Structure of a PL17 Family Alginate Lyase Demonstrates Functional Similarities among Exotype Depolymerases

Background: Exotype polysaccharide lyases can cleave glycosidic bonds to catabolize complex carbohydrates into simple monosaccharides. Results: The first characterization of a family 17 polysaccharide lyase (PL17) provides insights into the mechanism of alginate depolymerization. Conclusion: The structure of a PL17 enzyme illustrates unexpected similarities to other polysaccharide lyases despite a lack of sequence conservation. Significance: These studies demonstrate an unexpected evolutionary relationship among the polysaccharide lyases. Brown macroalgae represent an ideal source for complex polysaccharides that can be utilized as precursors for cellulosic biofuels. The lack of recalcitrant lignin components in macroalgae polysaccharide reserves provides a facile route for depolymerization of constituent polysaccharides into simple monosaccharides. The most abundant sugars in macroalgae are alginate, mannitol, and glucan, and although several classes of enzymes that can catabolize the latter two have been characterized, studies of alginate-depolymerizing enzymes have lagged. Here, we present several crystal structures of Alg17c from marine bacterium Saccharophagus degradans along with structure-function characterization of active site residues that are suggested to be involved in the exolytic mechanism of alginate depolymerization. This represents the first structural and biochemical characterization of a family 17 polysaccharide lyase enzyme. Despite the lack of appreciable sequence conservation, the structure and β-elimination mechanism for glycolytic bond cleavage by Alg17c are similar to those observed for family 15 polysaccharide lyases and other lyases. This work illuminates the evolutionary relationships among enzymes within this unexplored class of polysaccharide lyases and reinforces the notion of a structure-based hierarchy in the classification of these enzymes.

The increasing cost and limited reserve of fossil fuels have resulted in massive efforts in fields of research aimed at the discovery of efficient sources of renewable energy (1). Over the past decade, massive resources have been invested in the optimization of methods for extracting ethanolic precursors from terrestrial biomass, such as corn and perennial grasses (2). Comparatively, less effort has been directed toward the bioprocessing of marine algal biomass, which has shown potential as a replacement for land-based plant materials (3). Red, brown, and green algae contain an abundance of polysaccharide-derived sugars and, because of tolerance of changes in climate or soil conditions, represent the most sustainable source for fermentative biofuels (4). Moreover, large scale cultivation of algae can be carried out in controlled environments, such as photo bioreactors. The unique carbohydrate composition of algae presents a formative challenge for bioprocessing (5). As a consequence, there has been increasing interest in the discovery and development of enzymes that can depolymerize algal polysaccharides into fermentable monomeric sugars.
The most widely distributed component of brown algae is alginate, a cell wall polysaccharide composed of polymeric blocks of ␣-(1,4) O-linked ␤-D-mannuronate (M) 2 and its C5 epimer ␣-L-guluronate (G). These monosaccharides can form homopolymeric blocks of G residues (G blocks) or M residues (M blocks) or heteropolymers of alternating M and G (MG blocks). Depolymerization of alginate is catalyzed by diverse alginate lyases, which target the 134 O-linkage via ␤-elimination of the 4-O-glycosidic bond (6). Endotype alginate lyases, such as A1-I, A1-II, and A1-III from Sphingomonas sp. A1, depolymerize alginate into di-, tri-, and tetrasaccharides, which are then further processed by an exotype alginate lyase to yield monosaccharides that further undergo non-enzymatic conversion into 4-deoxy-L-erythro-5-hexoseulose uronic acid (Fig. 1). Within the Carbohydrate-Active EnZymes (CAZy) database (7), alginate lyases are categorized into one of 21 different polysaccharide lyase (PL) families, and exotype activity has been experimentally determined for the PL15 and PL17 families of enzymes (8). However, there is limited conservation of primary sequence between PL15 and PL17 family enzymes with less than 15% sequence identity between representative family members. In addition, PL17 enzymes lack nearly all the requisite catalytic and functionally critical residues identified through structural and biochemical studies of the PL15 enzyme Atu3025 from Agrobacterium tumefaciens (9). To date, there is The atomic coordinates and structure factors ( no structural or biochemical characterization of any PL17 family member. The marine bacterium Saccharophagus degradans 2-40 has been shown to degrade several complex polysaccharides, including alginate, and bioinformatics analysis of its genome identifies nine putative alginate lyases belonging to PL families 7, 14, 17, and 18 (10). Biochemical characterization of the S. degradans PL17 enzyme (Alg17c) demonstrates that this exotype enzyme can depolymerize alginate di-, tri-, and tetrasaccharides into monosaccharides, providing the necessary precursor for ethanol fermentation (11). However, given the lack of sequence similarity with PL15 and lack of conservation of active site features, little is known about the mechanism of any PL17 family enzyme. Given that there are 77 different polypeptides that are annotated as PL17 family members in the CAZy database (8), the structure and mechanism of a representative PL17 family enzyme should provide an understanding of the basis for alginate depolymerization. Here, we present several crystal structures of the PL17 enzyme Alg17c from S. degradans, including the wild-type enzyme (to 1.85-Å resolution), two active site variants (to 1.7-and 2.45-Å resolution), and the cocrystal structure of an inactive variant in complex with an alginate trisaccharide (to 1.9-Å resolution). Structure-guided analysis of several active site variants allowed identification of residues that are critical for substrate recognition and for the basis of the exolytic reaction mechanism of PL17 enzymes.

