Isolation, characterization, and genome assembly of Barnettozyma botsteinii sp. nov. and novel strains of Kurtzmaniella quercitrusa isolated from the intestinal tract of the termite Macrotermes bellicosus

Abstract Bioconversion of hemicelluloses into simpler sugars leads to the production of a significant amount of pentose sugars, such as d-xylose. However, efficient utilization of pentoses by conventional yeast production strains remains challenging. Wild yeast strains can provide new industrially relevant characteristics and efficiently utilize pentose sugars. To explore this strategy, we isolated gut-residing yeasts from the termite Macrotermes bellicosus collected in Comoé National Park, Côte d’Ivoire. The yeasts were classified through their Internal Transcribed Spacer/Large Subunit sequence, and their genomes were sequenced and annotated. We identified a novel yeast species, which we name Barnettozyma botsteinii sp. nov. 1118T (MycoBank: 833563, CBS 16679T and IBT 710) and two new strains of Kurtzmaniella quercitrusa: var. comoensis (CBS 16678, IBT 709) and var. filamentosus (CBS 16680, IBT 711). The two K. quercitrusa strains grow 15% faster on synthetic glucose medium than Saccharomyces cerevisiae CEN.PKT in acidic conditions (pH = 3.2) and both strains grow on d-xylose as the sole carbon source at a rate of 0.35 h−1. At neutral pH, the yeast form of K. quercitrusa var. filamentosus, but not var. comoensis, switched to filamentous growth in a carbon source-dependent manner. Their genomes are 11.0–13.2 Mb in size and contain between 4888 and 5475 predicted genes. Together with closely related species, we did not find any relationship between gene content and ability to grow on xylose. Besides its metabolism, K. quercitrusa var. filamentosus has a large potential as a production organism, because of its capacity to grow at low pH and to undergo a dimorphic shift.


Introduction
The biotechnology industry has historically relied on yeast species to carry out fermentation processes (Demain 2000). A major challenge today is the conversion of raw plant biomass directly into organic compounds of added value, such as organic acids or biogas (Qin et al. 2017; Martínez-Gutié rrez 2018; Lee et al. 2019). Lignocellulose material is mainly composed of cellulose (30-50%), hemicellulose (20-30%), and lignin (15-25%) (Kricka et al. 2014). To break down these complex polysaccharides, filamentous fungi, such as Trichoderma reesi, Aspergillus niger and Aspergillus oryzae, and their secreted enzymatic cocktails have been used with success (Mä kelä et al. 2014;Lange 2017). Because these fungi do not efficiently ferment pentose sugars such as D-xylose and L-arabinose produced during depolymerization of hemicelluloses, yeast species or genetically engineered Saccharomyces cerevisiae strains have been suggested for downstream processes (Bettiga et al. 2009). However, co-fermentation of hexoses and pentoses is still a major challenge during the efficient conversion of lignocellulosic plant biomass, mainly due to difficulties in integrating a new catabolic pathway within an existing metabolic network (Young et al. 2010). There is therefore a need to identify and optimize yeasts that can efficiently ferment both pentoses and hexoses. One strategy is to isolate novel wild yeast strains that can metabolize these sugars, which can either serve as a production organism on their own or as a genetic resource for metabolic pathways that can be engineered into industrial strains (Steensels et al. 2014).
The unicellular yeasts are commonly found in soil, plants, and flowers, and within the microbiome of insects (bees, beetles, termites, and ants), where they establish a variety of relationships, ranging from obligate to facultative mutualisms (Ganter 2006;Greig and Leu 2009;Kurtzman et al. 2011c;Buzzini et al. 2017). Described yeast species are found both in the Basidiomycota and Ascomycota phyla and are thought to represent only 1% of the expected yeast diversity worldwide (Robert 2006). In addition, yeast species from tropical rainforest, desert, or tundra ecosystems are underrepresented among characterized species (Kurtzman et al. 2015;Groenewald et al. 2017) and, therefore, serve as a reservoir for novel species with new characteristics. One example is the gut-residing yeasts from the genus of Candida, Pichia, Sporothrix, and Debratomyces, which have been isolated from a variety of termite species (Schä fer 1996;Stefanini 2018). Here, we explored yeasts within Macrotermes bellicosus termite guts to find novel yeast strains attractive from a biotechnological perspective and to shed light on insect-yeast relationships. M. bellicosus is a fungus-growing termite species (subfamily Macrotermitinae) that has an obligate symbiosis with the basidiomycete fungus Termitomyces (Basidiomycota: Lyophyllaceae) maintained on plant material harvested by the termites from outside the nest (Rouland-Lefèvre 2000). During a first passage through the termite gut, this plant biomass is mixed with asexual fungal spores and the resulting feces is used to build a so-called fungus comb (Leuthold et al. 1989). The maturation of the comb allows the fungus to grow and the plant material to be degraded. Termites will ingest the mature comb and during a second gut passage obtain nutrition (Poulsen et al. 2014;da Costa et al. 2019). The combined enzymatic activities from the termite, the gut microbiome, and the fungus account for the degradation of up to 90% of all dead wood in Kenya (Buxton 1981). Altogether, their foraging and digestive capabilities substantially contribute to carbon cycling in these ecosystems (Sugimoto et al. 2000) converting wood, leaves, and roots (Buxton 1981;Collins 1981;Smith et al. 2019).
In this study, we present and characterize two novel strains and one new species of yeast isolated from the gut of M. bellicosus. The two novel strains belong to the species Kurtzmaniella quercitrusa, phylum Ascomycota, and they were thus named K. quercitrusa var. comoensis (strain number in this work 1112) and K. quercitrusa var. filamentosus (strain number in this work 1120). We also identified a previously unknown species from the Ascomycota, which we name Barnettozyma botsteinii sp. nov. 1118 T (strain number in this work 1118). All three strains assimilate pentose sugars, which make them potentially suitable for biotechnological applications and interesting for further genomic and metabolic studies. Their degradation of plant-derived polymers suggests a potential role in assisting termite and Termitomyces metabolism and supports their potential for biotechnological production.

