Background

The Ustilaginales are an order of plant-parasitic fungi within the Ustilaginomycotina (Basidiomycota), commonly known as smut fungi of grasses. Although similar plant-parasitic strategies are implemented in other lineages, such as the Urocystidales (Ustilaginomycetes), Tilletiales, or even Doassansiales, Entylomatales (Exobasidiomycetes), and Microbotryaceae (Microbotryomycetes), the sheer number of species in the Ustilaginales express their ecological relevance in certain ecosystems [1]. Their adaptations to open, temperate grassland ecosystems include efficient dispersal of diaspores that remain capable of germination for many years [2] and a life cycle with two alternating phases, corresponding to seasonal change. Almost all species in this group share a dimorphic life cycle that comprises an asexual, saprotrophic yeast stage, followed by a filamentous sexual stage that is necessary to parasitize the host [1]. Infection is initiated after the mating of two cells of compatible mating-types (i.e., different alleles of mating-type genes). The mating-types are determined by two genetic loci, the pheromone/pheromone receptor locus (PR) and the homeodomain transcription factor locus (HD). The two mating-type loci can be physically linked (bipolar mating) or unlinked (tetrapolar mating) in the genome [3]. The pheromone/pheromone-receptor system controls the recognition of compatible mating partners by secretion of small pheromones or mating factors (MFa) that activate the g-protein coupled receptor (PRA) of the partner [4, 5]. Generally, one individual produces one or two different pheromone types (species with two (biallelic) or three (triallelic) different alleles in the PR locus) and one receptor type. The pheromones are unable to activate the mating receptors of cells of the same mating-type, but can activate receptors of strains with different mating-types (i.e., heterothallic), preventing selfing and promoting genetic variability [6]. After receptor activation, the cells form conjugation hyphae, proceeding to dikaryotic penetration hyphae to infect the plant [4]. This morphological switch is mediated by heterodimers of the homeodomain transcription factor, coded for by the two genes of the HD locus (bEast/bWest). Dimers can only be formed from gene products of two different strains that bear different alleles at the often multiallelic HD locus [7]. Several studies indicate that this homeodomain transcription factor not only maintains the dikaryotic growth and controls clamp development as in many Agaricomycotina (Basidiomycota), but it is also a major regulator of the parasitic (teleomorphic) phase in Ustilaginales [8]. After the infection of the plant, growth of the parasite is directed towards the site of sporulation, which is either leaf-tissue or, more often, flowers or complete inflorescences. This parasitic phase is characterized by a dikaryotic growth and is finally summarized in the formation of teliospores, which act as resting spores and are dispersed by wind or vectors such as insects. Karyogamy and meiosis take place during the germination of the teliospores, resulting in a septated basidia bearing four basidiospores, which can proceed to yeast-like growth of haploid sporidia (anamorphic) [1]. The haploid stage can be maintained for a long period of time [1], and initiation of a new infection can only be completed by mating of the sporidia that act as gametes, leading to the formation of the dikaryon and repeating the cycle.

With the advent of molecular phylogenetic tools, it has become clear that some members scattered in the Ustilaginales have only been isolated as yeasts [9]. Some of them are found frequently in diverse ecological niches and do not bear sexual or parasitic structures. Thus, the anamorphic genus Pseudozyma was used to accommodate the asexual species sharing physiological characteristics of other Ustilaginales [9], while being known for only reproducing asexually and producing compounds relevant for industry and agriculture [10,11,12,13]. However, the genus has since been found to be polyphyletic and many species have already been reassigned to other genera using molecular taxonomic studies and renamed accordingly [14]. In this study, we will use the new names when applicable, but to keep comprehensiveness of the texts, refer to the whole group as ‘Pseudozyma’. The reassignment based on molecular phylogenies suggests that asexuality would have evolved multiple times in the evolutionary history of the Ustilaginales [14, 15]. This could include the loss of mating and meiosis machinery that are only required in sexual reproduction, as well as the adaptation from a nutritional diverse metabolism [16] that is characterized by switching between saprobic and parasitic stages to purely saprobic growth.

