Molecular Basis for Substrate Recognition and Catalysis by a Marine Bacterial Laminarinase

Heterotrophic bacterial communities are key players in marine biogeochemical cycling due to their ability to remineralize organic carbon. Processing of complex organic matter requires heterotrophic bacteria to produce extracellular enzymes with precise specificity to depolymerize substrates to sizes sufficiently small for uptake. Thus, extracellular enzymatic hydrolysis initiates microbe-driven heterotrophic carbon cycling. In this study, based on biochemical and structural analyses, we revealed the depolymerization mechanism of β-1,3-glucan, a carbon reserve in algae, by laminarinase from an alga-associated marine Flavobacterium. The findings provide new insights into the substrate recognition and catalysis of bacterial laminarinase and promote a better understanding of how extracellular enzymes are involved in organic matter cycling.

ABSTRACT

Laminarin is an abundant algal polysaccharide that serves as carbon storage and fuel to meet the nutrition demands of heterotrophic microbes. Laminarin depolymerization catalyzed by microbial extracellular enzymes initiates remineralization, a key process in ocean biogeochemical cycles. Here, we described a glycoside hydrolase 16 (GH16) family laminarinase from a marine alga-associated Flavobacterium at the biochemical and structural levels. We found that the endolytic enzyme cleaved laminarin with a preference for β-1,3-glycoside linkages and showed transglycosylation activity across a broad range of acceptors. We also solved and compared high-resolution crystal structures of laminarinase in the apo form and in complex with β-1,3-tetrasaccharides, revealing an expanded catalytic cleft formed following substrate binding. Moreover, structure and mutagenesis studies identified multiple specific contacts between the enzyme and glucosyl residues essential for the substrate specificity for β-1,3-glucan. These results provide novel insights into the structural requirements for substrate binding and catalysis of GH16 family laminarinase, enriching our understanding of bacterial utilization of algal laminarin.

IMPORTANCE Heterotrophic bacterial communities are key players in marine biogeochemical cycling due to their ability to remineralize organic carbon. Processing of complex organic matter requires heterotrophic bacteria to produce extracellular enzymes with precise specificity to depolymerize substrates to sizes sufficiently small for uptake. Thus, extracellular enzymatic hydrolysis initiates microbe-driven heterotrophic carbon cycling. In this study, based on biochemical and structural analyses, we revealed the depolymerization mechanism of β-1,3-glucan, a carbon reserve in algae, by laminarinase from an alga-associated marine Flavobacterium. The findings provide new insights into the substrate recognition and catalysis of bacterial laminarinase and promote a better understanding of how extracellular enzymes are involved in organic matter cycling.

INTRODUCTION

Marine algal photosynthesis contributes half of global primary production (1). Most carbon fixation products are polymerized carbohydrates comprising up to 80% of the algal biomass (2). Marine algae can secrete extracellular polysaccharides to the phycosphere environment, which attracts heterotrophic bacterioplankton and fuels their nutrient demands. The high-molecular-weight dissolved organic carbon (HMW DOC; >1 kDa) in marine waters typically accounts for 15% to 40% of the total DOC (3). Because bacteria can only transport substrates with a molecular mass of <600 Da (4), HMW DOC needs to be cleaved by extracellular enzymes before uptake. Therefore, the decomposition of algal polysaccharides by bacterial enzymes is a key process in the marine carbon cycle.

Flavobacteriia are among the most abundant bacterial groups associated with marine algae (5), with their highest abundance typically observed during the decay phase of phytoplankton blooms (6) and indicating roles in the utilization of algal biomass. Genomic analysis of environmental representatives of Flavobacteriia revealed their innate ability for polysaccharide catabolism (7, 8). One unique feature of Flavobacteriia genomes is the presence of polysaccharide utilization loci (PULs) that orchestrate the expression of genes involved in the detection, digestion, and transport of complex carbohydrates (9). When a substrate glucan is encountered, the expression of the corresponding PUL is upregulated, and glycoside hydrolases (GHs) are secreted to depolymerize the glucan into smaller oligosaccharides that are then recognized and imported by SusCD-like complexes. Many polysaccharide-degrading enzymes from Flavobacteriia harboring both N-terminal Sec and C-terminal Por secretion system-type signal peptides are exported by the type IX secretary system (T9SS) associated with gliding motility (10, 11).

Laminarin is a soluble β-1,3-d-glucan with limited β-1,6-d-branching and involved in primary intracellular carbon storage present in most marine algae and phytoplankton (12, 13). Five to 15 billion metric tons of laminarin are estimated to be produced annually in the ocean (14). Laminarin can be released as HMW DOC to marine environments following algal cell lysis (15) and is rapidly degraded by bacterial extracellular carbohydrate-degrading enzymes in the pelagic system (16). Laminarinases (EC 3.2.1.39) are mostly members of GH16 and hydrolyze internal β-1,3-glucosyl linkages in laminarin, thereby playing crucial roles in the biogeochemical cycling of algal organic carbon. The environmental importance of laminarinases is suggested by the observations that GH16 enzymes dominate the algal bloom in the ocean and coincide with the peak abundance of Flavobacteriia (17, 18).

