Genetic Determinants of Xylan Utilization in Humisphaera borealis M1803T, a Planctomycete of the Class Phycisphaerae

Planctomycetes of the class Phycisphaerae are aerobic and anaerobic heterotrophic bacteria that colonize a wide range of marine and terrestrial habitats. Their functional roles in the environment, however, are still poorly understood. Humisphaera borealis M1803T is one of the very few characterized planctomycetes of this class. It is also the first described representative of the previously uncultured group WD2101, which is commonly detected in soils and peatlands. This work analyzed the genetic determinants that define the ability of Humisphaera borealis M1803T to grow on xylan, one of the plant cell wall polymers. The whole genome sequence analysis of this planctomycete resulted in identification of five genes encoding the proteins homologous to previously described endo-β-xylanases. For two of these proteins, evolutionarily closer experimentally characterized homologs with other substrate specificities were found. In a member of the GH10 family of glycoside hydrolases, the active center of the enzyme was destroyed. We consider two proteins from GH62 and GH141 families as the most likely candidates for the role of β-xylanase responsible for xylan utilization. Phylogenetic analysis of proteins of GH10, GH62, and GH141 families was carried out. The role of lateral transfers in the evolution of the genes for glycoside hydrolases and their close homologs is discussed.

The class Phycisphaerae includes aerobic and anaerobic heterotrophic planctomycetes that utilize a broad range of organic compounds and inhabit marine and terrestrial ecosystems (Fukunaga et al., 2009). Their key difference from the other class of planctomycetes, Planctomycetia, is the way of reproduction: members of Planctomycetia are budding bacteria, whereas Phycisphaerae propagate by binary fission (Ward and Dedysh, 2022). The number of cultured and characterized Planctomycetia members, as well as the body of data on their biology and organization of their cells, exceed considerably the amount of information available on Phycisphaerae. The latter class includes three orders: Phycisphaerales (Fukunaga et al., 2009), Sedimentisphaerales (Spring et al., 2018, and Tepidisphaerales (Kovaleva et al., 2015), with the total number of characterized members of less than a dozen.
Until recently, the order Tepidisphaerales included only one genus, Tepidisphaera, with a single described species, Tepidisphaera mucosa, represented by thermophilic polysaccharide-utilizing planctomycetes from terrestrial thermal springs (Kovaleva et al., 2015). One year ago, a mesophilic species isolated from an eutrophic lake of the boreal zone, Humisphaera borealis M1803 T , was added to the diversity of cultured members of the order Tepidisphaerales . This planctomycete is the first representative of the phylogenetic group WD2101 within the order Tepidisphaerales that has been obtained in culture. According to molecular studies of microbial diversity, planctomycetes of group WD2101 are typical inhabitants of peatlands and soils rich in organic matter (Ivanova et al., 2016;Dedysh and Ivanova, 2019;Dedysh et al., 2021). Genome analysis showed that the H. borealis M1803 T possesses a complete set of genes responsible for the synthesis of planctomycete-type microcompartments, specialized cell structures involved in utilization of monosaccharides of plant cell walls, e.g., rhamnose . Analysis of cellular ultrastructure by electron microscopy confirmed the presence of these compartments in cells grown on rhamnose, suggesting that these planctomycetes participate in degradation of plant debris in wetland ecosystems. It was found that a culture of H. borealis M1803 T can also grow on xylan, a polymeric constituent of the plant cell wall; however, its growth on this substrate has not been studied in any detail.
Planctomycetes have a great glycolytic potential, and their genomes encode a broad range of glycoside hydrolases (Ivanova et al., 2017;Kulichevskaya et al., 2020;Drula et al., 2022). Glycoside hydrolases, or glycosidases, (EC 3.2.1), are a vast group of enzymes that cleave O-glycoside bonds in various substrates, including heteropolysaccharides, glycoproteins, and glycolipids (Naumoff, 2011). Depending on substrate specificity, more than 200 variants of enzymatic activity are recognized: EC 3.2.1.1-3.2.1.217 (McDonald et al., 2009). In the CAZy database, most of the currently known glycoside hydrolases are combined into 165 families based on amino acid sequence homology, GH1-GH173 (except GH21, GH40, GH41, GH60, GH61, GH69, GH145, and GH155) (Lombard et al., 2014;Terrapon et al., 2017;Garron and Henrissat, 2019;Drula et al., 2022). As a rule, one family may include enzymes with different activities, while members of different families may exhibit the same substrate specificity. Using the results of phylogenetic analysis, families can be divided into largely monofunctional subfamilies, making it possible to predict fairly reliably the biological functions of proteins that have not been characterized experimentally. In the present work, this approach was applied to search for genes that encode putative β-xylanases in H. borealis M1803 T .
