A β-Galactosidase acting on unique galactosides: the structure and function of a β-1,2-galactosidase fromBacteroides xylanisolvens, an intestinal bacterium

Galactosides are major carbohydrates that are found in plant cell walls and various prebiotic oligosaccharides. Studying the detailed biochemical functions of β-galactosidases in degrading these carbohydrates is important. In particular, identifying β-galactosidases with new substrate specificities could help in the production of potentially beneficial oligosaccharides. In this study, we identified a β-galactosidase with novel substrate specificity from Bacteroides xylanisolvens , an intestinal bacterium. The enzyme did not show hydrolytic activity toward natural β-galactosides during the first screening. However, when α-D-galactosyl fluoride (α-GalF) as a donor substrate and galactose or D- fucose as an acceptor substrate were incubated with a nucleophile mutant, reaction products were detected. The galactobiose produced from the α-GalF and galactose was identified as β-1,2- galactobiose using NMR. Kinetic analysis revealed that this enzyme e ffectively hydrolyzed β-1,2- galactobiose and β-1,2-galactotriose. In the complex structure with methyl β-galactopyranose as a ligand, the ligand is only located at subsite +1. The 2-hydroxy group and the anomeric methyl group of methyl β-galactopyranose faces in the direction of subsite −1 and the solvent, respectively. This observation is consistent with the substrate specificity of the enzyme regarding linkage position and chain length. Overall, we concluded that the enzyme is a β-galactosidase acting on β-1,2- galactooligosaccharides. Synopsis The structural and functional analysis of β-galactosidase from an intestinal bacterium led to the discovery of a new β-galactosidase hydrolyzing unique β-1,2-galactooligosaccharides. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 5, 2024. ; https://doi.org/10.1101/2024.05.03.592134doi: bioRxiv preprint

Carbohydrate chains are structurally complex biopolymers indispensable for various processes, structures, and functions in organisms, including cell architecture, storage polysaccharides, immuno- stimulation, infection, and symbiosis 1–6. To synthesize and degrade carbohydrates with extremely diversified chemical structures, carbohydrate-associated proteins with a wide variety of functions and structures have evolved. The proteins in this category are currently registered in the Carbohydr ate- Active enZYmes Database (CAZy) (http://www.cazy.org/) and are classified into families primarily based on their amino acid sequence similarity 7. This database contains several enzyme classes, including glycoside hydrolases (GHs), glycosyltransferases, polysaccharide lyases, carbohydrate esterases, and others with auxiliary activities. The fact that the number of families keeps growing suggests that we are far from seeing the whole picture when it comes to the variety of carbohydrate - associated enzymes. Galactose is a major natural monosaccharide, and galactosides play important physiological roles in plants, animals, and microorganisms. For example, galactosides in plants are usually found in a cell wall component known as hemicellulose, which incorporates galactosides such as galactan and arabinogalactan 8. Rhamnogalacturonan-I has galactosides and galactans in its side chains 9. Galactose units are also found as side chain components in tamarind -xyloglucan. In mammals, human milk oligosaccharides (HMO) are some of the many important oligosaccharides that contain galactose units. Lacto-N-biose I (Gal-β-1,3-GlcNAc) and related oligosaccharides are well-known prebiotics that help bifidobacteria to grow predominantly in the large intestine 10. Some bacteria also produce galactooligosaccharides (GOS) from lactose using fermentation 11. Because of their potential as prebiotics, GOS are sometimes supplemented in various processed foods such as juice and powdered milk. Thus, analysis of the enzymes that degrade these carbohydrates is one of the keys to understanding the mechanism of action of prebiotics. β-Galactosidases are major glycoside hydrolases that release galactose from various galactosides as described above. One of the most famous β-galactosidases is a lactose-degrading enzyme called LacZ. However, substrate specificity in galactosides is divers ified among β-galactosidases. A wide variety of substrate specificity has been found in these enzymes, especially in Bifidobacterium 12–14. This is likely due to the variety of carbohydrate structures found in HMOs, including variations in the linkage positions of the galactosides, side chains, and other modifications 13. Phylogenetically, β- galactosidases are distributed mainly in the GH1, GH2, GH35 and GH42 families 15. In addition, several other β-galactosidase groups have been identified recently (GH147, GH154, GH165 and GH173) 16–19. The functional and phylogenetic diversity of β-galactosidases reflects the physiological .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 5, 2024. ; https://doi.org/10.1101/2024.05.03.592134doi: bioRxiv preprint importance in many organisms. Bacteroides xylanisolvens is an intestinal Gram -negative bacterium. In a recent report, the bacterium was identified as an effective nicotine degrader 20. As its name suggests, B. xylanisolvens was originally identified as a xylan-degrading bacterium, and it has many enzymes that belong to the GH family 21. This suggests that B. xylanisolvens has the potential to utilize a broad range of carbohydrates. However, little is known about its actual ability to do so. B. xylanisolvens has multiple genes encoding putative β-galactosidases. One of these genes (Bxy_22780) is located at a unique position in the genome. The gene encoding the Bxy_22780 protein in the GH35 family is a component of a gene cluster containing SusC -like and SusD -like proteins (Supplementary Fig. 1). Because the SusCD transporter system is a major component of the polysaccharide utilization locus (PUL), this gene cluster is annotated as a PUL 22. A gene encoding the GH144 protein (Bxy_22790) is another member of this gene cluster. GH144 is a family of β-1,2- glucanases (SGLs), which are endo -type glucanases that hydrolyze β-1,2-glucans to β-1,2- glucooligosaccharides 23. Given the biochemical function of GH144, the GH35 protein is presumed to be a β-1,2-glucan-associated enzyme. However, GH35 is a family that mainly contains β- galactosidases (Supplementary Fig. 2). Although β-glucosaminidases 24 and β-1,2- glucosyltransglycosylase (SGT) 25 are found in the phylogenetic tree, they are phylogenetically far from Bxy_22780 in the β-galactosidase group. These two conflicting facts make it difficult to arrive at any precise conclusions about the biochemical functions of Bxy_22780. In this study, we found that Bxy_22780 is a biochemically and structurally novel β-galactosidase that can carry out unique enzymatic reactions. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 5, 2024. ; https://doi.org/10.1101/2024.05.03.592134doi: bioRxiv preprint