The GH16 family includes several enzymes specific for algal polysaccharides, with at least 11 different known EC numbers (19), whereas laminarin hydrolysis is proposed to be the ancestral activity of the GH16 family (20) and is consistent with the ancient nature of β-1,3-glucan as a carbon reserve in eukaryotes (21). Analyses of GH16 family laminarinase structures from archaea (22), Actinomyces (23), and marine bacteria (19, 24, 25) reveal that these enzymes share a β-jelly roll fold and catalyze glycosyl hydrolysis in a retaining mechanism (26). In the active site, one glutamate residue serves as a nucleophile to attack the C1 atom, and another glutamate residue acts as a proton donor to complete the double-displacement mechanism. Although laminarinase catalysis is generally well understood, details of the structural interactions between enzymes and substrates/products are not as well characterized. To date, studies have provided enzyme-substrate complex structures of GH16 family laminarinases, including ZgLamA_GH16 (from Zobellia galactanivorans) in complex with laminaritetraose (25), ZgLamC_GH16 in complex with thio-β-1,3-hexaglucan (19), and Lam16A in complex with β-1,4-glucosyl-laminaribiose and β-1,6-glucosyl-laminaritriose (27). These studies established an overview on β-1,3-glucan recognition and focused less on the conformational changes or biochemical verification of substrate-binding residues.

Here, we systematically investigated how enzyme recognition and processing during laminarin depolymerization via biochemical and structural analyses of laminarinase from Aquimarina sp. SCSIO21287, a marine Flavobacterium recently isolated from the red alga Amphiroa ephedraea (Lamarck) Decaisne in the South China Sea. We solved high-resolution crystal structures of laminarinase and its inactive mutant in complex with β-d-glucan tetrasaccharides in the active cleft, allowing identification of structural features responsible for substrate specificity with β-1,3-glucan. Moreover, site-directed mutagenesis and enzyme kinetics analysis were performed to obtain a deeper understanding of specific enzyme-carbohydrate interactions. Our study has provided that (i) the laminarinase selectively endohydrolyzes β-1,3-glucan with simultaneous transglycosylase activities toward a unique broad spectrum of acceptors, (ii) a network of interactions in the negative subsites are responsible for substrate selectivity, and (iii) a substrate-induced change in conformation of the conserved tryptophan residue at the −2 subsite is a distinct feature in the catalytic process. These points offer novel insights into the molecular basis of laminarinase catalysis, as well as the biogeochemical cycling of laminarin driven by marine bacteria.

RESULTS

Roles of functional modules.

Full-length laminarinase from Aquimarina sp. SCSIO21287 (LamAQ) contains 560 amino acid residues and 5 putative modules, with an N-terminal signal peptide, followed by the catalytic GH16 family domain, a crystalline linker peptide, a carbohydrate-binding domain of family 6 (CBD IV), and a C-terminal PorSS tail (Fig. 1a). To investigate the function of each module, we heterologously expressed and characterized the four truncated forms of LamAQ, including Lam560, Lam472, Lam348, and LamCAT (Fig. 1b). The production of full-length enzyme was tried but without success. Among these derivatives, Lam560 exhibited <1% of the specific activity of Lam472 on laminarin under standard conditions, indicating an inhibitory role of the PorSS tail, which is recognized and cleaved by the type IX secretion machinery of clade Flavobacteriia (11). The noncatalytic carbohydrate-binding modules of glycoside hydrolases facilitate substrate accessibility for catalysis by selectively binding target carbohydrates (28), which was confirmed by the higher specific activity of Lam472 with a CBD IV module at its C terminus (Fig. 1c). However, we observed that the temperature adaptation capacity of the enzyme was reduced by the CBD IV module. LamCAT and Lam348 without the CBD module showed >50% and 30% of the maximum activities at 20°C and 90°C (Fig. 1d), respectively, whereas Lam472 activity at these temperatures was barely detected and had a significantly decreased thermal stability (Fig. 1d and e).

FIG 1.

Catalytic properties of laminarinase from the strain SCSIO21287. (a) Organization of the functional units of LamAQ and the module composition of the derivative proteins analyzed in this study. The enzyme contains a signal peptide (SP; residues 1 to 20), GH16 domain (GH16; residues 21 to 254), crystalline domain (residues 255 to 343), carbohydrate-binding type IV domain (CBD IV; residues 344 to 472), and a Por secretion signal (PorSS; residues 473 to 560). (b) Purity of the truncated LamAQ proteins according to SDS-PAGE. (c) Substrate preferences for β-1,3-glucans by laminarinase derivatives. (d) Temperature dependency of enzyme activities. (e) Thermal inactivation profile of laminarinase derivatives at the temperature of 50°C.

To test the substrate specificity of the enzymes, several polysaccharides, including starch (α-1,4), carboxymethyl cellulose (β-1,4), barley β-d-glucan (β-1,3 mixed with β-1,4), curdlan (β-1,3; triple helical), yeast β-d-glucan (β-1,3 mixed with β-1,6; insoluble), and laminarin (β-1,3; soluble), were used as the substrates. All enzymes selectively hydrolyzed substrates containing β-1,3-glycoside bonds (Fig. 1c) and showed no detectable activity toward other kinds of polysaccharides, starch, and carboxymethyl cellulose, indicating catalytic selectivity toward β-1,3-glucan.