Sequencing  and subsequent annotation (Drula et al., 2022) of the genome of H. borealis M1803 T showed that it encodes five proteins representing five glycoside hydrolase families from the above list: GH5, GH10, GH18, GH62, and GH141. In this context, the goal of the present work was to verify experimentally the ability of H. borealis M1803 T to grow on xylan and to identify the most likely candidates for the role of β-xylanase among the proteins encoded in its genome.

Strain M1803 T and its growth on xylan.
The study was conducted with the planctomycete H. borealis M1803 T (=KCTC 25269 T ) isolated and described previously by our research group . The culture was maintained on the MPYVG medium (modification of DSMZ no. 621) that contained (g/L distilled water): peptone (Fluka), 0.1; yeast extract, 0.25; NH 4 NO 3 , 0.1; glucose, 0.5; Hutner's basal salts, 20 mL, and Hutner's vitamin solution, 1 mL (Staley et al., 1992). To assess the ability of strain M1803 T to grow on xylan, it was cultured at 22°С in 160-mL serum bottles with 30 mL liquid nutrient medium containing 20 mL/L of Hutner's basal salts, 0.1 g/L NH 4 NO 3 , 1 mL/L vitamin solution, and 0.05 g/L yeast extract. Xylan (Sigma-Aldrich) was added to the medium at the concentration of 0.1%. At the same time, the growth of an M1803 T culture was assessed in a similar medium with glucose (0.05%) and in a control medium containing no carbon source. Experiments were performed in three replicates. To determine the cell numbers, suspension aliquots were collected at the beginning of the experiment and after incubation for 10 and 20 days. Planctomycete cells were counted by phase contrast microscopy on Teflon-laminated slides coated with 0.1% gelatin solution. Bacterial abundance in the samples was determined by counting cells in 50 fields of vision for each experimental variant and calculating cell numbers per 1 mL of culture.
Search for putative β-xylanases and their phylogenetic analysis. Using the CAZy database (Drula et al., 2022), the genome of H. borealis M1803 T was searched for the presence of genes that encode proteins from glycoside hydrolase families known to include experimentally characterized enzymes with the β-xylanase activity. The closest homologs of the identified proteins were searched for using the blastp algorithm on the NCBI site (http://www.ncbi.nlm.nih.gov/) and subsequently analyzed using the PSI Protein Classifier program (Naumoff and Carreras, 2009). Lists of experimentally characterized enzymes for the analyzed families of glycoside hydrolases were compiled based on the information available in the CAZy database.
To perform phylogenetic analysis, several dozens of closest homologs were selected for each of the candidate proteins from H. borealis M1803 T . Multiple alignment of amino acid sequences was performed manually with BioEdit (http://www.mbio.ncsu.edu/ BioEdit/bioedit.html) taking into account the results of pairwise comparisons obtained with blastp.
The outgroup used to study phylogeny of the GH62 family included the closest to IPV69_21845 (H. borealis) experimentally characterized enzyme (GenPept, DAC80243.1) and its 26 nearest homologs. For the GH10 family, the outgroup included nine proteins with the closest phylogenetic relationship to the planctomycete protein cluster (according to Naumoff et al., 2014), as well as additional 12 closest homologs. No special outgroup was selected for the GH141 family; to visualize the tree, the most divergent group was compiled of the proteins included in the analysis.