General properties Because Bxy22780 hydrolyzed p-nitrophenyl (pNP) β-D-galactopyranoside (pNP-Gal), pNP-Gal was used to further investigate its properties under different pH and temperature conditions. This enzyme functioned optimally at pH 5.5 –8.0 and 40 °C, and it was stable at pH 6.0 –9.5 and at temperatures of up to 30 °C (Supplementary Fig. 3). Bxy_22780 showed hydrolytic activity toward pNP-Gal and pNP β-D-fucopyranoside (pNP-D-Fuc) but not toward pNP β-D-glucopyranoside (pNP- Glc), pNP β-D-xylopyranoside, or pNP β-D-mannopyranoside (less than 0.025 U/mg in the presence of 5 mM substrate). Its kinetic parameters for the hydrolyzation of pNP -Gal and pNP -D-Fuc were comparable to those for GH enzymes (Table 1, Supplementary Fig. 4ab). It had a higher Vmax value and lower Km value for pNP-Gal than for pNP-D-Fuc, which led to Vmax/Km values for pNP-Gal that were approximately four times higher than those for pNP -D-Fuc. Thus, Bxy_22780 is intrinsically a β-galactosidase in that the enzyme shows a preference for β-galactoside over β-D-fucoside. Identification of a glycosynthase product To find the natural substrates of Bxy_22780, commercially available galactosides, including lactose, were incubated with the enzyme. Only a little hydrolytic activity was detected toward several examined substrates ( Supplementary Fig. 5). Therefore, a synthetic reaction (glycosynthase assay) using the E350G mutant (a nucleophile mutant) was used to identify the actual substrates of Bxy_22780. The reactions using the mutant were performed in the presence of various monosaccharides as acceptors and α-D-galactosyl fluoride (α-GalF) as a donor. Reaction products were detected only when galactose and D-fucose were present as acceptors (Fig. 1a left). In addition, reaction products were detected when melibiose, β-1,3(4)-galactobiose, allolactose, lactulose, and β- 1,6-galactobiose were selected from among the galactoside disaccharides for use as acceptors (Fig. 1a middle), suggesting that the reducing end moiety in a disaccharide is not important for substrate recognition when the substrate is an acceptor . When pNP -sugars were used as acceptors, reaction products were detected from pNP-Gal and pNP-D-Fuc (Fig. 1a right). In addition, pNP was not found to be a suitable acceptor. Oligosaccharides with higher degrees of polymerization (DPs) were preferably produced in the presence of acetone (Fig. 1b). The galactobiose and galactotriose produced during the reaction were purified using size exclusion chromatography. The linkage position of the galactobiose was analyzed using NMR. Heteronuclear Multiple Bond Correlation (HMBC) analysis showed a correlation between the C2 at the reducing end galactose unit (δ 77.9 in the α-anomer and δ 79.7 in the β-anomer) and the H1 at the non-reducing end galactose unit (δ 4.55 in the α-anomer and .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 5, 2024. ; https://doi.org/10.1101/2024.05.03.592134doi: bioRxiv preprint δ 4.67 in the β-anomer), indicating that the purified disaccharide was a β-1,2-linked galactobiose (Fig. 1cd and Supplementary Table 1 NMR, Supplementary Data NMR). Substrate specificity and kinetic analysis The pure β-1,2-galactobiose ( β-1,2-Gal2) and β-1,2-galactotriose (β-1,2-Gal3) produced by the E350G mutant as described above were used to investigate the substrate specificity of wild-type (WT) Bxy_22780. The WT enzyme completely hydrolyzed both β-1,2-Gal2 and β-1,2-Gal3 to galactose in a 3 h reaction (Fig. 2a). Contrarily, no hydrolytic activity toward the other substrates [lactose, melibiose, β-1,3(4)-galactobiose, allolactose, lactulose or β-1,6-galactobiose] was detectable even after an overnight reaction with the same enzyme concentration as that in the experiment that used β-1,2-Gal2 and β-1,2-Gal3 as substrates (Fig. 2b). This result indicated that Bxy_22780 is highly specific for galactooligosaccharides that have a β-1,2-galactosidic linkage. We determined the kinetic parameters of the hydrolytic activity of Bxy_22780 toward β-1,2-Gal2 and β-1,2-Gal3 (Table 1, Supplementary Fig 4c,d). The kinetic parameters for both were comparable with usual values for GH enzymes 26. This result suggests that β-1,2 linked galactooligosaccharides are the natural substrates of Bxy_22780. Specific activities toward the other galactoside disaccharides shown in Fig. 2 were less than 0.005 U/mg when the assay was performed using 5 mM concentrations of potential substrates. Structures of Bxy22780 To learn more about the binding modes of the natural substrates determined above, the ligand-free enzyme structure of the E350G mutant and the complex structures of the WT enzyme with galactose and methyl β-galactopyranoside (MeβGal) were determined at resolutions of 1.91 Å, 1.86 Å, and 1.94 Å, respectively (Supplementary Table 2 and Supplementary Fig. 6). All the structures in this study were tetramers based on PISA server analysis27, although the asymmetric units of these structures were found to contain two, one, and four molecules, respectively. The three overall structures were almost the same except that a loop region (insertion 2, residues 506 –539) was visible in the ligand -free structure (Supplementary Figs. 6, 7). The electron density of galactose in the α-anomer ( α-Gal) configuration was clearly observed only at subsite −1 in the WT enzyme complex with α-Gal, whereas galactose in the β-anomer (β-Gal) configuration was not observed at all (Fig. 3a top). The 4 - and 6- hydroxy groups in the galactose were found to form hydrogen bonds with the D494 derived from another subunit at the subunit interface, in the same way as they do in the GH35 enzymes in the phylogenetic group that contains Bxy_22780 28 (Figs. 3a and 4). .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 5, 2024. ; https://doi.org/10.1101/2024.05.03.592134doi: bioRxiv preprint In the WT-MeβGal complex, the electron density of MeβGal was clearly observed only at subsite +1 (Fig. 3b top). As shown in Fig. 3b, MeβGal is firmly recognized by Bxy_22780. In MeβGal, the 2- and 3-hydroxy groups form hydrogen bonds with Asn119 and Glu190, and its 3- and 4-hydroxy groups form water-mediated hydrogen bonds with the main chain atoms in Leu289, Gln291, Asp326 and Tyr328. The indole ring of Trp288 hydrophobically interacts with the six-membered ring in MeβGal. This binding mode of Me βGal accounts for the biochemical properties of Bxy_22780, especially linkage position specificity, well. The 2-hydroxy group of MeβGal is oriented toward subsite −1 and is located close to an anomeric hydroxy group in the superimpos ed galactose molecule at subsite −1 (Fig. 3c). Moreover, Glu190, a putative acid/base residue, interacts with the 2 -hydroxy group of MeβGal, suggesting that the residue can act on the hydroxy group catalytically. The methyl group in MeβGal faces the solvent (Fig. 3bc), which is consistent with the result that Bxy_22780 acts on β-1,2- Gal3 (Table 1, Supplementary Fig. 4cd). This observation is also consistent with the finding that the specificity of Bxy_22780 against reducing end glycoside moieties in β-galactoside disaccharides acting as acceptors was loose in the glycosynthase assay (Fig. 1a middle). The methylene group in MeβGal also faces the solvent without forming hydrogen bonds (Fig. 3bc), which is consistent with the finding that D-fucose (galactose without the C6 methylene group) is one of the acceptors in the glycosynthase assay (Fig. 1a left). Comparison of properties with W288A mutant The hydrophobic interaction between Trp288 and the six -membered-pyranose ring in Me βGal is likely to be important for substrate specificity because one potential position of the 4-hydroxy group in the glucose at subsite +1 is a little so close to the indole ring of Trp288 that steric hindrance can become a problem based on the position of Me βGal. Trp288 is replaced with Ala in many close homologs of Bxy_22780 (Supplementary Fig. 7). Among the homologs with the Ala residues, the β- galactosidase CjBgl35A from Cellvibrio japonicus Ueda107 acts on a galactosyl-β-1,2-xyloside unit in XLLG (Supplementary Fig. S8a) in tamarind -xyloglucan (TXG) 28. It has also been reported that XacGalD from Xanthomonas citri pv. citri str. 306 acts on the same linkages to hydrolyze XLXG and XXLG 29 (Supplementary Fig. S8a). Thus, W288A and W288A/E350G mutants were investigated to learn more about the possible roles of Trp288 in the enzyme. First, the xyloglucan degradation activity of Bxy_22780 was evaluated. The WT and W288A mutant enzymes were incubated with TXG-XEG (see Methods for abbreviations) and LG (Supplementary Fig. S8a). The WT enzyme showed no detectable activity toward these substrates after TLC analysis was performed (Supplementary Fig. 8b,c). The activity of the mutant enzyme on LG was higher than that .