Strains were isolated from the gut of I. typographus larvae collected within a study of its intestinal microbial diversity (Veselská et al., 2023). In brief, larvae were collected during September and October 2020 in the surroundings of forest area in Nižbor (Czechia, 49°59′09.9″N 13°56′47.5″E, 390 m.a.s.l.). Larvae were taken out of the galleries and surface sterilized by subsequent washing with 70% ethanol, 2% Tween 80 (Avantor, USA) and sterile distilled water. The larvae were dissected straight after collection, their guts were homogenized and plated in various dilutions (diluted with saline) on Petri dishes with YES (5 g of yeast extract, 30 g of glucose, 20 g agar, 1 l of distilled water) and antibiotics (streptomycine and chloramphenicol, 60 mg/l). Cultivation took place for 5–7 days at 24°C in dark. DNA was extracted by Nucleo Spin kit (Machery Nagel), the ITS sequence was used for determination based on the BlastN similarity search in NCBI Genbank database. Cultures were deposited in Culture Collection of Fungi (CCF, Department of Botany, Faculty of Sciences, Charles University, Prague) as Cryptococcus sp. CCF 6641, Kuraishia molischiana CCF 6642, Nakazawaea ambrosiae CCF 6643, Ogataea ramenticola CCF 6644, and Wickerhamomyces bisporus CCF 6645.

Genomic DNA (gDNA) of yeasts was obtained with Nucleo Spin kit (Machery Nagel) kit. DNA yield was quantified on Qubit 2.0 Fluorometer (Thermo Fisher Scientific, Waltham, MA, USA) using Qubit dsDNA BR Assay Kit and DNA quality was checked on Nanodrop (NanoDrop™ 2000 c, ThermoFisher scientific). gDNA samples were sent to Macrogen NGS sequencing service center (Amsterdam, Netherlands) for library preparation and sequencing. A TruSeq DNA PCR free (350) library was prepared and genomes were sequenced on Illumina NextSeq 500 system with 2 × 151 nt PE, resulting in a total of 38.96 Gb and roughly 6.49 Gb read bases per sample. The quality of raw data was analyzed with FASTQC v 0.11.9 (Andrews, 2010). Sequencing adapters and low-quality reads were trimmed with Trimmomatic v 0.39 (Bolger et al., 2014), k-mer = 5 was used to remove leading and trailing bases with Phred scores lower than 33, as well as reads shorter than 150 bp. De novo assembly was performed using SPAdes v 3.13.1 (Bankevich et al., 2012) in Python v 3.8.10 environment with –careful setting, k-mer = 121 was selected. The completeness of assemblies was evaluated with Benchmarking Universal Single-Copy Orthologs [BUSCO v5.2.1 (Seppey et al., 2019)] to search against the saccharomycetes_odb10 database for Kuraishia molischiana, Nakazawaea ambrosiae, Ogataea ramenticola, and Wickerhamomyces bisporus. The assembly of Cryptococcus sp. was searched against the tremellomycetes_odb10 database. The assessment of those assemblies was performed with Quast v 5.0.2 (Gurevich et al., 2013).

Gene prediction was performed using AUGUSTUS v 3.1.0 (Stanke et al., 2006) with Saccharomyces cerevisiae S288C as reference species. Gene Ontology (GO) annotation was carried out in eggnog-mapper v 2 (Cantalapiedra et al., 2021) with an e-value of 10^–3. The GO terms were distributed by R package GO.db (Carlson et al., 2019) and plotted by ggplot (Wickham, 2006). Kyoto Encyclopedia of Genes and Genomes (KEGG) annotation was carried out with BlastKOALA (Kanehisa et al., 2016). KEGG pathways were performed with KEGG mapper (Kanehisa and Sato, 2020).

Carbohydrate active enzymes (CAZymes) were identified in the predicted proteins of each genome with hmmsearch in HMMER 3.3.2 (Mistry et al., 2013) against the CAZy database dbCAN2 (Zhang et al., 2018)¹ version 8. The e-value was set as 1e–5.

