Parallels have been drawn between the evolution of nonrecombining regions in fungal mating-type chromosomes and animal and plant sex chromosomes, particularly regarding the stages of recombination cessation forming evolutionary strata of allelic divergence. Currently, evidence and explanations for recombination cessation in fungi are sparse, and the presence of evolutionary strata has been examined in a minimal number of fungal taxa. Here, the basidiomycete genus Microbotryum was used to determine the history of recombination cessation for loci on the mating-type chromosomes. Ancestry of linkage with mating type for 13 loci was assessed across 20 species by a phylogenetic method. No locus was found to exhibit trans-specific polymorphism for alternate alleles as old as the mating pheromone receptor, indicating that ages of linkage to mating type varied among the loci. The ordering of loci in the ancestry of linkage to mating type does not agree with their previously proposed assignments to evolutionary strata. This study suggests that processes capable of influencing divergence between alternate alleles may act at loci in the nonrecombining regions (e.g., gene conversion) and encourages further work to dissect the evolutionary processes acting upon genomic regions that determine mating compatibility.

Mechanisms responsible for the evolution of large nonrecombining regions on sex chromosomes receive extensive consideration in plant and animal systems, but the comparison to recently discovered similarities in mating-type chromosomes of fungi may shed new light on this phenomenon (Fraser et al. 2004; Bergero and Charlesworth 2009). Sex chromosomes in plants and animals have evolved from an autosomal pair by the expansion of the nonrecombining region around sex-determining genes (Bergero and Charlesworth 2009). Explanation for the expansion of nonrecombining regions in multiple steps, forming “evolutionary strata” for the age of linkage to the sex-determining locus in plants and animals (Lahn and Page 1999), has emphasized the recruitment of genes determining sexually antagonistic traits (i.e., traits beneficial in one sex and deleterious in the other) (Rice 1987; Bergero and Charlesworth 2009; but see Ironside 2010). Evidence for evolutionary strata takes the form of discrete blocks of the chromosomes, extending outward from sex-determining genes, where homologous sequences ceased recombination and homologous alleles started diverging from each other at different historical time points.

Similar to plants and animals, recombination suppression on fungal mating-type chromosomes also has been suggested to proceed in successive evolutionary strata (Fraser et al. 2004; Menkis et al. 2008; Votintseva and Filatov 2009). In fungi, evolutionary strata would be extending outward from the locus that determines mating types, and it remains unclear whether evolutionary forces responsible are similar to what has happened in plant and animal sex chromosomes. Fungi of opposite mating types exhibit few known ecological differences other than during the processes of syngamy (i.e., fusion of haploid gametes giving rise to fertilization) and formation of the zygote (Abbate and Hood 2010). Even when there is anisogamy (i.e., difference in size between two haploid gametes) in fungi, it is not associated with genetic determinants of mating types (Billiard et al. 2011). Therefore, the classic model of a multistep expansion of the nonrecombining regions by the recruitment of genes determining sexually antagonistic traits seems less plausible for fungi given our current state of knowledge. Alternative models, such as the role of accumulating recessive deleterious or lethal mutations, have received relatively little consideration compared to the role of sexually antagonistic traits but may, nonetheless, be relevant across organisms (Ironside 2010).

Some causes for an initial region of suppressed recombination around mating-type loci have been suggested for fungi, but these do not contribute directly to an expansion of the nonrecombining region. In some basidiomycete fungi, recombination suppression has been suggested to originate through linkage of the pheromone/pheromone receptor locus that controls gamete recognition to a second locus composed of homeodomain genes, which controls postsyngamy viability (Bakkeren and Kronstad 1994; Fraser et al. 2004). This linkage of loci involved in gamete recognition and postsyngamy compatibility is responsible for the derivation of a bipolar mating-type segregation from a tetrapolar system that has two independently segregating traits (but see James et al. 2006). Also, fungi that mate via automixis (intrameiotic tetrad mating) frequently exhibit linkage of mating type to the chromosome's centromere (Zakharov 1986; Hood and Antonovics 2000; Menkis et al. 2008). However, these singular linkage events would not be expected to create successive evolutionary strata of recombination cessation. In fact, in the three fungi where evolutionary strata have been posited, none provides a consistent pattern for discrete regions where the age of allele divergence is a positive function of physical distance from the pheromone receptor locus (Fraser et al. 2004; Menkis et al. 2008; Votintseva and Filatov 2009; Ellison et al. 2011). In these prior studies, both continued processes of localized rearrangements and the homogenization of alleles by gene conversion, rather than by meiotic recombination, have been suggested to obscure any clearer signals of evolutionary strata (Fraser et al. 2004; Ellison et al. 2011). Therefore, additional studies of fungi are warranted that can contribute to the evidence of recombination suppression on mating-type chromosomes and to the evaluation of possible explanations for the origins of these structures.