Sample collection
We collected M. bellicosus minor workers from the inside of wild nests in Comoé National Park, Cô te d'Ivoire during the month of November 2018. Specifically, the termite nests were located in the proximity of the Comoé National Park Research Station (8 46 0 11 00 N, 3 47 0 21 00 W), in the southwest end of the park next to the Comoé River. The gut of small workers (25 termites per nest) was dissected in situ and divided into midgut, hindgut, and foregut ( Figure 1) (Bos et al. 2020). Five to ten replicates were pooled together to increase the starting material and manual tissue dissociation was performed with a pestle in sterile 1Â phosphatebuffered saline (PBS) supplemented with Tween 80 (5 mg/ml). The resulting lysates were diluted (1/100 and 1/1000) and spread on potato dextrose agar plates. The medium was prepared by diluting 39 g/l of PDA (Sigma Aldrich, 70139) in water, autoclaved, and supplemented with 0.1 g/l of chloramphenicol, 0.05 g/l of streptomycin sulfate and 0.035 g/l of ampicillin before pouring it into plates. Plates were then incubated at 30 C until growth was observed. To screen for yeast species, single colonies were first streak out and subsequently incubated with YPD medium supplemented with 6% ethanol at room temperature until growth was observed. Isolates with a yeast-like morphology under the microscope were then subjected to genomic DNA extraction and rDNA sequencing.

Internal Transcribed Spacer and D1/D2 sequencing and phylogeny
For genomic DNA (gDNA) extraction, a standardized protocol for yeasts was applied to all the isolates. Briefly, yeast cell wall was broken using glass beads and vortexing. Buffers P1, P2, and N3 (Qiagen, Cat. No.: 19051, 19052, and 19064) were used to isolate the DNA fraction, which was further precipitated by the addition of isopropanol. Finally, after an ethanol wash, pellets were resuspended in TE buffer (10 mM Tris-HCl, pH ¼ 8, 1 mM EDTA) and DNA concentration was assessed by Nanodrop. Primers ITS4 (5 0 -TCCTCCG CTTATTGATATGC) and ITS1 (5 0 -CTTGGTCATTTAGAGGAAGTAA) (Kuiper-Goodman and Scott 1989) were used to amplify Internal Transcribed Spacer (ITS) regions while primers NL1 (5 0 -GCATATCA ATAAGCGGAGGAAAAG) and NL4 (5 0 -GGTCCGTGTTTCAAGACGG) were used to amplify the Large Subunit of the nuclear ribosomal RNA (LSU rRNA) gene region (Kurtzman and Robnett 1997). Both PCR products were purified, Sanger sequenced and results aligned to NCBI database using BLASTn (Altschul et al. 1990). Nucleotide sequences for a set of the most closely yeast species were retrieved from GenBank (Supplementary Table S1). Alignment was performed with Mafft v.7.471 (http://mafft.cbrc.jp/alignment/soft ware/) and a maximum-likelihood (ML) phylogeny based on ITS and LSU was constructed in MEGA v. X (Kumar et al. 2018). Branch support was obtained from bootstrap analysis with 1000 repetitions.

Genome sequencing and analyses
We sequenced and assembled the genomes of isolates 1118, 1112, and 1120 in the following steps. Code used is provided in Supplementary File S1.