The results of multiple sequence alignment (after removal of the most variable sequence fragments) were used to construct phylogenetic trees with the PROTPARS program (Protein Sequence Parsimony method, MP) of the PHYLIP software package (http://evolution.gs.washington.edu/phylip.html). Statistical significance of branching nodes was evaluated using bootstrap analysis (1000 pseudoreplicates for each tree). Graphic representations of the trees were obtained using TreeView Win32 (http://taxonomy.zoology.gla.ac.uk/rod/treeview.html).

Growth of Humisphaera borealis M1803 T on xylan.
Microscopy of H. borealis M1803 T cultures after 20 days of incubation in bottles with xylan revealed active development of planctomycetes on the surface of biopolymer particles (Fig. 1a). After 20 days of incubation with xylan, the number of M1803 T cells increased by more than an order of magnitude, whereas no significant change in the cell number was observed in the control culture (Fig. 1b). The specific growth rate for the strain M1803 T on xylan was 0.006 h -1 . As expected, the growth on glucose was more rapid.
According to the CAZy database (Drula et al., 2022), among 486 experimentally characterized enzymes of GH18 family of glycoside hydrolases, there is only one β-xylanase, which belongs to the fungus Purpureocillium lilacinum (GenPept, XP_018173934.1). Its comparison to protein IPV69_11765 (QOV91983.1) showed that their amino acid sequences had only 26% identity, which means that they belong to different subfamilies (according to the 30% criterion; Naumoff, 2005). This suggests that IPV69_11765 glycosidase does not possess β-xylanase activity. Glycoside hydrolase family GH62. Together with GH43 and GH117 families, glycoside hydrolase family GH62 forms the GH-F clan (Drula et al., 2022). On a higher hierarchical level, it belongs to the furanosidase superfamily (Naumoff, 2001(Naumoff, , 2011(Naumoff, , 2012Pons et al., 2004), characteristically featuring catalytic domains with the five-bladed β-propeller structure and three highly conservative acidic amino acid residues (Asp/Glu) that form the active center. It should be noted that GH43 glycosidases are predominantly characterized by two types of enzymatic activity: α-Larabinofuranosidase (EC 3.2.1.55) and β-xylosidase (EC 3.2.1.37). Both types of activity are frequently exhibited by members of the same subfamily, and sometimes even by one and the same protein (Mewis et al., 2016). These facts suggest that β-propeller catalytic domains can recognize both α-L-arabinofuranoside and β-D-xylopyranoside monosaccharide residues of their substrates. A vast majority of experimentally characterized members of GH62 family possess only α-L-arabinofuranosidase activity (Drula et al., 2022). However, two multidomain proteins exhibit several types of activity. In particular, a metagenomic glycosidase (GenPept, DAC80243.1) contains three catalytic domains that belong to the families CE1 (carbohydrate esterase with feruloyl-esterase activity, EC 3.1.1.73), GH62 (α-L-arabinofuranosidase), and GH10 (β-xylanase). Furthermore, subcloning and subsequent expression of its isolated GH62 domain showed that it also has a certain level of β-xylanase activity (Holck et al., 2019). Glycosidase from Ruminiclostridium josui (WP_024831741.1) contains two catalytic domains that belong to the families of GH62 glycoside hydrolases and CE6 carbohydrate esterases. Their subcloning showed that one of them mediated α-L-arabinofuranosidase and endo-β-xylanase activities, and the other one acted as an acetylxylan esterase (EC 3.1.1.72) (Wang et al., 2018). That is, GH62 domains of at least two known proteins possess the β-xylanase activity. Protein IPV69_21845 (QOV88842.1) from H. borealis contains only one domain, which represents GH62 family of glycoside hydrolases. A comparison of this protein to all known experimentally characterized members of this family showed that it had 45% sequence identity to the metagenomic glycosidase DAC80243.1 and only 30-39% identity to the other 24 proteins. Based on these data, IPV69_21845 glycosidase can be expected to possess not only arabinofuranosidase but also endo-βxylanase activity, which makes it a likely candidate for the role of β-xylanase in H. borealis.
