The Mating Type Locus () and Sexual Reproduction of : Insights into the Evolution of Sex and Sex-Determining Chromosomal Regions in Fungi

Summary

Mating in basidiomycetous fungi is often controlled by two unlinked, multiallelic loci encoding homeodomain transcription factors or pheromones/pheromone receptors. In contrast to this tetrapolar organization, Cryptococcus neoformans/Cryptococcus gattii have a bipolar mating system, and a single biallelic locus governs sexual reproduction. The C. neoformans MAT locus is unusually large (>100 kb), contains >20 genes, and enhances virulence. Previous comparative genomic studies provided insights into how this unusual MAT locus might have evolved involving gene acquisitions into two unlinked loci and fusion into one contiguous locus, converting an ancestral tetrapolar system to a bipolar one. Here we tested this model by studying Cryptococcus heveanensis, a sister species to the pathogenic Cryptococcus species complex. An extant sexual cycle was discovered; co-incubating fertile isolates results in the teleomorph (Kwoniella heveanensis) with dikaryotic hyphae, clamp connections, septate basidia, and basidiospores. To characterize the C. heveanensis MAT locus, a fosmid library was screened with C. neoformans/C. gattii MAT genes. Positive fosmids were sequenced and assembled to generate two large probably unlinked MAT gene clusters: one corresponding to the homeodomain locus and the other to the pheromone/receptor locus. Strikingly, two divergent homeodomain genes (SXI1, SXI2) are present, similar to the bE/bW Ustilago maydis paradigm, suggesting one or the other homeodomain gene was recently lost in C. neoformans/C. gattii. Sequencing MAT genes from other C. heveanensis isolates revealed a multiallelic homeodomain locus and at least a biallelic pheromone/receptor locus, similar to known tetrapolar species. Taken together, these studies reveal an extant C. heveanensis sexual cycle, define the structure of its MAT locus consistent with tetrapolar mating, and support the proposed evolutionary model for the bipolar Cryptococcus MAT locus revealing transitions in sexuality concomitant with emergence of a pathogenic clade. These studies provide insight into convergent processes that independently punctuated evolution of sex-determining loci and sex chromosomes in fungi, plants, and animals.

Introduction

Although many organisms, in particular microorganisms, can reproduce both asexually and sexually, the vast majority appears to undergo sexual reproduction during their life cycles. The hypotheses advanced as to why sex is so ubiquitous are myriad, but center on several recurrent themes [1], [2]. First, the admixture of genetic material from two genetically distinct individuals that occurs during sexual reproduction may give rise to novel gene combinations that result in offspring better suited to novel or changing environments. Second, sexual reproduction may serve to remove deleterious mutations, such as transposons, that have arisen within the genome. Finally, sex might serve both roles, facilitating combinations of successful alleles and simultaneously purging the genome of deleterious ones.

Until recently it has been difficult to test these models for the possible benefits of sex in experimentally tractable model organisms. However, recent studies with the model yeast S. cerevisiae have provided direct experimental evidence for a benefit of sexual reproduction. Isogenic diploid strains were engineered that are either capable of sporulation with meiotic recombination (sexual) or were able to sporulate but not undergo meiotic recombination (asexual, spo11,13 mutant) [3]. Under a variety of different stressful environments, the sexual strain had a competitive advantage compared to the asexual strain. A series of related studies in both S. cerevisiae and the model alga Chlamydomonas provide additional support for a benefit conferred by sex in response to novel or challenging environments [4]–[6]. In response to constant environments, asexual reproduction is also likely to be of relative benefit by relieving strains from the metabolic and genetic costs associated with sex [7]. Thus, a balance between sexual and asexual reproduction may be struck in response to different environmental conditions.

While the vast majority of sexually reproducing organisms occur as just two sexes or mating types, transitions in sexuality from two to multiple mating types, and vice versa, have occurred in the fungal kingdom. Sexual reproduction is common in fungi, and mating type occurs in two general patterns: bipolar and tetrapolar mating type (reviewed in [8]–[10]. In the bipolar systems, a single genetic locus known as the mating type locus (MAT) occurs in two alternate forms, known as idiomorphs (a or α, a or A, + or −, P or M), and these govern the identity of the cell [11]–[13]. Co-incubation of isolates of opposite mating type under suitable conditions leads to sexual reproduction. Species with bipolar mating systems are found in the ascomycete, basidiomycete, and zygomycete phyla, providing evidence that this is an ancestral organization. In the basidiomycete phylum, many species instead have a more complex sex determining system, known as tetrapolar, in which two unlinked sex determining loci are present [8], [10], [14], [15]. One locus encodes homeodomain transcription factors and the other locus encodes pheromones and pheromone receptors, and both loci must differ for sexual reproduction to occur. In many species, these loci are multiallelic, resulting in literally thousands of different mating types, which promotes outcrossing [16]. Transitions from tetrapolar to bipolar mating type determination have occurred multiple independent times, and in several examples result from fusions of the two unlinked loci to form one contiguous region (reviewed in [17], [18]), potentially illustrating common evolutionary pressures limiting mating-type/sexes to just two.