DNA extraction, sequencing, and assembly
Whole-genome sequencing of isolate 1118, 1112, and 1120 was done by first extracting DNA using a scaled-up CTAB extraction (Poulsen et al. 2014). Whole-genome sequencing was performed using a combination of 100/150 bp paired-end shotgun (BGIseq/ DNBseq) and long-read (PacBio sequel) sequencing by BGI. The resulting sequences were checked for quality using FastQC v0.11.9 (Andrews et al. 2012). Genomes were assembled using three different approaches and the best result was chosen for subsequent analyses. First, short and long reads were assembled together and used to generate scaffolds using SPAdes 3.14.1 (Nurk et al. 2013). Second, long reads were assembled into scaffolds with Canu 2.0 (Koren et al. 2017) and then polished by the short reads using NextPolish 1.3.1 (Hu et al. 2020). Finally, short reads were assembled into scaffolds alone, also using SPAdes 3.14.1 (Nurk et al. 2013). Contigs that were shorter than 500 bp were discarded in the final assemblies, and assembly quality was assessed by quantifying genome completeness based on the expected gene content of the Benchmarking Universal Single-Copy Orthologs (BUSCO), version 4.4.1, against the database for fungal genomes (Seppey et al. 2019).
Therefore, an initial MAKER analysis was carried out where gene annotations were generated directly from homology evidence without using the ab initio gene predictors. The resulting gene annotations supported by homology evidence were then used to train SNAP and AUGUSTUS. Once both ab initio gene predictors were trained, they were used together with homology evidence in a full MAKER analysis. Homology evidence was only used to inform gene predictions. Resulting gene models supported by homology evidence were used to re-train SNAP and AUGUSTUS. A second-round analysis was conducted using the newly trained SNAP and AUGUSTUS parameters, and once again the resulting gene models with homology supports were used to re-train SNAP and AUGUSTUS. Finally, a third round of MAKER analysis was performed using the new SNAP and AUGUSTUS parameters. All resulting gene models are reported, and these comprise the final set of annotated genes.
Functional gene annotation was performed using InterProScan version 5.48-83.0 (Jones et al. 2014) with annotation of gene ontology (GO) terms (Ashburner et al. 2000), KEGG pathways (Kanehisa et al. 2021), and Pfam 33 (Mistry et al. 2021). The predicted proteins sequences were blasted against NCBI-NR (O'Leary et al. 2016) and Swiss-Prot databases (Bateman 2019) with an E-value cutoff of 1e-5. Noncoding RNA (ncRNA) genes were identified by BAsic Rapid Ribosomal RNA Predictor v0.9, Barrnap (Seemann 2013) and tRNA genes were predicted using the tRNAScan-SE v2.0.5 (Lowe and Chan 2016) algorithm with default parameters. Repeat sequences were identified and classified using RepeatMasker v4.1.0 (Smit et al. 2019). The criterion used for LTR family classification was that the 5 0 LTR sequences should share at least 80% identity over at least 80% of their length for the same family.

Multi-locus-sequence-typing analysis
We performed multi-locus-sequence-typing (MLST) using a set of Universal Single-Copy Orthologous genes for all genomes that were identified with BUSCO v4 (Seppey et al. 2019) with default settings and using the fungi_odb10.2019-11-20 dataset. We collated 344 (of 758 potential) genes present in all 48 genomes and generated multi-fasta files of the orthologous genes (Supplementary Table S2). These multi-fasta files were aligned with Clustal Omega v1.2.4 (Sievers et al. 2011). Phylogenies were generated using RAxML-NG v0.9.9 (Kozlov et al. 2019), employing the -all mode, GTRþG model, and a seed of 2. Branch support based on bootstrapping and transfer distance were obtained, before generating a

Figure 1
Origin and identification of yeast isolates. Small workers termites were collected in Comoé National Park, in the north east of Cô te d'Ivoire (highlighted area in the map) from four Macrotermes bellicosus termite colonies (IC0021, IC0027, IC0031, and IC0034) (see Methods). Guts were dissected into foregut, midgut, and hindgut and yeast species were identified by comparing their ITS rDNA sequences to strains in the NCBI GenBank repository. According to their classification and origin, we grouped the isolates and assigned an internal ID. GenBank sequence IDs denoted with * were generated in this study. Otherwise, sequences from the reported isolates were found to be identical to the GenBank IDs provided. Map image modified from V C OpenStreetMap contributors.