The basidiomycete fungi in the genus Microbotryum represent a model system for numerous genetic and ecological studies, in particular regarding mating systems and mating-type chromosomes (Hood 2002; Hood et al. 2004; Giraud et al. 2008). In Microbotryum, mating is dependent on the union of haploid cells of two different mating types, named a1 and a2. Thus species of Microbotryum exhibit bipolar mating-type segregation (Kniep 1919), where it is assumed that the pheromone/pheromone receptor locus and the homeodomain locus are linked. Tetrad analysis and karyotype electrophoresis of Microbotryum lychnidis-dioicae revealed that the mating-type locus exhibits centromere linkage (Hood and Antonovics 1998) typical of fungi with an automictic mating system, and that the chromosomes containing this locus are of markedly different sizes (Hood 2002). In this species, the region of suppressed recombination between the mating-type chromosomes was originally estimated to extend to most of their lengths (Hood et al. 2004), whereas a more recent study argued that it encompasses a quarter of their lengths (Votintseva and Filatov 2009); resolving this discrepancy is the subject of ongoing research.

Devier et al. (2009) showed that the two pheromone receptor alleles of the a1 and a2 mating types have diverged from each other since before the split of the Microbotryomycetes from other basidiomycetes, representing the oldest trans-specific polymorphism known to date. The region of suppressed recombination in linkage to mating type in M. lychnidis-dioicae has been suggested to contain three evolutionary strata of different ages extending outward from the pheromone receptor region (Votintseva and Filatov 2009) using evidence based on DNA sequence divergence between alleles at mating-type linked loci. However, no physical order of the loci has been provided, and while not noted in that study, the variation in degree of allele divergence (i.e., as a surrogate measure of the age of recombination suppression) looked continuous rather than organized by discrete strata. In addition, the patterns of allele divergence in this species may not reflect the evolutionary histories of recombination suppression across related taxa. Abbate and Hood (2010) studied the phylogeny of alleles at one locus that is linked to mating type in M. lychnidis-dioicae, showing that this locus has experienced cessation of recombination with the mating type in multiple independent instances during the divergence of species in the genus Microbotryum. These results suggest that the nonrecombining regions of the mating-type chromosomes in Microbotryum species are experiencing dynamic evolutionary processes with regard to recombination suppression, which could help inform our understanding of genome structures determining compatibility for sexual reproduction.

In this study, we set out to use recently available genomic sequence data and phylogenetic methods to describe the genic content and history of recombination suppression associated with the mating-type locus in Microbotryum species. Recent work in the basidiomycete red yeasts Sporobolomyces sp. IAM 13481 and Sporidiobolus salmonicolor, also sexually reproducing species in the Microbotryomycetes, has identified genes localized in the mating-type locus that can be used for sequence similarity searches (Coelho et al. 2008, 2010). Sporobolomyces sp. IAM 13481 was chosen because it is the closest relative of Microbotryum for which a genome has been sequenced. Our specific goals were therefore to (1) identify genes in proximity to the mating pheromone/pheromone receptor locus in Microbotryum and (2) assess whether genus-level phylogenetic patterns for the divergence of homologous alleles provide evidence consistent with the proposed age of linkage to mating type in the species M. lychnidis-dioicae.

Materials and Methods


The fungal species used in this study were collected as diploid teliospores from natural populations in North America and Europe (Table 1). Recent studies have aimed to define species of Microbotryum, which had been commonly referred to by the epithet M. violaceum (Kemler et al. 2006; Denchev 2007; Lutz et al. 2008; Denchev et al. 2009). In general, the anther-smut fungi found on different host species form highly specialized and reproductively isolated Microbotryum species (Le Gac et al. 2007). For each species, haploid cultures of opposite mating types and originating from the same meiosis were obtained by micromanipulation of the postmeiotic yeast-like sporidia, which were grown on potato dextrose agar (Difco) and stored frozen under desiccation.

Table 1.  List of isolates used in this study and relevant information concerning them.
Species names1Host speciesCode used in the phylogenies2Origin
  1. 1As described elsewhere (Kemler et al. 2006; Denchev 2007; Lutz et al. 2008; Denchev et al. 2009). “M.” is for Microbotryum.

  2. 2Codes are abbreviations of the host species names and are used in the phylogenies.3The taxonomy of the species found on Dianthus is still under study and strains with the same species names (found on the same host) are sometimes different species.

M. violaceum sensu lato Atocion rupestris ArChambery, France, 2002
M. carthusianorum Dianthus carthusianorum Dc1 Sestriere, Italy, 2002
M. dianthorum 3 Dianthus neglectus Dn1St. Anna, Italy, 2003
M. dianthorum 3 Dianthus neglectus Dn2 Valle de Pesio, Italy, 2003
M. shykoffianum Dianthus sylvestris DsCesana Tor, Italy, 2002
M. violaceum sensu lato Lychnis flos cuculi Lfc Great Cumbrae Is, United Kingdom, 2000
M. violaceum sensu lato Lychnis flos jovis LfjValle de Pesio, Italy, 2003
M. saponariae Saponaria ocymoides So Cesana Tor, Italy, 2002
M. silenes-acaulis Silene acaulis SaValle de Pesio, Italy, 2003
M. violaceum sensu lato Silene caroliniana Sc Virginia Beach, Virginia, United States, 2004
M. silenes-dioicae Silene dioica Sd1Olivone, Switzerland, 2001
M. silenes-dioicae Silene dioica Sd2 France, 2002
M. violaceum sensu lato Silene italica SiItaly, 2006
M. lychnidis-dioicae Silene latifolia Sl1 Lamole, Italy, 2000
M. lychnidis-dioicae Silene latifolia Sl2Orsay, France, 2000
M. violaceum sensu lato Silene notarisii Sno Italy, 2006
M. violaceum Silene nutans SnuBois Carre, France, 2002
M. violaceum sensu lato Silene paradoxa Sp P. Nat. Gran Sasso, Italy, 2004
M. violaceum sensu lato Silene virginica SviCharlottesville, Virginia, United States, 2001
M. lagerheimii Silene vulgaris Svu Chambery, France, 2002