As proposed by Begerow et al. [9], some Pseudozyma species have been found to represent the anamorphic stage of described parasitic species, a teleomorph for Pseudozyma antarctica (Moesziomyces antarcticus) has been described recently [17], as well as cases of probable conspecificity in the Moesziomyces genus [15]. Adding on that, Sharma et al. [18] have annotated fast-evolving parasitic effector sets in several Pseudozyma species in silico and provided experimental evidence that the conserved effector Pep1 that is known from the model organism Ustilago maydis is present and functional in Pseudozyma. However, it remains unclear if this is true for all Pseudozyma species and whether their parasitic stages have been missed until now, or if there are truly asexual species which have lost their capacity for mating and meiosis. There are several scenarios which would support such hypotheses, such as the extinction of the host or the lack of a mating partner with the compatible mating genes. In addition, symptomless colonization could explain why the sexual, plant-parasitic stages have not been observed so far.

The presence of effector genes relevant for parasitism in Pseudozyma raises the question if they share the genomic makeup of sexual Ustilaginales species to proceed through the typical life cycle of this group. To answer this question, we sequenced the genomes of five former Pseudozyma species and six sexually reproducing parasitic species and complemented our dataset with six published Pseudozyma genomes and two genomes from parasitic species. We annotated the core genes for two central sexuality-related features in the life cycle of the fungi: mating and meiosis. This includes both mating loci (PR and HD), as well as 20 core meiosis genes described for Ustilaginomycotina [19,20,21]. We tested the conservation of the annotated genes in silico and assessed potential functionality. We subsequently inferred possible mating systems (bi-or triallelic, bi- or tetrapolar) in Pseudozyma and other related Ustilaginales species and assessed conservation of meiosis-related genes to hypothesize possible sexual reproduction in Pseudozyma. We found that the Pseudozyma species all display the genetic makeup for sexual reproduction, challenging their anamorphic species assignment.

Results

Genome sequencing of Ustilaginales

In this study, we sequenced and assembled 11 genomes from Ustilaginales strains (belonging to 11 different species from five genera including Pseudozyma). Genome statistics were computed to assess the genome structure and quality of newly sequenced and published genomes for comparison (Table 1). The genome size of the newly sequenced genomes ranged from 13.8 Mb in Farysia itapuensis (Farysizyma itapuensis) to 24.8 Mb in Sporisorium sorghi. These genomes extended the range of the published genomes incorporated in this study, comprising sizes from 17.3 to 20.8 Mb. The quality of the assembly of the Illumina reads varied between 71 scaffolds in Farysia itapuensis and 6454 scaffolds in Ustilago tritici, while published genomes were assembled in 22 to 12,840 scaffolds, respectively. N50 values of new genomes ranged between 18.8 kb and 473.1 kb. However, the assemblies contained less gaps than some of the already published genomes. This might have led to a higher scaffold count but reduces the risk of misassembly by scaffolding or problems due to masking.

Table 1 Genome and annotation statistics
Full size table

The BUSCO genome completeness assessment resulted in completeness values between 97.8% (Ustilago hordei) and 99.1% (Pseudozyma pruni), published genomes were between 95.9% (Moesziomyces aphidis) and 99% (Sporisorium graminicola) complete.

De novo gene prediction resulted in between 5,355 genes for the outgroup species Urocystis primulicola and 8,889 genes for Pseudozyma tsukubaensis. Resulting gene densities ranged between 297 genes/Mb in U. hordei and 421 genes/Mb in Farysia itapuensis, the smallest genome in the study. The lowest number of introns per gene was predicted in Ustilago maydis (0.51), the highest in Urocystis primulicola (1.7). All Ustilaginales genomes in this study contained less than one intron per gene (Table 1).

Analyzing the predicted genes with OrthoFinder resulted in 2,337 single-copy orthologous genes that were aligned for calculating a phylogenetic tree with Urocystis primulicola as outgroup (Fig. 1). This shows that Pseudozyma is a polyphyletic genus, validating reassignments of species to other genera, e.g., Anthracocystis, Sporisorium, and Moesziomyces [14].

Fig. 1
figure 1

Phylogenomic tree of Ustilaginales and annotation of mating genes. The phylogeny was calculated based on 2,337 single copy orthologous genes. Urocystis primulicola (Urocystidales) was specified as outgroup. Core mating genes and alleles present are highlighted. Pheromone nomenclature is given in dependence to pheromone receptor compatibility (e.g., MFaX.1 is compatible with PRA1). Pseudozyma hubeiensis (strain NBRC 105055) was not included in the calculation of the tree but added to the figure to visualize the mating genes. 1 Tetrapolar [28], 2 Bipolar [29], 3 Tetrapolar [2], 4 Bipolar [30], 5 Bipolar [31]

Full size image

Genomes of Pseudozyma show functional-appearing mating-type loci

Manual annotation by multiple sequence alignments revealed that all genomes in the study contained putative functional genes coding for one pheromone receptor and one or two pheromones, as well as for both homeodomain transcription factor subunits (bEast/bWest) (Fig. 1).

