Exploring a novel β-1,3-glucanosyltransglycosylase, MlGH17B, from a marine Muricauda lutaonensis strain for modification of laminari-oligosaccharides

Abstract The marine environment, contains plentiful renewable resources, e.g. macroalgae with unique polysaccharides, motivating search for enzymes from marine microorganisms to explore conversion possibilities of the polysaccharides. In this study, the first GH17 glucanosyltransglycosylase, MlGH17B, from a marine bacterium (Muricauda lutaonensis), was characterized. The enzyme was moderately thermostable with Tm at 64.4 °C and 73.2 °C, but an activity optimum at 20 °C, indicating temperature sensitive active site interactions. MlGH17B uses β-1,3 laminari-oligosaccharides with a degree of polymerization (DP) of 4 or higher as donors. Two glucose moieties (bound in the aglycone +1 and +2 subsites) are cleaved off from the reducing end of the donor while the remaining part (bound in the glycone subsites) is transferred to an incoming β-1,3 glucan acceptor, making a β-1,6-linkage, thereby synthesizing branched or kinked oligosaccharides. Synthesized oligosaccharides up to DP26 were detected by mass spectrometry analysis, showing that repeated transfer reactions occurred, resulting in several β-1,6-linked branches. The modeled structure revealed an active site comprising five subsites: three glycone (−3, −2 and −1) and two aglycone (+1 and +2) subsites, with significant conservation of substrate interactions compared to the only crystallized 1,3-β-glucanosyltransferase from GH17 (RmBgt17A from the compost thriving fungus Rhizomucor miehei), suggesting a common catalytic mechanism, despite different phylogenetic origin, growth environment, and natural substrate. Both enzymes lacked the subdomain extending the aglycone subsites, found in GH17 endo-β-glucanases from plants, but this extension was also missing in bacterial endoglucanases (modeled here), showing that this feature does not distinguish transglycosylation from hydrolysis, but may rather relate to phylogeny.


Introduction
Macroalgae from the marine environment include fast growing species with potential as resources for next generation biorefineries.Brown macroalgae are of interest, due to high bulk biomass production and high carbohydrate content.One of the main types of polysaccharides in brown macroalgae is laminarin, which is a storage β-glucan (Hakim and Patel 2020;Konstantin et al. 2023).Glucans are polysaccharides derived from d-glucose, linked by glycosidic bonds.Laminarin is a low molecular weight β-1,3-d-glucan polymer with sporadic β-1,6-linkages and a degree of polymerization (DP) of 20-35 glucose units (Becker et al. 2020).The reducing ends are capped with either a mannitol or a glucose moiety, and the extent of branching is important for the solubility of the polymer.Other types of β-1,3-linked glucans can be found in the cell walls of terrestrial plants, bacteria and fungi, including callose (from plants), and curdlan (from the bacterial family rhizobiaceae), which are not reported to contain 1,6linkages, while β-1,6-linkages can be found in β-1,3-linked glucans from baker's yeast (Sajna et al. 2015).Laminarin is mainly found in the brown seaweed genera Laminaria and Saccharina, but also to a lesser extent in other species in the genera Ascophyllum and Undaria (Kadam et al. 2015).Currently there is an increased interest in polymers from seaweeds, as it is a biomass that does not compete with arable land.In relation to this, the interest in laminarin, and laminarin-derived oligosaccharides, has also increased as these compounds can be expected to elicit the same healthpromoting responses as reported for other β-glucans (Jönsson et al. 2020).
Laminarin can be selectively converted to oligosaccharides by controlled enzymatic action.Endo-acting enzymes are in this aspect of interest, as they can be utilized in production of value-added laminari-oligosaccharides (Chesters and Bull 1963;Dobruchowska et al. 2016).Candidates catalyzing endohydrolysis of β-1,3 glucans (EC 3.2.1.39)have, based on sequence similarities, been classified under ten different glycoside hydrolase families (GHs) according to the CAZy database (cazy.org, visited 2023-10-11) (Drula et al. 2022), but only a few families have consistent laminarinase (or β-1,3 glucanase) activity.In GH5, GH16, and GH55, only few candidates display activity on laminarin, interspersed with Fig. 1.Nomenclature for sugar binding subsites in MlGH17B.The non-reducing end of the donor substrate is bound to subsite −3 while the reducing end is bound to subsite +2.After cleavage and release of the Glc 2 product, the donor substrate remains in the glycone subsites (−3,−2, and −1subsites) while the acceptor substrate is bound to the aglycone subsites (+1 and +2) in an alternative position for the transglycosylation reaction to happen.
Despite the large number of characterized enzymes in the GH17 family, the large majority are of eukaryotic origin and to date, only eight candidates of bacterial origin are listed in the CAZy database (visited 2023-10-11), which all originate from terrestrial habitats.Interestingly, among these eight enzymes, two distinct activity groups can be distinguished: glucan endo-1,3-β-glucosidases (EC 3.2.1.39)and transglycosylases .The transglycosylases are few, and to date the identified bacterial GH17-transglycosylases all originate from proteobacteria from soil habitats (Hreggvidsson et al. 2011;Linares-Pastén et al. 2021).In addition to this, two fungal transglycosylases can be found in GH17, originating from Aspergillus fumigatus (Okada et al. 1998) and Rhizomucor miehei (Qin et al. 2015).None of the previously identified transglycosylases, originate from marine environments.
The five listed GH17 transglycosylases cleave the donor substrate and transfer the remaining part to a new substrate molecule, but the type and position of the new linkage is dependent on the enzyme, and include: β-1,3-elongation (Hartland et al. 1996;Hreggvidsson et al. 2011), β-1,4elongation (Hreggvidsson et al. 2011), and β-1,6-elongation or branching (Gastebois et al. 2010;Hreggvidsson et al. 2011;Dobruchowska et al. 2016;Aimanianda et al. 2017).Elongation implies transfer of the bound glycone (the donor) to the non-reducing end of an incoming substrate molecule bound to the aglycone subsites (the acceptor) (Fig. 1), while branching implies transfer to a sugar moiety, inside the incoming aglycone chain.Due to the limited number of characterized transglycosylating enzymes, and the variations in product profiles of the so far characterized enzymes, more information on GH17 transglycosylases is needed, especially considering the lack of characterized marine candidates.
In this study, a novel GH17 glucanosyltransglycosylase, from Muricauda lutaonensis, a moderately thermophilic marine bacterium, classified as a member of the family Flavobacteriaceae (Arun et al. 2009a), known to contain many efficient biomass degraders, was subjected to thorough characterization.Selection of a moderately thermophilic species was made as enzymes from these sources often display stability for application purposes.The product profile of the recombinant enzyme was investigated in the presence of laminarin and laminari-oligosaccharides (Glc 2-6 ) by employing high-performance anion exchange chromatography with pulsed amperometric detection (HPAEC-PAD) and thin layer chromatography (TLC) methods.The enzyme's ability in branch formation and production of long chain oligosaccharides was investigated by matrix-assisted laser desorption ionization time-of-flight (MALDI-TOF) as well as one-dimensional and two-dimensional nuclear magnetic resonance (NMR) spectroscopy.Furthermore, bioinformatic analysis, including modeling of the three-dimensional structure, was performed to gain more insight into the structure and catalytic cleft of the enzyme, supporting it as the first GH17-transglycosylase isolated from a marine environment.