Yeast media and growth conditions
Strains were grown and kept on YPD medium (10 g/l of yeast extract, 20 g/l of bacto-peptone, and 20 g/l of glucose) at 30 C. To study growth dynamics and carbon source assimilation, strains were grown and kept on minimal medium (1.6 g/l of yeast nitrogen base without amino acids, 5 g/l of ammonium sulfate with a pH of 6, supplemented with 1% glucose or 1% D-xylose). To check D-xylose assimilation in different concentrations and aeration conditions, single colonies from minimal glucose agar plates were used to inoculate glass-test tubes (max. vol. 35 ml) at an OD 600 ¼ 0.15, containing minimal medium with 2% or 10% D-xylose at different volumes: 3 ml (high aeration), 7.5 ml (low aeration) or 20 ml (microaeration). The tubes were then kept at 30 C with shaking (150 rpm) for 14 days. To calculate the growth rate in D-xylose and D-glucose medium, starter cultures were prepared by growing single colonies from glucose or xylose minimal medium agar plates into corresponding liquid medium buffered to a pH ¼ 3.2 or pH ¼ 6 overnight at 30 C. The next day, the overnight cultures were used to inoculate 50 ml medium to an OD 600 $ 0.1 in 250 ml baffled Erlenmeyer flasks containing the same buffered medium. For both carbon sources, the medium was buffered to a pH ¼ 3.2 with 0.1 M final concentration of phosphoric buffer (1M KH 2 PO 4 adjusted to pH ¼ 3.2 with phosphoric acid) or to a pH ¼ 6 with 0.1 M final concentration of potassium phosphate buffer (pH ¼ 6). Cultures were incubated at 30 C and 150 rpm. To calculate each m, OD 600 measurements from two different single colonies were used. Each individual OD 600 measurement was obtained by the average of triplicates. As previously described, colonization patterns were made on semi-solid YPD 0.3% agar plates and cultures were grown for 14 days at room temperature (Kurtzman et al. 1991).

Carbon assimilation
Carbon-source assimilation was assessed using microplates (Kurtzman et al. 2011c) using Phenotypic Microarray plates (Biolog, PM1 # 12111 and PM2 # 12112). Briefly, the plates were inoculated with strains 1112, 1118, and 1120 in buffer PM containing Dye mix E to facilitate and improve detection, incubated at 30 C for 144 h and the absorbance measured at 550 nm, as described by the manufacturer. Values at þ144 h were used to assess growth on each of the carbon sources. For each time point, background absorbance was subtracted and the fold change (FC) over negative control (no carbon source) was calculated. Signals were assigned to a strong growth (FC >3 at 144 h, þþþ), intermediate growth (FC >2 and 3 at 144 h, þþ), weak growth (FC >1 and 2), and no growth (FC 1). Strains were seeded in duplicate plates in two independent experiments. Mean of the FC was calculated.

Microscopy
Yeast cells for microscopy were grown overnight in YPD medium at 30 C at 150 rpm. Cells were washed once with distilled sterile water and pellets were re-suspended in water. One to five microliters were mounted in a glass slide with a 0.17 mm thin cover slip. Samples were visualized using a Nikon Eclipse E600 and pictures taken with an Optronics MagnaFire CCD Microscope Camera.

Phylogenetic analysis of yeast isolates
We collected worker termites of the species M. bellicosus in Comoé National Park (Cô te d'Ivoire) and dissected their intestines in foregut, midgut, and hindgut for yeast isolation (Bos et al. 2020).
We identified a total of 20 individual yeast colonies spanning all three different gut compartments and four independent termite colonies ( Figure 1). To identify novel species and subspecies, we next characterized the isolates taxonomically by sequencing the ITS and D1/D2 rDNA regions of LSU and aligning the sequences to GenBank repository using BLASTN (Altschul et al. 1990) (Figure 1). From the set of 20 isolates, 11/20 were assigned to Meyerozyma caribbica. Next, 3/20 isolates were assigned to Candida sp. TMS-2011 and the rest (6/20 isolates) did not have a clear match with any NCBI sequence, showing similarities between 84.4% and 99% to their respective best matches (Figure 1). From the six uncategorized isolates, four (IDs: 1112, 1113, 1105 and 1106) showed three gaps with K. quercitrusa KMC-Y23, isolate 1120 showed one mismatch with K. quercitrusa WHFS-3 and isolate 1118 showed 12 gaps and 37 mismatches with Barnettozyma sp. EN30S02, leading to a similarity of 91%. Because 1112, 1113, 1105, and 1106 had identical rDNA D1/D2 sequences, we grouped them and selected 1112 as a representative of this group (Figure 1). The best ML model based on ITS and LSU was TN93þG. The ML tree allowed us to include all closely related yeast species within the genera (Figure 2). B. botsteinii sp. nov. (1118) clusters as an independent branch with Barnettozyma salicaria NRRL Y-6780 T (Figure 2A) while K. quercitrusa var. comoensis (1112) and var. filamentosus (1120) cluster with K. quercitrusa type strain CBS 4412 T ( Figure 2B).
To gain more confidence on its phylogenetic placement, we sequenced the genomes of isolates 1118, 1112, and 1120. We generated a Multiple-Locus-Sequence-Typing (MLST) analysis based on a set of 1706 BUSCO genes inferred to be Ascomycota-specific and single-copy in at least 85% of the available yeast genomes with high (<80%) genome completeness (for general genome features and completeness, see Table 1). The genome sizes and gene content of 1112 and 1120 are similar to Kurtzmaniella cleridarum, which has a 12.1 Mb large genome and contains 5478 genes, while 1118 is 11.1 Mb with 4888 genes; comparable to the closest relative with a genome available (Barnettozyma salicaria; 11.01 Mb, 5586 genes). The MLST phylogeny thus confirmed the placement of these new isolates in their respective genera and clades ( Figure 2).
We then compared the metabolic profile characterization of the 1118 with other known and closely related Barnettozyma and Candida species such as B. salicaria, B. siamensis and C. montana. The result revealed that it has different metabolic characteristics, as it can assimilate D-ribose, L-arabinose, gelatine, and maltose, but not salicin and sucrose (Supplementary Table S5). Similarly, the comparison of the physiology of 1112 and K. quercitrusa CBS 4412 showed that 1112 differed by being able to assimilate D-arabinose, D-xylose, inulin, L-arabinose, N-acetyl-D-glucosamine and salicin, but not D-glucosamine, D-mannitol, D-trehalose, glycerol, and xylitol. In contrast to K. quercitrusa CBS 4412, 1120 grows on D-arabinose, gelatine, L-arabinose and gelatine but not D-mannitol, maltose, sucrose, and L-sorbose (Supplementary Table S5).