For each specimen, haploid cultures of each mating type were identified by pairing with cultures of known mating types and examining the conjugation response that is elicited by the alternate mating pheromone (Day 1979). Also, polymerase chain reaction (PCR) primers that discriminate between a1 and a2 pheromone receptors (Devier et al. 2009) were used to test for mating type following DNA extraction with the DNeasy Plant Mini Kit (QIAGEN, Valencia, CA). By these methods, alternate alleles at the mating pheromone/pheromone receptor locus define the mating types of the isolated haploid cultures. For fungi in general, both mating types produce a pheromone product and possess a linked receptor for detection of the complimentary pheromone (Bakkeren et al. 2008; Raudaskoski and Kothe 2010; Lee et al. 2010).


Loci and PCR primers were identified using the translated tBLASTx search algorithm ( to compare genomic scaffold 9 of Sporobolomyces sp. IAM 13481, which contains the mating pheromone/pheromone receptor locus, and scaffold 7, which contains the homeodomain locus, to available Microbotryum DNA databases; Sporobolomyces sp. IAM 13481 regions included a 35 kbp DNA fragment from base pairs 617,000 to 652,000 of scaffold 9 and 300 kbp DNA fragment of scaffold 7 from base pairs 1,062,000 to 1,277,000 (genome sequence at the Joint Genome Institute Microbotryum DNA databases included EST databases (Microbase (Yockteng et al. 2007)) and genome sequences at the Broad Institute (

Microbotryum DNA sequences identified by the tBLASTx were then used as BLASTn search queries for shotgun DNA sequence libraries from several Microbotryum species to generate among-species nucleotide sequence alignments, and PCR primer sites were identified from these alignments (Table S1). For longer loci like the homeodomain locus and STE20 gene homolog, multiple primers spanning the locus that amplify overlapping regions were designed. Primers were designed using the web-based program Primer3 ( using the default settings.

In addition, PCR primers designed by Votintseva and Filatov (2009) for loci belonging to the proposed evolutionary strata of M. lychnidis-dioicae were also used (Table S1). In brief, Votintseva and Filatov (2009) identified 11 loci as localized within the nonrecombining region according to segregation analysis in different strains of M. lychnidis-dioicae. These loci had variable levels of divergence between alleles from cells of a1 and a2 mating types (from 0 to 8.6% nucleotide divergence). Based on the levels of divergence between alleles linked to a1 and a2, Votintseva and Filatov (2009) proposed that these loci were distributed into two to three evolutionary strata of cessation of recombination, extending from the oldest stratum that contains the pheromone receptor itself. Among the 11 loci, nine were used that could be amplified in the different Microbotryum species.

Separate PCR reactions using DNA from cultures of each mating type for each Microbotryum species were run using standard conditions appropriate for the primer combinations (Table S1), with modifications involving touchdown PCR or lowering of the annealing temperature to obtain products across various Microbotryum species. PCR products were sequenced using Sanger technology.

DNA sequences from M. lychnidis-dioicae with similarity to sequences in the mating-type locus of Sporobolomyces sp. IAM 13481 were used for in silico prediction of the gene order and direction using Augustus (Stanke et al. 2004). The DNA sequence of Sporobolomyces sp. IAM 13481 (Coelho et al. 2008, 2010) was also reanalyzed using the same in silico methodology. Putative functions of the predicted proteins were found by comparing them to the Conserved Domain Database ( and by BLASTp against the NCBI database.


Extended contiguous DNA sequences, obtained by joining overlapping forward and reverse sequences were edited using the software Sequencher (Genes Codes Corporation, Ann Arbor, MI), and sequences from both a1 and a2 mating types of the various Microbotryum species were aligned using the software program Clustal X Version 1.6 (Thompson et al. 1997) and corrected visually. When DNA was too divergent in coding regions to be aligned using Clustal default parameters, protein sequences were deduced from the DNA sequence using Augustus (Stanke et al. 2004) and the back-translated DNA was used to construct the tree.