Results
Sequence analysis of the 16S rRNA gene, and the identification of the gene encoding MlGH17B in strain ISCAR-4703 In order to find marine bacteria with a potential to produce enzymes with sufficient stability, sampling for candidate species was made at the coastal, intertidal, marine geothermal Yngingarlindir site, rich in seaweeds, located off the coast of the Reykjanes peninsula (Iceland).This was followed by isolation of aerobic species using marine agar, as detailed in materials and methods.Genomic DNA was then prepared from the isolate (termed ISCAR-4703) followed by sequencing and assembly into a draft genome.The amplified and partially sequenced 16S rRNA gene demonstrated >99% identity to the 16S rRNA gene (GenBank EU564844.1) of M. lutaonensis which is a moderately thermophilic member of the family Flavobacteriaceae (Arun et al. 2009b) and allowed classification of the strain as M. lutaonensis ISCAR-4703.The finding that the isolated bacterium was a member of the genus Muricauda was interesting as members from this genus are relatively understudied, To find GH17 candidates, the genome was subsequently searched for homologs to identified transglycosylases, allowing identification of the gene encoding MlGH17B.Sequence similarity search by blastp, with MlGH17B as query sequence against the non-redundant protein sequence database demonstrated that the encoded amino acid sequence of MlGH17B (GenBank WIW39500.1)was identical over the complete 295 residue sequence to a gene annotated as a putative glycosyl hydrolase (GenBank MBC31414.1) in the draft genome (Gen-Bank PARJ01000001.1) of Muricauda sp.SP11 identified from a marine metagenome from the Tara global ocean expedition (Tully et al. 2018).The enzyme lacks signal peptide, and suggests that it is cytoplasmic with a potential role in the cell's energy storage.
The blastp-search also revealed a large number of deposited genes that encoded homologous putative enzymes from the GH17 family.The deduced amino acid sequences of the 30 best matches, displayed sequence identities to MlGH17B ranging from 74.6% to 100%, over 100% query coverage.Most of these homologous genes were originating from various Muricauda species, demonstrating that this type of single module enzyme appears common among related bacteria in the marine environment.None of these genes has however, been cloned or its gene product characterized.
Sequence similarities to characterized enzymes were significantly lower.Candidates found by blastp in the Swiss-Prot database, for example resulted in a best match with a glucan-1,3-glucosidase (Swiss-Prot P15703.1)from Saccharomyces cerevisiae, with only 27% identity over 78% query coverage.Blastp search in the Protein Data Bank (PDB), demonstrated that the most similar structure determined candidate was the fungal GH17 enzyme designated as a 1,3β-glucanosyltransferase RmBgt17A (PDB 4WTP) from R. miehei with 33.8% identity over 77% query coverage.
Surprisingly, the blastp searches did not result in significant matches with the bacterial enzymes, listed as characterized in the CAZy database (cazy.org, 2023-10-11) (Drula et al. 2022), from which the sequences of the catalytic modules were originally used to find the gene in the genome sequence.The lack of hits is judged to be due to that these enzymes are deposited in their original multi-modular form, resulting in low query coverage.To further display similarities and differences, an alignment of the catalytic modules of the characterized bacterial enzymes, in the sequence-format used for recombinant production (Linares-Pastén et al. 2021), was generated.The alignment also included the fungal transglycosylase RmBgt17A, and the catalytic module of the bacterial transglycosylase Glt20 (originating from the soil-bacterium Bradyrhizobium diazoefficiens, despite not yet being indexed as characterized in the CAZy database, visited 2023-10-11) (Fig. 2).It is noteworthy to mention that both RmBgt17A and Glt20 show β-1,6-elongation or branching, in line with the activity found for MlGH17B (see below).
MlGH17B is shown to be a single-domain enzyme, consisting only of a catalytic module, whereas the characterized proteins from Proteobacteria (termed Glt1, Glt3, Glt7) (Hreggvidsson et al. 2011) have an N-terminal domain belonging to GH17 and a C-terminal domain belonging to the Leloir glycosyltransferase family GT2 (Hreggvidsson et al. 2011).The alignment demonstrated that sequence identities to MlGH17B were rather low, and in the same range for the catalytic modules of transglycosylases from proteobacteria and for the single module fungal transglycosylase RmBgt17A (Table 1), although the proteobacterial sequences shared higher sequence similarities with each other, than with either MlGH17B or RmBgt17A (Table 1), indicating some differences in conserved residues.Moreover, the sequence identity (16%-22%) was even lower between MlGH17B and bacterial laminarinases/endo-β-1,3 glucanases (Badur et al. 2020;Kumagai et al. 2022).

Production and purification of the recombinant MlGH17B
The relatively low sequence homologies to characterized enzymes, the low number of characterized transglycosylases, and the wide product spectrum of the previously characterized transglycosylases in GH17, motivated exploring the structure and function of the novel identified MlGH17B, which was subsequently produced using Escherichia coli as an expression host.
Production of a soluble entity was achieved by using an N-terminal MBP solubility tag, MBP-Smt3-MlGH17B, that comprised the productivity of approximately 0.16 g/Lh of the soluble recombinant protein fraction produced by the host cells.Cleavage of MBP with Ulp1 protease (recognizing the Smt3 motif between MBP and the enzyme sequence) was possible, however MlGH17B without MBP precipitated instantly and investigation of the recombinant enzyme properties necessitated use of the fusion-protein.
Purification of MBP-Smt3-MlGH17B by affinity chromatography ensured a stable fusion-protein of high purity, as assessed by SDS-PAGE (Supplementary Fig. S1).The molecular weight of MBP-Smt3-MlGH17B-6×His was estimated to 86 kDa, and comprised the MBP-Smt3 domain (54 kDa), the MlGH17B (31 kDa), and the 6×His-tag (0.8 kDa) confirming the expected molecular weight for the complete target construct.Activity was only detected in fractions where the MBP-domain was not cleaved or only partly cleaved off from MBP-Smt3-MlGH17B.Hence, the enzyme is not active after precipitation.