EXPERIMENTAL PROCEDURES
Materials-S. degradans 2-40 genomic DNA was purchased from the American Type Culture Collection. Low viscosity alginate was purchased from MP Biomedicals. Protein molecular weight markers for sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) were obtained from Bio-Rad. All . Exolytic depolymerization of these polysaccharides by alginate lyase yields a monosaccharide and a product containing a ⌬-(4,5)-unsaturated uronic acid moiety. B, thin layer chromatographic analysis of polymer specificity of Alg17c. Polysaccharide standards shown are for monomer (Lane 1, D-glucose) and dimer (Lane 2, D-trehalose). A mixture of alginate di-, tri-, and tetrasaccharides (Lane 3) are processed into mono-and disaccharides (Lane 4) in the presence of Alg17c. DP, degree of polymerization. other reagents were purchased from Fisher Scientific and were of the highest purity available.
Preparation of Alginate Oligosaccharide Substrate-Alginate oligosaccharides were generated by enzymatic depolymerization of low viscosity alginate using alginate lyase from Flavobacterium multivorum (Sigma CAS 40591-57-9). Briefly, 150 l of enzyme at a concentration of 1 unit/ml was added to a solution containing 0.1% sodium alginate at pH 6.3, and the mixture was incubated at 37°C for 30 min. Reaction mixtures containing up to 20 mg of digested polysaccharides were loaded onto a DEAE-Sephadex A-25 column (2.8 ϫ 20 cm), which was equilibrated with 25 mM ammonium acetate, pH 6.8. A gradient from the equilibration buffer to 1 M NH 4 OAc over 1 ml/min with a flow rate of 5 ml/min was used for separation at room temperature. Fractions containing digestion products were identified from their absorbance at 235 nm. Oligosaccharides were desalted using a BioGel P-2 column (1.5 ϫ 50 cm) equilibrated with 1 mM phosphate buffer, pH 6.8. Samples were then further purified using a Hypercarb column (100 ϫ 3 mm) on an Agilent 1100 Series HPLC system. The solvent flow rate was 0.3 ml/min with a temperature of 65°C. The products were eluted using the following gradient where solvent A is 0.2% TFA in water and solvent B is methanol: 30% B from 0 to 10 min and gradient from 30 to 90% B from 10 to 40 min. The final products were lyophilized, and masses for the lyophilized products were determined using an Agilent 1100 LC/MSD Trap XCT Plus mass spectrometer (Agilent Technologies, Inc., Palo Alto, CA).
Cloning and Site-directed Mutagenesis-Amplification of Alg17c (GenBank TM accession number ABD82539.1) from S. degradans genomic DNA was done using PCR, and the resultant insert was cloned into pET-28b. To facilitate heterologous expression, the expression construct lacks the first 24 residues that encode the signal peptide. Site-specific mutations in the coding sequence were carried out using a QuikChange sitedirected mutagenesis kit (Agilent Technologies, Inc., Santa Clara, CA). The PCR product (12.5 l) was digested with methylation-dependent restriction enzyme DpnI at 37°C for 8 h to digest the parental plasmid DNA. Escherichia coli DH-5␣ competent cells were transformed with the product from the DpnI restriction digestion and plated onto lysogeny broth (LB) solidified with Bacto agar (Difco) containing 25 g/ml kanamycin sodium salt, and the plates were incubated at 37°C overnight. Individual colonies were cultivated in LB medium supplemented with kanamycin until stationary phase was reached, and plasmids were extracted using a QIAprep Spin Miniprep kit (Qiagen, Valencia, CA). The integrity of all of the constructs was confirmed by DNA sequencing (W. M. Keck Center for Comparative and Functional Genomics, University of Illinois).