Phylogenetic analyses were conducted in MEGA5 (Tamura et al. 2011). The evolutionary history was inferred using the Neighbor-Joining method (Saitou and Nei 1987). The evolutionary distances were computed by the Kimura 2-parameter method (Kimura 1980) using the number of base substitutions per site. All positions containing gaps or missing data were excluded from the analyses. The bootstrap consensus tree inferred from 10,000 replicates is taken to represent the evolutionary history of the loci being analyzed (Felsenstein 1985).


Based on the phylogeny of each locus, a score was calculated that reflects how anciently it appears in linkage to mating type. Because the two alleles at the mating-type locus (i.e., the mating pheromone receptor alleles) are maintained across speciation events by strong balancing selection (Devier et al. 2009), alleles at loci linked to mating type are expected to diverge from each other, in contrast to alleles at loci that recombine with the mating type. The tree of the mating pheromone receptor (studied previously by Devier et al. (2009)) was chosen to represent the case of a gene in complete linkage to the mating type, whereas the tree of the gamma tubulin (studied previously by Le Gac et al. (2007)) was used to represent a gene in a recombining region. This expectation of divergence between homologous alleles due to cessation of recombination with mating type is the process and the nature of molecular data that underlie the evolutionary strata hypothesis. For the tree of the gamma tubulin, the mean nodal distance between all the taxon entries was computed. For each other locus, we identified the node with the most ancient linkage to the mating type (i.e., clustering of alleles by mating type rather than by species-of-origin) and computed the ratio of the largest genetic distance with the clade descendent from that node in the gamma tubulin tree over the largest genetic distance found in the gamma tubulin tree. In other words, the score was highest if linkage to mating type happened before the split of the most divergent species in the tree. If the score value is 1.00, then age of cessation of recombination with the mating type is ancestral to the genus Microbotryum; if the value is 0.00 then there is no instance where a1- and a2-derived alleles would have ceased recombining before the divergence of any two species considered in this study. The correlation between the phylogenetic tree score for the age of linkage to mating type and the amount of nucleotide divergence between homologous alleles was tested using Spearman's rho in PASW Statistics v18 (SPSS Inc., Chicago, IL).



Homologs to the majority of the genes previously identified within a 35-kbp region of the Sporobolomyces sp. IAM 13481 genome containing the mating pheromone/pheromone receptor locus were found in DNA sequence databases available for M. lychnidis-dioicae (Fig. 1). Homologs to the Sporobolomyces sp. IAM 13481 homeodomain genes HD1 and HD2 were also detected for M. lychnidis-dioicae but are positioned on a different scaffold from the one carrying the pheromone/pheromone receptor locus, as is the case with Sporobolomyces sp. IAM 13481 (Coelho et al. 2010). Assembly of the mating-type region in M. lychnidis-dioicae could, however, not be achieved to as great an extent as for Sporobolomyces sp. IAM 13481, possibly due to the presence of multiple transposable element copies that do not appear in Sporobolomyces sp. IAM 13481 when analyzed by the same in silico gene prediction methodology (Fig. 1). The a1 sequences could be assembled to a greater extent than the a2 sequences.

Figure 1.

Comparison between the gene synteny (with arrows indicating the direction of transcription) and composition of the mating-type loci a1 and a2 of Microbotryum lychnidis-dioicae with the locus in a1 of Sporobolomyces sp. IAM 13481. Contig numbers for the a1 correspond to the Broad institute code. Loci presented in Figure 2 can be found in the Microbotryum lychnidis-dioicae assembly but because they do not have matches in Sporobolomyces sp. IAM 13481 (to the exception of STE20 and ABC1) and the assembly does not give indications regarding their relative physical positions, they are not represented in this figure for clarity.

As evidence for the conserved structure of the mating-type region, the adjacency and directions of open reading frames of genes within contiguous sections of DNA sequence were most often the same in M. lychnidis-dioicae as in Sporobolomyces sp. IAM 13481 (six conserved adjacencies out of eight cases and seven conserved directions out of eight cases). Evidence that the genes in separate short assemblies shown in Figure 1 are linked to the mating pheromone/pheromone receptor locus in M. lychnidis-dioicae, as they are in Sporobolomyces sp. IAM 13481, was provided by divergence between alleles from a1- and a2-derived DNA from the same species (described below); however, the order of the separate assemblies could not be established. Also, STE20 and STE12 are mating-type specific in some species (e.g., Cryptococcus neoformans) but STE12 is not found in the Sporobolomyces sp. IAM 13481 assembly (Table S2). In addition, for the a1 pheromone precursor, which was identified by a conserved position next to the ABC1 gene homolog compared to Sporobolomyces sp. IAM 13481 and the presence of the characteristic C-terminal CAAX domain typical of fungal mating pheromones (Bölker and Kahmann 1993), PCR amplification was possible with template DNA of only one of the two mating types, suggesting its absence in cells of the opposite mating type or divergence at nucleotide positions targeted by the PCR primers. The pheromone precursor gene is present in three copies in Sporobolomyces sp. IAM 13481, but the copy lying between STE20 and KAP95 in Sporobolomyces was absent from the M. lychnidis-dioicae assembly that contained homologs of those genes. Additional sequences of putative genes were identified in the M. lychnidis-dioicae mating-type region that are lacking from Sporobolomyces sp. IAM 13481; these include transposable elements, 10 genes with predicted functions as well as some genes of unknown function. Table S2 details blast hits of these genes in different Basidiomycetes mating-type loci.