Glt7_ACO80054.1 Glt20
FaGH17A_CDF79584  melting point is in line with previous data, as estimated in the presence of 50 mM maltose, and it has been shown to vary between 59.3-72.1 • C as it is affected by buffer composition as well as buffer pH (Novokhatny and Ingham 1997).

pH and temperature optimum
Initial screening to find an optimal pH for activity was performed in the pH range 3.5-9 at room temperature using Glc 5 as the substrate.This enabled estimation of the pH optimum of MlGH17B to approximately pH 6.0.Further analysis was subsequently performed at a narrower pH range (pH 5, 5.5, 6, and 6.5) to confirm the pH optimum.Product formation was subsequently measured at a temperature range of 13-42 • C during 2 h reaction time, which resulted in estimated pH and temperature optima (Fig. 3) at pH 6.0 and 20 • C, respectively.The activity of MlGH17B was assayed with different oligo-and polysaccharides of varying length, branching, and glycosidic linkage types.The reaction mixtures were incubated at the optimum conditions for 24-72 h.Analysis by TLC and HPAEC-PAD, did not reveal any detectable activity on chitosan, xylan, lichenan, maltodextrin, amylose, maltooligosaccharides, or gentiobiose.However, activity toward laminari-oligosaccharides (Glc 4-6 ) and laminarin confirmed that the enzyme has high specificity toward glucan substrates with β-1,3-linkages.
The product profile obtained from HPAEC-PAD (Fig. 4) clearly illustrated that the enzyme was able to act on laminarioligosaccharides with a DP of 4-6 (Glc 4 , Glc 5 , and Glc 6 ).The activity on Glc 4 was, however, very low compared to the activity on Glc 5 and Glc 6 which were completely utilized by the enzyme in the selected incubation period.The increase in intensity of the peak corresponding to the laminaribiose standard (Glc 2 ) confirmed that the enzyme cleaved off two sugar units regardless of the size of the substrate.A Glc 2 product was also observed after incubation with a laminarin polysaccharide, and the remaining part was transferred to another molecule of the substrate (or extended product), acting as acceptor substrate.However, detection of the transfer products was not possible using laminarin as substrate due to the complexity and size distribution of the polysaccharide substrate (Supplementary Fig. S3).
After a 24 h incubation of the enzyme with the substrates, a trace amount of the DP3 product, resulting from hydrolysis of the Glc 5 substrate, and the DP4 product, resulting from hydrolysis of the Glc 6 substrate were detected.However, the amount of these products were negligible, compared to the respective transferase products.

MS analysis of transferase products of MlGH17B
The reaction mixture of MlGH17B with Glc 4 , Glc 5 , and Glc 6 , after 48 h incubation at the activity optimum condition, was analyzed by MS.MALDI-TOF mass spectra agreed well with the product profile of the enzyme, detected by HPAEC-PAD, demonstrating the release of Glc 2 and the transfer of the remaining donor molecule to another incoming substrate molecule functioning as acceptor molecule.
Although the activity of MlGH17B on Glc 4 was negligible compared to the activity on Glc 5 and Glc 6 , product spectra detected by MALDI-TOF, after 48 h incubation (Fig. 5A), demonstrated four predominant quasi-molecular ions [M + K] + at m/z 705.24, m/z 1029.4,m/z 1191.4,and m/z 1515.5 corresponding to the substrate Glc 4 and the transferase products Glc 6 , Glc 7 , and Glc 9 respectively.Additionally, a peak denoting Glc 10 was observed with quasi-molecular ions [M + K] + at m/z 1677.6,albeit less pronounced.Moreover, minor peaks associated with quasimolecular ion [M + K] + at m/z 867.5, m/z 1353.5, m/z 1839.7 and m/z 2001.7 representing Glc 5 , Glc 8 , Glc 11 , and Glc 12 were also detected.
The MS data showed a substantial quantity of unexpected products using Glc 4 as substrate, such as Glc 7 and Glc 9 , which were generated during the course of the reaction.As Glc 4 cannot occupy all putative five binding sites in the enzyme, it is likely that it binds in two alternative positions, specifically in the −3 to +1 subsites, as well as in the −2 to +2 subsites that is required for cleavage of Glc 2 .This may result in the cleavage of only one glucose molecule from the reducing end, instead of the predominantly formed Glc 2 .This alternative subsite binding could subsequently be confirmed by HPAEC-PAD product profile analysis in which a small peak corresponding to Glc was detected after 24 h incubation (Fig. 4).A DP4-oligo acceptor molecule will then enable the creation of a β-1,6linkage, leading to the production of Glc 7 .
Following 48 h incubation with Glc 5 , MALDI-TOF mass spectra (Fig. 5B) displayed four major quasi-molecular ions [M + K] + at m/z 867.2, m/z 1353.3,m/z 1839.5, and m/z 2325.8 corresponding to Glc 5 , Glc 8 , Glc 11 and Glc 14 .The minor presence of Glc 4 at m/z 705.14 can be attributed to a slight contamination of the Glc 5 substrate.The transferase products derived from Glc 5 exhibited a three glucose unit interval, denoting the bound donor, in which Glc 8 was the first major product, where Glc 5 served as acceptor.Lesser peaks at m/z 2812.4 and m/z 3301 corresponding to quasi-molecular ion [M + K] + of Glc 17 and Glc 20 were also observed.The minor peaks at m/z 1029 and m/z 1191.4,representing quasimolecular ion [M + K] + of Glc 6 and Glc 7 could be transferase product when Glc 4 acted as donor and/or acceptor substrate, respectively.
A similar pattern was observed when the enzyme was incubated with Glc 6 (Fig. 5C).The MALDI-TOF product profile of MlGH17B after 48 h incubation with Glc 6 illustrated four primary quasi-molecular ions [M + K] + at m/z 1029.3,m/z 1677.6,m/z 2326.1, and m/z 2974.8 corresponding to Glc 6 , Glc 10 , Glc 14, and Glc 18 , confirming the cleavage of Glc 2 and transferring the remaining donor part to another molecule of substrate.Nevertheless, minor peaks indicative of Glc 22 with quasi-molecular ion [M + K] + at m/z 3623.8 and Glc 26 with quasi-molecular ion [M + K] + at m/z 4271.8 were also detected.This long incubation time resulted in occurrence of a peak at m/z 705.3, corresponding to quasi-molecular ion [M + K] + of Glc 4 , which is the product of the cleavage, and a peak at m/z 2164 corresponding to quasi-molecular ion [M + K] + of Glc 13 .

NMR studies of the transferase product of MlGH17B
The one-dimensional 1 H spectra of purified transglycosylation products, Glc 8 , Glc 11 , and Glc 14 were very similar (Fig. 6).The one-dimensional 1 H NMR spectra of Glc 8 combined with the two-dimensional total correlation spectroscopy (TOCSY) (150 ms) (Fig. 7) and heteronuclear single quantum coherence (HSQC) measurements (spectra not shown) of Glc 8 , showed anomeric signals for
The NMR analysis indicated that Glc 7 likely forms following the binding of the donor substrate to subsites −3 to +1, with Glc acting as a leaving group.The presence of a laminaritetraose (Glc 4 ) acceptor molecule then facilitates the formation of a β-1,6-linkage, resulting in the production of Glc 7 .This unexpected product pattern, observed in the presence of Glc 4 , could be due to the substrate not occupying all the enzyme's subsites.This allows for two alternative positions (as previously mentioned), demonstrating that binding at either subsite −3 or +2 can be favored (Supplementary Fig. S4).