Recombinant Protein Production and Purification-E. coli BL21-(DE3)-Rosetta cells harboring a recombinant plasmid bearing either the Al17c gene or variants were cultured in LB containing 25 g/ml kanamycin for ϳ12 h at 37°C with vigorous shaking. Gene expression was induced when the absorbance at 600 nm reached Ϸ0.5 with the addition of 0.1 mM isopropyl 1-thio-␤-D-galactopyranoside, and the temperature was decreased to 16°C followed by incubation for an additional 16 h at 200 rpm. Cell pellets were harvested from the cultures by centrifugation at 5,000 ϫ g for 30 min at 4°C, washed with 10 ml of buffer (50 mM Tris, 300 mM KCl, pH 7.5), and frozen as cell pellets at Ϫ80°C. Preparation of selenomethionine (SeMet)labeled Alg17c was carried out by repression of methionine synthesis in defined media (12).
Frozen cell pellets were thawed, resuspended in 35 ml of ice-cold binding buffer, and passed through an EmulsiFlex C-3 homogenizer (Avestin, Ottawa, Canada) to rupture the cells. The cell lysates were clarified by centrifugation at 16,000 ϫ g for 30 min at 4°C, and the supernatant containing the soluble fraction was loaded onto a 5-ml immobilized metal ion affinity resin column (Hi-Trap nickel-nitrilotriacetic acid, GE Healthcare) pre-equilibrated with binding buffer (50 mM Tris-HCl, 300 mM KCl, 20 mM imidazole, pH 7.5). The column was extensively washed with lysis buffer supplemented with 30 mM imidazole and eluted by a linear gradient to 200 mM imidazole. The polyhistidine affinity tag was removed by overnight incubation with thrombin (1 unit/mg of protein). Samples were concentrated using a 10,000 molecular weight cutoff Amicon centrifugal filters and further purified using size exclusion chromatography (Superdex Hiload TM 75 16/60, GE Healthcare) in 20 mM HEPES, pH 7.5, 100 mM KCl buffer.
Phasing and Structure Determination-Flash cooled crystals of wild-type Alg17c diffracted x-rays to 1.85-Å resolution using a Mar 300 charge-coupled device detector at an insertion device synchrotron beam line (LS-CAT Sector 21 ID-G, Advanced Photon Source, Argonne, IL). These crystals belong to space group P2 1 with unit cell parameters a ϭ 85.5 Å, b ϭ 126.4 Å, c ϭ 87.6 Å, and ␤ ϭ 111.5°with two molecules in the crystallographic asymmetric unit. A 3-fold redundant data set was collected to 1.85-Å resolution with an overall R merge of 9.2% and I/(I) of 2.1 in the highest resolution shell. All data were indexed and scaled using either the HKL2000 (13) package or XDS (14). Crystallographic phases were determined using single wavelength anomalous diffraction data collected from a crystal of SeMet-labeled protein. Diffraction data were collected to 2.2-Å resolution and processed as described above. Heavy atom sites were determined using HySS and imported into SHARP for maximum likelihood refinement (15). The resultant phases were further improved by 2-fold non-crystallographic symmetry averaging and subjected to automated model building as implemented in ARP/wARP (16). The model was fitted using Coot (17) and further improved by rounds of refinement with REFMAC (18). Cross-validation was routinely used throughout the course of model building and refinement using 5% of the data in the calculation of the free R-factor (19).
The structure of Alg17c variants H202L (1.7-Å resolution) and Y258A (2.45-Å resolution) and the co-crystal structure of Y258A in complex with the alginate trisaccharide (1.9-Å resolution) were determined by molecular replacement using the coordinates of unliganded, wild-type Alg17c as a search probe.
Each of the structures was refined and validated using the procedures detailed above. For each of the structures, the stereochemistry of the model was monitored throughout the course of refinement using PROCHECK (20) and MolProbity (21). Relevant data collection and refinement statistics are provided in Table 1. The refined atomic coordinated have been deposited in the Protein Data Bank (22).