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 5, 2024. ; https://doi.org/10.1101/2024.05.03.592134doi: bioRxiv preprint of the WT enzyme, although its activity was still very low. This result indicates that the removal of the side chain of W288 contributes to galactoside specificity in xyloglucan. However, this substitution is far from sufficient for inducing a change in substrate specificity. The TLC analysis showed that the W288A mutant was still specific for β-1,2-Gal2 and β-1,2-Gal3 among the investigated galactosides, as was the WT enzyme (Supplementary Fig. 9a). However, a kinetic analysis of the two substrates showed that the Vmax values for both decreased over 10 times compared with that of the WT enzyme (Table 1, Supplementary Fig. 4e,f). Both Km values increased, especially for β-1,2-Gal3, which increased by more than 30 times compared with that of the WT enzyme. These changes led to remarkable decreases in the Vmax/Km values of the two substrates. In addition, the WT showed sli ght transglycosylation activity toward both β-1,2-Gal2 and β-1,2-Gal3, whereas W288A lost this transglycosylation activity (Supplementary Fig. 10). These results suggest that Trp288 is important for an effective catalytic reaction rather than substrate specificity. Indeed, a glycosynthase assay using W288A/E350G did not show any detectable activity even when galactose and D-fucose were used (Supplementary Fig. 9b). .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 5, 2024. ; https://doi.org/10.1101/2024.05.03.592134doi: bioRxiv preprint

Enzymological classification of Bxy_22780 As shown in the phylogenetic tree (Supplementary Fig. 2), most of the biochemically characterized GH35 enzymes are β-galactosidases. β-1,3-Galactosidases (EC 3.2.1. -) and exo -β-1,4-galactanases (EC 3.2.1.-) with narrow substrate specificity are also found in the GH35 family. In addition to β- galactosidases, exo -β-1,4-glucosaminidase (EC 3.2.1.165) and SGT (EC 2.4.1.391) have been reported 25,30–32. However, these enzymes belong to small groups far from Bxy_22780 in the GH35 family. In the Bxy_22780 group, CjBgl35A from Cellvibrio japonicus Ueda107 (KEGG locus tag, CJA_2707) and XacGalD from Xanthomonas citri pv. citri str. 306 have been found to be β- galactosidases that act on pNP -Gal 29. CjBgl35A acts on XLLG ( Supplementary Fig. 8a), a natural substrate, to release galactose but does not act on TXG 28. However, the kinetic parameters of the CjBgl35A-XLLG reaction have not yet been fully described and the expression of the CjBgl35A gene is not sharply induced by TXG 28. The characteristics of XacGalD have been mentioned in the context of xyloglucan metabolism a nd xyloglucan utilization loci in Xanthomonas species 29. Although its activity toward xyloglucan was addressed, the kinetic parameters and specific activity of the enzyme were not examined in that study. In this study, we found that Bxy_22780 specifically hydrolyzes β- 1,2-galactooligosaccharides. Although galactose units in TXG are linked to xylose units through a β- 1,2-linkage, Bxy_22780 does not act on LG or XEG-TXG, an oligosaccharide released from TXG by hydrolysis with xyloglucanase (Supplementary Fig. 8). Furthermore, the E350G mutant used for a glycosynthase did not show activity when xylose was provided as an acceptor (Fig. 1 left). These

indicate that the substrate specificity of Bxy_22780 is very different from that of either CjBgl35A or XacGalD. β-Galactosidases are distributed in various GH families. The GH42 family is another family that mainly contains β-galactosidases, 15 and GH59 is a family of galactosylceramidases 33. The GH2 family contains β-galactosidases as a major component 34. In the GH1 family, enzymes named β- glycosidase exhibit β-galactosidase activity 35. In addition, β-galactosidases have been recently reported in the GH147, GH154, GH165 and GH173 families 16–19. In exceptional cases, β- galactosidases can also be found in other GH families 36,37. Although β-galactosidases specific to β- 1,3- and β-1,4-galactooligosaccharides (β-1,3-galactosidase and exo-β-1,4-galactanase, respectively) have already been reported 38,39, a β-1,2-galactooligosaccharide-specific β-galactosidase has not yet been reported and was not found even after a search of the entire CAZy database. Overall, a new EC number should be provided for Bxy_22780. We propose β-1,2-galactooligosaccharide galactohydrolase as a systematic name and β-1,2-galactosidase as an accepted name for Bxy_22780. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 5, 2024. ; https://doi.org/10.1101/2024.05.03.592134doi: bioRxiv preprint Structural comparison of GH35 enzymes The hydrolytic activity of Bxy_22780 toward various natural β-galactosides was very low (Fig. 2). Similarly, it has been reported that CjBgl35A does not act on any of the other substrates that were included in the initial screen 28. However, Bxy_22780 was found to be a β-galactosidase that acted on β-1,2-galactooligosaccharides but not on XLLG or LG, whereas CjBgl35A acts on XLLG. Nevertheless, Bxy_22780 and CjBgl35A are classified into the same phylogenetic group (Supplementary Fig. 2). Thus, to understand the biochemical properties of Bxy_22780 in detail, structural comparison with CjBgl35A is needed. A β-galactose unit binds covalently to a xylose unit through a β-1,2-linkage, which is the same kind of linkage that the galactooligosaccharide substrates of Bxy_22780 have. Therefore, a xylose unit in an XLLG molecule is expected to bind at subsite +1 i n the same orientation as Me βGal in the Bxy_22780-MeβGal complex structure (Fig. 3b). In the Me βGal complex, the 6 -hydroxy group of MeβGal faces the solvent and does not form any hydrogen bonds with any of the residues, and a xylose has no 6-hydroxy group. However, Bxy_22780 showed very low activity toward LG (Supplementary Fig. 8c), suggesting that the presence of a xylose unit in the LG molecule means that it is not suitable as a substrate. A difference in the orientation of the 4 -hydroxy group in the galactose unit and the xylose unit of their respective substrates may account fo r the difference in the substrate specificities of Bxy_22780 and CjBgl35A. If a xylose molecule occupies the same position as the Me βGal in the Bxy_22780-MeβGal complex, the 4-hydroxy group of the xylose will be a little too close to the side chain of Trp288 to accommodate the xylose unit appropriately at subsite +1. Furthermore, Trp288 is replaced by Ala (Ala283), a residue with a small side chain, in CjBgl35A (Fig. 4). However, t he W288A mutant of Bxy_22780 did not show any remarkable improvement in hydrolytic activity on LG (Supplementary Fig. 8 c). The W288A mutant clearly retained hydrolytic activity toward β-Gal2 and β-Gal3, although its activity was much lower than that of the WT enzyme (Table 1). Trp288 is important for substrate recognition, but Trp288 alone does not affect substrate specificity. Gln291 is another important residue for substrate recognition at subsite +1 in Bxy_22780. This residue forms multiple hydrogen bonds with the 3 - and 4-hydroxy groups of the Me βGal molecule through water molecules (Figs. 3b bottom, 4). In CjBgl35A, this residue is conserved as a chemically similar residue (N286) in the multiple sequence alignment (Supplementary Fig. 7). However, the conformations of the loop regions including the residues are obviously different from each other (Fig. 4). The orientation of N286 side chain suggests that N286 cannot participate in substrate recognition of subsite +1 at all (Fig. 4). These observations suggest that Gln291 and the conformation of the loop .