The predicted enzymes associated with the detoxification process were identified with HMMER E-value 1e–5 (Mistry et al., 2013) against specific domain from Pfam database (Mistry et al., 2021): cytochrome P450 monooxygenase (PF00067), aldo-keto reductases (PF00248), carboxylesterase (PF00135), flavin-containing monooxygenase (PF00743), glutathione S-transferase (PF13409, PF14497, PF02798, PF13417, PF00043, and PF13410), multicopper-containing oxidases (PF07732, PF00394, and PF07731), ATP-binding cassette (ABC) transporter (PF00664, PF00005, PF06472, and PF00004), multidrug and toxic compound extrusion (MATE) (PF01554) and major Facilitator Superfamily (MFS) (PF07690, PF16983, PF05631, PF00083, PF12832, PF06779, PF13347, and PF05977). The results were verified and filtered with BlastP (e-value 1e-5) using SEG, a word-size six letters and BLOSUM62 matrix.

Multiple sequence alignment was carried out with MAFFT v 7.453 (Rozewicki et al., 2019) using maxiterate 1000. The phylogenetic trees were performed with PhyML v 3.0 (Guindon et al., 2010). The best evolutionary model for protein sequences was calculated with Akaike method and likelihood-based method was selected as aLRT SH-like and bootstrap analysis repeated for 100 times. According to the BlastP results and phylogenetic analysis, enzymes of aldo-keto reductases, cytochrome P450, and ABC transporter family were classified into specific groups. Phylogenetic trees were visualized and edited with iTOL (Letunic and Bork, 2021).

The subcellular localization of detoxification related enzymes was predicted with Protcomp-AN v 6.0 (Klee and Ellis, 2005), molecular mass isoelectric point (pI) was performed with Protein isoelectric point calculator (Kozlowski, 2016).

As our functional analysis predicts presence of pathways for the degradation of plant cell wall components, in the next step we verified them by in vitro degradation tests. The analyzed enzymes are listed in Table 1. Activity of endolytic enzymes was assessed using various chromogenic substrates (Megazyme, Bray, Ireland), whereas activity of exolytic enzymes was assessed using fluorogenic substrates (Sigma-Aldrich). For endolytic enzyme analyses, yeasts were cultivated in submerged media in 250 ml Erlenmeyer flasks on rotary shaker (200 rpm) at 24°C. Basic medium (BM) consisted of 1 g pepton, 1 g yeast extract, 1 g (NH₄)₂HPO₄, 0.5 g K₂PO₄, 0.5 g MgSO₄.7H₂O and substrates specific to the enzyme tested [3 g of cellulose, 6 g xylan birchwood, 15 g galactomannan (Locust beam gum), 6 g of starch, 1.5 g of amylopectin and 13.5 g of dextran] in 1 l of McIlvaine buffer, pH 4.5. The cultivation time was determined using the glass tube test. Glass tubes containing water agar (20 g agar in 1 L distilled water) were overlaid with a thin layer of a specific chromogenic substrate dissolved in water agar and the agar surface was inoculated with the tested fungi and cultured at 24°C. Enzymatic degradation was manifested by diffusion of the chromogenic substrate into the water agar. At this point, the submerged cultivation in the Erlenmeyer flask was stopped. If no staining of the water agar was evident after 14 days of cultivation in the glass tubes, submerged cultivation was also stopped and enzyme activities were evaluated. Enzymes were extracted into extraction buffers as it is recommended by manufacturer for 2 h at 4°C. After extraction, the mixture was centrifuged for 10 min at 3,000 rpm to exclude the medium and fungal cells. The supernatant containing extracellular enzymes was further filtered through filter paper to remove remaining solid parts. We used manufacturer’s protocols for measurement and calculation of enzymatic activities. Trichoderma viride CCF 4516 was used as a positive control for enzymatic degradation of cellulose, xylan and casein as it is known to degrade these substrates (Gomes et al., 1992; Kredics et al., 2005). Sterile media without inoculated fungi served as blanks.