Kinetic Analysis of Wild-type Alg17c-Kinetic characterization was carried out on wild-type Alg17c as well as the variants H202L and Y258A using low viscosity alginate as the substrate. For each assay, 0.613 M enzyme was added to 1 ml of mixture containing 50 mM Tris-HCl, pH 7.5 and varying concentrations of substrate (0.1-10 mg/ml). The mixture was incubated at 30°C for 1 h and then boiled for 5 min to stop the reaction, and the thiobarbituric acid assay method was used to quantify the amount of unsaturated uronic acids produced. Briefly, 40 l of the reaction mixture was added to 0.25 ml of 5 mM H 5 IO 6 (in 62.5 mM H 2 SO 4 ) and incubated at room temperature for 20 min followed by the addition of 0.50 ml of 2% sodium arsenite to the mixture and incubation for 2 min with vigorous shaking. 2 ml of 0.3% thiobarbituric acid at pH 2 was then added, and the mixture was heated to 100°C for 10 min. After cooling, the resultant condensation of ␤-formylpyruvic acid with thiobarbituric acid was monitored by measuring the absorbance at 548 nm. 1 unit of enzyme activity was defined as the amount required to liberate 1 M ␤-formylpyruvic acid/min.
Thin Layer Chromatographic (TLC) Analysis-Alginate oligosaccharides with degrees of polymerization of 2, 3, and 4 (0.5 mg/ml) were incubated with 0.6 M Alg17c in a buffer of 50 mM Tris-HCl, pH 7.5 at 30°C for 1 h. Reaction products were analyzed by TLC using a solvent of n-butyl alcohol:formic acid: water (4:6:1, v/v/v). The depolymerized products were visualized by developing the plates with 10% (v/v) sulfuric acid in ethanol followed by incubation at 100°C for 20 min. Standards consisted of either D-glucose (2 mM) monosaccharide or D-trehalose (2 mM) disaccharide.
Isothermal Titration Calorimetry-The MicroCal VP-ITC system from GE Healthcare was used to calculate dissociation constants for the binding of oligosaccharide to catalytically inactive Alg17c. Purified enzyme was concentrated to 15 mg/ml and diluted in 20 mM HEPES, pH 7.5, 100 mM KCl buffer. Oligosaccharide with a degree of polymerization between 2 and 4 was diluted in the same buffer to ϳ1 mM. Both Alg17c and substrate were degassed and brought to room temperature. The reference cell contained deionized water, and the sample chamber was washed with deionized water and methanol before the sample was injected. The protein sample (in a 1.44-ml reaction cell) was injected with 26 successive 10-l aliquots of ligand at 300-s intervals. Data were fitted by nonlinear regression using a single site model (MicroCal Origin).

RESULTS AND DISCUSSION
Kinetic Analysis of Wild-type Alg17c-To facilitate heterologous expression, the first 24 residues encoding the signal peptide were not included for any of the expression constructs. Determination of the kinetic parameters for degradation of low viscosity alginate by Alg17c was carried out using the thiobarbituric acid assay method (  (23) and Atu3025 from A. tumefaciens (9). Kinetic parameters for wild-type and site-specific active site variants of Alg17c are reported in Table 2 and discussed in the relevant sections below.
Determination of Polymer Specificity of Alg17c-The minimal polymer length that can be processed by Alg17c was determined by thin layer chromatographic analysis of Alg17c reaction products using alginate oligosaccharide substrates of defined length. Treatment of a mixture containing oligosaccharides with degrees of polymerization of 2, 3, and 4 resulted in complete processing only of the tri-and tetrasaccharide substrates. In contrast, the disaccharide substrate was only partially processed. These results are consistent with Alg17c having a minimal polymer specificity of 3. Kinetic analysis of wild-type Alg17c using a trisaccharide substrate yielded a k cat of 62.4 s Ϫ1 , k cat /K m of 8.2 ϫ 10 6 M Ϫ1 s Ϫ1 , K m of 7.7 M, and V max of 38.2 M s Ϫ1 . These kinetic values are similar to those determined using heteropolymeric low viscosity alginate as a substrate, consistent with the notion that an alginate trisaccharide represents the minimal length substrate for Alg17c.
Overall Structure of Alg17c-The structure of wild-type Alg17c was determined to 1.85-Å resolution using phases determined by single wavelength anomalous diffraction measurements collected using crystals of SeMet-labeled protein (relevant data collection statistics are provided in Table 1). In each of the structures, no electron density can be observed for the first 6 residues at the amino terminus, and no effort was made to model these residues. The final model for each of the structures consists of all residues spanning His 30 -Arg 734 .