Three-dimensional structure model and validation
To date, only one crystal structure of a GH17 transglycosylase is solved, which is the structure of the fungal enzyme RmBgt17A from R. miehei (Qin et al. 2015).Homology models are, however, available for three bacterial transglycosylases: Glt1 from P. aeruginosa, Glt3 from P. putida, and Glt20 from B, diazoefficiens (Linares-Pastén et al. 2021).Due to the low number of available structures, MlGH17B was modeled using AlphaFold2 (Jumper et al. 2021;Mirdita et al. 2022) using the Colab notebook to determine the structural organization (Fig. 9).
The predicted three-dimensional model of the enzyme, generated by AlphaFold2 indicated that the enzyme has a TIM-barrel ((β/α) 8 ) fold that spans the 295 amino acids.A significant portion of this model was characterized by a high Predicted Local Distance Difference Test (pLDDT) score, exceeding 90%.Two segments of the model displayed a moderate pLDDT score: the N-terminal segment (first 18 amino acid residues) and the region from Ser270 to Gly280 (SWKVGSEGDVG) corresponding to η6-α8-η7 (Fig. 9 and Supplementary Fig. S5), with pLDDT scores ranging from 70 to 90%.Moreover, the predicted position of Glu34 was categorized with a moderate confidence level, falling between 80-90% (these regions are shown in red boxes in Supplementary Fig. S5).
The conserved catalytic residues, Glu133 and Glu223 (shown in Fig. 10), were located at about two-thirds along the length of the catalytic cleft at an inter-residue distance of 5 Å.The length and geometry of the catalytic cleft makes it possible for the enzymes to accommodate longer-chain oligosaccharides and laminarin (as confirmed by the activity determinations).
A hybrid homology model was also generated in the YASARA program with the crystal structures of RmBgt17A from R. miehei (PDB 4WTP, 4WTR, and 4WTS, 28.73% sequence identity) and the laminarinase FbGH17A from Formosa sp.(PDB 6FCG) with sequence similarity 21.92% selected as main templates to create different fragments of the model (Table S1).The model's quality was assessed through ERRAT, VERIFY3D, and PROCKECK using the UCLA-DOE LAB-SAVES v6.0 server (saves.mbi.ucla.edu).The results indicated that the YASARA-generated model exhibited    satisfactory quality.Further details on the model's quality assessment can be found in the supplementary materials, specifically in the "Quality assessment of YASARA threedimensional model" section.Nevertheless, a comparison with the AlphaFold2 model revealed variations in the size and orientation of secondary structures' elements, particularly in the loops surrounding the active site (Supplementary Fig. S5).
Subsequently, the AlphaFold2 model was employed for further investigations involving the docking of Glc 5 to the enzyme's active site.
The docking of laminaripentaose (Glc 5 ) was carried out using local docking with AutoDock, which is implemented into YASARA.To evaluate the quality of the resulting docked model, it was compared to the structure of RmBgt17A cocrystallized with laminaribiose (PDB 4WTR) and laminaritriose (PDB 4WTS) (Fig. 11).This comparison involved assessing the alignment of the ligand, its conformation, glycosidic linkages within the sugar units, as well as the interactions between the ligand and the enzyme.
MlGH17B was lacking the additional subdomain (characteristic for plant endo-β-glucanases in GH17) located at the end of the catalytic cleft in the aglycone region.This subdomain has been associated with the extension of the aglycone subsites in the eukaryotic plant enzymes.Qin et al. (Qin et al. 2015) demonstrated that the fungal transglycosylase RmBgt17A lacks this subdomain, and Linares-Pastén et al. (Linares-Pastén et al. 2021) confirmed that proteobacterial enzymes (Glt1, Glt3, and Glt20) also lack this subdomain, indicating that this may play a role for transglycosylation.However, structural models, generated for bacterial GH17 endo-β-glucanases using Alphafold2 in this study (supplementary Fig. S6), including FaGH17A (GenBank CDF79584.1)from Formosa agariphila, FbGH17B (GenBank AOR29491.1)from Formosa sp.Hel1_33_131, VbGH17A  (GenBank OEF87991.1)from Vibrio breoganii, VvGH17 (GenBank ASM98089.1)from Vibrio vulnificus, as well as the crystal structure of FbGH17A (GenBank AOR29489.1),from Formosa sp.Hel1_33_131 consistently revealed absence of the subdomain at the aglycone part of the active site.As a consequence, all bacterial GH17 enzymes encompassing both laminarinases and transglycosylases (including MlGH17B) exhibit shorter active sites within the aglycone region in comparison to the eukaryotic plant endo-β-glucanases.
The active site of MlGH17B is surrounded by 8 loops, which may influence the temperature optimum for activity, while the core of the MlGH17B exhibits higher temperature stability according to DSF as the T m of the protein is significantly higher than the activity optimum.Subsite −3 is surrounded by loops β1-α1, β2-α2 and β3-α3; subsites −2 is surrounded by the β8-α8 loop; subsite −1 is surrounded by the β4-α4, β7-α7 and β8-α8 loops; subsite +1 is surrounded by loops β4-α4, and β6-α6, and subsite +2 is surrounded by loops β5-α5 and β6-α6, with the proton donor (Glu133) and nucleophile (Glu223) residues located in loop β4-α4 and at the C-terminal of β7, respectively.
The potential substrate interacting residues, were indicated as those located within 5 Å around the Glc 5 ligand (Fig. 11A) with conserved positions of interacting residues between MlGH17B and RmBgt17A from R. miehei (Fig. 11B) which was co-crystallized with laminaribiose (PDB 4WTR) and laminaritriose (PDB 4WTS) (Qin et al. 2015).Active site comparison between MlGH17B and RmBgt17A (Fig. 11B) revealed 13 conserved residues between the enzymes, excluding the proton donor (Glu133 in MlGH17B and Glu99 in RmBgt17A) and nucleophile (Glu233 in MlGH17B and Glu189 in RmBgt17A) residues.Among these residues, Glu112 (Glu75 in RmBgt17A), Arg137 (Arg103 in RmBgt17A), Tyr167 (Tyr135 in RmBgt17A) and Tyr168 (Tyr136 in RmBgt17A) are conserved between both enzymes and occupy corresponding positions in the superimposed structures (Fig. 11B), indicative of common functions, although these residues at corresponding positions, were not reported by Qin et al. (Qin et al. 2015) as directly involved in the hydrolysis or transglycosylation events of RmBgt17A.The potential ligand interactions between MlGH17B and the Glc 5 ligand are visually presented in Fig. 12 and detailed in Table 3, while a comparison of the interactions in MlGH17B and RmBgt17A is shown in Table 4.
At the glycone subsites, Phe76 (Phe38 in RmBgt17A), Trp101 (Trp65 in RmBgt17A), Glu269 (Glu235 in RmBgt17A), and Phe264 (Phe230 in RmBgt17A) are conserved and superimpose well.Glu276 (Glu242 in RmBgt17A) is also conserved but did not superimpose with the residue in RmBgt17A.The position of Glu276, however, falls within a region of the model with a pLDDT score ranging from 70 to 90%.The variation in the structural position of Glu276 compared to Glu242 in RmBgt17A could be attributed to the model's lower confidence in this specific region (Fig. 9 and Supplementary Fig. S5).In the glycone region, Asn132 and Lys272 in MlGH17B were replaced by Ser98 and Arg238 in RmBgt17A, respectively (Fig. 11).At subsites −2 and −3, aromatic residues including Phe76 (Phe38 in RmBgt17A), Trp101 (Trp65 in RmBgt17A), and Trp271 stacked against the glucose units in the docked ligand, forming a hydrophobic sugar-binding platform.Potential hydrogen bonds were observed between Lys272 and OH6 of the glucose in −2 glycone subsite, and between Glu276 and OH3 and OH6 of the glucose in −1 subsite.Moreover, Glu133 potentially forms hydrogen bonds with OH2 in the −1 subsite, and Trp191 could potentially form a hydrogen bond with OH6 in the same subsite.
The conserved residues in the aglycone subsites also superimpose well in the two structures (Fig. 11B).Three residues, Trp191 (Trp157 in RmBgt17A), Glu192 (Glu158 in RmBgt17A) and Tyr136 (Tyr102 in RmBgt17A) are conserved and reported to play an important role in the transglycosylation transition and β-1,6-linkage formation (Qin et al. 2015).Moreover, Asn186 (Asn152 in RmBgt17A) is also conserved between MlGH17B and RmBgt17A.However, Phe154 in RmBgt17A is replaced by Tyr188 in MlGH17B.The residues Tyr167 (Try135 in RmBgt17A), Tyr168 (Tyr136 in RmBgt17A), and Glu192 (Glu158 in RmBgt17A) were found near the reducing end of the glycosyl at the +2 subsite (Fig. 11A and B), and potentially constrain both the entrance position and the orientation of the acceptor substrate within the catalytic cleft.Potential hydrogen bonds were observed between Glu192 and OH1 of the glucose moiety in the +2 aglycone subsite, between Ser275 and OH6 of the glucose moiety in the +1 subsite, and between the catalytic residue Glu133 with OH2 and OH3 of the glucose moiety in +1 subsite.
As evident from the data presented in Table 4, there is a notable conservation of hydrogen bond interactions between MlGH17B and RmBgt17A within the −1, +1, and +2