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 5, 2024. ; https://doi.org/10.1101/2024.05.03.592134doi: bioRxiv preprint are important for substrate specificity. CjBgl35A contains two loops; Leu386 is present in one loop and Trp480 is present in the other (Fig. 4, Supplementary Fig. 7). These loops are located close to the catalytic pocket. However, Bxy_22780 does not contain any corresponding loops. Although these loops do not recognize subsite +1 directly, they may be at least partly responsible for differences in substrate specificity beyond subsite +2 in Bxy_22780 and CjBgl35A. Concluding remarks Galactans are mainly found as polymers with β-1,3-, β -1,4- and/or β-1,6-linkages. β- Galactooligosaccharides with the same linkage types are found in plant cell wall polysaccharides such as side chains of arabinogalactan and pectin 40. Enzymes degrading these glycans endolytically and exolytically are found in bacteria such as Bifidobacterium 8,41–47. However, there have been no reports of β-1,2-galactans or β-1,2-galactooligosaccharides derived from plants to the best of our knowledge. Although β-1,2-Gal2 is found in some GOS produced by bacterial fermentation as a prebiotic, it is just a minor component in a mixture of various GOS produced from lactose as a starting material by Bacillus circulans 11,48. Trypanozoma cruzi, a protozoan parasite, produces O-linked glycan containing a β-1,2-Gal2 moiety 49. Chagas disease is caused by this parasite, and it is known to seriously harm humans. Viscumin, a major toxic lectin from Viscum album, shows strong affinity for β-1,2-Gal2 50. However, β-1,2-galactooligosaccharides in nature and proteins related to the oligosaccharides have rarely been reported. Recently, it was reported that colonization of B. xylanisolvens reduces concentrations of nicotine in the large intestine and prevents the progression of smoking-related non-alcoholic fatty liver disease in nicotine-exposed mice 20. This suggests that B. xylanisolvens is a beneficial bacterium for health. Learning more about the new β-galactosidase described in this study is the first step toward characterizing and adding these unique β-1,2-galactooligosaccharides to the list of new and potentially beneficial carbohydrates. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 5, 2024. ; https://doi.org/10.1101/2024.05.03.592134doi: bioRxiv preprint

Cloning, expression, and purification Genomic DNA of B. xylanisolvens (DSM18836) was purchased from Leibniz Institute DSMZ - German Collection of Microorganisms and Cell Cultures (Braunschweig, Germany). The amino acid sequence of Bxy_22780 (Genbank accession number, CBK67349.1) in KEGG database (https://www.genome.jp/kegg/kegg2.html) was analyzed using the SignalP5.0 server (https://services.healthtech.dtu.dk/services/SignalP-5.0/) 51 to find an N -terminal signal peptide. No putative signal peptide was found. Then, N -terminal signal peptide was searched using Bxy_22780 with an upstream amino acid sequence of Bxy_22780 (14 residues). As a result, a putative signal peptide with 22 residues was clearly found. Residue numbers are based on the amino acid sequence in KEGG database . Primer pair s used for cloning and preparation of mutants are shown in Supplementary Table 3. First, the whole gene encoding Bxy_22780 was amplified by PCR using KOD plus (TOYOBO, Osaka, Japan ) according to a manufacturer’s instruc tion. The amplified DNA fragment was inserted in pET30a using NdeI and XhoI to fuse C-terminal His6-tag according to the

described in Shimizu et al. 52. Then, removal of N-terminal signal peptide and introduction of mutation was performed based on the manufacturer’s instruction using PrimeSTAR MAX (Takara Bio, Shiga, Japan). The resulting plasmids were introduced into E. coli BL21(DE3). The transformants were cultured at 37 ºC overnight in Luria -Bertani medium containing 30 μg/ml kanamycin. Each culture medium was seeded to the same new medium and cultured until OD 600 reached 0.8 at 37 ºC. After addition of 0.1 mM isopropyl 1 -thio-β-D-galactopyranoside (final concentration), each transformant was incubated at 20 ºC overnight. The cells were collected by centrifugation at approximately 7000 g for 5 min and suspended with 50 mM MOPS (pH 7.5) containing 500 mM NaCl (buffer A). The suspended cells were disrupted by sonication using a Branson model 450 sonifier. Each supernatant was collected by centrifugation at 33000 g. The supernatants were loaded onto a HisTrapTM FF crude column ( 5 ml, Cytiva, MA, USA) equilibrated with buffer A. After unbound compounds were washed with buffer A containing 10 mM imidazole, target proteins were eluted by linear gradient of imidazole (10 –500 mM) in buffer A. The proteins were concentrated and simultaneously imidazole was removed in 50 mM MOPS (pH 7.5) containing 300–500 mM NaCl by a Vivaspin15 (MWCO 10000 Da, Cytiva) or Amicon Ultra 10000 (Merck, NJ, USA). Protein concentrations were calculated using the absorbance at 280 nm, theoretical molecular weight (62506.983 Da) and an extinction coefficient (130345 M −1cm−1) based on Pace et al. 53 for the recombinant wild -type Bxy_22780. The purity of the enzymes w as checked by SDS -PAGE (Supplementary Fig. S11). DynaMarker Protein MultiColorⅢ (BioDynamics Laboratory Inc., Tokyo, .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 5, 2024. ; https://doi.org/10.1101/2024.05.03.592134doi: bioRxiv preprint Japan) was used as protein markers. Carbohydrates and materials for assay 2-Deoxy-D-galactose, D-talose, D-fucose, Me βGal and pNP -D-Fuc were purchased from Tokyo Chemical Industry (Tokyo, Japan). D-Galactose, D-glucose, D-xylose, D-fructose, L-arabinose, D- mannose, L-rhamnose, L-fucose and pNP-Gal, pNP-β-D-xylopyranoside were purchased from nacalai tesque (Kyoto, Japan). pNP-Glc, D-gulose and D-galacturonic acid were purchased from FUJIFILM Wako Chemical Corporation (Osaka, Japan), Toronto Research Chemicals (Ontario, Canada) and Merck (NJ, USA), respectively. pNP-β-D-Mannopyranoside was purchased from Biosynth (Staad, Switzerland). T-XG was purchased from Neogen (MI, USA). Xyloglucan oligosaccharides (TXG - XEG) were prepared by hydrolysis of T-XG with GH74 xyloglucanase from Paenibacillus sp. strain KM21 54. LG (D-galactopyranosyl-β-1,2-D-xylopyranosyl-α-1,6-D-glucopyranosyl-β-1,4-D-glucose) was prepared as described previously 55.