The overall architecture of each Alg17c monomer consists of two domains: an amino-terminal imperfect ␣-barrel (consisting of residues His 30 -Glu 365 ) and the carboxyl-terminal ␤-sheet domain (composed of residues Pro 370 -Arg 735 ) (Fig. 2,  A and B). There are 13 helices in the amino-terminal domain, resulting in a topological arrangement that is similar to an ␣ 6 /␣ 6 -barrel but with an extra helix that buttresses against the side of the barrel. The arrangement of the ␣ 6 /␣ 6 -helices produces an open barrel structure, and this additional helix serves to maintain rigidity of the barrel. The carboxyl-terminal domain is composed of three co-planar layers of antiparallel ␤-strands arranged as sheets and further supported by four small helices near the top layer that interject between the two domains (Fig. 2, A and B). Sequence conservation among other PL17 family members extends throughout all of the secondary structural elements (Fig. 2C), suggesting that the Alg17c structure is representative of all PL17 enzymes. In contrast, all of the alginate lyases in nearly all other PL families adopt either an ␣-barrel or a jelly roll fold (8), which is distinct from the combination of the ␣ 6 /␣ 6 -barrel and antiparallel ␤-sheet arrangement observed in the PL17 structure.
The general domain structure observed in Alg17c is reminiscent of that observed in the structure of the PL15 family enzyme Atu3205 from A. tumefaciens (Protein Data Bank code 3A0O) (9). Although the two enzymes are topologically distinct as evidenced by an r.m.s.d. value of 3.4 Å upon least square alignment of 567 C␣ atoms, the PL15 family Atu3205 also consists of an ␣ 6 /␣ 6 -barrel joined to a carboxyl-terminal antiparallel ␤-sheet. A recent structure-based classification of polysaccharide lyases terms this topological class as that of a multidomain (␣/␣) n toroid (Fig. 3) (29). Each of these enzymes is composed of an ␣-barrel domain fused to a domain consisting of ␤-strands, and this structural superfamily has been designated as the multidomain (␣/␣) n  (24). Although the helical dispositions are quite distinct, the ␣ 6 /␣ 6 -barrel domain of heparinase II also contains an additional helix. The structural homology to PL15 enzymes and to other multidomain (␣/␣) n toroid class enzymes is especially notable due to the fact that there is no significant conservation between the primary sequence of Alg17c and any of its structural homologs. Given the similar exolytic mode of action on alginate substrates, the structural homology between the PL17 family Alg17c and the PL15 family Atu3205 is noteworthy. The PL15 enzyme contains an additional amino-terminal domain of roughly 100 residues composed of seven ␤-strands but otherwise shares similar structural features (9). This conservation is unexpected given the low 13% sequence identity shared between the two enzymes. Given the lack of sequence conservation, it is not surprising that the active site of Alg17c is quite distinct from that of Atu3205. Nearly all of the residues that have been demonstrated to play a critical role in alginate depolymerization by Atu3205 are divergent in Alg17c, and the steric confines of the two active sites are also different. A thorough description of these differences is detailed under "Co-crystal Structure of an Alg17c Variant in Complex with an Alginate Trisaccharide" (see below).
Oligomerization State of Alg17c-Analytical size exclusion chromatographic studies are consistent with the formation of Alg17c homodimers in solution (data not shown), and this dimer is recapitulated in the slightly different crystal forms of wild-type and variant Alg17c (Fig. 2B). Dimerization results in the burial of roughly 1300 Å 2 of surface area per monomer (measured using a probe radius of 1.4 Å). Formation of the homodimer results in the insertion of an antiparallel ␤-turn, encompassing residues Gly 660 -Ala 672 , from the carboxyl-ter-minal ␤-sheet domain of one monomer into the ␣ 6 /␣ 6 -barrel of the other monomer (Fig. 2B). Alg17c elutes as a dimer in analytical size exclusion experiments even at low concentrations near the limit of detection. Given the experimental data and the observation of intersubunit interactions in the Alg17c crystal structure, it is likely that the homodimer represents the biologically relevant form of the enzyme.