Conserved residues
a Residues are involved in water mediated hydrogen bound.
subsites.Many of these interactions were recognized as crucial for the hydrolysis and transglycosylation activities of RmBgt17A (Qin et al. 2015).Specifically, the hydrogen bond interaction involving Trp191 (Trp157 in RmBgt17A) with OH6 of the glucose moiety in subsite −1 assumes a critical role in stabilizing the transition state for both transglycosylation and hydrolysis reactions.Additionally, Trp191 (Trp157 in RmBgt17A) and Tyr136 (Tyr102 in RmBgt17A) serve as essential sugar-binding platforms within the +1 subsite.Furthermore, Glu192 (Glu158 in RmBgt17A) exhibits a conservation of hydrogen bond formation with OH1 of the glucose moiety in the +2 subsite, enabling it to establish direct hydrogen bonds with the acceptor substrate within the +2 subsite for the transglycosylation reaction (Qin et al. 2015).

Discussion
This study provides detailed insights into the structure and function of the GH17 transglycosylase MlGH17B of marine origin which is the first transglycosylase from this environment to be characterized.The enzyme is encoded in the genome of the marine bacterium Muricauda lutaonensis strain ISCAR-4703, and related candidates has in this study been discovered, encoded in the genomes of other Muricauda species, suggesting that the enzyme likely plays a significant role in the degradation and/or storage of laminarin in these related marine microorganisms.Moreover, sequence similarity analysis using blastp, demonstrated presence of numerous homologs in deposited genomes of other marine bacteria, indicating this to be a common type of activity in the marine environment.This may not be surprising as laminarin is a common carbohydrate polymer in different marine brown algal species (Hreggviðsson et al. 2020).Currently, the genus Formosa, contains the more investigated bacterial species concerning characterization of carbohydrate converting enzymes active against seaweed polymers.However, no enzyme has yet been characterized as a GH17 transglycosylase from any species of this genus, although the catalytic module of the exoglucanase FbGH17B from Formosa sp.Hel1_33_13 that is releasing glucose from the reducing end of laminarin (Unfried et al. 2018), is sharing 53% sequence identity with MlGH17B.A difference is that FbGH17B is connected to a membrane transporter, but whether that has implications on its activity is not known.The homology to MlGH17B, however makes it possible that FbGH17B may exhibit a yet undetected transglycosylase activity or that this is missing due to missing crucial residues important for transglycosylation, as discussed below.
The to date characterized GH17 bacterial transglycosylases all originate from soil environments.Glt1 from Pseudomonas aeruginosa, Glt3 from Pseudomonas putida, Glt7 from Azotobacter vinelandii, all originate from soil bacteria.Opposed to MlGH17B, these enzymes cleave the substrate from the non-reducing end during the first step and then, dependent on enzyme, perform transglycosylation activities such as β-1,3elongation, β-1,4-or β-1,6-elongation and β-1,6-branching (Hreggvidsson et al. 2011).Only Glt20 from the soil bacterium B. diazoefficiens (Linares-Pastén et al. 2021) and RmBgt17A from the compost thriving fungus R. miehei (Qin et al. 2015), show transglycosylation activity resembling that of MlGH17B: Cleaving the substrate from the reducing end, producing Glc 2 , and transferring the remaining part of the donor substrate to the non-reducing end of the acceptor molecule (Linares-Pastén et al. 2021).MlGH17B is in this work shown to be structurally homologous to RmBgt17A, with a similar number of subsites (−3, −2, − 1, +1, +2), while the modeled structure of Glt20 is predicted to have an additional −4 glycone subsite.The high resemblance between MlGH17B and RmBgt17A shows that enzymes of similar structures have been adapted to participate in conversions of β-1,3-linked glucans from their respective habitat.
MlGH17B has a distinct specificity toward β-1,3-linkages, and a minimum substrate length of DP4 (confirmed using Glc 4 as substrate) which, however, displays low activity compared to Glc 5 and longer substrates where all five subsites in the catalytic cleft are filled.Activity on polymeric laminarin proves that the substrate chain is allowed to extend beyond subsite −3, but with reduced specific activity compared to the activity on laminari-oligosaccharides (DP5 or DP6).Steric hindrance by the MBP, negatively affecting binding of larger substrate molecules to the enzyme, can however not be excluded at this stage.
RmBgt17A and MlGH17B shared 13 potential substrate binding residues (excluding the two catalytic residues Glu133 and Glu233) over five subsites, corroborating similar functions.HPAEC-PAD, MALDI-TOF, and NMR analysis unambiguously showed that MlGH17B cleaved off two residues from the reducing sugar end (the aglycone subsites +1 and +2), in principal regardless of the substrate size (the only exception being Glc 4 where small amounts of glucose were released), followed by transfer of the remaining part to an acceptor substrate molecule, generating a β-1,6linkage.Among the conserved substrate binding residues, those in the aglycone subsites, have been proposed to be of special importance for the transglycosylation transition, especially the three residues, Trp191 (Trp157 in RmBgt17A), Glu192 (Glu158 in RmBgt17A) and Tyr136 (Tyr102 in RmBgt17A) (Qin et al. 2015).The corresponding residues have been mutated to Ala in RmBgt17A (Tyr102Ala, Trp157Ala and Glu158Ala) resulting in completely abolished transglycosylation ability of the enzyme (Qin et al. 2015), while approximately 14% of the transglycosylation activity was retained when Trp was replaced with the smaller aromatic residue Phe (Trp157Phe) (Qin et al. 2015).Trp191 is proposed to both hydrogen bond with OH6 of the glucose in the −1 subsite (stabilizing the transition state) and together with Tyr136 serve as a sugar-binding platform within the +1 subsite, while Glu192 is proposed to hydrogen bond with OH1 in the +2 subsite.The presence of