoI site underlined) using TaKaRa ExTaqHS (Takara Bio, Shiga, Japan) and the genomic DNA as a template. The genomic DNA of C. pinensis was extracted from the cells with InstaGene Matrix (Bio-Rad). The amplified Cpin_6279 gene was inserted between the XhoI and NdeI sites of the pET-30a vector (Novagen, Madison, WI) to add a His 6 tag to the target protein at the C terminus. For preparation of Cpin_6279rN, the Cpin_6279 gene was amplified by PCR (30 cycles of 94°C for 15 s, 60°C for 30 s, and 68°C for 2 min) with a forward primer (as described above) and a reverse primer 5Ј-ATCAACTCGAGTCACTTCAGGTAAGGGCT-3Ј (XhoI site underlined) using KOD Plus (TOYOBO, Osaka, Japan) and pET-30a inserted with Cpin_6279 as a template. The amplified gene was inserted to the corresponding site of the pET-28b vector (Novagen).
The constructed plasmid was transformed into E. coli BL21(DE3), and the transformant was cultured in 1 liter of Luria-Bertani medium containing 30 mg/liter kanamycin at 37°C until the absorbance at 600 nm reached 0.6. Then, expression was induced with 0.1 mM isopropyl ␤-D-1-thiogalactopyranoside, and the transformant was further cultured at 20°C for 24 h. The cells were centrifuged at 3,900 ϫ g for 10 min and then suspended in 20 mM MOPS-NaOH buffer (pH 7.0) containing 500 mM NaCl (buffer C). The suspended cells were disrupted by sonication and centrifuged at 27,000 ϫ g for 15 min to obtain a cell extract. The cell extract was applied onto a HisTrap FF crude column (5 ml; GE Healthcare) pre-equilibrated with buffer C containing 20 mM imidazole. After the column had been washed with the same buffer, the target protein was eluted with a linear gradient of 20 -300 mM imidazole in buffer C. Amicon Ultra 30,000 molecular weight cutoff (Millipore) was used for the concentration of a portion of the fractionated protein and for exchanging the buffer in the protein solution to 20 mM MOPS-NaOH (pH 6.5).
As for the crystallization sample of Cpin_6279rC, ammonium sulfate was added to the peak fractions obtained on nickel affinity chromatography to 40% saturated concentration after the fractions had been collected and diluted to 0.5 mg/ml with 20 mM MOPS-NaOH (pH 6.5). The enzyme solution was applied onto a HiTrap Butyl HP column (5 ml; GE Healthcare) pre-equilibrated with 20 mM MOPS-NaOH (pH 6.5) containing 40% saturated ammonium sulfate. After the column had been washed with the same buffer, the target protein was eluted with a linear gradient of 40 to 0% saturated concentration of ammonium sulfate. Appropriate fractions were concentrated and buffered with 5 mM MOPS-NaOH (pH 6.5).
For preparation of Cpin_6279rN for crystallization, hydrophobic interaction chromatography was performed as described above except that 20 mM MOPS-NaOH (pH 7.0) and 30% saturated ammonium sulfate were used for purification instead of 20 mM MOPS-NaOH (pH 6.5) and 40% saturated ammonium sulfate, respectively. Then, Cpin_6279rN was applied onto a Superdex 200 pg 16/60 column (GE Healthcare) pre-equilibrated with 20 mM MOPS-NaOH (pH 7.0) containing 150 mM NaCl. Appropriate fractions were concentrated and buffered with 10 mM MOPS-NaOH (pH 7.0). Protein concentrations were determined by measurement of the absorbance at 280 nm, and calculation was based on the theoretical extinct coefficients of Cpin_6279rC and Cpin_6279rN (134,815 M Ϫ1 ⅐cm Ϫ1 ). The purity was checked by SDS-PAGE, and Dyna-Marker Protein MultiColor Stable (BioDynamics Laboratory Inc., Tokyo) was used as a protein marker.