for assay system was described as follows . Galactose 1 -dehydrogenase/galactose mutarotase (E -GALMUT) and galactose oxidase were purchased from Neogen and Merck, respectively. Thio-NAD+, NADH and pyruvate kinase were purchased from Oriental Yeast (Tokyo, Japan). ATP and sodium pyruvate, and phosphoenolpyruvate were purchased from nacalai tesque and FUJIFILM, respectively. Galactokinase (GalK) and L-lactate dehydrogenase (LDH) from Thermotoga maritima (KEGG locus tags, TM1190 and TM1867, respectively) 56,57 were prepared to assay for β- 1,6-galactobiose. Genomic DNA from T. maritima NBRC 100826 was purchased from NITE Biological Resource Center (Chiba, Japan). The genes encoding these two enzymes were subcloned into pET30a by a usual method as described above and SLiCE method 58. Only the GalK was fused with His6-tag derived from the vector at C-terminus. These two plasmids were transformed into E. coli Rosetta2(DE3) (Merck) and both enzymes were produced based on a usual procedure. GalK was purified from the collected cells almost the same way as Bxy_22780. For LDH preparation, after the suspended cells were disrupted, the supernatant was incubated at 70 ºC for 30 min. The sample was centrifugated and then the supernatant was purified using a HiTrap TM Butyl HP (5 ml, Cytiva) , a hydrophobic chromatography. Finally, buffers for both enzymes were exchanged to 50 mM MOPS (pH 7.0). Temperature and pH profiles To investigate optimum pH, Bxy_22780 with an appropriate concentration was added to a solution containing pNP-Gal and each buffer (5 mM and 20 mM in a reaction mixture, respectively). The .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 5, 2024. ; https://doi.org/10.1101/2024.05.03.592134doi: bioRxiv preprint reaction mixture (100 µl) was incubated at 30 ºC for 10 min. Then, 20 µl of the reaction mixture was mixed with 180 µl of Na2CO3 for termination of the reaction and colorization of the reaction solution. Absorbance at 405 nm was measured to calculate hydrolytic activity using 18500 M −1cm−1 as an extinction coefficient of pNP. To evaluate pH stability, Bxy_22780 (180 µg/ml) was incubated in each 20 mM buffer at 30 ºC for 1 h. Enzymatic reaction was performed in the solution containing 5 mM pNP-Gal, 50 mM sodium acetate (pH 5.5), and 24 µg/ml of the incubated enzyme at 30 ºC for 10 min. Optimum temperature was determined by assaying the enzyme in the solution containing 5 mM pNP- Gal and 20 mM sodium acetate (pH 5.5) at each temperature for 10 min. To investigate thermostability of Bxy_22780, the enzyme (180 µg/ml) was incubated in 20 mM sodium acetate (pH 5.5) at each temperature for 1 h. The reaction was performed the same way as optimum temperature except that all temperatures are 30 ºC. Detection of pNP for optimum pH was adopted to the other profiles. Kinetic analysis To determine kinetic parameters of Bxy_22780 toward various galactosides, coupling assay was performed as described below. After each reaction mixture (100 µl) containing 20 mM HEPES (pH 7.5) and appropriate concentrations of the substrate and the enzyme was incubated at 30 ºC for 30 min, the reaction was terminated by heat treatment at 99 ºC for 5 min. Each reaction solution (20 µl) was mixed with 180 µl of a solution containing galactose dehydrogenase and galactose mutarotase (E- GALMUT), thio -NAD+ and HEPES (pH 8.0) (1 U/ml, 0.021 mg/ml, 0.25 mM, and 20 mM , respectively, as final concentrations). After the solution was incubated at 25 ºC for 5 min, absorbance at 398 nm was measured. Specific activity of the enzymes was calculated using extinction coefficient of thio-NADH at 398 nm (11900 M−1cm−1). Only in the case of β-1,6-galactobiose as a substrate, this assay system was not adopted because galactose dehydrogenase acts on β-1,6-galactobiose. The reaction by Bxy_22780 and termination of the reaction were performed in the same way as the other substrates. Each reaction solution (20 µl) was mixed with the assay solution (180 µl) containing GalK, LDH, pyruvate kinase, ATP, NADH, phosphoenolpyruvate, KCl and potassium phosphate (pH 7.0). Final concentrations of the components in the assay solution were GalK, 1.0 mg/ml; LDH, 1.0 mg/ml; pyruvate kinase, 210 U/ml; ATP, 0.53 mM; NADH, 0.11 mM; phosphoenolpyruvate, 1.5 mM; KCl, 71 mM; potassium phosphate, 55 mM (pH 7.0). After the solution was incubated at 25 ºC for 10 min, absorbance at 340 nm was measured. Specific activity of the enzymes was calculated using extinction coefficient of NADH (6300 M−1cm−1). Crystallography .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 5, 2024. ; https://doi.org/10.1101/2024.05.03.592134doi: bioRxiv preprint Initial screening of crystallization condition was performed by sitting drop vapor diffusion method using PACT primerTM HT-96 and JCSG -plusTM HT-96 (Molecular Dimensions, UK) as reservoirs. Reservoir solution (1 μl) and 1 μl of Bxy_22780 E350G (10 –11 mg/ml) in 50 mM MOPS (pH 7.5) containing 300 mM NaCl was mixed and then incubated at 20 ºC o n 96-wells CrystalQuick plates (Greiner Bio-One, Germany). To obtain complex structures, galactose or MeβGal (20 mM as a final concentration) was added to the WT enzyme solution before the enzyme solution was mixed with the reservoir solutions. After optimization of crystallization conditions by hanging drop vapor diffusion

using VDXTM plate (Hampton Research, CA, USA), ligand-free crystals were prepared using a reservoir containing 0.2 M sodium citrate tribasic dihydrate, 20% PEG 3350 and 100 mM bis -Tris propane (pH 7.5). To obtain the complex structures with galactose and Me βGal, a reservoir solution containing 0.35 M potassium thiocyanate, 12% PEG 3350 was mixed with the WT enzyme solution containing galactose or Me βGal as described above . Before X-ray data collection, protein crystals were transferred to the reservoir solutions supplemented with 30% glycerol for the ligand -free structures and 25% 2-methyl-2,4-pentanediol for the complex structures as cryoprotectants. Although the ligand-free crystals were soaked with the cryoprotectant containing pNP -Gal, the ligand was not observed and this structure is treated as the ligand-free structure. The crystals were kept at 100 K under a nitrogen gas stream during data collection. The X -ray diffraction data were collected using a CCD detector (ADSC Q210) on beamline BL-5A at the Photon Factory. The diffraction data sets were processed using X-ray Detector Software ( http://xds.mpimf- heidelberg.mpg.de/) 59. The initial phase of the first E350G mutant structure were determined by molecular replacement using MOLREP (https://www.ccp4.ac.uk) 60 and the enzyme structure (PDB ID, 3U7V) as a model. Automated model building was performed with Buccaneer 61 and ArpWarp 62. For the other E350G mutant structures, the first determined structure was used as a model for molecular replacements. Automated and manual structure refinements were performed using Refmac5 63 and Coot 64, respectively (https://www.ccp4.ac.uk). The figures were drawn using PyMOL (http://www.pymol.org). TLC analysis We examined hydrolase and glycosyltranferase activity of Bxy_22780 toward various substrates. For detection of hydrolytic activity except xyloglucan-associated substrates, the reaction mixtures comprising 5 mM substrate [β-1,2-galactobiose, β-1,2-galactotriose, lactose, melibiose, β-1,3(4)- galactobiose, allolactose, β-1,6-galactobiose or lactulose] and the WT enzyme (0.02 mg/ml or 0.1 mg/ml) or W288A mutant (0.02 mg/ml) in 50 mM MOPS (pH7.0) were incubated at 30 ºC for 3 h or .