Almost all Pseudozyma genomes contained one pheromone receptor allele and two different pheromones at the PR locus (Fig. 1). The only exception was Kalmanozyma brasiliensis that only contained one putatively functional pheromone copy, with the other one lacking the stop codon. Some sexual species (Ustilago hordei, Ustilago tritici, and Ustilago maydis), also showed a second pheromone gene that seemed to be pseudogenized (i.e., strongly resembling the overall sequence of the respective pheromone, but lacking the stop codon). Other species, such as Sporisorium reilianum, Sporisorium sorghi, Sporisorium scitamineum, and Testicularia cyperi, maintained only one pheromone without showing evidence for a pseudogenized pheromone. For Pseudozyma hubeiensis, strain SY62 carried a pheromone receptor type 3 (PRA3) and pheromones compatible with PRA1 and PRA2, and strain NBRC 105055 carried a PRA1 receptor and pheromones compatible with PRA2 and PRA3.

We were able to resolve the allele of each receptor and pheromone gene based on phylogenetic analyses (Supplementary Figs. 1 and 2), and therefore identified the mating-type for each strain based on the receptor. For all genomes, we identified that the pheromone receptor coding genes belonged to one allele, while the pheromone genes only were compatible to other alleles.

The HD locus, including functional domains of the genes (Supplementary Tables 1 and 2), could also be annotated for all species and was structured in the same way in almost all genomes, with the genes being adjacent and divergently transcribed (Fig. 1). In Kalmanozyma brasiliensis, the structure of the HD locus deviated, as both genes were located on the same strand. The two sequenced Pseudozyma hubeiensis strains showed differences in the HD locus genes that imply different alleles of the bEast and bWest genes. The genes of the HD locus were well-conserved within the Ustilaginaceae, while the similarity to the genes from Ustilago maydis decreased outside of the family (Supplementary Table 1).

Additionally, we assessed bipolarity and tetrapolarity in the species (Fig. 1). Both mating-type loci were located on the same scaffold in Testicularia cyperi, Tranzscheliella williamsii, and Urocystis primulicola (genome was sequenced from a culture containing both mating-types) with a genomic distance of less than 520 kb. For non-chromosomal assemblies, the determination of tetra- and bipolarity remains challenging if flanking regions of the mating-type loci were not correctly assembled and different scaffolds could be part of the same chromosome. Addressing this, telomeres were predicted for the scaffolds containing both mating-type loci (for comparative purposes, we included chromosomal assemblies). Moesziomyces antarcticus, Sporisorium graminicola, and Kalmanozyma brasiliensis showed telomeres at one or both scaffolds containing mating-type loci on both ends (Supplementary Table 3), establishing the tetrapolarity of these species. For Pseudozyma hubeiensis and Pseudozyma/Anthracocystis flocculosa, only one telomere could be annotated on the distal end of one mating scaffold, not providing information exceeding the minimal total flanking regions. Adding on this, the PR locus in Pseudozyma/Anthracocystis flocculosa was annotated at the ends of two different scaffolds, split between the pheromone receptor and the pheromones.

Sporisorium sorghi and Ustilago tritici were assessed to be bipolar, as the loci were located on different scaffolds with minimal total assembled flanking regions smaller than 150 kb. Although Ustilago hordei, Ustilago tritici, and Sporisorium scitamineum were previously identified as bipolar species [29,30,31], we could not retrieve this result in our study due to assembly quality. Farysia itapuensis, Moesziomyces parantarcticus, Pseudozyma hubeiensis, Pseudozyma thailandica, and Pseudozyma tsukubaensis could be either pseudobipolar (mating-type loci located on the same chromosome but inherited independently) or tetrapolar, as their mating-type loci were located on two different scaffolds and the minimal total assembled flanking regions were between 150 and 520 kb. All other species are tetrapolar according to definition (located on two different scaffolds and more than 520 kb distance minimal total assembled flanking regions or located on two different scaffolds and tetrapolar according to previous studies [2, 28]. Many of the species in our study that only had one pheromone or a pseudogenized pheromone are bipolar. However, Farysia itapuensis could be tetrapolar, as well as Kalmanozyma brasiliensis. Ustilago maydis, which also contained a pseudogenized pheromone, is a known tetrapolar species [32].