A Zinc-binding Site in Alg17c-Experimental electron density maps from data collected on crystals of SeMet-labeled Alg17c near the selenium absorption edge ( ϭ 0.9786) showed an electron-dense feature near the vicinity of the amino-and carboxyl-terminal interface. The side chains of His 415 , Asp 433 , and His 464 along with three solvent molecules are located around this density with the nitrogen and oxygen atoms engaging in near octahedral coordination geometry (Fig. 4). Given the coordination distance and geometry, this peak has been tentatively assigned to a bound Zn 2ϩ ion. This electron density is enhanced in a data set that was collected at ϭ 1.0, which would correspond to an fЉ value of 2.55 at this wavelength for zinc. A similar peak was also observed in the structure of P. heparinus heparinase II at nearly the same location and was also modeled as a bound zinc ion (26) (Fig. 3C). Coordination of the metal joins the small helices located at the juncture of the two domains with the top ␤-sheet layer of the carboxyl-terminal ␤-strand domain. As in heparinase II, the metal ion is located over 15 Å away from the active site and likely only plays a structural role. Specifically, His 415 and Asp 433 are both part of different loop regions that include residues that may function in binding to substrate oligosaccharides. Presumably, the metal ion establishes the orientation of His 415 and Arg 438 , which would otherwise be mobile, to set these residues for interactions with substrate. and antiparallel ␤-sheet (green) domains. The zinc ion is colored in red. B, structure of the Alg17c biological dimer. C, structure-based multiple sequence alignments of characterized PL17 family enzymes, including Alg17c, AlgL (from Sphingomonas sp. MJ3), and AlyIII (from Pseudomonas sp. OS-ALG9). The first 24 amino acids encode the signal sequence and are enclosed in the green rectangle. Secondary structural elements are derived from the Alg17c structure, and residues involved in substrate binding (green triangles), reaction chemistry (red diamonds), and metal ion binding (blue circles) are indicated as described. To probe the relevance of this zinc ion in catalytic activity, one of the metal-binding protein ligands, His 415 , was replaced, and the H415A mutant was kinetically characterized. The catalytic activity of this mutant was slightly compromised compared with the wild type with a negative effect on k cat /K m (2.6 ϫ 10 3 M Ϫ1 s Ϫ1 , which is nearly a 1,000-fold lower than that of the wild type). These results are consistent with the proposed role of the metal ion in establishing the orientation of residues that may be important for engaging oligosaccharide substrates.
Despite the structural similarities of Alg17c and the PL15 family of enzymes, no metal ion is observed at a similar location in the structure of Atu3205 (9). Moreover, only one of the three metal coordination residues is conserved (His 533 in place of His 415 ), and the other Asp 433 and His 464 are Gln 551 and Lys 573 , respectively. The cavity created by the lack of a metal ion in Atu3205 is partially compensated by the insertion of a hydrophobic Phe 414 into the putative binding site. The presence of a bound metal is likely a distinguishing feature of PL17 family of enzymes as equivalent metal coordination residues are also observed in the sequences of other PL17 members, such as AlyIII (from Pseudomonas sp. OS-ALG9) (30) and AlgL (from Sphingomonas sp. MJ3) (31) that have been partially characterized (Fig. 2C).

Co-crystal Structure of an Alg17c Variant in Complex with an
Alginate Trisaccharide-The nomenclature for substrate binding in lyase active sites defines subsites that engage each saccharide with cleavage occurring at the glycosidic bond that links the sugars at subsites Ϫ1 and ϩ1. In the canonical exolytic mechanism, the negative charge on the C6 carboxylate of the saccharide at subsite ϩ1 is stabilized, facilitating the removal of the C5 proton by a general base and subsequent donation of a proton to the O4 atom by a general acid (32). Structural and biochemical studies on the PL15 family Atu3205 suggest that His 311 and Tyr 365 function as the catalytic base and acid, respectively. A structure-based alignment identifies His 202 and Tyr 258 in Alg17c situated at equivalent positions to His 311 and Tyr 365 in Atu3205. Based on this alignment, the H202L and Y258A mutations were generated in Alg17c for co-crystallization studies with intact substrate oligosaccharides. To determine whether Tyr 258 and His 202 are indeed the catalytic acid and base, respectively, in Alg17c, the Y258A and H202L variants were subject to kinetic characterization. Although the Y258A variant was devoid of any activity, the H202L variant has a k cat of 10.9 s Ϫ1 and K m of 92 M. Consequently, although the H202L variant is compromised in activity, it is still active, and therefore, His 202 is not likely to participate in acid/base chemistry. However, as the H202L Alg17c yielded crystals that diffracted to a slightly higher resolution that the wild type, further structural characterization of this variant was carried out. To ensure that the mutations had not resulted in significant changes to the enzyme, the structure of H202L Alg17c and the structure of Y258A Alg17c were determined to 1.7-and 2.45-Å resolution, respectively. The structures of the mutant enzymes were essentially identical to that of the wild-type enzyme with all of the differences constrained to minor movements of loop regions in the carboxyl-terminal ␤-sheet domain.