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 5, 2024. ; https://doi.org/10.1101/2024.05.03.592134doi: bioRxiv preprint overnight. Each sample solution or marker (1 μl) was spotted onto a TLC plate (silica gel 60 F 254, Merck). The TLC plates were developed appropriate times with 75% (v/v) acetonitrile in water. After soaking in 5% (v/v) sulfuric acid in methanol, the TLC plates were heated until bands were visualized sufficiently. In the case of glycosynthase activity, the reaction mixtures comprising 100 mM acceptor substrate [D-glucose, D-galactose, D-fructose, D-xylose, L-arabinose, L-fucose, L-fucose, D-mannose, L-rhamnose, D-talose, 2 -deoxy-D-galactose, D-galactosamine N-acetyl-D-galactosamine, lactose, melibiose, β-1,3(4)-galactobiose, allolactose, β-1,6-galactobiose, lactulose, pNP-Gal, pNP-D-Fuc or pNP-Glc], 20 mM α-GalF as a donor substrates, and E350G mutant (1.1 mg/ml) or W288A/E350G mutant (1.1 mg/ml) in 100 mM MOPS (pH7.0) were incubated at 30 ºC for 3 h. After the enzymatic reactions were performed, the samples were transferred onto ice to stop the reaction. Reaction products were visualized by TLC analysis as described above except that a concentration of acetonitrile was changed to 85% for development. To investigate hydrolytic activity toward TXG-XEG and LG, the reactions were performed in 20 mM HEPES (pH 7.5), 1% TXG -XEG, and 2 mg/ml WT or W288A mutant 30 ºC overnight, and 20 mM MOPS (pH 7.5), 5 mM LG, and 0.2 mg/ml WT or W288A mutant 30 ºC for up to 3 h. The reactions were terminated by heat treatment at 98 ºC for 10 min. The reaction mixture (1 μl) was spotted onto the glass TLC plate (silica gel 60 F 254, Merck) The TLC plates were developed appropriate times with the solution (isopropyl alcohol : acetic acid : water = 4:1:1, v/v). Bands were visualized using 2.5% p-anisaldehyde, 3.5% sulfuric acid, and 1% acetic acid in ethanol. Purification of glycosynthase products Reaction mixtures comprising 5 mg/ml E350G mutant (a nucleophilic catalytic residue mutant), 20 mM α-GalF, and 100 mM galactose in 50 mM MOPS (pH7.0) were incubated at 30 ºC for 4 h. The enzyme in the reaction mixtures was denatured by heat treatment at 100 ºC and was removed by centrifugation. The supernatants were loaded onto gel permeation chromatography using a Toyopearl HW-40F column (approximately 2 L gel) to fractionate the products by DPs. The reaction products were detected using TLC analysis. Fractions containing dis accharide or trisaccharide were collected separately and then desalted using ion exchange resin Amberlite MB-4 (Organo, Tokyo, Japan). After the deionized samples were filtered and then lyophilized. NMR analysis The purified disaccharide produced by glycosynthase reaction was dissolved in D2O, and acetone was added as a standard for calibration of chemical shifts. The chemical shifts were recorded relative .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 5, 2024. ; https://doi.org/10.1101/2024.05.03.592134doi: bioRxiv preprint to the signal of methyl group of the internal standard acetone. One-dimensional NMR spectra ( 1H NMR and 13C NMR) and two-dimensional NMR spectra (COSY, HMQC and HMBC) were recorded using a Bruker Avance 400 spectrometer (Bruker, MA, USA) with acetone (δ 2.22 ppm for 1H, and δ 29.92 ppm for 13C) as an internal standard. Safety statement No unexpected or unusually high safety hazards were encountered. Notes The authors declare no competing financial interest. Acknowledgments We appreciate the help of all the staff at the Photon Factory for X -ray data collection (proposal nos. 2018G506, 2020G527, and 2022G523). We thank Dr. Katsuro Yaoi and Dr. Takehiko Sahara (National Institute of Advanced Industrial Science and Technology) for providing xyloglucan oligosaccharides. We thank Jennifer Parker, PhD, from Edanz (https://jp.edanz.com/ac) for editing a draft of this manuscript. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 5, 2024. ; https://doi.org/10.1101/2024.05.03.592134doi: bioRxiv preprint

1. Yamada, C. et al. Molecular insight into evolution of symbiosis between breast-fed Infants and a member of the human gut microbiome Bifidobacterium longum. Cell Chem Biol 24, (2017). 2. Zeeman, S. C., Kossmann , J. & Smith, A. M. Starch: its metabolism, evolution, and biotechnological modification in plants. Annu Rev Plant Biol 61, 209–34 (2010). 3. Ruiz-Herrera, J. & Ortiz-Castellanos, L. Cell wall glucans of fungi. A review. The Cell Surface 5, 100022 (2019). 4. Popper, Z. A. et al. Evolution and diversity of plant cell walls: From algae to flowering plants. Annu Rev Plant Biol 62, 567–590 (2011). 5. Kondo, Y . et al. Cytokine-related immunopotentiation activities of paramylon, a β-(1→3)-D- glucan from Euglena gracilis. J Pharmacobio-Dyn 15, 617–621 (1992). 6. Haag, A. F., Myka, K. K., Arnold, M. F., Caro-Hernández, P. & Ferguson, G. P. Importance of Lipopolysaccharide and Cyclic β-1,2-Glucans in Brucella -Mammalian Infections. Int J Microbiol 2010, 124509 (2010). 7. Drula, E. et al. The carbohydrate-active enzyme database: Functions and literature. Nucleic Acids Res 50, D571–D577 (2022). 8. Sakamoto, T. & Ishimaru, M. Peculiarities and applications of galactanolytic enzymes that act on type I and II arabinogalactans. Appl Microbiol Biotechnol 97, 5201–13 (2013). 9. Kondo, T. et al. Characterization of three GH35 β-galactosidases, enzymes able to shave galactosyl residues linked to rhamnogalacturonan in pectin, from Penicillium chrysogenum 31B. Appl Microbiol Biotechnol 104, 1135–1148 (2020). 10. Nishimoto, M. & Kitaoka, M. Identification of N-acetylhexosamine 1-kinase in the complete lacto-N-biose I/galacto-N-biose metabolic pathway in Bifidobacterium longum. Appl Environ Microbiol 73, 6444–6449 (2007). 11. Torres, D. P. M., Gonçalves, M. do P. F., Teixeira, J. A. & Rodrigues, L. R. Galacto - Oligosaccharides: Production, Properties, Applications, and Significance as Prebiotics. Compr Rev Food Sci Food Saf 9, 438–454 (2010). 12. Yoshida, E. et al. Bifidobacterium longum subsp. infantis uses two different β-galactosidases for selectively degrading type -1 and type -2 human milk oligosaccharides. Glycobiology 22, 361–8 (2012). 13. Ambrogi, V . et al. Characterization of GH2 and GH42 β-galactosidases derived from bifidobacterial infant isolates. AMB Express 9, 9 (2019). 14. Viborg, A. H. et al. A β1-6/β1-3 galactosidase from Bifidobacterium animalis subsp. lactis Bl- .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 5, 2024. ; https://doi.org/10.1101/2024.05.03.592134doi: bioRxiv preprint 04 gives insight into sub-specificities of β-galactoside catabolism within Bifidobacterium. Mol Microbiol (2014) doi:10.1111/mmi.12815. 