Hydrolysis and transglycosylation activities of LamCAT.

We then used the catalytic domain LamCAT as a model enzyme to investigate laminarin hydrolysis mechanisms by matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) and nuclear magnetic resonance (NMR). Mass spectra of the laminarin hydrolysis products showed the peaks of oligosaccharides: laminaritriose plus Na at m/z 527, laminaritetraose plus Na at m/z 689, laminaripentaose plus Na at m/z 851, laminarihexaose plus Na at m/z 1,013, laminariheptaose plus Na at m/z 1,175, and laminarioctaose plus Na at m/z 1,337 (Fig. 2a), with the final smallest sodium adduct peak of the trisaccharide chain, suggesting an endolytic mechanism. Laminarin comprises 20 to 30 glucose units joined by β-1,3-linkages and ∼5% β-1,6-linkages as 6-O side-branching chains. To confirm specificity for β-1,3-linkages, we analyzed the ¹H NMR spectra of laminarin and the hydrolytic products and found that the expected chemical shifts of ¹H signals for pure laminarin corresponded to a β-1,3-d-glucopyranose-linked backbone chain at 4.52 ppm, a reducing terminus at 4.36 ppm, and a β-1,6 side chain at 4.27 ppm (13). Following laminarin incubation with LamCAT, we observed a significant decrease in the backbone signal and a simultaneous increase in signals corresponding to reducing termini formed upon laminarin degradation (Fig. 2b).

FIG 2.

Spectral analyses of enzymatic products of LamCAT. (a) MALDI-TOF MS of the products of laminarin hydrolysis and transglycosylation by LamCAT. The molecular weight for each detected mass peak is indicated. The addition of different acceptors (final concentration, 1 M) is shown (magenta, water; blue, ethylene glycol; green, glycerol; red, erythritol; and yellow, xylitol). (b) ¹H-NMR spectra of laminarin (top) and laminarin hydrolysis products (bottom). The chemical shifts for the β-1,3-glycosidic backbone chain at 4.52 ppm, reducing terminal at 4.36 ppm, and β-1,6 side chain at 4.27 ppm are labeled.

Laminarinases of the GH16 family perform hydrolysis with a retaining mechanism and can potentially exhibit transglycosylation activity, allowing the transfer of a glycosyl residue from laminarin (donor) to molecules containing hydroxyl groups (acceptor) instead of water (29). Incubation of LamCAT with laminarin and 1 M glycerol as an alternative acceptor resulted in new peaks (m/z 601, 763, 925, 1,087, and 1,249) associated with glycerol in addition to those related to hydrolysis (Fig. 2a). Interestingly, LamCAT exhibited an unexpectedly broad specificity with other sugar alcohols as acceptors, including ethylene glycol, erythritol, and xylitol. Although members of the GH16 family reportedly exhibit transglycosylase activities (29, 30), few studies have indicated the use of a broad spectrum of accepters other than glucosyl. The bifunctional feature of the enzyme complicates the identification of an exact physiological function but invites further research opportunities in the field of oligosaccharide synthesis.

The LamCAT structure.

To unravel the molecular mechanisms of substrate specificity of LamCAT, we solved the crystal structure of LamCAT at a resolution of 1.54 Å (Table S1 in the supplemental material). The structure contains one molecule in the asymmetric unit and belongs to the C222₁ space group, comprising two leaflets of antiparallel β-sheets in a jelly roll topology with lengthy loop connections, segmented β-strands, and one small-helix segment (Fig. 3a). One metal ion is found on the convex side of the protein and coordinated by Asp245 (main chain and side chain), Glu30 (main chain), Gly68 (main chain), and one water molecule (Fig. S1a). These residues are the conserved calcium-binding motif among GH16 laminarinases, except for ULam111 from Flavobacterium sp. strain UMI-01, whose coordinated glutamic acid is positioned with histidine. The ULam111 mutation of H12E results in higher thermal stability (24), thereby confirming the essential role of calcium in the stability of GH16 family enzymes (31). The enhancement of thermostability by calcium ion was also observed in LamCAT (Fig. S1b). We clearly observed a cleft spanning ∼30 Å, bound by loops extending from the β-strand. This cleft contains residues Glu135 and Glu140 (Fig. 3a), which are highly conserved among GH16 laminarinases and represent part of the catalytic motif (Fig. S2). Substitution of either site with alanine completely inactivated the enzyme, suggesting its importance in catalysis. These two active carboxyl groups are in close proximity (∼6.7 Å), supporting the space requirement for retaining enzymes (32). Based on a previously proposed catalytic mechanism (26), Glu140 protonates the glycosidic oxygen with concomitant C-O breakage of the β-1,3 glycosidic bond, whereas deprotonated Glu135 acts as a nucleophile to stabilize the oxocarbenium intermediate by forming a covalent glycosyl-enzyme intermediate. Trapping experiments using substrate analogues suggest that the covalent intermediate is more likely formed in retaining glycosidases (33, 34). The second deglycosylation step involves the attack of a water molecule assisted by the conjugate base of Glu140, which allows the free sugar to retain its overall configuration and the enzyme to return to its initial protonated state (Fig. S3).