In addition to two of the genes identified by the methods discussed above (STE20 and ABC1), nine loci among the 11 previously proposed to belong to different evolutionary strata of cessation of recombination in M. lychnidis-dioicae (Votintseva and Filatov 2009) could be amplified for phylogenetic analyses across Microbotryum genus. The previously identified nine loci were located on different contigs in the M. lychnidis-dioicae assembly, thus, providing no information about their relative positions. Moreover, these nine loci were not detected in the Sporobolomyces sp. IAM 13481 assembly; therefore, they could not be shown in Figure 1.

Phylogenetic trees constructed for genes matching those in proximity to the mating-type locus of Sporobolomyces sp. IAM 13481 and for the loci previously characterized as belonging to evolutionary strata in M. lychnidis-dioicae revealed a wide range of divergence patterns for a1- and a2-derived DNA sequences (Fig. 2). As experimental control measures for extreme patterns of linkage histories to mating type, alleles for the mating pheromone receptor (STE3) showed an ancient and complete bifurcation according to mating types of the cells of origin (i.e., highly divergent alleles), whereas the gamma tubulin gene (i.e., a freely recombining region similar to the pseudo-autosomal region) produced a phylogeny reflecting the Microbotryum species tree with a1- and a2-derived DNA sequences from the same species being homozygous (i.e., identity between alleles). The calculated tree scores for the mating pheromone receptor and the unlinked gamma tubulin locus were 1.00 and 0.00, indicating ancestral linkage to mating type and continued recombination with mating type, respectively.

Figure 2.

Figure 2.

Phylogenies of loci hypothesized to belong to the mating-type chromosome and ranked according to the score of likeliness to be ancestrally linked to the mating type. Species name are coded as presented in Table 1. Scores of likeliness are indicated in bold after the name of the locus of each tree. The percentages of replicate trees in which the associated taxa clustered together in the bootstrap test are shown next to the branches (when >60%).

Figure 2.

Figure 2.

Phylogenies of loci hypothesized to belong to the mating-type chromosome and ranked according to the score of likeliness to be ancestrally linked to the mating type. Species name are coded as presented in Table 1. Scores of likeliness are indicated in bold after the name of the locus of each tree. The percentages of replicate trees in which the associated taxa clustered together in the bootstrap test are shown next to the branches (when >60%).

Figure 2.

Figure 2.

Phylogenies of loci hypothesized to belong to the mating-type chromosome and ranked according to the score of likeliness to be ancestrally linked to the mating type. Species name are coded as presented in Table 1. Scores of likeliness are indicated in bold after the name of the locus of each tree. The percentages of replicate trees in which the associated taxa clustered together in the bootstrap test are shown next to the branches (when >60%).

For the STE20 and ABC1 and nine loci previously assigned to evolutionary strata, divergence between alleles from a1 and a2 cultures from the same species revealed contrasting evolutionary histories with regard to mating-type linkage. The phylogenies of some loci suggest that recombination with mating type was ancestral to the Microbotryum genus and cessation of recombination was derived multiple times (e.g., loci 516 and 404 in Fig. 2); this pattern is similar to the previous finding with the locus MTL1 (Abbate and Hood 2010). The phylogenies of other loci suggested that linkage to mating type was ancestral to the Microbotryum genus, but no locus produced evidence of ancestral linkage to mating type as complete across all species as the pheromone receptor gene (Fig. 2). Ancestral linkage to mating type, but with some occurrence of derived allelic homogenization (i.e., a lineage with a more recent common ancestor for both alleles in cells of opposite mating types than observed for the genus as a whole), is suggested by DNA sequences from the most distantly related Microbotryum species being clustered by their mating type-of-origin, whereas alleles in a subsection of the tree being clustered by the species-of-origin (e.g., STE20 and ABC1 homologs in Fig. 2). Notably, the events of derived allelic homogenization for the STE20 and ABC1 homologs did not occur in the same species (Fig. 2), indicating independent departures from the pattern of ancestral mating-type linkage at these two loci.

Because more ancient cessation of recombination would produce higher calculated tree scores (i.e., being more like the tree produced by the mating pheromone receptor alleles, Fig. 2), a positive correlation might be expected between the tree score for a locus and the relative age of its proposed evolutionary strata. However, there was no significant relationship between the phylogeny scores and the suggested evolutionary strata calculated in M. lychnidis-dioicae (Votintseva and Filatov 2009). When considering the amount of sequence divergence between homologous alleles within the species M. lychnidis-dioicae as defined in Votintseva and Filatov (2009) rather than the strata categories, the correlation with phylogenetic tree scores was not statistically significant (Spearman's rho, r= 0.290, P-value = 0.45). The locus 457 could be considered an outlier, and if excluded with necessary caution over the interpretation of this result, the divergences between homologous alleles and tree scores were positively correlated (Spearman's rho, r= 0.779, P-value = 0.02). The two additional loci, STE20 and ABC1, with divergence at silent sites (used because these genes are being compared to noncoding loci from the previous study) of 1.0% and 6.5%, respectively, can be combined with the loci analyzed in Votintseva and Filatov (2009) for assessing the relationship to phylogenetic age of mating-type linkage (Fig. 3). Inclusion of these loci in the correlation test also produces a nonsignificant result (Spearman's rho, r= 0.49, P-value = 0.15); again caution is needed because STE20 and ABC1 were not part of the work on a strata assignment in Votintseva and Filatov (2009), and as genes rather than random noncoding loci may be subject to different selective pressures.