15. Hidaka, M. et al. Trimeric crystal structure of the glycoside hydrolase family 42 beta - galactosidase from Thermus thermophilus A4 and the structure of its complex with galactose. J Mol Biol 322, 79–91 (2002). 16. Cabral, L. et al. Gut microbiome of the largest living rodent harbors unprecedented enzymatic systems to degrade plant polysaccharides. Nat Commun 13, 629 (2022). 17. Cheng, J. et al. Functional metagenomics reveals novel β-galactosidases not predictable from gene sequences. PLoS One 12, e0172545 (2017). 18. Luis, A. S. et al. Dietary pectic glycans are degraded by coordinated enzyme pathways in human colonic Bacteroides. Nat Microbiol 3, 210–219 (2018). 19. Hameleers, L., Pijning, T., Gray, B. B., Fauré, R. & Jurak, E. Novel β-galactosidase activity and first crystal structure of Glycoside Hydrolase family 154. N Biotechnol 80, 1–11 (2024). 20. Chen, B. et al. Gut bacteria alleviate smoking-related NASH by degrading gut nicotine. Nature 610, 562–568 (2022). 21. Chassard, C., Delmas, E., Lawson, P. A. & Bernalier-Donadille, A. Bacteroides xylanisolvens sp. nov., a xylan-degrading bacterium isolated from human faeces. Int J Syst Evol Microbiol 58, 1008–13 (2008). 22. Terrapon, N. et al. PULDB: The expanded database of Polysaccharide Utilization Loci. Nucleic Acids Res 46, D677–D683 (2018). 23. Abe, K. et al. Biochemical and structural analyses of a bacterial endo-β-1,2-glucanase reveal a new glycoside hydrolase family. J Biol Chem 292, 7487–7506 (2017). 24. Liu, B. et al. Cloning, expression and characterization of a thermostable exo -β-D- glucosaminidase from the hyperthermophilic archaeon Pyrococcus horikoshii. Biotechnol Lett 28, 1655–1660 (2006). 25. Kobayashi, K. et al. Characterization and structural analyses of a novel glycosyltransferase acting on the β-1,2-glucosidic linkages. J Biol Chem 298, 101606 (2022). 26. Saburi, W. et al. Function and Structure of Lacticaseibacillus casei GH35 β-Galactosidase LBCZ_0230 with High Hydrolytic Activity to Lacto -N-biose I and Galacto -N-biose. J Appl Glycosci (1999) 70, 43–52 (2023). 27. Krissinel, E. & Henrick, K. Inference of macromolecular assemblies from crystalline state. J Mol Biol 372, 774–797 (2007). 28. Larsbrink, J. et al. A complex gene locus enables xyloglucan utilization in the model .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 5, 2024. ; https://doi.org/10.1101/2024.05.03.592134doi: bioRxiv preprint saprophyte Cellvibrio japonicus. Mol Microbiol 94, 418–33 (2014). 29. Vieira, P. S. et al. Xyloglucan processing machinery in Xanthomonas pathogens and its role in the transcriptional activation of virulence factors. Nat Commun 12, 4049 (2021). 30. Mine, S. & Watanabe, M. Structural insights into the molecular evolution of the archaeal exo- β-d-glucosaminidase. Int J Mol Sci 20, 2460 (2019). 31. Tanaka, T., Fukui, T., Atomi, H. & Imanaka, T. Characterization of an exo -β-D- glucosaminidase involved in a novel chitinolytic pathway from the hyperthermophilic archaeon Thermococcus kodakaraensis KOD1. J Bacteriol 185, 5175–5181 (2003). 32. Mine, S., Watanabe, M., Kamachi, S., Abe, Y . & Ueda, T. The Structure of an Archaeal β- Glucosaminidase Provides Insight into Glycoside Hydrolase Evolution. Journal of Biological Chemistry 292, 4996–5006 (2017). 33. Hill, C. H., Graham, S. C., Read, R. J. & Deane, J. E. Structural snapshots illustrate the catalytic cycle of β-galactocerebrosidase, the defective enzyme in Krabbe disease. Proc Natl Acad Sci U S A 110, 20479–84 (2013). 34. Møller, P. L., Jørgensen, F., Hansen, O. C., Madsen, S. M. & Stougaard, P. Intra - and extracellular β-galactosidases from Bifidobacterium bifidum and B. infantis: molecular cloning, heterologous expression, and comparative characterization. Appl Environ Microbiol 67, 2276– 83 (2001). 35. Santangelo, T. J., Cubonová, L. & Reeve, J. N. Thermococcus kodakarensis genetics: TK1827- encoded beta-glycosidase, new positive-selection protocol, and targeted and repetitive deletion technology. Appl Environ Microbiol 76, 1044–52 (2010). 36. Temuujin, U. et al. Identification and characterization of a novel β-galactosidase from Victivallis vadensis ATCC BAA-548, an anaerobic fecal bacterium. J Microbiol 50, 1034–40 (2012). 37. Tang, Y., Zhang, X., Yin, Y., Hardwidge, P. R. & Fang, W. Streptococcus suis type 2 SSU0587 protein is a beta -galactosidase that contributes to bacterial adhesion but not to virulence in mice. J Vet Med Sci 76, 1055–9 (2014). 38. Cheng, W. et al. Structural insights into the substrate specificity of Streptococcus pneumoniae β(1,3)-galactosidase BgaC. Journal of Biological Chemistry 287, 22910–22918 (2012). 39. Sakamoto, T., Nishimura, Y ., Makino, Y ., Sunagawa, Y . & Harada, N. Biochemical characterization of a GH53 endo -β-1,4-galactanase and a GH35 exo -β-1,4-galactanase from Penicillium chrysogenum. Appl Microbiol Biotechnol 97, 2895–906 (2013). 40. Munoz, J., James, K., Bottacini, F. & Van Sinderen, D. Biochemical analysis of cross-feeding .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 5, 2024. ; https://doi.org/10.1101/2024.05.03.592134doi: bioRxiv preprint behaviour between two common gut commensals when cultivated on plant -derived arabinogalactan. Microb Biotechnol 13, 1733–1747 (2020). 41. O’Connell Motherway, M., Fitzgerald, G. F. & van Sinderen, D. Metabolism of a plant derived galactose-containing polysaccharide by Bifidobacterium breve UCC2003. Microb Biotechnol 4, 403–16 (2011). 42. Hinz, S. W. A., Pastink, M. I., van den Broek, L. A. M., Vincken, J. -P. & Voragen, A. G. J. Bifidobacterium longum endogalactanase liberates galactotriose from type I galactans. Appl Environ Microbiol 71, 5501–10 (2005). 43. Zavaleta, V . & Eyzaguirre, J. Penicillium purpurogenum produces a highly stable endo -β- (1,4)-galactanase. Appl Biochem Biotechnol 180, 1313–1327 (2016). 44. Fujita, K. et al. Degradative enzymes for type II arabinogalactan side chains in Bifidobacterium longum subsp. longum. Appl Microbiol Biotechnol 103, 1299–1310 (2019). 45. Fujita, K., Sakaguchi, T., Sakamoto, A., Shimokawa, M. & Kitahara, K. Bifidobacterium longum subsp. longum Exo-β-1,3-Galactanase, an enzyme for the degradation of type II arabinogalactan. Appl Environ Microbiol 80, 4577–84 (2014). 46. Cartmell, A. et al. A surface endogalactanase in Bacteroides thetaiotaomicron confers keystone status for arabinogalactan degradation. Nat Microbiol 3, 1314–1326 (2018). 47. Ichinose, H. et al. Characterization of an exo -beta-1,3-galactanase from Clostridium thermocellum. Appl Environ Microbiol 72, 3515–23 (2006). 48. Yanahira, S. et al. Formation of oligosaccharides from lactose by Bacillus circulans beta - galactosidase. Biosci Biotechnol Biochem 59, 1021–6 (1995). 49. Sheikh, M. O. et al. Trypanosoma cruzi 13C-labeled O -Glycan standards for mass spectrometry. Glycobiology 29, 280–284 (2019). 50. Lee, R. T., Gabius, H. J. & Lee, Y . C. The sugar-combining area of the galactose-specific toxic lectin of mistletoe extends beyond the terminal sugar residue: comparison with a homologous toxic lectin, ricin. Carbohydr Res 254, 269–76 (1994). 51. Almagro Armenteros, J. J. et al. SignalP 5.0 improves signal peptide predictions using deep neural networks. Nat Biotechnol 37, 420–423 (2019). 