To measure exolytic activity, yeast strains were also cultured in submerged medium in 250 ml Erlenmeyer flasks on a rotary shaker (200 rpm) at 24°C. The medium consisted of yeast extract (5 g/l), peptone (2 g/l), glucose (2 g/l), casein (2 g/l), cellulose (3 g/l), lignin (3 g/l), starch (3 g/l), lime phloem (63 g/l). The yeast grew for 14 days. Sodium acetate, pH 4.5, was used for enzyme extraction. The fluorescence of the released reaction products was measured as previously described (Baldrian, 2009) using a method adapted from Vepsäläainen et al. (2001). The fluorescence value of each MUF fluorogenic substrate was corrected by subtracting the background fluorescence of the growth medium. Laccase activity (EC 1.10.3.2) was measured by monitoring the oxidation of ABTS [2,2-azinobis-3-ethylbenzothiazoline-6-sulfonic acid (Bourbonnais and Paice, 1990) in citrate-phosphate buffer (100 mM citrate, 200 mM phosphate, pH 5.0)]. The formation of the resulting green dye was evaluated spectrophotometrically at 420 nm.

General genome features, e.g., genome size, scaffolds number, N50, L50, number of predicted genes, GC content, completeness of assembly and NCBI accession numbers are summarized in Table 2. The number of reads generated by Illumina sequencing ranged from 39,806,602 to 47,849,884 reads with an average count of 43,006,421. Established genomes range in size from 9.86 to 18.3 Mb. Genomes assembled with Spades resulted in 315–799 scaffolds and were predicted to contain 5,314–7,050 genes, and GC contents from 34.5 to 59.3%. The N50 of five genomes range from 197,085 to 1,183,430, while L50 align from 5 to 26. BUSCO assessment results showed that these assembles are in high quality with around 96% of genes were covered for all five species.

Gene Ontology enrichment analysis indicated that about 63.06, 32.96, 59.14, 63.54, 70.76, and 70.65% genes from K. molischiana, Cryptococcus sp., N. ambrosiae, O. ramenticola, and W. bisporus were mapped to GO terms (Figure 1 and Supplementary Table 1). They were matched to three ontologies: molecular function, cellular component, and biological process. They were highly enriched in cellular process, metabolic process, intracellular anatomical structure, organelle, cytoplasm, binding and catalytic activity. Cryptococcus sp. differed from the other considered species in several functions, such as antioxidant activity, biological adhesion, multicellular organismal process, carbon utilization, immune system process, biological phase and rhythmic process.

To investigate the function of proteins in those five species, the annotation based on BlastKOALA (Kanehisa et al., 2016) against the KEGG database was performed. It was possible to annotate 58.2, 57.4, 60.1, 56.3, and 35.6% of conserved sequences from K. molischiana, N. ambrosiae, O. ramenticola, W. bisporus, and Cryptococcus sp., respectively, Figure 2. As a result, the highest enriched system was “Genetic information processing,” followed by “Carbohydrate metabolism,” and “Cellular processes.” Vitamin B6 compounds, are synthetized and metabolized in all five species (Supplementary Figure 1). Additionally, essential amino acids such as arginine, lysine, phenylalanine, tyrosine, valine, leucine, and isoleucine are biosynthesized in these yeast strains as well (Supplementary Figure 2).

The carbohydrate-active enzyme (CAZyme) gene contents were predicted for all five genomes (Figure 3A, Supplementary Figure 3 and Supplementary Table 2) in this study. These CAZymes are classified into six classes: auxiliary activity (AA), carbohydrate-binding module (CBM), carbohydrate esterase (CE), glycoside hydrolase (GH), glycosyltransferase (GT), and polysaccharide lyases (PL). Nakazawaea ambrosiae has more CAZymes (221) than the other four studied species, while the Cryptococcus sp. has the smallest CAZymes count (157). GH accounted for almost half of the identified enzymes (83–98) in those species except for Cryptococcus sp. (41), followed by GH (63–79), AA (13–28), CE (14–24), and CBM (4–9). PL (0–5) comprised only a few genes and is absent in Nakazawaea ambrosiae.

The number of identified genes regulating plant cell wall-degrading enzymes varies in five yeast species (Figure 3B). Wickerhamomyces bisporus has the highest (71) and Cryptococcus sp. has the least number of genes (35). The number of genes related to cellulose degradation is lower than those related to hemicelluloses and pectin degradation in all genomes.

More than 200 detoxification related genes were identified from each species, K. molischiana (225), N. ambrosiae (235), O. ramenticola (200), and W. bisporus (276), except for Cryptococcus sp. (110) (Table 3 and Supplementary Table 3).