FIG 3.

Structures of LamCAT and oligosaccharide complexes. (a) Ribbon presentation of the overall LamCAT structure. The catalytic residues Glu135 and Glu140 are indicated as red sticks. Catalytic cavities of apo-E135A (b), E135A-c3g (c), and E135A-lam4 (d) reveal conformational changes upon oligosaccharide binding.

Conformational changes upon oligosaccharide binding.

To elucidate the mechanism associated with β-1,3 glucan recognition of LamCAT, we determined the crystal structures of an E135A mutant of LamCAT, an inactive enzyme variant, in complex with 1,3-β-cellotriosyl-glucose (c3g) and laminaritetraose (lam4) at resolutions of 2.21 Å and 1.90 Å, respectively (Fig. 3b to d). Linked by distinct glycosidic bonds from the 2nd to the 4th glucosyl unit, these two tetrasaccharides showed different affinities to the enzyme, with K_D (equilibrium dissociation constant) values for lam4 and c3g of 0.8 μM and 28.7 μM, respectively (Fig. 4a and b). Each protein-carbohydrate complex displayed three well-defined glucosyl moieties (subsites −1 to −3), whereas the glucosyl unit in subsite −4 showed consistently weaker electron density and higher B-factor values (Fig. 4c and d; Table S2), which can be attributed to fewer protein contacts and increased conformational flexibility. In positive subsites, no electron density was observed, and a well-defined hydrogen-bond network formed by a number of occupied water molecules was established. All the glucosyl units are in ⁴C₁ chair conformation, and dihedral angle analysis showed both tetrasaccharides near the global energy minimum (Table S3) (35, 36). The apo protein E135A and the complexes share the same backbone conformations, with overall Cα root mean square deviation values of <0.2 Å. Nevertheless, the enzyme-ligand complexes possessed clearly expanded catalytic clefts for ligand binding with cavity volumes of 240 ų, 275 ų, and 314 ų for E135A, E135A-c3g, and E135A-lam4, respectively (Fig. S4). A structural comparison shows substrate-induced structural changes of the residues Val128-Ala133 (Fig. S5). The loop moved against the ligands to form open states for tetrasaccharide accommodation. Among those residues, Trp130 forms the most significant orientation change to interact with the subsite −2 glucosyl units of both bound saccharides, which may serve as a key structural feature in β-1,3-glucan recognition.

FIG 4.

Oligosaccharide binding within the LamCAT active site. (a and b) Results of isothermal titration calorimetry (ITC) experiments on E135A titrated with lam4 (a) and c3g (b). Binding of lam4 by E135A resulted in a K_D value of 0.80 ± 0.02 μM, ΔH (enthalpy change) of −12.55 ± 0.03 kcal/mol, and −TΔS (entropy change) of 4.37 kcal/mol for N (number of sites) of 1.26 ± 0.01. Binding of c3g by E135A resulted in a K_D value of 28.65 ± 0.71 μM, a ΔH (enthalpy change) of −8.50 ± 0.13 kcal/mol, and a −TΔS (entropy change) of 2.41 kcal/mol for N (number of sites) of 1.01 ± 0.01. The F_o-F_c omit map (1.5 σ) for lam4 (c) and c3g (d) bound to the LamCAT active cleft.

Interactions between enzymes and oligosaccharides.

The enzyme-ligand interactions are summarized in Fig. 5 and Table S4. The glucosyl units in subsites −3 and −4 of lam4 and c3g showed substantially fewer contacts with the enzyme than in subsites −1 and −2, with only one residue found to interact with Glc(−3) and Glc(−4) each. The Glc(−4) subsites of lam4 and c3g form a water-mediated hydrogen bond with Asn53 and a hydrogen bond with Asn47, whereas the Glc(–3) units of both oligosaccharides interact with the main-chain NH group of Trp130 via water-mediated hydrogen bonding, respectively. The ligand subsites −1 and −2 of the two tetrasaccharides, both connected by β-1,3 glycosidic bonds, share similar conformations; therefore, the residues interacting with Glc(−1) and Glc(−2) can be assumed to be crucial for β-1,3-glycosidic bond selectivity by the enzyme. The subsite Glc(−2) forms a hydrogen bond with residues Asn52 and Arg88 and undergoes hydrophobic stacking with Trp130. The glucose unit of lam4 located at subsite −1 is coordinated by various interactions, including hydrogen bonding formed between Asn52 and O³ and O⁴, water-mediated hydrogen bonding between Trp115 and O⁶, salt bridge between His153 and O¹, water-mediated hydrogen bonding between Tyr161 and O², and water-mediated hydrogen bonding between Asn221 and O⁵. Moreover, we observed that both Asn52 and Trp130 participate in substrate binding as bridges for interactions with subsites −1 and −2 and subsites −2 and –3, respectively. With the observed conformational changes in the indole group of Trp130 resulting in an enlarged catalytic cleft for saccharide accommodation (Fig. S5), the stacking effect is likely essential for substrate binding.