Tyr136, Trp191, and Glu192 in MlGH17B thus leads to the possibility of the enzyme to bind acceptor molecules (>DP2) in an alternate position, possible when the donor is already bound in the catalytic cleft of the enzyme, in line with the proposed binding in RmBgt17A.Binding of the acceptor has been proposed to occur on top of the catalytic site, which would promote β-1,6-linkage specificity (Qin et al. 2015) based on the structure of the laminari-oligosaccharide.These findings underscore the pivotal role played by rather few residues in enhancing the enzyme's transglycosylation activity.Their absence may also serve as an alternative explanation for lack of transglycosylation in otherwise related enzymes, such as FbGH17B.The alignment (Fig. 2) for example shows that Tyr136 (Tyr102 in RmBgt17A) is replaced by a Met in FbGH17B, while the residues corresponding to Trp191 and Glu192 (MlGH17B numbering) are conserved.
The high number of conserved residues in subsites −1, +1 and +2 subsites, combined with the similarities in the observed transglycosylation pattern, makes it likely that MlGH17B and RmBgt17A share the same two-step catalytic mechanism as proposed by Qin et al. (Qin et al. 2015).In the first step, the donor substrate (laminari-oligosaccharide) occupies the active site (−3 to +2 subsites) of the enzyme and the departure of the Glc 2 leaving group (from subsites +1 and +2) is mediated by the proton donor (Glu133) donating a proton to the O atom between the glycosyl moieties located at sites −1 and +1, while the nucleophile Glu223 forms a covalent intermediate with the remaining part of the donor substrate.In the second step, the acceptor substrate molecule occupies the +1 and +2 subsites, in a different way than the original substrate, promoting subsequent formation of the β-1,6-linkage.The deprotonated proton donor (Glu133) activates the C-6 hydroxyl of a glucose moiety of the acceptor, which carries out a nucleophilic attack on the enzymesubstrate intermediate complex, leading to formation of a transglycosylation product with a β-1,6-linkage.No potential hydrogen bonds were detected between Glu233 and the substrate in the current model.It is, however, noteworthy that the docking simulations conducted in this study do not account for the presence of structural water molecules within the catalytic cleft.In the crystal structure of RmBgt17A, three water molecules have been identified (Table 4), which increased the number of hydrogen bonds formed between the ligand and active site residues of the enzyme (Qin et al. 2015).
In both MlGH17B and RmBgt17A, the reducing end of the donor substrate is always bound in subsite +2.In addition, the −3 subsite must be important for binding and activity of the enzyme, as Glc 4 was a relatively poor substrate.In the case of Glc 4 , the substrate could, based on the collected data, bind in alternative positions, either in the −3 to +1 subsites or in the −2 to +2 subsites, resulting in either Glc or Glc 2 as leaving groups.The NMR-data revealed that a β-1,6-linkage is formed between the glucose moiety of the donor substrate bound in the −1 subsite and either the non-reducing end sugar unit of the acceptor (creating a kink) or the second or third glucose from the non-reducing end in the acceptor molecule (creating a branch).This will lead to synthesis of oligosaccharides with either branched or kinked structure.
Comparison of the modeled structure of MlGH17B with the corresponding structure of RmBgt17A (Qin et al. 2015) showed similarities in catalytic cleft architecture, specifically in subsites −1, +1 and +2.In MlGH17B, the three glycone subsites are surrounded by polar residues, as was also observed for Rmbgt17A, however, several aromatic residues in the glycone subsites, namely Phe76, Trp101, Phe264, Trp271 and Trp283 are involved in hydrophobic interactions with the donor substrate.The two aglycone subsites are surrounded by hydrophobic residues providing a hydrophobic sugar binding platform, which could be the reason for the release of Glc 2 from the reducing end while the rest of the substrate remains bound in the active site for the transfer reaction.The MALDI-TOF mass spectrometry results also demonstrated the ability of MlGH17B to synthesize long-chain, multiple branched, oligosaccharides such as Glc 20 and Glc 26 from Glc 5 and Glc 6 substrates, respectively, using long reaction time, allowing us to conclude that branched oligosaccharides (dependent on their respective concentration) can also be either donors or acceptors in the transglycosylation reaction.

Conclusion
In this study, a novel β-1,3-glucanosyltransglycosylase MlGH17B from the marine Muricauda lutaonensis strain ISCAR-4703 was characterized in terms of substrate specify and product profile.DSF analysis showed that the enzyme displayed a thermostable core, while being active at lower temperature, a feature that may be beneficial at geothermal marine sites, where temperature gradients can be expected.The three-dimensional structure of the enzyme was modeled and Glc 5 was docked into the catalytic cleft revealing five potential subsites (−3, −2, −1, +1, +2).Structural comparison with RmBgt17A from R. miehei revealed presence of three conserved residues Trp191, Glu192, and Tyr136 in the aglycone subsites proposed to be crucial for the transferase reaction (and not conserved in the homologous FbGH17B, reported as a glucose releasing exo-glucanase), allowing these enzymes to participate in transglycosylation reactions of β-1,3-linked glucans from their respective habitat.
The experimental data demonstrated a distinct specificity of MlGH17B toward β-1,3 linked substrates (DP > 4).The substrate was cleaved two residues from the reducing end of the substrate resulting in Glc 2 as a leaving group, followed by transfer of the remaining part of the donor substrate to another acceptor molecule of the substrate making a β-1,6-linkage.The resulting products were branched or kinked oligosaccharides, up to sizes exceeding DP20, when using laminaripentaose or laminarihexaose substrates at the activity optimum conditions (pH 6.0 and 20 • C).