Hydrolytic activity of the recombinant enzyme
The substrate specificity for various polysaccharides was determined by measuring the increase in reducing ends using PAHBAH. The enzymatic reaction was performed in a reaction mixture (200 l) containing 4.0 g/ml Cpin_6279rC, 0.2% (w/v) ␤-1,2-glucan (average DP 64), and 50 mM MES-NaOH buffer (pH 6.0) at 30°C for 10 min. These conditions were defined as the standard conditions. To investigate the substrate specificity of Cpin_6279rC, the enzymatic reaction was performed under the standard conditions using an appropriate concentration of the enzyme, by substituting 0.2% ␤-1,2-glucan with 0.2% barley ␤-glucan, 0.0125% lichenan, 0.0125% laminarin, 0.006% arabinogalactan, 0.05% CM-curdlan, 0.025% CM-pachyman, 0.1% konjac glucomannan, 0.2% xyloglucan (tamarind), 0.1% polygalacturonate, 0.1% CM-cellulose, 0.05% pustulan, or 0.2% arabinan. Then, an aliquot of the reaction mixture (35 l) was mixed with 105 l of a 1% (w/v) PAHBAH-HCl solution and incubated at 100°C for 5 min, followed by measurement of the absorbance at 405 nm. Because Sop n s do not reduce alkaline copper or dinitrosalicylic acid (20,22) and show low sensitivity toward PAHBAH, we used Sop 4 as a standard for calculation of the enzymatic activity toward ␤-1,2glucan. Sop 4 is the main hydrolysate of ␤-1,2-glucan at the early stage of the reaction catalyzed by Cpin_6279rC. The enzymatic activity toward other polysaccharides was determined based on a Glc standard curve. One unit was defined as the amount of enzyme that released Sop 4 or Glc equivalents/1 min under the standard conditions.
The action patterns for ␤-1,2-glucan and Sop n s were analyzed by TLC. The enzymatic reaction was performed under the standard conditions with 0.2% each substrate. After the reaction had been stopped by heat treatment, an aliquot of the reaction mixture was analyzed by TLC.
Specific activity toward Sop 5-7 was determined by HPLC analysis. The enzymatic reaction was performed in the presence of 10 mM each substrate at 30°C for 5 and 10 min and then was stopped by heat treatment at 80°C for over 3 min. The concentrations of the enzyme used for degradation of Sop 5 , Sop 6 , and Sop 7 were 50, 10, and 5 g/ml, respectively. The samples (25 l) were injected onto a Shodex Asahipak NH2P-50 4E column (particle size, 5 m; inner diameter 4.6 mm ϫ 250 mm; Showa Denko K. K.) equilibrated with 70% (v/v) acetonitrile/water. Elution of the reaction products was performed with the same solution at the flow rate of 1 ml/min at 40°C for 40 min by HPLC (Prominence; Shimadzu, Kyoto, Japan). A refractive index detector (RID-10A, Shimadzu) was used for detection of the eluates. The concentrations of the reaction products (Sop 2-5 ) were determined using mixtures of Sop 2-5 (0, 0.5, and 2 mM each) as standards. Specific activity was calculated from velocity of release of each reaction product (Sop 2-3 , Sop 2-4 , and Sop 2-5 for Sop 5 , Sop 6 , and Sop 7 substrates, respectively).

ESI-MS analysis
A substrate solution (20 l) containing 5 mM MES-NaOH (pH 6.0) and 0.2% (w/v) Sop 7 was mixed with 70 l of 18 Olabeled water (97%, Sigma) or unlabeled water as a control. The enzymatic reaction was conducted by adding 10 l of 40 g/ml Cpin_6279rC in 20 mM MOPS-NaOH (pH 6.5) to the substrate solution at 30°C for 1 h. After the reaction, the reaction mixture was applied to Amicon Ultra 10,000 molecular weight cutoff (Millipore), and the flow-through solution was collected. An aliquot of the solution was dried in a centrifugal evaporator, dissolved in 50% (v/v) methanol/water, and then subjected to positive ESI-MS analysis using a JMS-T100LC AccuTOF spectrometer (JEOL, Tokyo, Japan). Reference spectra of Sop 3 and Sop 4 were measured with the same procedure as for the reaction solution.