Figure 3.

Relationship between the phylogenetic age of linkage of loci to mating type across the genus Microbotryum and the percent divergence between homologous alleles in M. lychnidis-dioicae. Values on the y-axis were produced by Votintseva and Filatov (2009) to define evolutionary strata, as identified on the figure by dotted horizontal lines; divergence for the loci STE20 and ABC1 was calculate from the current study. The oldest stratum containing the pheromone receptor included loci with >9% divergence between homologous alleles; the next younger stratum includes loci with 5%–9% divergence; the youngest stratum includes loci with 1%–4% divergence; the pseudoautosomal region include loci with less than 1% divergence.

Because the homeodomain locus (important because it is the second mating-type-determining locus in most basidiomycetes) was neither placed relative to the pheromone receptor locus nor in a proposed evolutionary stratum of M. lychnidis-dioicae (Votintseva and Filatov 2009), nor was PCR amplification possible in many Microbotryum species, it is considered separately in this section. The divergence of alleles of the HD1 gene at the homeodomain locus does not match the ancient bifurcation seen for the pheromone receptor despite that linkage between the homeodomain and pheromone receptor has been assumed to provide the bipolar mating-type segregation throughout the Microbotryum genus. The overall structure of the homeodomain gene trees (Fig. 4) suggests ancient linkage to mating type, but the proximity of a1- and a2-derived alleles for HD1 of Microbotryum saponariae (So), Microbotryum dianthorum (Dn), and Microbotryum carthusianorum (Dc) (Fig. 4) indicates more recent common ancestors for the alleles from these species. For M. dianthorum and M. carthusianorum the a1- and a2-derived alleles for HD1 were identical through the length of this putative gene (723 bp), whereas in M. saponariae the alleles show 4% divergence in nucleotide sequence. The HD2 alleles for these species could not be obtained with the available PCR primers.

Figure 4.

Phylogenies of HD1 and HD2. Species name are coded as presented in Table 1. The percentages of replicate trees in which the associated taxa clustered together in the bootstrap test are shown above the branches (when >60%).


By combining an assessment of gene content in linkage with mating type and a phylogenetic approach to determine the histories of recombination cessation across the Microbotryum genus, this study provides insights into the evolution of suppressed recombination on chromosomes determining mating compatibility. Because the literature on sex chromosomes relates strongly to the idea of sexually antagonistic traits as a primary driver for recombination cessation (see Ironside 2010), and yet such antagonistic traits are less likely to play an important role in fungi, the interpretation of evidence for evolutionary strata in fungi presents a challenge for the comparison to canonical models. However, the similarity of biological function and the shared properties of the chromosomes related to the mating process across diverse organisms should encourage the further study of fungi for evolutionary genomics in sexual eukaryotes.


For multiple sequence assemblies of M. lychnidis-dioicae containing putative genes linked to the mating-type locus, both gene order and direction are remarkably similar to the mating-type locus of its Sporobolomyces sp. IAM 13481 relative (Coelho et al. 2008), which belongs to a different order (Sporidiobolales) within the Microbotryomycetes class (Sampaio et al. 2003). Yet, additional genes and transposable elements were identified in the assemblies of M. lychnidis-dioicae, which would suggest that the mating-type region encompassing the pheromone receptor, pheromone precursor, and immediately neighboring genes (e.g., STE20, etc.) is much larger than in Sporobolomyces sp. IAM 13481 (greater than 100 kbp in M. lychnidis-dioicae compared to a 35 kbp region in Sporobolomyces sp. IAM 13481). Whether genes and transposable elements have been acquired in M. lychnidis-dioicae or lost in Sporobolomyces sp. IAM 13481 would require broader phylogenetic studies that could reconstruct the common ancestor to these fungi. Knowledge of the presence and activity of transposable elements in Microbotryum is not new, in particular with the report of their transcription (Yockteng et al. 2007) and biased accumulation on the fungal mating-type chromosomes relative to the autosomes (Hood 2005). Yet, the location of multiple transposable element copies in such close proximity to the mating locus itself suggest possibly a more direct role in local restructuring of this genomic region than has been previously reported.