Pseudozyma species have the same genomic capability of meiosis as sexual species of Ustilaginales

Most core meiosis genes from Ustilago maydis were detected and identified as potentially functional in all tested species using manual annotation with BLAST and multiple sequence alignments (Fig. 2, Supplementary Table 1). Potentially functional gene models were lacking in Moesziomyces aphidis (SMC1, SMC2) due to gaps in the assembly within the potential gene models. However, the parts of the gene that flanked the gap could be aligned with the reference gene and coding sequence of Ustilago maydis without premature stop codons. Treating the gaps as introns, a gene model could be produced. In Ustilago hordei and Sporisorium sorghi, MER3 and SPO11, respectively, were located at the ends of scaffolds and could therefore not be modeled completely, despite being detected in the genomes. In the outgroup species Urocystis primulicola, SMC6 and SPO11 could not be annotated, as even extensive tblastn searches did not produce an alignment specific to these genes. In summary, all Pseudozyma species besides Moesziomyces aphidis produced reliable gene models for core meiosis genes, resembling sexual species in the Sporisorium and Ustilago genera, and other, less closely related sexual species, such as Testicularia cyperi and Tranzscheliella williamsii. Alignments with the meiosis specific genes MER3, MSH4, REC8, and SPO11 produced gene models in all Pseudozyma and most sexual species in Ustilaginales that were uninterrupted by stop codons. Meiosis gene loci were syntenic between the analyzed species corresponding to their phylogenetic relations, showing little rearrangements between closely related species. No rearrangements could be specifically linked to species that have been described to be asexual (Supplementary Fig. 3).

Fig. 2
figure 2

Heatmap showing presence of potentially functional core meiosis genes from Ustilago maydis in other species. Curated gene models were created by multiple sequence alignments

Full size image

To further assess possible functionality of meiosis and mating gene models, a HMMER protein domain search was performed on the translated coding sequences of the genes using the Pfam database. Compared to Ustilago maydis data from Pfam and InterPro as a reference, the gene models produced in this study show all specific functional domains of the respective reference genes (Supplementary Tables 1 and 2). The location of the domains in the genes was syntenic to their homologs in Ustilago maydis in all tested strains. However, the SMC3 gene model in Moesziomyces aphidis and the SMC4 gene model in Urocystis primulicola each showed a shortened SMC_N-domain. Despite these, there were no length or location irregularities in the functional domains of the annotated genes in the Pseudozyma genomes.

We then investigated the relaxation of selection on genes involved in or specific to meiosis in Pseudozyma using RELAX. The program compares selection in a test group (Pseudozyma) to a reference group (sexually reproducing species) and computes a coefficient k which suggest relaxation (k < 1), no change (k = 1), or intensification (k > 1) of selection in the test group (Pseudozyma). We detected intensified selection on 13 genes that are not specific for but involved in meiosis (Table 2). These genes are known to be involved in processes like mitosis and DNA repair additionally to meiosis (Table 3). Especially strongly intensified selection could be detected for RAD51, that is involved in DNA repair and recombination. Nine other genes showed no difference in selective pressure compared to sexual species, including MSH4 and REC8, which are meiosis-specific genes. Slightly relaxed selection (k = 0.91) was observed for MER3, a meiosis specific gene. SPO11, which is essential for meiosis by creating double strand breaks during crossing-over [33], showed more strongly relaxed selection (k = 0.59). However, SPO11 could not be detected in the sexual outgroup species Urocystis primulicola.

Table 2 Selection on meiosis genes in Pseudozyma compared to sexual Ustilaginales
Full size table
Table 3 Core meiosis genes from Ustilago maydis and their function during meiosis
Full size table

Discussion

To assess the capability of mating and meiosis of members of the asexual (former) genus Pseudozyma, we used available and newly sequenced genomes of members of the order Ustilaginales. These genome sequences enabled us to support the recent, marker-based phylogenies that were used to position Pseudozyma species in other genera [14] with a multigene phylogeny. The overall topology of Ustilaginales as proposed by Begerow et al. [9] or Wang et al. [14] received higher support through our phylogenomic approach (Fig. 1), and most clades were well-resolved. More recent phylogenomic studies on Ustilaginales showed similar results as ours [37, 38]. However, the placement of the Pseudozyma/Anthracocystis flocculosa strain sequenced in our study seems to deviate from the placement of the strain PF-1, which clustered as a sister group to other Ustilaginales species in a phylogenomic study by Benevenuto et al. [38]. The strain utilized in our study clustered more closely to Sporisorium, which would be expected for Anthracocytis [14].