Materials and methods
All materials were purchased from Sigma-Aldrich (Merck) unless otherwise specified.Laminari-oligosaccharides were purchased from Megazyme (Neogen).
For genome sequencing, DNA libraries were prepared using the Nextera XT method (Illumina) and sequenced with the MiSeq System (Illumina).Raw sequences were trimmed for quality using the Trimmomatic v0.39 program (Bolger et al. 2014).A draft genome was assembled using trimmed paired reads and the SPAdes v3.15.2 assembly algorithm (Bankevich et al. 2012).The genome sequences were annotated using the prokaryotic genome annotation server Rapid Annotations using Subsystems Technology (RAST) (Aziz et al. 2008).The number of rRNA and tRNA genes were predicted by the RNAmmer v1.2 tool (Lagesen et al. 2007) and the tRNAscan-SE v2.0 tool (Lowe and Chan 2016) respectively.

Cloning, protein production and purification
The gene encoding a putative GH17 glycosyl hydrolase annotated in the draft genome of the bacterial strain ISCAR-4703, designated MlGH17B, was deposited in the NCBI GenBank with accession number OQ297286.The MlGH17B gene was amplified by PCR with Phusion High-Fidelity DNA Polymerase (New England Biolabs).
The forward primer MlGH17B-f2: 5´-GCCAGCAAGGGC GAGATGACCACTAAAGAATTGAGAAG-3 , containing a 5 -linker sequence (bold), and the reverse primer, MlGH17Bbam-h-r: 5´-CGCGGATCCAAATTTTAGTTTTTCATTCTT ATCCC-3 containing a BamHI restriction site (underlined) were used for the amplification.The amplified sequence was cut with BamHI and ligated into a SfoI (blunt end) -BamHI digested pJOE4905 vector (Motejadded and Altenbuchner 2009).Thereby, the 5 end of the MlGH17B encoding sequence, excluding 58 bp, with the short 5linker, was fused in-frame with the vector sequence encoding MBP, and a S. cerevisiae ubiquitin-like protein motif (Smt3) for proteolytic cleavage with Ulp1 (Motejadded and Altenbuchner 2009).Furthermore, the 3'end of the sequence was fused with the vector sequence encoding the C-terminal 6 × His-tag.The resulting expression clone, verified by sequencing, was transformed into Escherichia coli BL21(DE3).
The expression strain was shake-flask cultivated in LB-Lennox medium at 37 • C inducing heterologous expression at OD 600 0.5 with 0.25% (w/v) L-rhamnose for 4 h at 37 • C. The cells were harvested by centrifugation and lysed in lysis buffer (20 mM Tris-HCl pH 7.4, 200 mM NaCl) by ultra-sonication using a UP400S homogenizer (Hielscher Ultrasonics).The lysate was collected after centrifugation at 26,000 × g for 20 min at 4 • C, filtered through regenerated cellulose 0.2 μm pore size filters (GE Healthcare Life Sciences), and subjected to an MBP-Trap HP 1 mL (7 × 25 mm) column (GE Healthcare Life Sciences) for affinity purification using ÄKTA start FPLC purification system (GE Healthcare Life Sciences).Bound protein was eluted with maltose gradient up to 10 mM in lysis buffer over 10 column volumes.Fractions containing the recombinant protein were immediately assembled, diluted to final concentration below 1 mg/mL, and stored at 4 • C. The integrity and purity of the protein were analyzed by 4%-15% glycine-SDS-PAGE.The protein concentration was determined considering the absorption coefficient (135,680 M −1 cm −1 ) by measuring A 280 using a NanoDrop 1000 spectrophotometer (Thermo Scientific).

Melting temperature determination
To determine the thermal unfolding transition, the MBP was cleaved from MBP-Smt3-MlGH17B with Ulp1 and subsequently purified.The pure MBP-Smt3-MlGH17B was incubated with Ulp1 with a fusion-protein/protease ratio of 25:1 (w/w) in the reaction buffer (20 mM Tris-HCl pH 7.4, 10 mM imidazole and 500 mM NaCl) at 30 • C for 1 h.The MBP cleavage reaction after incubation was subjected to an HisTrap HP 1 mL (7 × 25 mm) column (GE Healthcare Life Sciences) collecting purified MBP in a flow through fraction.The integrity and purity of the protein were analyzed by 4%-15% glycine-SDS-PAGE.The protein concentration was determined considering the absorption coefficient (66,350 M −1 cm −1 ) by measuring A 280 .
The melting temperature of MBP-Smt3-MlGH17B and MBP at 0.2 mg/mL concentration in lysis buffer with 10 mM maltose were measured applying nanoscale differential scanning fluorimetry (nanoDSF) with Prometheus NT.48 instrument using standard grade capillaries (NanoTemper Technologies), and the results were processed with the PR.ThermControl software (NanoTemper Technologies).The 350/330 nm fluorescence intensity ratio was monitored at 40% excitation power with a temperature gradient 20-95 • C at a ramp rate of 1 • C/min.The inflection point, indicative of a thermodynamic phase change, was considered as melting temperature (T m , • C).Furthermore, the onset of the transition (T onset , • C) was measured by estimating light scattering, measuring back-reflection light intensity change.The measurements were performed in triplicates.

Activity optimum determination
The catalytic activity of the enzyme was initially evaluated visualizing reaction products obtained using laminaripentaose (Glc 5 ) as the substrate in a broad range of pH 3.5-9 by TLC analysis.Then, a narrower pH range pH 5, 5.5, 6, and 6.5 close to the pH with highest amount of transfer products was selected to confirm the pH optimum.Reaction mixtures of 50 μL were prepared by mixing 5 μL of 20 mg/mL Glc 5 , 5 μL of 500 mM acetate/phosphate buffer, pH 5-6.5, 20 μL of the suitably diluted enzyme in lysis buffer (20 mM Tris-HCl pH 7.4, 200 mM NaCl), and 20 μL of ultrapure water (Milli-Q grade).The reaction mixtures were incubated in the thermal cycler with a temperature gradient 13-49 • C ramping for 2 h, followed by reaction termination at 95 • C for 5 min.Product formation including oligosaccharides with DP2 and DP8 (Glc 2 and Glc 8 ) was analyzed applying HPAEC-PAD.Duplicate samples were analyzed for each condition.The results were plotted, and figures were illustrated using the MODDE v12.1 program (Sartorius).

MALDI-TOF MS
MALDI-TOF mass spectra were acquired with an autoflex speed MALDI-TOF/TOF (Bruker Daltonics) in positive reflector mode.Samples were diluted with ultrapure water (Milli-Q grade) to receive a total salt concentration less than 10 mM.One microliter of diluted sample was mixed with 0.5 μL of aqueous 10 mg/mL dihydroxybenzoic acid matrix solution.All spectra were externally calibrated using Peptide Calibration Standard II (Bruker Daltonics).

