AcXyn30B (Clocl_2746, AEV69300.1, G8M2Z1) consists of 673 amino acids. The signal peptide is 28 amino acids long (SignalP) (Teufel et al. 2022) and the GH30 catalytic module (amino acids 29-446) is followed by a CBM6 module (amino acids 469-593), and a dockerin module (amino acids 602-673). A comparison of the full-length amino acid sequence to a protein database using the BlastP revealed that among the first 100 hits, there was only one characterized enzyme CpXyn30A from Ruminiclostridium papyrosolvens (WP_004618990.1, also containing CBM6 domain) (St John et al. 2014) which had 31.51% identity with AcXyn30B. The most similar proteins were uncharacterized GH30 proteins from the genera Clostridium, Anaerobacterium, and Bacillota. A pairwise sequence comparison of AcXyn30B catalytic domain with catalytic domains of several members of each GH30 subfamily showed that AcXyn30B is most similar to GH30_8 subfamily; however, the identity and similarity are quite low, about 24–27% and 38–44%, respectively. In the CAZy database, the enzyme is classified in the GH30 family but is not assigned to any subfamily. To reveal a relationship of AcXyn30B with other GH30 members, a phylogenetic tree was constructed (Fig. 1). Amino acid sequences of catalytic domains of characterized GH30 members from all known subfamilies and 14 amino acid sequences of proteins most similar to AcXyn30B according to BlastP search were initially aligned in Clustal Omega and the alignment was then analyzed in MEGAX. The phylogenetic tree (Fig. 1) shows that AcXyn30B and related proteins form a separate clade within group 2 (explained below) but clearly distinct from other subfamilies. Therefore, we propose AcXyn30B to be a founding member of a new GH30 subfamily, GH30_12.

Phylogenetic relationship of GH30 characterized enzymes and the enzymes having similarity to AcXyn30B. The alignment performed using Clustal Omega was analyzed in MEGAX. The evolutionary history was inferred by using the Maximum Likelihood method and JTT matrix-based model. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (500 replicates) are shown next to the branches

The overall structure of GH30 enzymes is formed by (β/α)8 barrel which is linked to a side β-structure that is composed of 9 β-strands (Puchart et al. 2021). Based on the arrangement of the β9-domain relative to the (β/α)8 barrel, the GH30 family is divided into two groups: group 1 (subfamilies 1, 2, 3, 9) and group 2 (subfamilies 4, 5, 6, 7, 8, 10, 11) (St John et al. 2010). In group 1 the first three β-strands of the β9-domain precede the (β/α)8 barrel (Supplemental Fig. S1, highlighted in yellow) while in group 2 just one β-strand of the β9-domain is located in front of the (β/α)8 barrel. From the amino acid sequence alignment (Supplemental Fig. S1), it is obvious that in AcXyn30B, there is only one β-strand of the β9-domain preceding the (β/α)8 barrel, thus placing AcXyn30B into the group 2. Based on the sequence alignment Glu171 (an acid/base) and Glu279 (a nucleophile) are predicted to be catalytic residues (Supplemental Fig S1, highlighted in magenta). AcXyn30B and related enzymes do not contain a prokaryotic Arg which is responsible for glucuronoxylan specificity of the GH30_8 members (Supplemental Fig S1, highlighted in green). A different arginine residue, which often plays a similar role in eukaryotic GH30_7 glucuronoxylanases, is also absent in the sequence of AcXyn30B where Trp52 is found (Supplemental Fig S1, highlighted in gray). In this aspect, AcXyn30B resembles a non-specific GH30_7 xylanase TcXyn30C having Phe47 in the corresponding position (Nakamichi et al. 2020).

There are several differences in primary structure between GH30_7 and GH30_8 sequences (Puchart et al. 2021). The most obvious is the presence of much longer β2-α2 loop in the eukaryotic GH30_7 enzymes. The corresponding region of AcXyn30B is shorter, more similar to GH30_8 representatives (Supplemental Fig S1, highlighted in cyan). Moreover, the GH30_7 members contain additional short β-strands β8A and β8B in the β8-α8 segment. This region of AcXyn30B more resembles GH30_7 enzymes, but actually it is even longer (Supplemental Fig S1, highlighted in blue). It seems that in contrast to GH30_7 members, AcXyn30B does not lack the α6 helix (which is present in GH30_8 members), but the α7 helix is shorter and most similar in length to the non-specific GH30_8 enzymes CpXyn30A and CaXyn30A (Supplemental Fig. S1, highlighted in orange). Based on the primary structure comparison we can conclude that AcXyn30B and related enzymes have a special combination of structural features found in the enzymes from both GH30_7 and GH30_8 subfamilies.