The whole genome data revealed no tendency for Pseudozyma species to harbor smaller or bigger genomes compared to their parasitic relatives (Table 1). The number of predicted genes and low amount of introns per gene was also typical for Ustilaginales and does not differentiate Pseudozyma species from plant-parasitic species [30, 39]. Furthermore, members of Pseudozyma did not show changes in their genetic makeup regarding meiosis and mating, two important keys to the parasitic life cycle that is typical for smut fungi.

Mating

We were able to detect the core mating genes of both the PR locus and the HD locus in the Pseudozyma genomes and assess the functionality of the genes by comparing them to the mating genes of known parasites (Fig. 1). Our findings support previous detections of mating genes in some Pseudozyma strains that have been published [10, 40] with the information from newly sequenced Pseudozyma species (Fig. 1), assignment of mating-types (Fig. 1, Supplementary Figs. 1 and 2) as well as an assessment of possible functionality (Table 2, Supplementary Tables 1 and 2, Supplementary Fig. 3).

However, mating genes have previously been detected in representatives of the putatively asexual human pathogenic species of Malassezia (Malasseziales) [19], which could imply that mating related genes are maintained despite the potential loss of sexual reproduction. Assuming that members of Pseudozyma are able to mate, we showed that they would be self-sterile (heterothallic), containing either one or two pheromones compatible with other mating-types but not their own receptor, as all other known members of Ustilaginaceae [41]. In the two Pseudozyma hubeiensis strains, we were able to identify two different alleles for the pheromone receptor gene, inter-strain compatible pheromones, and divergent HD genes, which would allow those two strains to conjugate, enter, and maintain filamentous growth. This strain pair could therefore be a candidate for culture-based mating assays to show the capability of sexual reproduction independently of the unknown potential host plant.

In Kalmanozyma brasiliensis, the genes at the HD locus were organized adjacently and in the same orientation rather than being divergently transcribed, which is a deviation of the usual organization in Basidiomycota [41] and in all other species in this study. Nevertheless, both genes were annotated with full functional domains and are therefore potentially functional in sexual reproduction, i.e., potential filamentous growth that is maintained by the HD dimers [8].

The presence of two different pheromones in most analyzed Pseudozyma genomes hints to the existence of three different pheromone receptor alleles in these species, as found in many other species of Ustilaginaceae [6]. The possible pseudogenized pheromone in Kalmanozyma brasiliensis, as well as in some sexual species of Ustilaginales, is congruent to the situation in Ustilago maydis that displays a tetrapolar and biallelic mating system with a pseudogenized pheromone [5, 32]. Non-chromosomal assemblies of most of our data do not allow a thorough analysis of bipolarity and tetrapolarity, which is best shown in the examples of Farysia itapuensis and Pseudozyma/Anthracocystis flocculosa, where even the PR locus seems to be incorrectly assembled and split on the ends of two different scaffolds. This displays the need for more long-read sequencing data in the Ustilaginales order to study mating system evolution, also taking into account the accumulation of transposable elements that is expected for bipolar species evolving from the tetrapolar ancestral state [29, 31, 41], as recently shown in the anther smut genus Microbotryum (Microbotryales) [42]. Nevertheless, our data support the plesiomorphic situation with three alleles of the PR locus as previously suggested [6] and hint towards a situation of scattered bipolarity and tetrapolarity. This would mirror the situation in the lineage of Microbotryum, in which multiple independent events are responsible for bipolar mating systems in most species [42, 43].

