Glycoside Hydrolase family 30 harbors fungal subfamilies with distinct polysaccharide specificities

Efficient bioconversion of agro-industrial side streams requires a wide range of enzyme activities. Glycoside Hydrolase family 30 (GH30) is a diverse family that contains various catalytic functions and has so far been divided into ten subfamilies (GH30_1-10). In this study, a GH30 phylogenetic tree using over 150 amino acid sequences was contructed. The members of GH30 cluster into four subfamilies and eleven candidates from these subfamilies were selected for biochemical characterization. Novel enzyme activities were identified in GH30. GH30_3 enzymes possess β -(1 → 6)-glucanase activity. GH30_5 targets β -(1 → 6)-galactan with mainly β -(1 → 6)- galactobiohydrolase catalytic behavior. β -(1 → 4)-Xylanolytic enzymes belong to GH30_7 targeting β -(1 → 4)-xylan with several activities (e.g. xylobiohydrolase, endoxylanase). Additionally, a new fungal subfamily in GH30 was proposed, i.e. GH30_11, which displays β -(1 → 6)-galactobiohydrolase. This study confirmed that GH30 fungal subfamilies harbor distinct polysaccharide specificity and have high potential for the production of short (non-digestible) di-and oligosaccharides.

selectively stimulating the growth and activity of beneficial bacteria in the colon and promote the health of the host [16].However, these enzymes can also be used in other biotechnology applications.For example, β-(1→6)-glucanases from GH30_3 could be included in antifungal products due to their fungal cell wall degrading capabilities [17,18], while β-(1→4)-xylanolytic enzymes from GH30_7 enzymes can be applied for biomass saccharification for the production of biofuels and fine chemicals [19].
In this study, to verify the functional specificity of these identified fungal GH30 subfamilies and deepen understanding of the fungal members in GH30, a GH30 phylogenetic tree was constructed using 161 amino acid sequences from bacterial, fungal and animal species, from which eleven candidates covering different fungal subfamilies were selected for biochemical characterization using polysaccharides and crude plant biomass.This revealed a new fungal subfamily in GH30.After discussion with curators from the CAZy Database, this is now established as GH30_11, which displays β-(1→6)-galactobiohydrolase activity.In addition, biochemical characterization revealed that the enzymes from different GH30 subfamilies exhibited distinct substrate specificities.Furthermore, to investigate the catalytic mechanism of GH30 fungal enzymes, the potential residues involved in substrateprotein interactions were analyzed by homology modelling, from which some possible amino acid residues affecting substrate specificity were highlighted.In conclusion, this study facilitates the industrial use of fungal GH30 enzymes, as the knowledge of different fungal subfamilies will help to select optimal candidates for the desired Fig. 1.Phylogenetic relationship among GH30 members from fungi, bacteria, and animals based on their amino acid sequences.The phylogenetic analysis was performed as previously reported [22].The main subfamilies were collapsed.Statistical support for phylogenetic grouping was estimated by 500 bootstrap re-samplings, only the bootstrap above 50 % were shown on the clades.Five mannanases from GH5 were used as an outgroup.GH30_10 containing bacterial xylobiohydrolases was identified by Crooks et al. [7], which is not shown in the current GH30 tree.(a) Whole tree; (b) GH30_3; (c) GH30_11; (d) GH30_5; (e) GH30_7.Selected candidates are highlighted and the characterized enzymes are in boldface.The full phylogenetic tree can be found in Supplementary Figure S1.
X. Li et al. application, be that for the production of specific short NDOs or for other biorefinery process.

Table 1
Substrate specificity of characterized enzymes and selected candidates from fungal subfamilies of GH30.Cloning, protein production and purification of the selected candidates The selected genes (Fig. 1, Table 1) without predicted signal peptides and introns were codon optimized and synthesized into pPicZαA plasmid (Genscript Biotech, Leiden, the Netherlands), which was then transformed into Escherichia coli DH5α (Invitrogen, Thermo Fisher Scientific, Carlsbad, CA) and subsequently transformed into Pichia pastoris strain X-33 (Invitrogen) according to the manufacturer's recommendation.The positive colonies were selected as described earlier [22].
The selected P. pastoris transformants were grown and induced for the production of enzyme as described previously [28].Culture supernatants were harvested (8000 × g, 4 • C, 20 min).One portion was filtered (0.22 mm; Merck), aliquoted and stored at − 20 • C (as crude enzyme).The other part was purified by ÄKTA FPLC System (GE Life Sciences, Uppsala, Sweden).Crude enzymes were loaded onto a HisTrap FF 1 mL column (Cytiva, Marlborough, USA) equilibrated with 20 mM HEPES, 0.4 M NaCl, 20 mM imidazole, pH 7.5, and eluted using a linear gradient of 22− 400 mM imidazole in buffer mentioned above at a flow rate of 1.0 mL/min.Fractions containing enzyme were collected, concentrated and buffer-exchanged to 20 mM HEPES (pH 7.0) using 10 kDa cut-off ultrafiltration units Amicon (Millipore, Bedford, MA).All purification steps were performed at 4 • C [26].The concentration of purified protein was evaluated as described earlier [22].