Linkage between the pheromone receptor and the homeodomain loci has been particularly important to the evolution of sexual systems in basidiomycete fungi (Bakkeren and Kronstad 1994), with these two loci determining pre- and post-syngamy compatibility, respectively. Even though homologous sequences to the homeodomain locus could be found in M. lychnidis-dioicae, the physical distance to the pheromone-receptor locus remains unknown, unlike in the case in Sporobolomyces sp. IAM 13481 where the distance is estimated to 1.2 Mbp (Coelho et al. 2010). The ancestry of bipolar mating-type segregation in the genus Microbotryum and its occurrence in Sporobolomyces sp. IAM 13481 suggest that the event establishing linkage between these two loci was ancient, and the inability to join these loci in a common assembly raises the possibility that the cessation of recombination may have encompassed a relatively large portion of the mating-type chromosome. In the red yeast S. salmonicolor, the mating system sometimes deviates from the strict bipolar behavior (“pseudo-bipolar”; see Coelho et al. 2010), which is due to rare recombination events between the homeodomain and the receptor loci. Yet, like the region proximal to the pheromone receptor gene, the homeodomain locus appears to have some localized rearrangements between M. lychnidis-dioicae and Sporobolomyces sp. IAM 13481. In particular, the gene XRN is 0.2 kbp from the homeodomain gene in M. lychnidis-dioicae but is 200 kbp away in Sporobolomyces sp. IAM 13481 and appears to be on the opposite side relative to the orientation of HD1 and HD2 sequences. This difference between M. lychnidis-dioicae and Sporobolomyces sp. IAM 13481 contributes to the growing number of observations of inversions of the HD1/HD2 locus or in the neighboring regions reported in other fungi (James et al. 2011; van Peer et al. 2011).


The evolutionary history of linkage to mating type within the genus Microbotryum revealed that alleles at loci expected to exhibit complete linkage, due to their proximity, do not produce phylogenies that are fully bifurcating according to mating type like the pheromone receptor. Instead, a range of patterns was observed, suggesting recombination with the mating type (i.e., the pheromone receptor) was ancestral for some loci and that there have been instances of derived linkage to mating type; whereas for other loci evidence suggests ancestral linkage to mating type and the derived resetting to a common allelic form in a subset of species (possible patterns diagramed in Fig. S1).

The scenario of derived linkage to mating type has been previously reported for another locus (Abbate and Hood 2010) and is consistent with the evolution of expanding recombination suppression on the mating-type chromosomes. However, a mechanism is less apparent for the latter scenario of ancestral linkage and derived homogenization of alleles at a mating-type-linked locus, as suggested by the phylogenies of the ABC1 and STE20 homologs. Similarly, the homeodomain alleles appear to have an ancestral linkage to mating type but some cases of allelic homogenization, even though there can be constraints against the homeodomain locus becoming homozygous throughout the length of HD1 and HD2. This constraint is because viability of the postsyngamy stage is achieved by the formation of a heterodimer between HD1 and HD2 from cells of different genotypes at these loci. In three species of Microbotryum, alleles at HD1 are homozygous or nearly so, with homogenization appearing to have occurred toward the a1-linked ancestral allele, suggesting the derived events of recombination or gene conversion. The sequence of HD2 could not be isolated from these species, and further studies are needed to ascertain whether HD2 alleles retain the variation necessary to achieve functional heterodimer formation with the ancestrally a1-linked HD1 allele.

One possible explanation for derived homogenization of alleles linked to mating type is a rare crossing over event between a given locus and the pheromone receptor that would produce a1 and a2 haploid genotypes in the population with the same allele by descent at the given locus and all distal loci up to the next crossing over point. Mating between such haploid genotypes would produce homozygous descendants at the given locus and reset the process of allelic divergence if linkage to mating type continued. Reinstitution of divergence following a derived common ancestor for alleles can be seen by the relatively small but positive measures of phylogenetic distance between the alleles from a1 and a2 cells, for example, from M. lychnidis-dioicae in the STE20 tree or M. lagerheimii in the ABC1 tree.

Gene conversion, the nonreciprocal transfer of allelic sequence (Szostak et al. 1983), is an alternative explanation that would produce a pattern similar to resetting homozygosity via recombination. In this case the homogenization would occur in a much more localized manner, in the few hundred basepairs of a gene conversion tract, and would not influence more distal loci relative to the pheromone receptor. Previously, gene conversion has been suggested as a potential influence upon the divergence of sex chromosomes (Charlesworth et al. 2005) and even as a cause of strata-like structure where its occurrence varies along the length of sex chromosomes (Ironside 2010). A remaining possibility, that the loci under consideration have experienced transposition to an autosomal locus, seems unlikely particularly where the loci that have begun again to accumulate some differentiating mutations since the derivation of a common ancestor allele (e.g., STE20, ABC1).

A comparison of phylogenetic trees for separate loci may help distinguish between the alternate possibilities of a rare crossing over and gene conversion for derivation of allelic common ancestors for loci that otherwise suggests linkage to mating type is ancestral to the Microbotryum genus. STE20 and ABC1 may likely be on the same side of the pheromone receptor locus (if the order is conserved from Sporobolomyces sp. IAM 13481), but the resetting of divergence through the derivation of a common allelic ancestor does not involve the same species for both loci. If homogenization was caused by a recombination event, whatever affected the gene more proximal to mating-type should affect the gene more distal to mating type in the same species. Thus, the independent resetting of a common allelic ancestor for loci involving different Microbotryum species is more consistent with the gene conversion process, although better physical mapping of loci is needed to support this interpretation.