FIG 5.

Enzyme-carbohydrate interactions in subsites −1 through −4. Schematic representation of Lam-E135A bound to lam4 (a) and c3g (b). The dotted lines and double arrows represent hydrogen bonding and hydrophobic stacking interactions, respectively. Water molecules mediating hydrogen bonding are depicted as gray balls.

Based on the structural observations, the substrate selectivity on β-1,3-glucan of LamCAT can be speculated to be associated with a network of hydrogen bonding and hydrophobic stacking interactions formed between several substrate pocket residues and saccharides. To elucidate the roles of residues within the active pocket in catalysis, seven residues (Asn52, Arg88, Trp115, Trp130, His153, Tyr161, and Asn221) were selected for mutagenesis. The kinetics of wild-type LamCAT and its variants against laminarin were determined and compared. We first generated alanine mutations for each site, which resulted in a barely detectable activity against laminarin in all cases (specific activities of <0.1% that of wild-type LamCAT). We then mutated the residues to other amino acids having similar R group structures (Table 1). The catalytic efficiency (k_cat/K_m) on laminarin of all the mutants decreased by factors ranging from 20- to 1,000-fold, suggesting essential roles of all the investigated residues for β-1,3-glucan recognition and hydrolysis. Therefore, this suggests that these catalytic pocket residues are evolutionarily optimized for laminarin hydrolysis, which is also supported by their high conservation among GH16 family laminarinases (Fig. S2). Of all these substrate-binding residues, Trp130 was noted for its significant conformational change, leading to an enlarged catalytic cavity. Moreover, Trp130 can promote substrate positioning to form an efficient Michaelis complex state for catalysis, which was confirmed by the analysis of its mutants with significantly reduced k_cat values by 1,400-fold. Similarly, the interactions of Arg88 and His153 with subsites −2 and −4 might also influence the substrate positioning other than substrate binding for the largely unchanged K_m values of their mutants. We observed that the residues interacting with the O⁶ of the glucosyl unit at subsite −1 are crucial for substrate binding, given that mutants W115F and N221L exhibited 7- and 5-fold increases in the K_m values, respectively. These two residues both interact with the site via water-mediated hydrogen bonding, which is a part of the water network channel in the catalytic cleft of glycoside hydrolases deemed to control substrate preference and activity (37, 38).

TABLE 1.

Catalytic activity of LamCAT and its variants on the substrate laminarin^a

Enzyme	Km (g/liter)	kcat (min−1)	kcat/Km (liter/g/min)	Subsite
LamCAT	2.51 ± 0.27	3,193.90 ± 233.52	1,274.40 ± 43.31	—
N52Q	12.52 ± 3.33	23.21 ± 4.81	1.88 ± 0.14	−1, −2
N52H	6.96 ± 0.62	8.51 ± 0.16	1.23 ± 0.10	−1, −2
R88M	1.83 ± 0.45	102.57 ± 15.37	57.04 ± 6.61	−1, −2
W115F	20.05 ± 4.42	447.94 ± 90.89	23.16 ± 2.39	−1
W130Y	2.16 ± 0.25	2.17 ± 0.19	1.01 ± 0.06	−2, −3
W130H	11.97 ± 2.81	56.55 ± 10.38	4.76 ± 0.30	−2, −3
H153F	1.65 ± 0.08	1.94 ± 0.07	1.17 ± 0.10	−1
Y161F	5.30 ± 1.13	107.11 ± 15.73	23.44 ± 4.21	−1
N221L	15.25 ± 4.47	47.24 ± 14.36	3.10 ± 0.13	−1

^aThe K_m, k_cat, and k_cat/K_m data are means and standard deviations from the means (n = 3). —, not applicable.

DISCUSSION

The depolymerization of algal polysaccharides driven by marine bacterial enzymes is a cornerstone of biogeochemical cycling of organic matter in the ocean. Laminarin is a particularly important component of marine polysaccharides; however, its hydrolyzing mechanisms by bacterial laminarinases are not well understood. In this study, we chose the GH16 laminarinase LamAQ from an alga-associated Flavobacterium as the model to probe the mechanism of bacterial hydrolysis of marine polysaccharides. Different from most bacterial GH16 laminarinases with a single catalytic domain (22, 23, 39, 40), LamAQ comprises several functional modules, including N-terminal signal peptide, catalytic domain, carbohydrate-binding domain, and PorSS tail (Figure 1a). Laminarinases from members of the phylum Bacteroidetes share similar multiple-molecule architecture, among which the C-terminal PorSS domain is unique associated with type IX secretion. Interestingly, we found that the PorSS domain showed inhibitory effects on the laminarinase activity, indicating the removal of PorSS domain via type IX secretion pathway is an essential maturing process for releasing catalytic activity of the enzyme. The C-terminal PorSS, which is only found in the Bacteroidetes phylum, functions as an outer membrane translocation signal recognized and cleaved by type IX secretion system for exporting proteins to the cell surface (41, 42). The N-terminal signal peptide of LamAQ appears to be processed by the Sec system passing through the inner cytoplasmic membrane to the periplasm, and the PorSS tail is then cleaved by sortase, allowing the enzyme to export across the outer membrane. Therefore, the mature form of LamAQ might be absent in both the N-terminal signal sequence and C-terminal PorSS tail.