Small oligosaccharide substrates were generated from crude alginate polymer and purified using size exclusion chromatography for biochemical studies. To determine the affinity of Alg17c for alginate substrates, isothermal titration calorimetry was carried out using the Y258A variant. This variant binds to the trisaccharide ligand with 1:1 stoichiometry with an association affinity of 9.9 ϫ 10 3 M Ϫ1 and is largely enthalpically driven (Fig. 5).  . Difference Fourier maps of the bound metal ion. A stereoview of the Alg17c metal-binding site derived from the refined coordinates of Alg17c is shown. The Alg17c carbon atoms are shown in yellow ball-and-stick representation, metal-bound solvent molecules are colored in black, and the metal is shown in red. Superimposed is a difference Fourier electron density map (contoured at 5 over background in blue and 15.0 in red) calculated with coefficients ͉F obs ͉ Ϫ ͉F calc ͉ and phases from the final refined model with the coordinates of the metal deleted prior to one round of refinement. Based on the coordination chemistry and the metal assignment in hyaluronate lyase, the metal is putatively assigned as a zinc ion.
To further define active site residues that may participate in both substrate recognition and the exolytic mechanism, small oligosaccharide substrates with a degree of polymerization between 2 and 4 were generated and purified from crude alginate polymer and then used to soak into crystals of Y258A Alg17c. Co-crystals of Y258A Alg17c diffracted to 1.9-Å resolution. Although molecular replacement using the coordinates of the unliganded wild-type Alg17c gave a satisfactory solution, the resultant model could not be refined to any acceptable criteria. Inspection of the resultant electron density maps revealed significant conformational changes at the carboxyl-terminal ␤-sheet domain, and manual rebuilding of the region followed by crystallographic refinement resulted in a major decrease in the free R-value to below 0.3. At this point, clear and continuous electron density for the bound oligosaccharide could be observed in the vicinity of the putative active site residue, Tyr 258 . The high resolution of the data and the high quality of the corresponding electron density map reveals the bound trisaccharide as ⌬MMG where ⌬M denotes unsaturated D-mannuronate and M and G denote saturated D-mannuronate and L-glucuronate, respectively (Fig. 6). Following refinement of the model, the bound trisaccharide has crystallographic B-values (a relative measure of mobility) that are similar to those of the protein main chain, consistent with well ordered binding of the ligand (Table 1).
A superposition of the structures of unliganded and trisaccharide-bound structures of Alg17c reveals that the enzyme undergoes conformational changes upon binding of substrate (Fig. 6B). Specifically, an alignment of the ␣ 6 /␣ 6 -barrel domains demonstrates that the carboxyl-terminal ␤-sheet domain undergoes a rigid body rotation of roughly 12°upon ligand binding that results in the displacement of various active site residues to be poised for interactions with the bound trisaccharide. This domain movement is necessary to position several critical active site residues in catalytically competent orientations (Fig. 6C).
A similar domain rotation movement has also been observed upon ligand binding by the PL15 family Atu3205 (in complex with ⌬GGG) (9), and computational studies suggest that active site flexibility plays a significant role in substrate engagement in the PL8 hyaluronate lyase (28). Despite the fact that none of these three enzymes demonstrate any appreciable similarities in primary sequence, they share the common architecture of an ␣-barrel linked to a ␤-sheet domain, suggesting that conformational flexibility may be a common feature of all enzymes from this multidomain (␣/␣) n toroid structural class. The bound ⌬MMG trisaccharide occupies subsites Ϫ1, ϩ1, and ϩ2 with the nonreducing mannuronate occupying the deep cleft of the binding pocket where the pyranose ring stacks against the aromatic side chain of Tyr 261 . Hydrogen bond interactions that account for binding specificity include those from Asn 149 , Tyr 257 , Arg 260 , His 413 , Arg 438 , and Glu 667 . Additional polar residues engage the sugar indirectly through solvent-mediated interactions. At subsite ϩ1, the saturated D-mannuronate is engaged through a more limited set of hydrogen bond interactions with Gln 146 and Asn 201 , indicative of flexibility in accommodating different bound saccharides at this position. The saturated L-glucuronate at the ϩ2 position similarly interacts in a minimal manner with only enzyme residues Lys 136 and Lys 198 located within hydrogen bonding distance.