pH and temperature profiles
To investigate the pH and temperature profiles, the hydrolytic activity of Cpin_6279rC was determined by measuring Glc after degradation of the ␤-1,2-glucan-hydrolysate produced by Cpin_6279rC into Glc by LiBGL for enhancement of sensitivity. The enzymatic reaction was carried out under the standard conditions, and an aliquot of the reaction mixture (20 l) was incubated at 100°C for 3 min to stop the reaction. The solution (20 l) was mixed with 20 l of 0.2 mg/ml LiBGL in 50 mM MES-NaOH (pH 6.0) and then incubated at 25°C for 1 h. As for the effect of pH on Cpin_6279rC, 0.5 M MES-NaOH was used instead of 50 mM MES-NaOH (pH 6.0). The precipitate generated during heat treatment was removed by centrifugation. The supernatant (20 l) was mixed with 140 l of GOPOD Format kit (Megazyme) and then incubated at 40°C for 20 min. The absorbance was measured at 510 nm, and the Glc concentration was determined based on a Glc standard curve.
The effect of pH on Cpin_6279rC activity was determined with 4.0 g/ml Cpin_6279rC under the standard conditions by replacing 50 mM MES-NaOH buffer (pH 6.0) with 20 mM each buffer. The effect of temperature on Cpin_6279rC activity was determined with 2.0 g/ml Cpin_6279rC, and the reaction was performed at each temperature for 7.5 min. pH and temperature stability were determined from the residual activity under the standard conditions after incubation of 0.4 mg/ml Cpin_6279rC in each buffer at 30°C for 1 h and 0.04 mg/ml Cpin_6279rC in 20 mM MOPS-NaOH (pH 6.5) at each temperature for 1 h, respectively.

Kinetic analysis
To determine the kinetic parameters for ␤-1,2-glucan, the enzymatic reaction was performed in a 250-l reaction mixture containing 3.0 g/ml Cpin_6279rC and 0.019 -0.39 mM ␤-1,2glucan (average DP 64) under the standard conditions. The concentration of ␤-1,2-glucan was calculated based on its average DP. An aliquot of the reaction mixture (40 l) was mixed with 120 l of a 1% (w/v) PAHBAH-HCl solution, followed by spectrophotometric measurement and calculation of the enzymatic activity as described above. Kinetic analysis for Sop 5 was performed in a reaction mixture containing 10 g/ml Cpin_6279rC and 0.3-10 mM Sop 5 . An aliquot of the reaction mixture (10 l) was mixed with 10 l of 200 mM sodium acetate buffer (pH 5.0) containing 1.0 mg/ml ␤-glucosidase from almonds (Oriental Yeast, Tokyo, Japan) and then incubated at 50°C for 2 h to hydrolyze sophorose released by Cpin_6279rC. Because the ␤-glucosidase from almonds did not act on Sop n s with DP Ն3, 4 only sophorose was decomposed to Glc. An aliquot of the reaction mixture (15 l) was mixed with 105 l of GOPOD Format kit (Megazyme), and the Glc concentration was determined based on a Glc standard curve as described above. Sop 5 -hydrolyzing activity was calculated by halving the Glc concentration.