Kurtzmaniella quercitrusa var. comoensis
Strain ID in this work: 1112. Microscopic examination of this isolate confirmed that it grew as yeast. Cells appear as black, oval shaped and were 3-4 mm long and 2-3mm wide during exponential growth in rich (YPD) and minimal glucose medium (see Methods section). Cells divide mitotically by a small bud that fully separates from the mother cell ( Figure 3A). Colonies on YPD agar are white, bright, and round, while they appear as a flat and round with an even rim on medium with a low concentration of agar, 0.3% YPD agar ( Figure 3D).

Kurtzmaniella quercitrusa var. filamentosus
Strain ID in this work: 1120. Yeast cells of this strain are clearly distinguishable from K. quercitrusa var. comoensis. Being 5-6 mm in length and 1-2 mm wide, cells appear as longer and thinner than 1112 ( Figure 3B). Moreover, during exponential growth in glucosebased medium, YPD, and minimal glucose medium (see Methods section), some cells do not fully separate after division leading to formation of 5-6-cell-long pseudohyphal structures. After 5 days of growth in liquid 1% xylose medium, formation of pseudohyphae is strongly enhanced, leading to truly septated hyphae. Hyphae first appear as germ tubes ( Figure 4A), which are 2-7 cell long chains growing from a single round cell. They later elongate and mature to true hyphal structures showing bifurcations (Figure 4, C and C'). Colonies on rich glucose-based medium (YPD) and minimal glucose medium show a rough, butyrous, and ivory-colored phenotype. In addition, their colonization pattern on 0.3% agar plates is noneven with visible edges and with some growth elevated from the agar (see Methods section) ( Figure 3E).  They are commonly observed as clusters of 3-5 cells forming pseudohyphal structures ( Figure 3C). No sign of true hyphae formation was observed in any condition. Although colonies on YPD or minimal glucose medium appear as smooth, the colonization on glucose 0.3% YPD agar leads to an edgy pattern ( Figure 3F). The colonies are ivory-colored and flat. The relatively low growth rate observed on liquid medium (Table 2) was also reflected in smaller colonies on solid medium (both on YPD and on SC-1% glucose).