Contrasting phylogenetic patterns between alleles at the mating receptor and genes in close physical distance to it have been reported in other fungi: Neurospora tetrasperma (Menkis et al. 2010), Cryptococcus (Fraser et al. 2004), S. salmonicolor (Coelho et al. 2010). Fraser et al. (2004) presented genes from Cryptococcus where some portion of each gene was species-specific and the other portion was mating-type specific and hypothesized that gene conversion was the mechanism responsible for this pattern. Similarly, Metin et al. (2010) invoked gene conversion to explain the tree topologies among Cryptococcus species for loci in proximity to the pheromone receptor where alleles for some genes clustered according to the mating type-of-origin in one part of the phylogeny but according to species-of-origin elsewhere in the tree. In N. tetrasperma, Menkis et al. (2010) proposed gene conversion to explain the contrasting phylogenetic patterns for genes assumed to be in close vicinity of mating type, based on assumed synteny with the related Neurospora crassa genome. Coelho et al. (2010) also allude to the idea of gene conversion operating in the mating-type locus in S. salmonicolor to explain allele distributions for genes in the mating-type locus that are partially species-specific and partially mating-type specific. Gene conversion might be a more broadly applicable phenomenon operating in nonrecombining regions of fungi than previously considered, and recent studies indicate that gene conversion may be able to operate freely in eukaryotic genomes in regions with complete suppression of recombination (Shi et al. 2010).


The potential influences of gene conversion upon allelic divergence may belie more complex patterns of linkage to mating type that have formed over long periods of evolutionary history. For example, even if there was a single event that blocked recombination over the majority of the chromosome (i.e., large-scale inversion), variation in the rates of gene conversion could create regions with differing levels of divergence (Charlesworth et al. 2005), a scenario that may lead to the appearance of “evolutionary strata” (Ironside 2010) but would not fit with the conceptual model that describes a succession of events that add to the nonrecombining region. Furthermore, the affect of gene conversion to reset allelic homozygosity indeed means that estimates based on only pairwise allelic divergence within a species can be misleading. In particular, the ages of the proposed evolutionary strata (Votintseva and Filatov 2009) does not consistently agree with how anciently these loci appeared to be linked to the mating type at the genus level. If genes very close to the mating pheromone receptor (i.e., STE20 and ABC1), or that assumedly require heterozygosity for their function (HD1/HD2), can show a relatively recent ancestor for the alleles in both a1 and a2 cells, the age of cessation of recombination would be difficult to detect using one species. Thus, the growing support for the influence of gene conversion in regions of recombination suppression could be a reason for the lack of a correlation between relative ages of recombination cessation based on the phylogenetic assessment and the previously suggested evolutionary strata.

In addition, the influence of local rearrangements, possibly mediated by transposable element insertions, has been suggested to explain the complex patterns of reshuffled loci among four evolutionary strata in C. neoformans, C. gattii, and N. tetrasperma (Fraser et al. 2004; Ellison et al. 2011), and this process is not unexpected in Microbotryum as well due to the elevated transposable element content of the mating-type chromosomes (Hood et al. 2004). More generally, the establishment of age-ordered and discrete evolutionary strata of recombination cessation as a genomic feature of fungi requires further evidence than is currently available.


This study illustrates that evolutionary strata of cessation of recombination in fungi are less apparent than previously proposed, and it exposes the potential confounding influence of species-specific histories of homogenization events through rare recombination or gene conversion. Measuring the rates of gene conversion would be essential to properly test the latter theory. More work, including a larger numbers of loci and species comparisons is needed in Microbotryum and other lineages of fungi to establish the linkage histories between loci and genes determining mating compatibility. However, such advances in understanding of the tenable analogies between fungi and other eukaryotes will broadly inform the role of sexual reproduction upon important genomic features. Sequence data from this article have been deposited with the GenBank Data Libraries under accession nos. JQ423462-JQ423895.

Associate Editor: J. Mank


The work was supported under the award NSF-DEB 0747222 to MEH. We acknowledge grant ANR-09–0064-01. Sequencing was funded by the ‘‘Consortium National de Recherche en Génomique’’ and the ‘‘service de systématique moléculaire’’ of the Muséum National d’Histoire Naturelle (CNRS IFR 101). It is part of the agreement no. 2005/67 between the Genoscope and the Muséum National d’Histoire Naturelle on the projects ‘‘Macrophylogeny of life’’ and “SPEED ID” directed by G. Lecointre and J.-Y. Rasplus. The genome sequence and assembly for M. lychnidis-dioiciae strain Sl1 (Lamole strain, a1) was accomplished through support from National Science Foundation (NSF) grant 0947963 to MHP, DJS, and CC. We acknowledge S. S. Toh, D. J. Schultz for their help with the genome sequence and assembly.