We performed a phylogenetic analysis of the amino acid sequences of the GH62 domains of putative glycosidase IPV69_21845 and metagenomic glycosidase DAC80243.1, as well as of their closest homologs (altogether, 103 proteins) (Fig. 2). All proteins included in the analysis were found to feature the triad of catalytically important amino acid residues (in case of IPV69_21845, Asp-64, Asp-179, and Glu-227), which indicates that they possess a glycosidase activity. The closest homologs of the metagenomic glycosidase are a vast group of proteins encoded in the genomes of various planctomycetes. Moreover, two proteins have a very high level of sequence identity to its GH62 domain fragment (positions 406-714): NLK41752.1 (100%) and HDS84024.1 (98%). This invites a hypothesis that the polyfunctional metagenomic glycosidase DAC80243.1 also belongs to some planctomycete of the class Phycisphaerae.
Putative glycosidase IPV69_21845 from H. borealis and its closest homolog from an uncharacterized planctomycete SH-PL14 (AMV21057.1) are singledomain proteins, and their amino acid sequences have 61% identity. They belong to a large cluster that mainly includes proteins from planctomycetes and a significantly lower number of proteins from Armatimonadetes and Verrucomicrobia (Fig. 2); other bacterial phyla are represented by very few members (Acidobacteria, Chloroflexi, Proteobacteria, and other) or are not represented at all (e.g., Actinobacteria, Bacteroidetes, and Firmicutes). The results of the phylogenetic analysis (Fig. 2) suggest that the last common ancestor of the genes encoding all GH62 proteins studied belonged to some planctomycete. Subsequent evolution of this gene involved multiple lateral transfers into genomes of other bacterial phyla and even reverse transfers from Verrucomicrobia back into planctomycetes (one or two cases in the cluster marked in blue; Fig. 2).
Glycoside hydrolase family GH10. Almost all enzymatically characterized members of GH10 family possess endo-β-xylanase activity (EC 3.2.1.8 and EC 3.2.1.32). In our previous work, a comparative analysis of planctomycete proteins of this family showed that they could be divided in two principal groups Naumoff, 2014Naumoff, , 2016. Proteins of the first group were present only in some planctomycetes, usually as several paralogs, and in the phylogenetic tree they did not form a separate cluster among proteins from other prokaryotes, which is a typical pattern for many glycoside hydrolase families (if members of one given high-rank bacterial taxon are considered). Proteins of the other group were represented in all heterotrophic planctomycetes, always as only one paralog, and they formed a single cluster in the phylogenetic tree where their positions were consistent with the taxonomy of the respective bacterial species (i.e., the gene inheritance was strictly vertical without lateral transfers, deletions, or duplications). Multiple alignment of amino acid sequences showed that all proteins of this group had a disrupted active center that lacked the two key Glu residues acting as a proton donor and a nucleophile (see Fig. S1 in Rakitin et al., 2021); therefore, it is impossible for them to exhibit any enzymatic activity. For this reason, we indicated the lack of enzymatic activity for the corresponding proteins when annotating the genomes of the planctomycetes Paludisphaera borealis (GenPept, APW63099.1), Limnoglobus roseus (QEL20001.1), Frigoriglobus tundricola (QJX00392.1), and Telmatocola sphagniphila (QVL33104.1). Disrupted active centers were also detected in a number of other prokaryotic GH10 proteins (unpublished data), but they were not in a close evolutionary relationship to the proteins of the planctomycete cluster. This fact suggests that the loss of enzymatic function occurred repeatedly in the course of evolution of this protein family.
In the present work, a phylogenetic analysis of the amino acid sequences of GH10 domains (altogether, 181 proteins; Fig. 3) provided more detailed information on the structure of the planctomycete cluster, which corresponds to the second group of GH10 proteins. It is composed of two subclusters corresponding to the classes Phycisphaerae (49.1% bootstrap support) and Planctomycetia (99.9%). Furthermore, the subcluster Planctomycetia contains very few proteins from other bacterial taxa (see caption to Fig. 3) and is in turn composed of four clusters corresponding to the orders Gemmatales (82.5%), Isosphaerales (99.9%),  Glycoside hydrolase family GH141. According to the CAZy database (Drula et al., 2022), only two proteins of GH141 family have been characterized enzymatically: α-L-fucosidase (EC 3.2.1.51) from Bacteroides thetaiotaomicron (GenPept, AAO76109.1) and β-xylanase (EC 3.2.1.8) from Hungateiclostridium thermocellum (ABN53397.1). These proteins belong to different subfamilies but both feature two Asp residues in homologous positions; these residues acting as proton donors and nucleophiles are highly conserved across GH141 family (Heinze et al., 2017;Ndeh et al., 2017). Protein IPV69_20890 (QOV88669.1) from H. borealis does not belong to either of these subfamilies and has less than 27% amino acid sequence identity with the above proteins, but it possesses both conserved Asp residues (positions 489 and 525). This makes it another possible candidate for the role of β-xylanase in this species.