Enzyme activity assays Activity screening using crude enzymes
The activity assays of crude enzymes were performed in 600 μL reaction mixtures containing 500 μL 1% (w/v) of the substrates (Fig. 2 and Supplementary Table S2) in 50 mM NaOAc buffer (pH 5.5) and 100 μL crude enzymes at 30 • C, 110 rpm for 16 h.The culture supernatant from P. pastoris harboring pPicZαA plasmid without insertion was used as a negative control for crude enzymes.

Activity assays using purified enzymes
The activity of the purified enzymes was assayed using 3 mg/mL of the selected polysaccharides in 50 mM NaOAc (pH 5.5) incubated with 6 μg/mL purified enzyme.HEPES buffer was used as a negative control for purified enzymes.The reaction was performed at 30 • C, 110 rpm, for up to 16 h, and aliquots of 500 μL were removed from the mixtures at different time points (5 min, 15 min, 30 min, 45 min, 1 h, 2 h, 6 h and 16 h) for time course analysis.
To investigate the effect of xylan substitutions on GH30_7 enzyme activity, MGX and WAX were used as representative substrates for 4-O- methylglucuronic acid (MeGlcA) substituted and arabinofuranose substituted xylans, respectively [29]. 3 mg/mL MGX and WAX were incubated with 0.17 mg/mL AGU and ABF, respectively, in 50 mM NaOAc (pH 5.5) and 0.02 % NaN 3 at 30 • C, 110 rpm, for 72 h.The reaction was stopped by heating at 95 • C for 10 min.These are referred to as pre-treated substrates.For the activity assay, the reaction mixtures containing 550 μL 3 mg/mL MGX, WAX or pre-treated substrates and 50 μL purified GH30_7 enzymes (equal to approximately 3 μg protein) were incubated under the above-mentioned conditions.
The released glucuronic acid was quantified using HPAEC-PAD in the same way as for monitoring monosaccharide release.The chromatographic separation was performed with a multi-step gradient according to [26]

Fungal GH30 enzymes belong to four subfamilies
Phylogenetic analysis of GH30 updated the fungal subfamilies included in the CAZy database (Fig. 1a and Supplementary Fig. S1) (http://www.cazy.org/GH30_characterized.html).Fungal members clustered into the previously defined GH30_3, 5 and 7 and also into the newly classified GH30_11 subfamily (this work).In the GH30 phylogenetic tree (the order of description is based on the location of subfamilies in the tree), GH30_3 members were further divided into two main clades based on different kingdoms, i.e. fungi and bacteria.The fungal clade was predicted to possess β-(1→6)-glucanase (ENG) activity based on earlier characterization [8][9][10][11] (Fig. 1a and b).The fungal GH30_11 shares the same node with the bacterial GH30_4, which suggests that they were more related to each other than to the other subfamilies.However, the sequence alignment showed low amino acid sequence identity (<28 %) between members from GH30_11 and GH30_4, indicating that they might differ in catalytic function.GH30_11 contained only sequences from basidiomycetes (Fig. 1c), but no characterized enzyme from this subfamily has been reported until now.A previous study found two β-D-fucosidases in GH30_4 [32].Similar to GH30_3, members of GH30_5 were classified into two clades (fungi and bacteria).However, the fungal clade of GH30_5 only contained sequences from ascomycetes (Fig. 1d).The fungal and bacterial GH30_5 showing similar activity (β-(1→6)-galactanase) shared more than 50 % amino acid sequence identity [12,13,33].GH30_7 contained sequences from both ascomycetes and basidiomycetes (Fig. 1e) and the characterized enzymes showed several xylanolytic activities (Table 1).
To evaluate different fungal subfamilies systematically and compare their activities and substrate specificity, eleven candidates from these four subfamilies were selected for biochemical characterization: four from GH30_3, two from the newly established GH30_11, one from GH30_5, and four from GH30_7.