The activity of AcXyn30B was tested on different polysaccharides. The enzyme was not active on cellulose, hydroxyethyl cellulose, starch, laminarin (β-1,3-glucan with β-1,6 branches), and pustulan (β-1,6-glucan) but depolymerized glucuronoxylan (GX), rhodymenan (β-1,3-β-1,4-xylan, Rho), and arabinoxylan (AraX). The specific activity of AcXyn30B on GX, Rho, and AraX was quite low and very similar, 3.5, 2.9, and 2.7 U/mg, respectively. AcXyn30B exhibited slightly higher specific activities of 4.8 and 9.3 U/mg on chromogenic substrates NP-Xyl2 and NP-Xyl3, respectively. Similar activity on the polysaccharides and aryl glycosides of linear xylooligosaccharides indicates a wide substrate specificity, which is in contrast with narrow specificities of GH30 glucuronoxylanases and xylobiohydrolases and suggests that AcXyn30B is a non-specific xylanase. The differences in catalytic properties also support the phylogenetic classification of AcXyn30B into the separate clade. Kinetic parameters (Table 1) showed that GX is a little bit better substrate than AraX due to lower Km value, and that the enzyme prefers longer oligosaccharides because the catalytic efficiency on NP-Xyl3 is higher than on NP-Xyl2.

The activity of AcXyn30B was also qualitatively examined on linear xylooligosaccharides (XOs) Xyl2 – Xyl6 (Fig. 2). Xyl2 was not attacked by the enzyme while Xyl3 was slowly converted to Xyl and Xyl2. Xyl4 was hydrolyzed to approximately equal amount of Xyl, Xyl2, and Xyl3. The enzyme thus has no significant preference for binding the substrate in subsites -2 and +2 (generating Xyl2) over the subsites either −3 to +1 or −1 to +3 (yielding Xyl3 and Xyl). Xyl2 and Xyl3 were the major products formed from Xyl5, and Xyl6 was cleaved to Xyl2, Xyl3, and Xyl4, so in both cases, at least two subsites are occupied on both sides from the catalytic amino acids (from −2 to +2). Longer XOs seem to be hydrolyzed faster than shorter ones and at least four catalytic subsites need to be occupied for effective cleavage. In the 5-h hydrolysates, tiny amounts of XOs longer than a substrate were observed as a result of transglycosylation reaction.

TLC analysis of the products formed from linear XOs (Xyl2 – Xyl6) by AcXyn30B after 10 min, 1 h, 5 h, and 24 h. St, standards of linear XOs

To reveal the mode of action of AcXyn30B on GX, AraX, and rhodymenan, the hydrolysates were analyzed by TLC (Fig. 3) and MALDI-TOF MS (Fig. 4). The detectable products of different lengths were produced from GX already after 10 min of hydrolysis. After 24 h they were shortened to linear oligosaccharides Xyl – Xyl6 and acidic XOs MeGlcAXyl2 – MeGlcAXyl4. To determine the structure of the released acidic XOs, either GH3 β-xylosidase or GH67 α-glucuronidase was applied to the 24-h hydrolysate (Figs. 3 and 4). The GH3 β-xylosidase is able to release non-substituted xylopyranosyl residue from the non-reducing end of XOs (Biely et al. 2016). α-Glucuronidases from GH67 family are known to release the (4-O-methyl-)glucuronic acid (GlcA/MeGlcA) only from the non-reducing end xylopyranosyl residue (Biely et al. 2016). The application of β-xylosidase on the GX hydrolysate did not affect the amount of MeGlcAXyl2, while the application of α-glucuronidase resulted in its disappearance accompanied with an increase in xylobiose amount (Figs. 3 and 4). This means that MeGlcA is attached to the non-reducing end xylose moiety of this acidic XO and its structure is MeGlcA2Xyl2. However, in the case of MeGlcAXyl3 and MeGlcAXyl4, the results were not so straightforward. Most of MeGlcAXyl3 was cleaved by β-xylosidase to MeGlcAXyl2 but a smaller part remained in the hydrolysate indicating that the predominant form of MeGlcAXyl3 is MeGlcA2Xyl3, but MeGlcA3Xyl3 is also present. In contrast, most of MeGlcAXyl4 was hydrolyzed by α-glucuronidase to MeGlcA and Xyl4, meaning that MeGlcA4Xyl4 is the main MeGlcA-substituted Xyl4, but there are also other isomers (e.g., MeGlcA3Xyl4) present in the hydrolysate. To further inspect the hydrolysis of acidic XOs, AcXyn30B was applied on MeGlcA3Xyl3 and MeGlcA3Xyl4. After 24 h, MeGlcA3Xyl3 was not attacked, while MeGlcA3Xyl4 was hydrolyzed to MeGlcA2Xyl3 and Xyl. Prolonged incubation (3 days) led to a very slow cleavage of MeGlcA3Xyl3 to MeGlcA2Xyl2 and Xyl, while MeGlcA2Xyl3, a degradation product of MeGlcA3Xyl4, was slowly further converted to MeGlcA2Xyl2 (Supplemental Fig. S2).