Purification of transglycosylation products
Ten milliliters of the reaction mixture in 50 mM phosphate buffer, pH 6.0, containing 2 mg/mL of Glc 5 and 0.4 mg/mL of the enzyme was incubated at 20 • C for 6 h under shaking.The reaction was terminated by incubation at 95 • C for 5 min and the reaction mixture was subsequently filtered through PTFE membrane 0.2 μm pore size filters (Pall).Transglycosylation products were purified applying preparative size-exclusion chromatography (SEC) using Superdex 30 PG size exclusion chromatography media (GE Healthcare Life Sciences) in HiLoad 26/600 column (GE Healthcare Life Sciences).Aliquots of 2 mL of the reaction mixture was loaded and carbohydrates were eluted with ultrapure water (Milli-Q grade) at a flow rate of 0.3 mL/min.Fractions of 0.6 mL were collected, and the carbohydrate content of individual fractions was analyzed by TLC and HPAEC-PAD.Fractions containing a single compound were combined and lyophilized by freeze drying (Labconco).

NMR spectroscopy
Resolution-enhanced one-and two-dimensional NMR spectra were recorded in D 2 O on an Avance Neo spectrometer (Bruker) equipped with a TCI Prodigy CryoProbe (Bruker) (Utrecht University, The Netherlands) at a temperature of 311 K. Prior to analysis, samples were exchanged twice in D 2 O with an intermediate lyophilization, and then dissolved in 0.5 mL D 2 O. Suppression of the HOD signal was achieved by applying a water-eliminated Fourier transform (WEFT) pulse sequence for one-dimensional NMR experiments.The two-dimensional TOCSY spectra were collected using a composite pulse devised by M. Levitt (MLEV) mixing sequence with 40-150 ms spin-lock times.Multiplicity-edited 1 H-13 C HSQC was recorded with 1,536 data points in F2 and 256 in F1.Chemical shifts (δ) are expressed in ppm by reference to internal acetone (δ 2.225 for 1 H and δ 31.07 for 13 C).
Thin layer chromatography (TLC) TLC was performed using TLC Silica gel 60 F 254 plates (Merck) loading 1 μL of sample and subsequently allowing loaded samples to air-dry.Prepared plates were developed twice with an intermediate drying in a solvent system nbutanol/acetic acid/water, at the solvent ratio 2:1:1 (v/v/v) for Glc 1-8 oligosaccharides separation or at the solvent ratio 3:2:2 (v/v/v) for Glc >8 oligosaccharides separation.Developed plates were air-dried and subsequently treated by spraying with staining solution 5% (v/v) sulfuric acid in methanol supplemented with 1 mg/mL of orcinol.Oligosaccharides were visualized by heating the stained plates at 110 • C on a hotplate.

Bioinformatics analysis
Multiple sequence alignment (MSA) Sequence similarity searches were made in NCBI BLAST suite (Altschul et al. 1990) with blastp against the non-redundant protein sequence database, UniProtKB/Swiss-Prot database and PDB database using default algorithm parameters.Multiple sequence alignment was performed in Jalview v2.11 program using Clustal Omega v1.2.2 alignment tool (Waterhouse et al. 2009) and graphically presented with the ESpript v3.0 program (Robert and Gouet 2014).
The model confidence was assessed by the pLDDT score (Jumper et al. 2021) and the top ranked models were used for subsequent analysis.The figure was made using PyMOL v2.5.2 program (Schrödinger).
The three-dimensional structure of MlGH17B was also modeled by homology modeling using the YASARA v21.12.19 program (YASARA Biosciences) as described in Linares-Pastén et al. (Linares-Pastén et al. 2021).The crystal structure of RmBgt17A (PDB 4WTP, 4WTR, and 4WTS) and FbGH17A (PDB 6FCG) were used as main templates and modeling parameters were set as were applied by Linares-Pastén et al. for modeling the three-dimensional structure of glycosyltransferase enzymes, Glt1 from P. aeruginosa, Glt3 from P. putida, and Glt20 from B. diazoefficiens (Linares-Pastén et al. 2021) (Table S2).The generated hybrid model was subjected to model refinement using default parameters in YASARA program.A simulation cell was designed 2 × 7.5 Å larger than the model along each axis.The simulation cell was filled with water and Na + and Cl − as counter ions.The simulation was run for 500 ps using YAMBER03 force field, and snapshots were stored every 25 ps.The quality of the generated models was assessed through ERRAT, VERIFY3D, and PROCKECK using the UCLA-DOE LAB-SAVES v6.0 server (saves.mbi.ucla.edu).The model was compared with the model generated by AlphaFold2 program.

Substrate docking
The top ranked model obtained from AlphaFold2 was subjected to docking experiment with Glc 5 .
Laminaripentaose ligand was built using oligosaccharides building tool available in the YASARA program.The ligand was subjected to MD simulation at physiological pH 7.4 and 298 K for 2 ns using AMBER03 force field in a simulation cell 20 Å larger than the molecule with explicit molecules of water and Na + and Cl − as counter ions.The conformer with the lowest force field energy was selected for docking studies.To perform docking, after the superimposition of the MlGH17B model with the crystal structure of RmBgt17A (PDB 4WTP) and learning about the location of the active site, AutoDock implemented in YASARA was performed.The quality of the docked model was assessed by checking the conformation of the ligand in the active site, assessing any abnormalities in glycosidic linkages of the ligand, and evaluating fluctuations in RMSD Cα or increasing binding energy during a 50 ns MD simulation.
The three-dimensional model of the enzyme-ligand complex and the active site were graphically presented using the PyMOL program.

Fig. 10 .
Fig. 10.Modeled structure of MlGH17B from AlphaFold2, represented as a ribbon diagram.The TIM-barrel ((β/α) 8 ) structure of the enzyme is illustrated from (A) the front view and from (B) the side view.The catalytic proton donor (Glu133) and nucleophile (Glu223) residues are indicated in the model.

Fig. 11 .
Fig. 11.Active site architecture of MlGH17B.A) the residues within 5 Å of Glc 5 located from −3 to +2 subsites of MlGH17B; conserved proton donor and nucleophile residues are outlined in purple, and the ligand is shown in magenta.B) Conserved residues between MlGH17B (green) and RmBgt17A (dark blue), the ligand belonging to MlGH17B and RmBgt17A are shown in magenta and yellow, respectively; conserved residues, which do not participate in interactions within the active site of RmBgt17A, are represented in silver.

Table 2 .
1 H and 13 C NMR chemical shifts (δ) of reaction products obtained after incubation of laminari-oligosaccharides with MlGH17B enzyme, recorded in D 2 O at 311 K.

Table 4 .
Interaction of the sugar residues of Glc 5 and the active site residues of MlGH17B and RmBgt17A.