D-Xylose metabolism
After observing that the novel isolates grow on pentose sugars, we next characterized their D-xylose utilization, using different aeration and concentrations of sugar in the growth medium. Strains K. quercitrusa var. comoensis (strain 1112) and K. quercitrusa var. filamentosus (strain 1120) show an aeration-dependent growth preference on D-xylose ( Figure 5). From the conditions tested, the highest biomass is reached with high aeration in combination with 2% D-xylose, where the OD 600 after 14 days of growth reaches 16.6 6 0.6 and 14.9 6 1.0, respectively. For both K. quercitrusa strains, we observed a correlation between aeration levels and OD 600 . A decrease in aeration level correlates with a decrease in OD 600 both with 2% D-xylose and 10% D-xylose. For example, in low aeration conditions 1112 grows to OD 600 of 5.8 6 0.2 with 2% D-xylose and 5.8 6 0.2 with 10% D-xylose. Similarly, 1120 grows to an OD 600 of 5.4 6 0.8 with 2% D-xylose and 2.3 6 0.6 with 10% D-xylose. We observed a further reduction in the OD 600 in microaeration conditions, especially in 1112 strain, where it reaches only 3.2 6 0.29 in 2% D-xylose and 2.4 6 0.1 in 10% D-xylose. B. botsteinii sp. nov. 1118 T , on the other hand, did not efficiently grow on D-xylose. Only marginal to very limited growth was observed with microaeration and high D-xylose concentration ( Figure 5). Interestingly, K. quercitrusa var. filamentosus (strain 1120) undergoes a dimorphic shift and forms hyphae in 2% D-xylose medium combined with high or low aeration ( Figure 5). The dimorphic shift consisted of formation of pseudohyphae and hyphae of various lengths. No changes in morphology were observed with the others strains.
We next compared the growth rates of K. quercitrusa when propagate in D-xylose and D-glucose at low and neutral pH ( Table 2). K. quercitrusa var. comoensis (strain 1112) cells grew in 1% synthetic glucose medium with a doubling time of 2.0 h at pH ¼ 3.2. At pH ¼ 6, the doubling time was similar at 1.9 h. In 1% xylose medium, cells grew at a doubling time of 19.3 h, while they exhibited no growth at pH ¼ 3.2. Cells grew as single yeast cells and divided by budding through exponential phase at both carbon source medium (Table 2). For comparison, a S. cerevisiae CEN.PK type strain grew only in synthetic glucose medium at a doubling time of 2.3 h at pH ¼ 3.2 while at pH ¼ 6 it grew with a doubling time of 2.1 h.
K. quercitrusa var. filamentosus (strain 1120) grew on 1% glucose with a doubling time of 2.0 h at both tested pH (3.2 and 6.0). In xylose, the strain showed pH-dependent morphologic differences (Table 2). At pH ¼ 3.2, the strain grew as single yeast cells with a doubling time of 19.6 h, while at pH ¼ 6.0 medium, pseudohyphae accumulated and matured into true hyphae (Table 2 and Figure 5). Both strains of K. quercitrusa (strains 1112 and 1120) showed very similar growth rates in 1% glucose medium and slightly lower than S. cerevisiae CEN.PK T strain (Table 2).
B. botsteinii sp. nov. 1118 T grew in 1% glucose medium at pH ¼ 6 with doubling time of 2.8 h (Table 2). Interestingly, the strain was sensitive to acidic pH that drastically increased its doubling time of 25.7 h at pH ¼ 3.2. Compared to the K. quercitrusa strains and S. cerevisiae CEN.PK T , this species showed the slowest growth rate at both pH (Table 2). Under this experimental growth condition, the strain was not able to grow on 1% xylose medium at either of the tested pH (Table 2).
Copy number changes of genes coding for enzymes in metabolic pathways are known to affect metabolic pathway efflux (Steenwyk and Rokas 2018). Therefore, we investigated the copy number of genes involved in xylose and pentose metabolism. Our comparative genomics analysis for XYL and PPP genes revealed that the vast majority of yeasts contained all the orthologues necessary for xylose metabolism ( Figure 6). The lack of growth of some of the species on xylose thus does not appear to be as a result of the absence of xylose-metabolizing genes ( Figure 6).

Taxonomy
Description of B. botsteinii sp. nov. 1118 MycoBank: MB 833563 Barnettozyma botsteinii (bOt'StaIni, N.L. gen. n. botsteinii, named after scientist David Botstein) isolated from the hindgut of termite M. bellicosus in Cô te d'Ivoire. Holotype is preserved at the Westerdijk Institute as CBS 16679 T and an isotype at the IBT Culture Collection of Fungi as IBT 710. Cells appear as spherical and small (3-4 mm Â 2-3 mm), typically forming clusters of 3-5 cells in pseudohyphal structures. No sign of true hyphae formation was observed in the conditions tested. The colonies appear ivory colored, flat and circular in YPD or minimal 1% glucose medium. However, at low agar (0.3% agar) glucose medium the edge becomes undulate (Figure 3). Cells are able to grow in liquid or solid YPD at 22 C-30 C. The cells grow in 1% glucose medium at pH ¼ 6 with a growth rate of 0.24 h À (60.02) and doubling time of 2.8 h. Their growth rate is decreased to 0.03 h À (60.005) corresponding to a doubling time of 25.7 h at pH ¼ 3.2 (Table 2). Based on its genome, phylogeny, and physiology analysis, we confirm that isolate 1118 is clearly distinct from any other known Growth rate (m max , h À1 ) and doubling times (DT, h) of CEN.PK, B. botsteinii sp. nov. 1118, K. quercitrusa var. nov. 1112, and K. quercitrusa var. nov. 1120 in liquid SC-1% glucose and SC-1% xylose at pH 3.2 or 6. All growth rates are averages of two independent experiments from two independent clones.
Barnettozyma species, we assign the status of sp. nov. to isolate 1118 and thus, propose B. botsteinii sp. nov. 1118 T as the founder and type strain of a novel yeast species. We name the species after Professor David Botstein, a true inspirer, great geneticist, and important figure for the yeast research community (Botstein et al. 1980;Botstein and Fink 1988).