The co-crystal structure provides further insights into residues that likely function in acid/base chemistry and substrate stabilization for alginate depolymerization (Fig. 7). Assignment of the roles of functional residues is inferred based on their location in the proximity of the uronic acid carboxylate, the glycosidic bond, and the acidic C5 proton. In the co-crystal structure, the C6 carboxylate is located adjacent to Asn 201 (3.2 Å from N␦2) and His 202 (2.7 Å from N⑀2), and these residues likely function to stabilize the carboxylate charge to lower the pK a of the C5 proton. Tyr 450 is poised axially to the presumed location of the C5 proton (3.4 Å from O) where it likely functions as the general base. Lastly, Tyr 258 is positioned in line with the glycosidic bond between subsites Ϫ1 and ϩ1 (2.4 Å from O) and is likely the general acid that protonates O4 during bond cleavage.
Kinetic Characterization of Structure-based Active Site Mutants-To further confirm the presumed functions of the active site residues in catalysis, kinetic analyses were carried out on site-specific variants at each of the amino acids identified in the co-crystal structure. Consistent with the role of Tyr 450 as the general base and Tyr 258 as the general acid, the Y258A and Y450A single mutants were both completely devoid of activity. Parenthetically, we note that our assumption of Tyr 258 as a participant in acid/base chemistry was correct; however, this residue is likely the general acid and not the general base as initially assumed. The Y261A mutant is also inactive, consistent with the proposed role for this residue in engaging the nonreducing mannuronate through stacking interactions. Based on the structure of the unliganded enzyme and structure-based alignments with Atu3025, His 202 was presumed to be the general base. However, the H202L mutant was catalytically active (although compromised), and our co-crystal structure provides a rational for this observation. The function of His 202 is not as a general base but rather to work in concert with Asn 201 to stabilize the charge on the C6 carboxylate. Consequently, the H202L mutant should not result in a complete loss of activity, consistent with our kinetic analysis. Lastly, mutations at residues that stabilize the binding of the oligosaccharide (Asn 149 , Arg 260 , and His 413 ) are significantly impaired with k cat /K m values that are orders of magnitude lower than that of the wild type. However, these variants still show some activity, reflective of the fact that the function of each of these residues is largely redundant.
Mechanistic Basis for the Exolytic Mode of Alginate Depolymerization by Alg17c-The mechanism for Alg17c likely follows the canonical mechanism for oligosaccharide lyase that necessitates active site residues that participate in acid/base chemistry and act to stabilize the negative charge on the C6 carboxylate (Fig. 7). The stepwise mechanism utilizes at subsite ϩ1 a basic residue that lowers the pK a of the C5 proton to facilitate abstraction by a general base and a general acid to protonate the O4 atom during cleavage of the glycosidic bond. In many lyases, such as chondroitin AC lyase and hyaluronate lyase, the C6 carboxylate is not stabilized by an acidic residue but rather by an amide and/or His that forms a strong hydrogen bond with this moiety. The co-crystal structure of Alg17c reveals that Asn 201 and His 202 are near the vicinity of the C6 carboxylate and likely function in neutralizing the negative charge of the uronic acid carboxylate, and this was confirmed by mutational analysis at these residues. Tyr 450 functions as the general base to abstract the C5 proton to facilitate glycosidic bond cleavage. Lastly, Tyr 258 functions as the general acid to donate a proton to the O4 atom of the glycosidic bond during cleavage. The role of both of these residues is further borne out by our kinetic characterization of site-specific mutations.
The utilization of two tyrosine residues as both the general acid (Tyr 258 ) and the general base (Tyr 450 ) is unusual within the multidomain (␣/␣) n toroid lyase superfamily. Given that the phenolic side chain has a pK a of ϳ10, tyrosine is an atypical choice to function as a general base. For example, in the homologous PL15 family alginate lyase Atu3205, His 311 functions as the catalytic base. Presumably, microenvironmental tuning of the pK a of both tyrosine residues can accommodate their function in acid/base chemistry. Analogously, a single tyrosine residue (Tyr 234 ) functions as both the general acid and general base in chondroitin AC lyase (33).
Conclusions-Our studies provide the first crystallographic structures of a PL17 family enzyme, Alg17c from S. degradans. We also present the first detailed structure-function character- ization of active site variants identified based on our co-crystal structures that establish the role of various residues in the chemistry of alginate depolymerization. Our combined results demonstrate a convergence both in structure and in function among the large class of lyases that fall within the multidomain (␣/␣) n toroid structural class. Despite global similarities, numerous differences also exist, including the identity of the catalytic residues, and these cannot be identified by simple sequence or structure alignments. Additional studies of enzymes within this structural class will likely shed new insights into the mechanism of oligosaccharide depolymerization.