Stereochemical analysis of reaction products
The anomeric configurations of the reaction products arising on enzymatic hydrolysis of ␤-1,2-glucan were determined by polarimetry and 1 H NMR. Polarimetric analysis was performed with a JASCO P1010 polarimeter (JASCO Co., Tokyo, Japan). Prior to the addition of the enzyme, the control optical rotation was measured (0 min). The enzymatic reaction was conducted at room temperature in a reaction mixture (1.52 ml) containing 15.8 mM sodium phosphate (pH 6.0), 1.97% (w/v) ␤-1,2-glucan (average DP 25), and 0.64 mg/ml Cpin_6279rC, and the time course of the optical rotation was recorded (1.5-10 min, with 14 recordings). At 6 min, 100 l of a 35% ammonia solution was added to the reaction mixture to accelerate mutarotation. 1 H NMR analysis was performed with a Bruker Advance 600 spectrometer (Bruker Biospin, Rheinstetten, Germany) at 20°C. The enzymatic reaction was conducted in a reaction mixture (600 l) containing 10 mM sodium acetate (pH 5.0), 1.5% (w/v) ␤-1,2-glucan (average DP 25), 98.5% (v/v) D 2 O, and 0.2 mg/ml Cpin_6279rC. Before addition of the enzyme, a reference spectrum was recorded. After addition of the enzyme, the time course of the spectral change was recorded (3 min to 24 h, with 14 recordings). Sodium acetate was used as an internal standard for peak area calculation. Acetone was added after the reaction for use as an external standard for chemical shift adjustment. Reference spectra of Sop 2-5 and ␤-1,2-glucan were obtained under the same conditions as for the reaction solution with a Bruker Advance 400 spectrometer. ␣-Anomers of the compounds were assigned based on the assignment of NMR spectra of ␤-1,2-glucan, Sop 2 , Sop 3 , and Sop 4 in the previous study (18,46).

TLC analysis
An aliquot of the reaction mixture (1 l) after the reaction was stopped was spotted onto a TLC plate (7.5 ϫ 5 cm, Kieselgel 60 F 254 ; Merck, Darmstadt, Germany). The TLC plate was developed with a 75% (v/v) acetonitrile/water solution. The TLC plate was soaked in a 5% (w/v) sulfuric acid/ethanol solution and then heated in an oven until sugars were sufficiently detected on the plate.

Crystallography
Crystals of ligand-free Cpin_6279 were grown at 25°C using the hanging drop vapor diffusion method in a 500-l aliquot of a reservoir solution by mixing 1 l of a 10 mg/ml Cpin_6279rC solution in 5 mM MOPS-NaOH (pH 6.5) with an equal volume of the reservoir solution containing 0.1 M ammonium iodide and 7% (w/v) PEG 3350. The crystals were soaked and cryoprotected with a solution containing 0.1 M ammonium iodide, 16% (w/v) PEG 3350, 25% (w/v) trehalose, and 5 mM Sop 3 . Despite the supplementation of Sop 3 , the crystal structure was obtained as a ligand-free form (see "Results"). Co-crystals of Cpin_6279 with Sop 3 were produced at 25°C using the sitting drop vapor diffusion method in a 70-l aliquot of the reservoir solution by mixing 0.5 l of a 28 mg/ml Cpin_6279rN solution containing 10 mM Sop 3 in 9 mM MOPS-NaOH (pH 7.0) with an equal volume of the reservoir solution from the JCSG core I suite (Qiagen, Hilden, Germany), number 51 (0.2 M sodium chloride, 0.1 M sodium/potassium phosphate (pH 6.2), and 10% (w/v) PEG 8000). Crystals appropriate for X-ray diffraction experiments grew at least in 1 week. The crystals were soaked in the reservoir solution supplemented with 30% (v/v) PEG 400 for cryoprotection. The crystals were flash-cooled at 100 K in a stream of liquid nitrogen. X-ray diffraction data were collected using a charge-coupled device camera on beamline AR-NW12A at the Photon Factory of the High Energy Accelerator Research Organization (KEK, Japan). The data set was indexed, integrated, and scaled using HKL2000 (47). The initial phase was determined by the molecular replacement method using Molrep (48), and BF9343_0330 from B. fragilis NCTC9343 (PDB code 3EU8) was used as a search model. Automated model building was carried out using Buccaneer (49). Manual model building and refinement were carried out using Coot (50) and Refmac5 (51), respectively. The refined structures were validated using Molprobity (52) and Rampage (53). A model structure of ␤-Sop 3 was built using JLigand (54). The structural figures were prepared using PyMOL (DeLano Scientific, Palo, Alto, CA).