Meiosis

We were able to detect and annotate all core meiosis gene as defined for Ustilago maydis [19] in the genomes from Pseudozyma strains (Fig. 2, Supplementary Table 1) and showed that most of them displayed no relaxed selection in the examined genomes (Table 2). In addition, all core meiotic genes have shown syntenic locus organization in closely related species throughout all meiotic functions (Supplementary Fig. 3) and maintained functional protein domains (Supplementary Tables 1 and 2). The MSH3 genes appeared shortened in some species of Pseudozyma as well as the sexually reproducing species Testicularia cyperi and Urocystis primulicola but maintains a full-length conserved functional domain (Supplementary Table 1). The maintenance of meiosis genes strongly suggests the relevance of meiosis in Pseudozyma, as it is likely that genes that are not utilized in an organism will be lost or pseudogenized eventually [44]. Another possibility would be a neofunctionalization, which would include increased or delayed mutation rates in these genes. In this case, we would expect a gain or loss of functional domains in the genes and a striking difference in gene conservation for Pseudozyma species compared to sexual species, as well as genomic rearrangements at the meiosis gene loci that cannot purely be explained by species borders. The comparison of selection strength on the meiosis-specific genes SPO11 and MER3 showed a relaxation of selection in the Pseudozyma group (Table 2) compared to the other species. SPO11 is widely considered to be essential for meiosis [33], although it has been lost in a lineage of amoeba that performs meiosis [45, 46] and could not be successfully annotated in the sexually reproducing outgroup species Urocystis primulicola. This could be a sign that it is non-essential in smut fungi, but, more likely, the sequence conservation is less striking than in other meiosis-related genes and the gene could not be identified by our strategy. Several genes showed intensification of purifying selection, but these genes are expressed during meiosis and mitosis. In species reproducing only asexually, genes involved in mitotic division may be under stronger selection. Intensification of selection in these genes could be caused by their role in mitotic division. Meiosis-specific genes REC8 and MSH4 and other meiosis-related genes showed no change in selection in Pseudozyma compared to other species, leading to the conclusion that the biological function of meiosis is conserved over all analyzed Ustilaginales, as it is in eukaryotes in general [20, 47, 48]. In contrast, Malassezia species lack some relevant core genes such as MSH5, or meiosis-specific genes MSH4 and MER3, while Pseudozyma species typically retained all 20 core meiosis genes found in Ustilago maydis [19, 20, 48]. Thus, we conclude that the analyzed species of Pseudozyma seem to be capable of meiosis, as members of the genus Malassezia have probably lost a considerable part of core meiosis genes, while still showing some evidence for recombination and hybridization [19, 49]. If there was a loss of function in meiosis genes in Pseudozyma, we hypothesize it to be on protein-structure level and probably very recent, as the functional domains of the genes are conserved and seem functional. Additionally, such a change would have occurred multiple times throughout evolution while maintaining syntenic locus organization without any hints for increased genomic rearrangement activity around the core meiosis genes.

Conclusion

The genetic makeup of Pseudozyma species in mating and meiosis loci suggests not only that these species are able to sexually reproduce, but, given the close linkage of sexual reproduction and parasitism in Ustilaginales, could colonize plants causing the smut syndrome. This is supported by core effector genes that could be annotated for some of the species included in this study, and by the functional conservation of the Pep1 effector that is responsible for establishing interaction zones with the host plant in Ustilago maydis [18]. It could be possible that most species labeled as Pseudozyma can be assigned as the yeast stage of known parasites utilizing genomic and other molecular data in the future [15, 50], or new collection trips may reveal their parasitic stage of so far unknown host plants [15, 17]. A different scenario would be that these smuts cause little to no symptoms when colonizing the host plant, maybe as endophytes. That can make it difficult to sample the sexual stage, although some Pseudozyma species can be found in association with plants [13]. Another possibility are parasitic species with a strongly dominating yeast stage, as previously suspected for the genus Moesziomyces [15, 51]. This would, much like the loss of sexual stages, have evolved multiple times throughout the phylogenetic tree of the Ustilaginales, but it might be favorable for the fungi to avoid immune responses of the host, while the complete loss of sexual reproduction is characterized by a loss of genetic variability which would probably lead to extinction in the long run [52].

Material and methods

Sequenced strains and incorporated published genomes

For this study, we sequenced 11 genomes from Ustilaginales species (Table 1). For each species, a haploid strain was selected. All strains were obtained from existing pure cultures deposited at the fungal culture collection of the Westerdijk Fungal Biodiversity Institute (CBS). Five of these strains belong to species in the former Pseudozyma genus (Pseudozyma/Antracocystis flocculosa, Moesziomyces parantarcticus, Pseudozyma pruni, Pseudozyma tsukubaensis, Pseudozyma thailandica). We sequenced the genomes of additional strains of Ustilago hordei, Ustilago tritici, Sporisorium scitamineum, and Ustilago maydis to be able to compare results with results from similar quality sequences of intensively studied parasitic species. Additionally, we sequenced genomes of the so-far unsequenced species Sporisorium sorghi, Tranzscheliella williamsii, and Farysia itapuensis (Farysizyma itapuensis). For further comparison of sexual and asexual species, we incorporated published genomes of former Pseudozyma species (Kalmanozyma brasiliensis, Moesziomyces aphidis, Moesziomyces antarcticus, Pseudozyma hubeiensis, Sporisorium graminicola) and sexual species (Sporisorium reilianum, Sporisorium scitamineum, Testicularia cyperi) in the analysis, and used a sexually reproducing representative from the sister group Urocystidales (Urocystis primulicola, strain sequences from a culture containing cells from both mating-types) as outgroup. Moesziomyces aphidis and Moesziomyces antarcticus were classified in the Pseudozyma group despite the possibility of conspecificity with sexual morphs and the recent discovery of a teleomorph, respectively, to examine them for possible signs of reduced sexual reproduction as proposed for the genus [51].