Different fungal subfamilies target different substrates
The activity of culture supernatants was first screened towards several substrates (Fig. 2 and Supplementary Table S2): β-glucans with different linkages, arabinogalactan II (AG-II), and xylan and pectin from different sources.The screening results showed that selected candidates exhibited different linkage specificities, with GH30_3 candidates acting on glucosyl linkages, GH30_11 and GH30_5 candidates hydrolyzing galactosyl linkages, and GH30_7 candidates breaking xylosyl linkages.The corresponding mono-and di-saccharides released from hydrolyzing the tested substrates by different candidates are shown in Fig. 2.
GH30_7 were only active on xylan substrates, i.e.Beech wood xylan (BeWX) and WB, and showed a high preference for BeWX (Fig. 2d).The candidates from GH30_7 all released xylobiose as the major product from BeWX, while the production of other xylooligosaccharides (XOS) was enzyme dependent.This is discussed in more detail below.
Based on the initial screening results of the crude enzymes, β-(1→6)glucan, LWAG-II and BeWX were selected as the main polysaccharides for more detailed characterization of candidate enzymes from GH30_3, GH30_11&GH30_5, and GH30_7, respectively.
Of the GH30_11 and GH30_5 enzymes tested, ShGbhA was the most active and released 2-and 4-fold higher amounts of β-(1→6)-galactobiose than TvGbhI and PsGbhA, respectively (Fig. 4c).In addition, ShGbhA continuously released β-(1→6)-galactobiose over a period of 16 h, whereas PsGbhA and TvGbhI reached a saturation level after 1 h.This suggests different substrate specificity or stability among these enzymes.GH30_11 and GH30_5 enzymes (Fig. 4c) showed a limited ability to hydrolyse LWAG-II.This might be related to the structure of LWAG-II, which consists of a β-(1→3)-galactan backbone with β-(1→6)galactooligosaccharide side chains.The side chains are highly variable in length and some are further substituted by arabinofuranose, which could affect the enzyme activity [36].

BeWX degradation revealed various activities of GH30_7 enzymes
BeWX was selected as the substrate for purified GH30_7 enzymes.TtXbhA and TeXbhA released almost exclusively xylobiose from BeWX, while PsExlA released different MeGlcA substituted XOS (Fig. 5a), indicating that they possess xylobiohydrolase and endoxylanase activities, respectively.The product pattern of TtXbhA and TeXbhA in this study differed slightly from the previously reported xylobiohydrolases (e.g.TtXyn30A, TcXyn30B, AaXyn30A), although they all released xylobiose as the major product from BeWX after a long time incubation.TtXbhA and TeXbhA could release high-purity xylobiose from BeWX, whereas other oligosaccharides (e.g. × 4, X5) and a set of MeGlcA substituted XOS were also detected from the BeWX hydrolysis with TtXyn30A, TcXyn30B, and AaXyn30A [3,4,39].TtXyn30A and TcXyn30B were reported to be bifunctional MeGlcA appendage-dependent xylobiohydrolases/endoxylanases [3,39], while AaXyn30A is a more strict fungal xylobiohydrolase [4].AaXyn30A also shows slight endoxylanase activity, which was likely to be the result of excessive enzyme loading in the experimental setup [4,7].
PsExlA showed a similar hydrolytic product pattern to T. reesei XYN VI [40], both of which resembled the bacterial endoxylanases from GH30_8 [27,[41][42][43].TtExlA predominantly released xylobiose, but also different (MeGlcA substituted) XOS (Fig. 5a), similar to the earlier report [4], which shows that it is an endoxylanase with xylobiohydrolase activity.Among the four tested enzymes, TtXbhA was the most suitable for xylobiose production as it released the highest amount of xylobiose (Fig. 5b).
Considering the different product profiles from BeWX (glucuronoxylan) hydrolysis, the effect of MeGlcA substitution on GH30_7 activity was investigated.Given the low level of MeGlcA substitutions in BeWX (mole ratio MeGlcA:Xyl = 1:15, [44]), 4-O-methyl-D-glucurono-D-xylan (MGX; mole ratio MeGlcA:Xyl = 1:5 [45]) was used instead of BeWX in this experiment.GH67 α-glucuronosidase (AGU) from Geobacillus stearothermophilus could partially hydrolyze the α-1,2-glycosidic bond between MeGlcA and terminal non-reducing xylosyl residues of xylooligosaccharides and xylan [46].MGX contains approximately 0.1 mg total uronic acid per mg polysaccharide.Only around 8% MeGlcA (of the total uronic acid) was released from MGX by GH67 AGU.The comparison of the amount of xylobiose released from AGU-treated and untreated MGX revealed that the activity of TeXbhA was most affected.For this GH30_7 enzyme, a decrease in the yield of xylobiose was observed with the AGU-treated MGX, while the activity of the other enzymes was only slightly or not affected (Fig. 5d).It is not understood how such a minor decrease in the MeGlcA substitution could decrease the yield of xylobiose for TeXbhA.Further studies are required to address the fungal GH30_7 functional requirements for the MeGlcA and the yield of xylobiose.