Discussion
In this study, we isolated novel and biotechnologically relevant yeast strains from termite guts. We describe two novel strains of K. quercitrusa and a new species of yeast, which we name B. botsteinii based on genomic information. The criteria to classify these three novel taxa follows Kurtzman et al. (2011c), in which conspecific strains differ by no more than three nucleotides, whereas distinct species show six or more nucleotide differences (Kurtzman et al. 2011b).
Here, the novel yeast species B. botsteinii sp. nov. 1118 T is described based on a single isolate. Novel single-strain fungal taxa descriptions are debated as complete intraspecific variability (including genetic and phenotypic) and its ecological role cannot be fully captured using one strain. However, nearly one-third of described yeast species is based on single isolates (Kurtzman et al. 2011a). The documentation and publishing of single strains novel taxa serve as a nucleation point for further studies and the possibility of combining information from strains independently isolated in time. While it was not possible for us to collect new material or find any other available strain with a complete match in GenBank, it is possible that more isolates of B. botsteinii sp. nov. are found in the future. This debate was recently reviewed in Brysch-Herzberg et al. (2020).
We find that both novel strains of K. quercitrusa are able to grow on D-xylose, and their overall metabolic profile makes them interesting for further applied studies. The complex phenotypic change that K. quercitrusa var. filamentosus (strain 1120) undergoes (filamentation when growing on certain carbon sources) reveals the existence of a metabolic shift that can control alterations in protein secretion and enzyme production similar to that of other yeasts with a dimorphic shift (Gauthier 2017; Celi nska and Nicaud 2019).
We identified two novel strains and a new species of yeast by studying a relatively small number of individuals (around 100 termites from only one termite species, M. bellicosus). Other yeast strains, without a clear match to described species, have previously been isolated from Macrotermes subhyalinus (Yoro et al. 2014) and other fungus-growing termites (Odontotermes formosanus) (Mathew et al. 2012). Therefore, the Macrotermitinae termite subfamily, comprising 330 described species, appears as a potentially extensive biological reservoir for new yeast species and strains of biotechnological relevance.
K. quercitrusa was first isolated from the insect frass on an oak tree (Quercus sp.) (Meyer and Yarrow 1998), type strain CBS 4412 T . Other isolates of the same species have been collected from other insects (beetles) and from flowers around the world (e.g., Oceania, South America, and South-east Asia) (Crous et al. 2004;Molná r et al. 2008). Although generally considered nonpathogenic, adult and pediatric human infections have been reported in immunocompromised patients, widening the spectrum of nonalbicans candidiasis (Xiao et al. 2014;Westblade et al. 2015).
Yeasts are widely used in the biotechnological production of metabolites and protein. However, most do not metabolize pentose sugars well. The two new K. quercitrusa strains (var. comoensis 1112 and var. filamentosus 1120) grow well on hexose sugars, such as glucose, fructose, mannose, and galactose, and are also able to assimilate pentose sugars such as ribose, xylose, arabinose, and lyxose. They show differences in growth on short oligosaccharides in which var. comoensis grows on maltose and sucrose while var. filamentosus grows on trehalose and adonitol. Furthermore, they both can grow in the presence of complex plant-derived polysaccharides such as mannan, dextrin, and laminarin. The metabolic profile resembles that of K. quercitrusa-type strain CBS 4412 T , including xylose assimilation (Kurtzman et al. 2011c). Strain var. filamentosus is different to var. comoensis in its ability to differentially metabolize both monosaccharides (adonitol) and disaccharides (D-trehalose). The filamentous growth of var. filamentosus has been reported in another member of K. quercitrusa, strain NRRL Y-5392, although described as pseudohyphae (Kurtzman et al. 2011c). In addition, other members of the Kurtzmaniella clade, such as Candida railenensis, also show hyphae formation under certain conditions (Ramírez and Gonzá lez 1984;Kurtzman et al. 2011c). We observed that this phenotype is highly variable and dependent on factors that are common, for example, among Candida species, namely: carbon source and concentration, pH, and aeration level (Caplice and Moran 2015).  Orhtologus gene count Figure 6 Multiple-locus-sequence-typing analysis. In bold, species and strains described in this study. Consensus ML distance tree based on 344 singlecopy orthologous genes with nonparametric bootstrapping with individual gene trees built in RAxML-NG using a GTRþG model. Heatmap of identified number of genes involved in the assimilation and metabolism of xylose across selected yeast species. The heat map indicates that the novel strains and species are comparable to other yeast species (see text for details). Color code indicates number of orthologous genes (top gene panel) from four to none. Positive (þ), negative (À), weak (W), variable (V), and slow (S). Details and references of strains used are found in Supplementary Table S6. *The species was transferred from Candida to Kurtzmaniella genus (Lopes et al. 2019).