DNA extraction

Ustilaginales strains were incubated in liquid medium in a 250 mL Erlenmeyer flask at 25 °C. After 48–72 h, we retrieved the cells, then snap froze them in liquid nitrogen. We extracted genomic DNA using the CTAB method [53, 54]: We used a sterile mortar and pestle to disrupt the frozen cells, then rapidly transferred the resulting fine powder to a 1.5 mL microfuge tube and added DNA lysis buffer (100 mM Tris (pH 8.0), 20 mM EDTA, 2% CTAB, 1.2 M NaCl, and 0.1% β-mercaptoethanol) preheated in a 60 °C water bath. We vortexed the sample thoroughly and incubated it at 60 °C for 1 h. The mixture was cooled to room temperature, then admixed with an equal volume of phenol: chloroform: isoamyl alcohol (25:24:1, v/v/v), and centrifuged at 10,000 rpm for 15 min. We transferred the supernatant to a new microtube, and added an equal volume of cold absolute isopropanol for precipitating total DNA at -20 °C for 20 min; we then centrifuged the mixture at 10,000 rpm for 10 min. We washed the pellet twice with 75% ethanol and centrifuged it at 10,000 rpm at 4 °C for 10 min. We discarded the supernatant, and dried the pellet at RT. We resuspended the DNA pellet in 500 μl high salt TE (10 mM Tris (pH8.0), 2 mM EDTA, 1 M NaCl) with 2.5 µl RNase A at 60 °C for 1 h. We added an equal volume of chloroform:isoamyl alcohol (24:1, v/v) and centrifuged the sample at 10,000 rpm for 15 min. We transferred the supernatant to a new microtube and added 2 volumes of precooled ethanol. We inverted and then centrifuged the tube at 10,000 rpm for 10 min to pellet the DNA. We removed the supernatant and dried the pellet at RT, before resuspending it in 100 µL TE buffer. We stored the extracts at − 4 °C for immediate use or at − 20 °C for long-term storage.

Genome sequencing and assembly

All sequencing libraries were constructed using the Illumina TruSeq library kit. Paired-end sequencing (150 bp/read) was performed using an Illumina NovaSeq 6000 sequencer.

We trimmed adapters of the raw reads and filtered low-quality sequences using fastp v0.19.5 [55] with default options. We then de novo assembled the genome sequences then with SPAdes v3.13.1 [56] using the “– careful” and “– cov-cutoff auto” parameters.

We applied BUSCO version 5.1.3 [57] to assess genome completeness using the included Basidiomycota lineage data (basicdiomycota_odb10) as a reference.

Phylogenetic tree reconstruction

We predicted genes in assembled genomes using AUGUSTUS version 3.3.2 [58] pre-trained for Ustilago maydis and obtained the corresponding translated sequences as fasta files with an integrated script. We identified orthologous genes using OrthoFinder version 2.3.1 [59] and individually aligned the resulting 2,337 single copy orthologuous sequences present in all species using MAFFT version 7.427 [60]. We concatenated the alignments with the seqkit package [61] and calculated a multigene phylogeny in RaxML version 8 (PROTGAMMAWAG substitution model, rapid Bootstrap analysis, 500 runs [62]), specifying Urocystis primulicola as outgroup.

Manual meiosis and mating gene annotation

To identify core meiosis and mating genes, we retrieved reference genes from Ustilago maydis (meiosis genes, Table 3) and other species (mating genes, Supplementary Table 4) from NCBI database as whole gene sequences, coding sequences, and amino acid sequences. We applied the tblastn algorithm version 2.4.0 [63, 64] to identify these reference genes in the genomes by curating the alignments manually (pre-filtered for e ≤ 0.0001). We performed multiple sequence alignments of the candidate loci including their immediate flanking regions using MAFFT version 7.427 [60] with the full gene and coding nucleotide sequences of the respective reference genes. Based on these alignments, we performed several translation steps and multiple alignment steps in MEGA version 5 [65], utilizing the integrated MUSCLE algorithm [66] to manually model start and stop codons as well as introns and exons based on the reference genes’ reading frames and local gene prediction with AUGUSTUS online tool pre-trained for Ustilago maydis [67]. To measure the conservation between these genes, we used blastp version 2.12.0 + [63, 64] to align the translated coding sequences of each species’ meiosis and mating genes with the respective gene in Ustilago maydis.