We performed a phylogenetic analysis of the amino acid sequences of the putative glycosidase IPV69_20890 and its 109 closest homologs (Fig. 4). Both enzymatically characterized GH141 proteins were not included in the analysis, because their considerable divergence made it impossible to obtain an unambiguous multiple sequence alignment. Protein IPV69_20890 was placed in a stable cluster composed mainly of proteins belonging to bacteria of the phylum Armatimonadetes (100% bootstrap support; top right in Fig. 4). The fact that proteins representing other phyla (Lentisphaerae and Planctomycetes) did not exhibit a trend to co-clusterization within this cluster suggests that the corresponding genes might have appeared due to a series of independent lateral transfers from Armatimonadetes. The remaining GH141 members included in the analysis formed eight more clusters in the tree. One of them was formed exclusively from planctomycete proteins, mainly representing the genus Paludisphaera (100% bootstrap support; top left in Fig. 4). All other clusters (I-VII) also included a certain number of planctomycete proteins. Cluster IV included a stable subcluster (100% bootstrap support) of seven planctomycete proteins (MBC8871159.1, MBM4047734.1, NLF73417.1, NLY00979.1, NOX55446.1, NOZ20755.1, and NUQ63536.1); however, all these proteins are encoded by metagenomes, and for this reason it was difficult to determine more accurately the taxonomic affiliation of their species of origin.
Five of these seven clusters (II-V, and VII) included several proteins of the phylum Lentisphaerae; Armatimonadetes were represented in four clusters (II, III, VI, and VII), and so were Verrucomicrobia (II, IV, V, and VII); Kiritimatiellaeota were found in three clusters (II, VI, and VII). Three bacterial phyla were represented in only one cluster each: Nitrospirae (I), Candidatus Hydrogenedentes (I), and Candidatus Nealsonbacteria (V). These facts indicate that the last common ancestor of the genes encoding all GH141 proteins studied in this work must have belonged to some planctomycete. Subsequent evolution of this gene involved multiple lateral transfers into bacterial genomes of several other phyla and even a reverse transfer from Armatimonadetes back into planctomycetes.
The Paludisphaera cluster included a protein from Paludisphaera borealis PX4 T (APW58834.1), which was initially annotated as "putative beta-solenoid-type carbohydrate-active enzyme (GH, PL, or CE) of unknown function" based on remote evolutionary relationships (family PF13229 in the Pfam database; Mistry et al., 2021), since the GH141 family of glycoside hydrolases was not recognized at that time (Ivanova et al., 2017). All proteins of this family analyzed Fig. 3. Phylogenetic tree of the GH10 family of glycoside hydrolases constructed using the maximum parsimony method. Statistical significance of the tree was assessed by bootstrap analysis; the number of supporting pseudoreplicates (out of 1000) is indicated next to each branching site. Letters indicate the taxonomic affiliation of the source organisms: C, Chloroflexi; D, Gammaproteobacteria; G, Gemmatimonadetes; I, Acidobacteria; J, Candidatus Dependentiae; M, bacterial metagenome; T, Actinobacteria; and Y, Cyanobacteria. All unmarked proteins belong to planctomycetes of the class Phycisphaerae. Protein IPV69_18965 (QOV88313.1) from Humisphaera borealis M1803 T is shown in red. Strain names are indicated for protein IPV69_18965 and four its closest homologs. Yellow triangles indicate the clusters corresponding to the orders Gemmatales, Isosphaerales, Pirellulales, and Planctomycetales. The bootstrap support of each cluster is given inside the triangle, and the number of proteins it includes is indicated to the right. The cluster Pirellulales (marked with an asterisk) additionally includes a protein from an aquatic metagenome annotated as Candidatus Nealsonbacteria (NQU20226.1). The outgroup (not shown) includes 21 proteins representing five prokaryotic phyla: Crenarchaeota (RLF19374.1), Cyanobacteria (ABW28526. in our work (with an only exception of protein HFH06191.1 from Candidatus Hydrogenedentes; cluster I) possess both highly conserved Asp residues, which suggests that all of them probably exhibit some enzymatic activity. Evolution of glycoside hydrolases of the families studied. Glycoside hydrolases are among those proteins that perform functions of facultative significance and therefore are not obligate components of the proteome (none of the glycoside hydrolase families is present in most biological species, and some microorganisms do not possess them at all). Vertical inheritance is not typical for prokaryotic genes of glycoside hydrolases: they are easily acquired by lateral transfer and subsequent duplications and also rapidly lost as soon as their presence provides no selective advantages (Naumoff, 2011).