We predicted presence of 115 aldo-keto reductase enzymes, with lengths from 107 to 599 amino acids, pIs from 4.62 to 9.00, and predicted molecular weights between 12.02 and 65.57 kDa. They were predicted to be localized in the cytoplasm, plasma membrane, nucleus, mitochondria, vacuoles, and the bonding membranes of the Golgi apparatus. They can be classified into 9 subfamilies (Figure 4), aryl-alcohol dehydrogenases, D-arabinose 1-dehydrogenases, oxidoreductases, and NAD(P)H-dependent D-xylose reductases. They were found in all five genomes with 22, 7, 24, and 10 members, respectively. There are 9, 14, 10, and 16 genes identified in subgroups pyridoxal reductases, NAD(P)H-dependent reductases, D-arabinose 1-dehydrogenases (NAD(P)+) heavy chain, and glycerol 2-dehydrogenases (NAD(P)+), respectively. These subgroups were absent in Cryptococcus sp. Three genes putatively encoding for aldehyde reductase I were found in Cryptococcus sp. and N. ambrosiae.

We identified 29 cytochrome P450 (CYP) monooxygenases with amino acids lengths from 126 to 591, molecular weights from 13.78 to 66.48 kDa, and pIs from 4.9 to 8.51. They were predicted to localize in cytoplasm, plasma membrane, mitochondria, bounding membrane of mitochondria and endoplasmic reticulum. The topology tree of CYP monooxygenases showed that they can be classified into six subfamilies (Figure 5), with 7, 6, 5, 2, 4, and 5 members for CYP51, CYP61, CYP501, CYP56, CYP52, and CYP5252, respectively. Sequences in the CYP51 and CYP61 subfamilies were observed in all five species. CYP501 and CYP5252 were absent in Cryptococcus sp. CYP52 only existed in Cryptococcus sp., while CYP56 was present in W. bisporus genome.

ATP binding cassette (ABC) transporters were predicted from those five genomes, they can be classified into seven subfamilies (Figure 6): 35 pleiotropic drug resistance, 10 eye pigment precursor transporter, 62 drug conjugate transporter, 7 heavy metal transporter, 13 α-factor-pheromone transporter, 43 mitochondrial peptide exporter, and 11 peroxisomal fatty acyl CoA transporter. Their amino acid lengths varied from 120 to 2058 aa, molecular weights from 13.14 to 230.256 kDa, pIs from 4.23 to 9.06. Their predicted subcellular localizations are cytoplasmic, mitochondrial, plasma and endoplasmic reticulum membranes, peroxisomal and vacuolar.

We predicted the presence of 21 carboxylesterases with 109 to 566 amino acids and molecular weight of 12.10 to 62.49 kDa. Then we predicted 46 flavin-containing monooxygenases with 137–647 amino acids, and molecular weights 15.86–72.63 kDa. Glutathione S-transferases were predicted with number of amino acids varying from 111 to 1006 and molecular weights 12.21–110.52 kDa. We also predicted 21 multicopper oxidases with 152–722 amino acids and 17.48–76.62 kDa, 38.10% of them were predicted as extracellular enzymes. The number of amino acids of multidrug and toxic compound extrusion (MATE) ranged from 361–1018 with molecular weights 39.43–113.68 kDa. Major Facilitator Superfamily (MFS), which contains 720 sequences, has 92–1863 amino acids with molecular weights 10–207.86 kDa (Supplementary Table 3).

We measured the activities of endo- and exolytic enzymes degrading complex components of plant tissues (see Table 4 for results) as functional analysis predicted the presence of these genes in yeast genomes. We found that the yeast species studied do not produce endolytic enzymes, with the exception of endo-1,4-β-D-mannanase, which cleaves pectin. However, all yeasts were strong producers of exolytic enzymes. Kuraishia molischiana was the only species producing all the enzymes tested, especially β-glucosidase, cellobiohydrolase, β-xyloside and β-mannosidase. Nakazawaea ambrosiae was the strongest producer of acid monophosphatase, lipase and β-mannosidase. Wickerhamomyces bisporus mainly secreted β-glucosidase and lipase, and O. ramenticola β-glucosidase and β-mannosidase. Cryptococcus sp. was the weakest of the yeasts in enzyme activities and produced mainly β-glucosidase and lipase. We successfully measured the enzyme activities of the positive control Trichoderma viride, suggesting that the absence of endolytic enzymes in the yeast species was not a methodological error.