GH16 is a large and diverse family of glycosidases that adopt a compact β-jelly roll fold and often exhibit activity on β-1,4- or β-1,3-glycosidic bonds in various glucans and galactans. GH16 family members were recently delineated into 23 subfamilies by large-scale analysis of sequences and functions, with several different activities detected among the ∼8,000 sequences in the CAZy database (43). Most GH16 members, including keratan-sulfate endo-1,4-β-galactosidase (EC 3.2.1.103), laminarinase (EC 3.2.1.39), endo-1,3(4)-β-glucanase (EC 3.2.1.6), licheninase (EC 3.2.1.73), β-agarase (EC 3.2.1.81), κ-carrageenase (EC 3.2.1.83), xyloglucanase (EC 3.2.1.151), endo-β-1,3-galactanase (EC 3.2.1.181), β-porphyranase (EC 3.2.1.178), hyaluronidase (EC 3.2.1.35), and endo-β-1,4-galactosidase (EC 3.2.1.103), exhibit GH activity toward plant and marine polysaccharides (44). Notably, some GH16 members, such as yeast chitin β-1,3/1,6-glucanosyltransferase (EC 2.4.1.-) (45), plant xyloglucan:xyloglucosyl transferases (EC 2.4.1.207) (46), and fungal exo-β-1,3-glucosyltransferase/elongating β-transglucosylase (EC 2.4.1.–) (30), are predominant transglycosylases involved in cell wall remodeling. GH16 enzymes are retaining enzymes that utilize breakdown of a covalent glycosyl-enzyme intermediate by transferring glycosyl to water or a carbohydrate acceptor during hydrolysis or transglycosylation reaction, respectively (37). In the present study, we found that LamCAT displayed transglycosylase activity with a broad spectrum of sugar alcohols as acceptors (Fig. 2a), which makes it a promising enzyme for synthesis of diverse oligosaccharides according to its ability to incorporate β-1,3-glucan derivatives into different sugar acceptors. To the best of our knowledge, this unique feature has not previously been reported in GH16 family members, despite similar observations in some carbohydrate hydrolases of other GH families (47–49). In vitro synthesis of complex carbohydrates is of considerable importance for fundamental research in glycoscience and for the preparation of commercially valuable products. Compared with glycosyltransferases using nucleotide sugars as donor substrates, retaining GHs are more stable and have a much wider range of substrate specificities (37). Therefore, these findings suggest the potential benefit of engineering LamCAT for suppressed hydrolase activity and enhanced transglycosylase activity in future studies.

Laminaritriose is the smallest product from laminarin hydrolysis by LamCAT, and we expected to capture the structure of oligosaccharide substrate spanning both the negative and positive subsites using tetrasaccharides for cocrystallization. However, electron density was not observed in the positive subsites, and the four glucose moieties of both tetrasaccharides were modeled at subsites −1 to −4. Aurore et al. (19) also tried to crystallize a hexasaccharide with ZgLamC_GH16, which resulted in the electron density in positive subsites being too disordered for modeling as sugar units, indicating weak binding affinity in positive subsites. Indeed, oligosaccharides at positive subsites are rarely observed in complex structures of enzymes from the GH16 family (27, 50). Previous studies suggest that the distorted skew-boat conformation of sugar unit in subsite −1 results in the preferred axial orientation for the leaving group, making it difficult to obtain simultaneous occupation of negative and positive subsites (25, 51). In the present study, we found that the glucopyranosyl rings in subsite −1 of both lam4 and c3g were parallel to the two conserved tryptophans (Trp115 and Trp119). The position of the −1 glucosyl unit is similar to those reported in ZgLamA_GH16 and Lam16A, whereas the glucose in ZgLamC_GH16 binds perpendicularly to the corresponding aromatic residues Trp117 and Trp121 due to less hydrogen bonding with O⁶ (19).

Structural comparison between apo and complexed enzymes revealed an enlarged catalytic cavity of LamCAT following oligosaccharides binding. Although the switch from “closed” to “open” state was not observed in members of GH16_3 laminarinases, it was reported that the catalytic tunnel of GH16_17 carrageenase PcCgkA adopted a “closed” conformation in the presence of the substrate (52). Among the shifted loop (Val128 to Ala133) promoting open-state catalytic cleft, the indole ring of Trp130 underwent significant conformational changes via a stacking interaction with the −2 glucosyl unit (Fig. S5 in the supplemental material). Trp130 in LamCAT is part of the conserved sequence “WPA…WXX…WPX” (X represents M or L for the second motif and A, K, R, M, or L for the third motif) in the GH16_3 subfamily, members of which can all recognize β-1,3-glucan (43). We subsequently confirmed the role of Trp130 in catalysis by mutagenesis and enzyme kinetic analysis (Table 1). The hydrophobic platform induced by the tryptophan residue interacting with the −2 glucose unit has previously been reported in laminarinases (19, 25, 27, 53). Nonetheless, the tryptophan in neither ZgLamA_GH16 nor ZgLamC_GH16 (both apo and complexed structures) exhibited similar conformational shifts, and the open-closed structural transition is likely to be distinct to LamCAT. The other two conserved tryptophan residues (Trp115 and Trp119) bordering subsite −1 also involved in substrate binding in GH16 laminarinases (Fig. S6). These findings support the premise that the tryptophan residues located in the loop of negative subsites are of functional and evolutionary importance (43).