a TLC analysis of hydrolysis products released by AcXyn30B from beechwood glucuronoxylan (GX), wheat arabinoxylan (AraX) and rhodymenan (Rho) after 10 min, 1 h, 5 h, and 24 h, and after subsequent addition of either β-xylosidase (x) or α-glucuronidase (g). St, standards of linear XOs. b Action of β-xylosidase and α-glucuronidase on different acidic XOs which are produced from GX by AcXyn30B

MALDI-TOF MS analysis of hydrolysis products released by AcXyn30B from beechwood glucuronoxylan after 24 h, and after subsequent addition of either α-glucuronidase or β-xylosidase

AraX was hydrolyzed by AcXyn30B to a mixture of linear and Ara-substituted XOs which were difficult to identify (Fig. 3a). However, the mode of action of AcXyn30B on arabinosylated substrates can be assumed from the hydrolysis of short Ara-XOs of defined structure. After 24 h, the enzyme did not attack A3X, A2XX, and A2+3XX, but it released Xyl from XA3XX, XA2XX, and XA2+3XX (Fig. 5, Supplemental Fig. S3). When the mixtures were incubated for a prolonged time (3 days), a very low amount of Xyl was also liberated from A2+3XX (Supplemental Fig. S3). In all cases, Xyl was released from the reducing end of the substrates. This conclusion is based on experiments using the GH3 β-xylosidase or various α-arabinofuranosidases and is depicted and explained in Supplemental Fig. S4. The release of Xyl from the reducing end of XA3XX, XA2XX, and XA2+3XX means that these substrates are bound in the catalytic subsites −3, −2, −1, and +1 (cleavage occurring between the subsites −1 and +1). Singly or doubly Ara-substituted Xylp residue is then accommodated in the −2 subsite and unsubstituted Xylp residues are located in the subsites −3, −1, and +1. If we assume that one Xylp shorter substrates (A2XX vs XA2XX and A2+3XX vs XA2+3XX) are accommodated in a similar way, they occupy the subsites −2, −1 and +1, but they are not cleaved or are cleaved very slowly. This indicates that the occupation of the −3 subsite by Xylp residue promotes the hydrolysis of substituted XOs. In other words, substituted substrates are effectively cleaved only when at least four catalytic subsites of AcXyn30B are occupied (which is in consonance with hydrolysis of neutral and acidic XOs), Xylp residue in the −1 subsite is non-substituted, the −3 subsite is occupied and the substituted Xylp residue (at position 2 and/or 3) is located in the −2 subsite.

Various branched XOs tested as the substrates for AcXyn30B. The cleavage site is indicated by an arrow

Hydrolysis of Rho by AcXyn30B yielded a mixture of β-1,4-linked and β-1,3-1,4-linked XOs (Fig. 3). Similarly to AaXyn30A, the enzyme was able to release small amount of isomeric xylotriose having the structure of β-d-Xylp-1,3-β-d-Xylp-1,4-β-d-Xyl (X3X4X) as the shortest mixed linkage oligosaccharide. However, the amount of longer XOs (degrees of polymerization (DP) 4-6) was higher, and β-1,3-1,4-linked XOs were prevailing.