WAX degradation confirmed the BeWX preference of GH30_7 enzymes
To validate the activity of GH30_7 enzymes towards different types of xylan, they were also analyzed using wheat arabinoxylan (WAX) as a substrate.TtXbhA released around 10-fold less xylobiose from WAX than from BeWX, while the other enzymes showed only trace amounts or no xylobiose release from WAX (Fig. 5b and c).This agreed with most of the previously characterized enzymes from GH30_7, which also showed much lower activity against WAX than BeWX [3,4,40,39].This could be due to the high arabinosyl substitution of WAX, which might hinder the binding of these enzymes to WAX.
To confirm the effect of arabinosyl substitution towards GH30_7 activity, WAX was treated with GH51 α-arabinofuranosidase (ABF) from A. niger, which partially hydrolyses mono-arabinosyl substitutions at O-3 and di-arabinosyl substitutions at O-2 and O-3 from WAX [47].WAX contains about 0.38 mg arabinose per mg polysaccharide, of which around 30 % arabinose was released after incubation with GH51 ABF.The removal of arabinosyl substitution in WAX improved the release of xylobiose by all enzymes (Fig. 5e).TtXbhA released 3-fold more xylobiose from ABF-treated WAX, and the other GH30_7 enzymes could also release detectable amount of xylobiose.These results indicate that the high degree of arabinosyl substitution hinders the accessibility of the enzyme to the xylan backbone.

Differences in the catalytic region of GH30_7 affect their substrate specificities
To investigate the catalytic mechanism of GH30 fungal enzymes, homology modelling analysis of selected candidates was used in this study.Currently, there is no structure available with >35 % amino acid sequence similarity to be used as a reliable template for the homology modeling of GH30_3, GH30_11 and GH30_5 enzymes (Supplementary Table 3).Hence, only the homology models of the enzymes from GH30_7 were created (Supplementary Table 3 and Fig. S3).T. cellulolyticus TcXyn30B (Supplementary Fig. S3a, PDB ID: 6KRN) [25] was used as a template for homology models of GH30_7 enzymes (Supplementary Fig. S3b-e).The putative subsites -1 to -3 are predicted based on the structure of TcXyn30B as well as the E. chrysanthemi EcXyn30A (Supplementary Fig. S3f; PDB ID: 2Y24) [27].
A comparison of the putative catalytic amino acids of GH30_7 enzymes showed that the residues in subsite -1 are highly conserved, while those in subsites -2 and -3 are less conserved (Table 2).Instead of the -3 subsite of PsExlA, a short loop was observed with TtXbhA, TeXbhA and TtExlA (Supplementary Figs.S2 and S3b-e).This loop forms a steric barrier close to the catalytic site to accommodate two xylosyl residues, at subsites -1 and -2, explaining the xylobiose release [3,39].Within the loop, two amino acids were reported to possibly contribute to xylobiohydrolase activity, i.e.N93 in TcXyn30B (corresponding to D88 of TtXbhA, D78 of TeXbhA, and D78 of TtExlA) and W101 in AaXyn30A (corresponding to F89 of TtXbhA, F79 of TeXbhA, and H79 of TtExlA) (Supplementary Figs.S2 and S3) [3,25].X. Li et al.

Fig. 2 .
Fig. 2. Mono-and di-saccharides released from different substrates by crude GH30 enzymes.Hydrolysis was performed at 30 • C for 16 h.Saccharide release was detected by HPAEC-PAD.AG-II, arabinogalactan type II.(a)-(d) the type and amount of specific products released from different substrates by crude enzymes of each subfamily.The amount of glucobiose, galactobiose and xylobiose released was calculated based on cellobiose, β-(1→6)-galactobiose, and β-(1→4)-xylobiose standard curves, respectively.All assays were performed in duplicate.