In conclusion, our study provides an in-depth characterization of the structure and function of a GH16 family laminarinase in the bacterial degradation or remodeling of polysaccharides in marine biomass. The structural and biochemical findings suggest an updated catalytic mechanism of β-1,3-glucan depolymerization and transglycosylation catalyzed by the laminarinase LamCAT from marine Flavobacterium SCSIO21287 (Fig. 6). The solved structures allowed the first reported visualization of conformational changes in LamCAT upon β-1,3-glucan binding, as well as substrate selectivity. Additionally, structural analyses, mutagenesis, and kinetics studies provide novel insights into the residues forming an extended substrate-binding cleft. Moreover, the detected transglycosylase activity and the unique broad spectrum of acceptors shed light on the potential utilization of the enzyme for active oligosaccharide and glycoconjugate synthesis.

FIG 6.

Schematic diagrams of the proposed hydrolytic/transglycosyl mechanisms of the marine bacterial laminarinase LamCAT. The active residue Glu140 protonates the glycosidic oxygen and breaks the β-1,3 glycosidic bond of laminarin (top left), producing an oxocarbenium intermediate that is stabilized by generating a covalent glycosyl-enzyme complex with nucleophile Glu135 (bottom left). The covalent intermediate is then broken down by transferring glycosyl to water or a carbohydrate acceptor during hydrolysis or transglycosylation reaction, respectively (bottom right). Donor-acceptor incorporation releases the generated product from the catalytic cleft (top right). The acceptor for the enzyme can be water, ethylene glycol, glycerol, erythritol, or xylitol. The catalytic cleft accommodates the substrate via an open-state conformation upon β-1,3-glucan binding.

MATERIALS AND METHODS

Cloning, overexpression, and purification.

The gene encoding laminarinase was amplified from the genomic DNA of strain SCSIO21287, with the target GH16 family laminarinase representing a polysaccharide-degrading component of the PUL (Fig. S7 in the supplemental material). A 15-bp overhang on each primer, complementary to the vector sequence, was designed for seamless cloning (TransGen Biotech, Beijing, China). Four truncated derivatives of LamAQ (designated Lam560, Lam472, Lam348, and LamCAT) contained different modules and were designed based on analyzing the conserved domains (Fig. 1a). The target genes were cloned into the pET22b(+) vector (Novagen, Madison, WI, USA) between the restriction sites NdeI and XhoI along with a C-terminal His tag. Recombinant proteins were expressed in the Escherichia coli strain BL21(DE3) (Novagen). Cells (1:200) were inoculated in LB media supplemented with ampicillin (50 μg/ml) and cultured overnight. Upon reaching an optical density of 0.6 at 600 nm, the temperature was decreased from 37°C to 16°C, and the culture was supplemented with 0.5 mM isopropyl β-d-1-thiogalactopyranoside. Protein expression was performed for 16 h, after which cells were harvested by centrifugation at 5,000 rpm for 10 min, followed by resuspension in lysis buffer (20 mM Tris-HCl at pH 8.0, 500 mM NaCl, and 5 mM imidazole). Cells were lysed by sonication using 1-s pulse/1-s pause cycles, and cell debris was removed by centrifugation at 40,000 × g for 30 min. The supernatant fraction was loaded onto an equilibrated Ni-nitrilotriacetic acid (NTA) resin (Qiagen, Hilden, Germany). After extensive washing with wash buffer (20 mM Tris-HCl at pH 8.0, 500 mM NaCl, and 20 mM imidazole), the recombinant proteins were eluted with elution buffer (20 mM Tris-HCl at pH 8.0, 500 mM NaCl, and 200 mM imidazole). The eluate was further purified by loading onto a Mono Q column (GE Healthcare, Chicago, IL, USA), and elution was performed with a linear gradient of 0 M to 1.0 M NaCl. All purified proteins were concentrated with Vivaspin 20 devices (5,000 molecular weight cutoff [MWCO] polyethersulfone [PES]; Sartorius, Göttingen, Germany).

Enzyme assay.