Phaffomyces antillensis
We also describe a new yeast species, named B. botsteinii sp. nov. 1118 T , closely related to Barnettozyma salicaria NRRL Y-6780 T . Metabolically, B. botsteinii grows on an array of hexose and pentose sugars. Notably, the strain showed positive growth on short organic acids: acetic and glyoxylic acids (2 carbons), pyruvic acid (3C), fumaric and malic acids (4C), a-ketoglutaric acid (5C), and citric acid (6C). Because many of these are part of the Krebs cycle, this metabolic profile suggests that this species could be classified as Krebs positive (Barnett and Kornberg 1960;Casal et al. 2008). The isolated strain grew in culture conditions although at relatively low rates, suggesting that we did not find the best growth conditions for this species. B. botsteinii could be dependent on the environment provided by the termite for optimal growth like some other known gut-associated yeasts, such as Coccidiascus legeri, which have an obligate relationship with their host insect and are unable to grow on any tested laboratory conditions (Kurtzman et al. 2011c;Stefanini 2018).
The annotation and analyses of genes involved in xylose metabolism did not spur differences in the novel strains and other known yeasts, suggesting that the phenotypic variation in growth in this group of yeasts is unlikely to be driven by differences in the gene content. Although no effect of gene copy number was observed, other genetic factors could influence gene expression and explain the observed phenotypic diversity such as singlenucleotide polymorphisms (SNPs), small inversions-deletions, or translocations. In addition, differences in the balance of key metabolic components and co-factors can influence enzyme activities, for example, NADH/NADþ, which are important for xylose metabolism (Riley et al. 2016). This suggests that the capacity to assimilate and metabolize xylose may be ancient and conserved in yeasts.
A question that remains open is whether these yeast species stably colonize the termite gut or are ingested during foraging. K. quercitrusa strain CBS 4412 T was isolated from insect frass and we isolated K. quercitrusa var. filamentosus from the termite hindgut (i.e., last compartment before faces deposition), suggesting that at least this particular yeast species can survive gut passage, a feature not common to all yeasts nor insect guts (Madden et al. 2018). Gut passage is exploited by other yeast species for their dispersal and breeding (Reuter et al. 2007;Stefanini et al. 2012;Madden et al. 2018) and it is possible that K. quercitrusa and B. botsteinii are also dependent on gut passage for growth, mating, and dispersal.
Biotechnologically, the strains and species characterized here show interesting features that warrant further exploration. K. quercitrusa growth on xylose makes them interesting for industrial applications, especially if they are able to co-ferment xylose in the presence of hexoses (Hahn-Hä gerdal et al. 2007;Ceccato-Antonini et al. 2017). Especially, the characteristics of K. quercitrusa var. filamentosus make it suited for biotechnologically production, since growth at low pH, metabolism of pentose sugars, and a controllable filamentous state are particularly desired traits. In addition, all three strains appear to be able to metabolize polysorbate, also known as Tween. This is a complex molecule formed by polyexyethylenated sorbitol molecules with lauric acid and used as a detergent and surfactant. Other yeasts such as Candida albicans or Candida parapsilosis sustain growth on Tween as a sole carbon source, a trait considered as a virulence factor as its lipid content triggers the expression of extracellular phospholipases, including LIP1 (Fu et al. 1997;Slifkin 2000).
In addition, other metabolic processes other than pentose consumption are also biotechnologically attractive. The cellobiose respiration by B. botsteinii sp. nov. 1118 T suggests the presence of ß-glucosidases enzymes (EC 3.2.1.21), which are able to break down the cellobiose molecule into single glucose molecules. During the depolymerization of cellulose, cellobiose is produced which inhibits cellulose-degrading enzymes (cellulose 1,4-ß-cellobiosidase and endoglucanases), reducing the overall efficiency (Murphy et al. 2013;Sørensen et al. 2013). Thus, the addition of ßglucosidases to the process can greatly improve its efficiency (Woodward and Wiseman 1982).
In conclusion, the intestine of fungus-growing termites appears to be a promising niche for the identification of new yeast strains and species. The metabolic profiles of the yeasts isolated from termite gut reflect the adaptations to a plant-based diet found either within the wild or within the gut of the termite, and these gut-residing yeasts of Macrotermes termites have interesting traits of relevance to the biotechnology industry. Supplementary material is available at figshare: https://doi. org/10.25387/g3.14899278. .