In this aspect, the group of planctomycete proteins within family GH10 analyzed in our work represents a highly unusual case (Fig. 3), which requires an explanation. Apparently, these proteins lack enzymatic activity due to the loss of the active center. However, this group of orthologs must have acquired a currently unidentified novel function that is obviously obligate for bacteria of this phylum, which underlies the consistent vertical inheritance of their genes.
Two other groups of proteins studied, those representing GH62 and GH141 families, exhibited the pat-tern of evolution typical for glycoside hydrolases and characterized by numerous events of lateral gene transfer among different bacterial phyla. Apparently, for both families, the transfer events originated in planctomycetes and were directed exclusively to other gram-negative bacteria: predominantly members of the phyla Armatimonadetes and Verrucomicrobia in case of GH62 family (Fig. 2); and members of Armatimonadetes, Lentisphaerae, and Verrucomicrobia in case of GH141 family (Fig. 4).
It should be mentioned that the two other families studied, GH5 and GH18, also followed the general pattern; however, in this case, planctomycetes more frequently acted as acceptors of laterally transferred genes. The closest homologs of protein IPV69_13055 from GH5 family belonged to bacteria of the phylum Chloroflexi, and the closest homologs of protein IPV69_11765 (GH18) predominantly belonged to Verrucomicrobia (data not shown).
On the whole, the data obtained in this work are consistent with our previous observations suggesting that lateral transfer of glycoside hydrolase genes among various groups of prokaryotes occurred in nonrandom directions . In particular, gene exchange between gram-negative and gram-positive bacteria is a relatively rare event, and Proteobacteria are separated from other members of the gram-negative group. The highest intensity of transfers was observed, on the one hand, within a group of phyla including Armatimonadetes, Planctomycetes, and Verrucomicrobia, and on the other hand, among members of the phyla Acidobacteria, Bacteroidetes, and Balneolaeota. The results of our work indicate that proteins IPV69_21845 (family GH62) and IPV69_20890 (GH141) are the most probable candidates for the role of β-xylanase in Humisphaera borealis M1803 T . Protein IPV69_21845 can be expected to exhibit both α-L-arabinofuranosidase and β-xylanase activity, whereas IPV69_20890 probably possesses either α-Lfucosidase or β-xylanase activity. Further investigation of their characteristics is of substantial interest for biochemical research and can provide deeper insights into the functional role that planctomycetes of the class Phycisphaerae play in degradation of plant debris in natural ecosystems.

COMPLIANCE WITH ETHICAL STANDARDS
Conflict of interests. The authors declare that they have no conflict of interest.
Statement on the welfare of animals. This article does not contain any studies involving animals or human participants performed by any of the authors.

AUTHORS' CONTRIBUTIONS
The idea of the study was proposed by S.N.D. Experiments were designed by S.N.D. and D.G.N. Planctomycete cultures and cell count were performed by I.S.K. Identification of putative β-xylanase genes and their phylogenetic analysis were conducted by D.G.N. The manuscript was prepared by D.G.N. and S.N.D. All authors participated in the discussion of the results.

OPEN ACCESS
This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit ttp://creativecommons. org/licenses/by/4.0/.