Enzymatic hydrolysis of laminarin was performed by incubating wild-type and mutant laminarinases (5 nM to 50 μM) with laminarin (0.1 to 10 g/liter; Sigma-Aldrich, St. Louis, MO, USA) in 20 mM Tris-HCl buffer (pH 8.0) at 50°C for 30 min. The reactions were stopped by the addition of 3,5-dinitrosalicylic acid (Sigma-Aldrich), and catalytic rates were determined according to the release of reducing sugar in linear 1-min intervals. One unit of laminarinase activity was defined as the amount of enzyme required to liberate reducing sugar equivalent of 1.0 μmol glucose per minute under standard conditions. Data for the initial reaction rate (v₀; μM/min) versus [S]₀ (g/liter) were fitted to the Michaelis-Menten model to calculate kinetics parameters.

MS and NMR analysis.

For spectrum analysis of laminarinase products, 1 ml of 1% laminarin solutions with 100 μg purified recombinant laminarinase and 20 mM Tris-HCl (pH 8.0) were incubated at 50°C for 1 h, and reactions were stopped by transferring samples to a boiling water bath for 10 min. To determine whether the enzyme exhibits transglycosylation activity, putative acceptors (glycerol, ethylene glycol, erythritol, xylitol, and mannitol) at 1 M were added to the reaction solution. MALDI-TOF MS (UltrafleXtreme; Bruker, Billerica, MA, USA) was used to analyze the mass/charge ratios of laminarin hydrolysis products using an accelerating voltage of 25 kV. Before NMR analysis, the reaction mixtures were dried using a refrigerated speed vacuum and then dissolved in 0.5 ml of dimethyl sulfoxide (DMSO)-d₆ (Sigma-Aldrich). High-field ¹H-NMR spectroscopy was performed using an Avance 700-MHz spectrometer instrument (Bruker). All ¹H-NMR spectra were analyzed using Mnova software (Mestrelab Research, Santiago de Compostela, Spain).

Site-directed mutagenesis.

Primers were designed with substituted codons at target sites to generate mutants by one-step site-directed mutagenesis (54). PCR amplification was performed with PrimeStar HS DNA polymerase (TaKaRa Biotechnology, Chiba, Japan) using the following program: 95°C for 10 min, followed by 32 cycles of 95°C for 30 s, 50°C for 30 s, and 72°C for 7 min, and then 72°C for 10 min. PCR products were digested with the restriction enzyme DpnI (MBI Fermentas, Vilnius, Lithuania) to remove the methylated parent plasmid and then purified with a PCR purification kit (Qiagen). The resulting linearized plasmid derivatives were used to transform E. coli XL1-Blue competent cells to obtain mutant plasmids, after which target plasmids were used to transform E. coli BL21(DE3) cells for the production of mutant enzymes.

Isothermal titration calorimetry.

Oligosaccharide-binding assays were performed using a MicroCal iTC₂₀₀ isothermal titration calorimeter (GE Healthcare). The sample cell was filled with 200 μl of inactivated enzyme solution to a final concentration of 100 μM in 20 mM Tris-HCl (pH 8.0) and 200 mM NaCl. Nineteen consecutive 2.0-μl aliquots of 1 mM laminaritetraose (Megazyme, Bray, Ireland) or 1,3-β-cellotriosyl-glucose (Megazyme) were injected into a prepared protein solution at 150-s intervals at 20°C. The first injection volume was 0.4 μl, and the observed thermal peaks were excluded from data analysis. Data fitting was performed in the “one set of sites” mode, and K_D values were calculated using Origin software (MicroCal, Northampton, MA, USA).

Crystallization, X-ray diffraction, and structure determination.

All crystallization trials were performed using the sitting-drop vapor diffusion method. Crystals of the catalytic domain LamCAT and its inactive mutant E135A were obtained by mixing 0.4 μl of the purified protein (25 mg/ml) with 0.4 μl of reservoir solution and allowing the drop to equilibrate at 20°C in 96-well protein crystallization plates (Violamo, Osaka, Japan). The reservoir solutions for the crystallization of LamCAT and Lam-E135A were 100 mM morpholineethanesulfonic acid (MES) (pH 7.0), and 15% (vol/vol) ethanol and 100 mM MES (pH 7.0), 15% (vol/vol) ethanol, and 28% (mass/vol) PEG 20000, respectively. To obtain protein-ligand complexes, E135A crystals were soaked overnight in reservoir solution with 1 mM lam4 or c3g. A single-protein crystal was picked and soaked in reservoir solution containing 22.5% (vol/vol) ethylene glycol as a cryoprotectant, and a diffraction data set was collected on beamline BL-5A at Photon Factory (Tsukuba, Japan). The diffraction data were indexed, integrated, and scaled using XDS (55), and the structure of the catalytic domain of laminarinase ZgLamC from Zobellia galactanivorans (PDB ID: 4CRQ; amino acid sequence identity, 64%) was used as a model for molecular replacement by Phaser (56). Iterative refinement cycles were performed using REFMAC (57), phenix.refine (58), and Coot (59). Data collection and refinement statistics are presented in Table S1. All structures were depicted using PyMOL (v.2.3; Schrödinger, LLC, New York, NY, USA).

Data availability.

The atomic coordinates and structure factors for apo-LamCAT, apo-E135A, E135A in complex with lam4, and E135A in complex with c3g have been deposited in the Protein Data Bank under accession codes 6JH5, 6JHJ, 6JIA, and 6M6P, respectively.