52. Shimizu, H. et al. Characterization and structural analysis of a novel exo-type enzyme acting on β-1,2-glucooligosaccharides from Parabacteroides distasonis. Biochemistry 57, 3849 – 3860 (2018). 53. Pace, C. N., Vajdos, F., Fee, L., Grimsley, G. & Gray, T. How to measure and predict the molar absorption coefficient of a protein. Protein Sci 4, 2411–2423 (1995). .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 5, 2024. ; https://doi.org/10.1101/2024.05.03.592134doi: bioRxiv preprint 54. Yaoi, K., Nakai, T., Kameda, Y., Hiyoshi, A. & Mitsuishi, Y. Cloning and Characterization of Two Xyloglucanases from Paenibacillus sp. Strain KM21. Appl Environ Microbiol 71, 7670– 7678 (2005). 55. Tuomivaara, S. T., Yaoi, K., O’Neill, M. A. & York, W. S. Generation and structural validation of a library of diverse xyloglucan-derived oligosaccharides, including an update on xyloglucan nomenclature. Carbohydr Res 402, 56–66 (2015). 56. Wrba, A., Jaenicke, R., Huber, R. & Stetter, K. O. Lactate dehydrogenase from the extreme thermophile Thermotoga maritima. Eur J Biochem 188, 195–201 (1990). 57. Liebl, W., Wagner, B. & Schellhase, J. Properties of an alpha-galactosidase, and structure of its gene galA, within an alpha -and beta -galactoside utilization gene cluster of the hyperthermophilic bacterium Thermotoga maritima. Syst Appl Microbiol 21, 1–11 (1998). 58. Motohashi, K. A simple and efficient seamless DNA cloning method using SLiCE from Escherichia coli laboratory strains and its application to SLiP site-directed mutagenesis. BMC Biotechnol 15, (2015). 59. Kabsch, W. XDS. Acta Crystallogr D Biol Crystallogr D66, 125–132 (2010). 60. Vagin, A. & Teplyakov, A. Molecular replacement with MOLREP. Acta Crystallogr D Biol Crystallogr 66, 22–25 (2010). 61. Cowtan, K. The Buccaneer software for automated model building. 1. Tracing protein chains. Acta Crystallogr D Biol Crystallogr 62, 1002–1011 (2006). 62. Morris, R. J., Perrakis, A. & Lamzin, V . S. ARP/wARP’s model-building algorithms. I. The main chain. Acta Crystallogr D Biol Crystallogr 58, 968–75 (2002). 63. Murshudov, G. N., Vagin, A. A. & Dodson, E. J. Refinement of macromolecular structures by the maximum-likelihood method. Acta Crystallogr D Biol Crystallogr 53, 240–255 (1997). 64. Emsley, P. & Cowtan, K. Coot : model-building tools for molecular graphics. Acta Crystallogr D Biol Crystallogr 60, 2126–2132 (2004). 65. Krissinel, E. & Henrick, K. Secondary-structure matching (SSM), a new tool for fast protein structure alignment in three dimensions. Acta Crystallogr D Biol Crystallogr 60, 2256–2268 (2004). .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 5, 2024. ; https://doi.org/10.1101/2024.05.03.592134doi: bioRxiv preprint Table 1. Kinetic parameters of Bxy_22780. Vmax (U/mg) Km (mM) Vmax/Km (U/mg/mM) Wild type pNP-Gal 21.2 ± 1.4 3.49 ± 0.53 6.08 ±0.58 pNP-D-Fuc 14.0 ± 0.45 7.95 ± 0.46 1.76 ± 0.05 β-1,2-Gal2 14.6 ± 0.3 1.40 ± 0.05 10.4 ± 0.3 β-1,2-Gal3 24.1 ± 0.9 1.65 ± 0.14 14.6 ± 0.8 W288A mutant β-1,2-Gal2 1.18 ± 0.05 3.22 ± 0.25 0.367 ± 0.015 β-1,2-Gal3 2.08 ± 0.49 54.9 ± 13.7 0.0378 ± 0.0007 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 5, 2024. ; https://doi.org/10.1101/2024.05.03.592134doi: bioRxiv preprint Figure 1. Glycosynthase reaction of E350G mutant of Bxy_22780. a, Glycosynthase reaction of E350G mutant of Bxy_22780. Asterisks represent origins of the TLC plates. Reaction time is shown above the bar for each acceptor. “+” and “−” represent that α-GalF was incubated with and without the mutant, respectively. Arrows represent position of spots of α-GalF. A parenthesis represent that the sample diluted by twice was used. b, Effect of acetone on synthesis of galactooligosaccharides. Arrows represent presumed DPs. Asterisk s represent origin s of the TLC plates. Percentages represent concentrations of acetone in the reaction solutions. c, Enlarged view of signals in HMBC representing the glycosidic bond of the galactobiose product. Red and blue labels indicate signals for α and β anomers, respectively. Apostrophes represent non-reducing end galactose units. d, The chemical structure of β-1,2-Gal2. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 5, 2024. ; https://doi.org/10.1101/2024.05.03.592134doi: bioRxiv preprint Figure 2. Substrate specificity of Bxy_22780 toward β-galactosides. Origins of the TLC plates are indicated with an asterisk. Concentrations of substrates used were 5 mM. The reaction times (h) were shown above the bold lines and overnight is abbreviated as O/N. Galactose, glucose, and fructose (10 mM each) were used as markers. a, Activity toward β-1,2- galactooligosaccharides in 3 h of reaction. b, Activity toward galactosides in the reaction for overnight. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 5, 2024. ; https://doi.org/10.1101/2024.05.03.592134doi: bioRxiv preprint Figure 3. Complex structures of Bxy_22780 with ligands. a–b (top), Electron densities of α-Gal (a) and MeβGal (b) in the Bxy_22780 complex structures. The Fo–Fc omit maps for the ligands are shown at the 3 σ contour level and are shown as gray meshes. Main substrate recognition subunits and the subunits that D494 is depicted are shown as cyan and green cartoons, respectively. The symmetric unit and its symmetry mate for the α-Gal complex (a) and subunits C and B for the MeβGal complex (b) are used, respectively. Residues around the ligands are .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 5, 2024. ; https://doi.org/10.1101/2024.05.03.592134doi: bioRxiv preprint labelled and shown in sticks. α-Gal and MeβGal are shown as yellow and white sticks, respectively, with their subsite positions. a–b (bottom), Recognition of α-Gal (a) and MeβGal (b) in the Bxy_22780 complex structures. Hydrogen bonds are shown as orange dashed lines. Water molecules forming hydrogen bonds with both residues and ligands are shown as red spheres. e, (left) Overall structure of the MeβGal complex. Subunits A–D are shown as light gray, pale cyan, light green, dark gray surfaces, respectively. MeβGal are shown as a white stick and indicated with a red circle. (right) Close-up view of the substrate pocket of the MeβGal complex. α-Gal in the α-Gal complex is superimposed. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 5, 2024. ; https://doi.org/10.1101/2024.05.03.592134doi: bioRxiv preprint Figure 4. Comparison of substrate pockets between Bxy_22780 and CjBgl35A. Color usage of the Me βGal complex is the same as Fig. 3. The complex of CjB gl35A with 1 - deoxygalactonojirimycin (DGJ) are shown in light purple and light brown. DGJ is shown as a gray stick. The complex of Bxy_22780 with MeβGal (PDB ID, 8Z47) and the complex of CjBgl35A with DGJ (PDB ID, 4D1J) are superimposed by PDBeFold (https://www.ebi.ac.uk/msd-srv/ssm/) 65. α-Gal is superimposed the same way as Fig. 3 c right. Residues and ligands are labelled with black bold letters for Bxy_22780 and plain letters in parentheses for CjBgl35A. Subsite positions are shown in the ligands. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 5, 2024. ; https://doi.org/10.1101/2024.05.03.592134doi: bioRxiv preprint