We extended the identification of the mating genes to a second sequenced strain of Pseudozyma hubeiensis (strain NBRC 105055, Accession GCA_001736105.1, 55) to assess the possibility of two different mating-types for this species.

Assignment of mating-types and systems

After modeling the pheromone receptor and pheromone genes, we translated the sequences into amino acid sequences. For the PRA and MFa genes, respectively, we aligned the modeled genes together with published, mating-type assigned sequences from NCBI database (Supplementary Table 4) using MAFFT [60] and calculated gene trees using RAxML v8 (PROTGAMMAWAG substitution model, rapid Bootstrap analysis, 1000 runs [62]). From the location of the modeled genes in the tree clusters, we inferred the alleles of the genes. For a better visualization, we recalculated the trees using only gene models from the studied genomes after assignment of mating-types.

We assessed the PR and HD loci linkage by i) scanning the scaffolds containing mating genes for telomeric regions and ii) calculating the minimal total assembled flanking regions between the loci if they were located on different scaffolds, assuming that the scaffolds could be connected to form a chromosome (in absence of telomeric repeat sequences). We predicted telomeres by scanning the output of a Tandem Repeats Finder version 4.10.0 [68] analysis (recommended parameters) for the telomeric repeat pattern known for Ustilago maydis TTAGGG [69] at the beginning or end of scaffolds.

We assumed non-linked mating-type loci (tetrapolar) when the minimal total assembled flanking regions were larger than 520 kb (largest known distance of mating-type loci in a bipolar Ustilaginomycete (Ustilago hordei [29, 41], excluding flanking regions that contained a telomeric sequence) or one or both scaffolds containing the mating-type loci were assembled with both telomeres. We assumed a potential for tetrapolarity for regions of a size between 150 and 520 kb, and a potential for bipolarity for regions shorter than 150 kb (distance of mating-type loci in bipolar species such as Malassezia sympodialis [19]). We assigned bipolarity only when the loci were located on the same scaffold with a distance of less than 520 kb [28, 38].

Conservation of meiosis genes

To assess the functionality of meiosis and mating genes, we detected functional domains using the standalone Pfam database v 33.0 [70] and HMMER version 3.3 [71]. We analyzed the resulting annotated protein domains in comparison to the respective reference genes from Ustilago maydis in the Interpro [72] and Pfam [70] online databases. We checked functional protein domains for full length and relative location within the sequence and performed alignments of the functional domains of all species with those of Ustilago maydis using blastp version 2.12.0 + [63, 64].

We established conservation of meiosis gene loci over synteny plots of the meiosis genes and 10 kb flanking regions upstream and downstream of the respective genes using EasyFig version 2.2.2 [73] between closely related species.

Furthermore, we compared possible relaxation of selection in the modeled meiosis genes of Pseudozyma (test set) to their homologs in sexually reproducing species (reference set) using RELAX (branch site model, version 3.1.1 [74]). Using gene-coding sequences and a species phylogeny, RELAX detects relaxation or intensification of selection between two different taxa by comparing the distribution of the ω values (dN/dS, ratio of non-synonymous substitutions and synonymous substitutions) estimated from a random effects branch site model (BS-REL) and classified in three categories (purifying, neutral and diversifying selection) from the test set in relation to a reference set. To reduce the complexity of the model, the values of ω in the test branches are set as ωT = ωRk, with k the selection intensity parameter, so that k < 1 indicates relaxed selection, k = 1 no change in selection and k > 1 intensification of selection in the test taxa. A likelihood ratio test is performed to assess whether the ω distributions of test and reference branches are significantly different, meaning that the test set would be under relaxed or intensified selection compared to the reference. We conducted this analysis only for the meiotic genes, because the different PR-locus mating-type alleles have evolved before speciation, reducing the sample size to one specific mating-type and mating genes could be influenced by loss of the third allele in some species, while the HD locus genes show a too diverse allelicity in different species.