Although many fungi are currently classified as asexual, genomics has revealed that many such fungi retain the mating type locus and machinery necessary for both mating and meiosis [19]–[24]. In some cases, population genetics studies also reveal an equal distribution of opposite mating types in nature, and evidence for recombination [25]–[28]. In some notable recent examples, such as Aspergillus parasiticus and Aspergillus fumigatus, an extant sexual cycle has been discovered [29], [30]. Thus, there may be few truly asexual fungi, and many fungi, and perhaps most, for which cryptic sexual cycles remain to be discovered [31], [32].

Sexual reproduction in the basidiomycetous Cryptococcus pathogenic species complex involves a well-defined bipolar mating type system with two opposite mating types, a and α [33]–[38]. Cell-cell fusion leads to the production of dikaryotic hyphae with fused clamp connections, and ultimately the hyphal tips differentiate to form basidia in which nuclear fusion and meiosis occur. Budding from the basidial surface then produces long chains of basidiospores, the suspected infectious propagules [39], [40] for reviews see [41]–[43].

Molecular analysis of the Cryptococcus mating type locus reveals it to be unusual, spanning >120 kb and encompassing >20 genes [44]–[49]. With the exception of the homeodomain genes, SXI1α and SXI2a, which govern cell type identity, the a and α mating type alleles are otherwise composed of divergent alleles of a common gene set [50], [51]. A genomic approach comparing the extant, diverged mating type alleles in C. neoformans var. grubii (serotype A), C. neoformans var. neoformans (serotype D), and C. gattii revealed that the MAT locus is composed of four gene strata of distinct evolutionary ages, including an ancient set of ancestral genes, two gene strata of more intermediate evolutionary origin, and a set of apparently recently acquired genes that are species- but not mating type-specific [47]. The finding that the Cryptococcus bipolar MAT locus contains all of the genes normally found in the two unlinked tetrapolar loci suggests that a fusion of the two loci might have been involved in the origins of this unusual gene cluster linked to virulence and differentiation. The hypothesized evolutionary model posits that an ancestral tetrapolar mating system expanded by a series of acquisitions of genes of related function to form two large gene clusters that subsequently underwent fusion via a chromosomal translocation to form an unstable tripolar intermediate stage that subsequently underwent recombination or gene conversion to form the bipolar state. A series of more recent gene conversions and inversions then occurred, giving rise to the extant bipolar mating type alleles and resetting the molecular clock for some genes within distinct strata and even evicting genes from MAT in other examples [47]. The recent discovery that the MAT locus is flanked by and contains recombination hotspots provides evidence that the expansion and fusion of these loci may have been driven by activation of recombination [52]. In a recent study, the homeodomain genes were relocated to an unlinked genomic location, providing direct experimental support that Cryptococcus can be engineered to complete a tetrapolar sexual cycle and that the tripolar intermediate may have conferred an evolutionary disadvantage and therefore have been transitory [53]. A central goal of the studies presented here was to further rigorously examine the proposed MAT evolutionary model by comparative studies with related but divergent fungal species.

In a recent study, a multi-locus gene phylogeny was developed with species closely related to the pathogenic Cryptococcus species complex, which includes C. neoformans var. neoformans, C. neoformans var. grubii, and C. gattii [54]. In this study, the yeasts analyzed defined two clades. One contains the pathogenic species complex as well as Cryptococcus amylolentus, Tsuchiyaea wingfieldii, and Filobasidiella depauperata, and was termed the sensu stricto clade. The other clade encompasses yeasts more distantly related to the pathogenic species complex, and was termed the sensu lato group. C. heveanensis lies in the sensu lato clade, together with Bullera dendrophila, Cryptococcus bestiolae, Cryptococcus dejecticola, and a recently described sexual species, Kwoniella mangroviensis [55]. The fact that C. heveanensis is located in a sister clade to the pathogenic Cryptococcus species complex makes it an excellent candidate to provide insights into the evolution of the MAT locus and sexual reproduction in pathogenic Cryptoccocus species.

The closest yeast with a defined sexual cycle to C. heveanensis is K. mangroviensis, which is also located in the sensu lato clade containing yeasts more distantly related to the pathogenic Cryptococcus species [54]. The basidial structure of K. mangroviensis differs from that in C. neoformans and C. gattii and contains septa (phragmobasidia) whereas the pathogenic Cryptococcus species have nonseptate basidia (holobasidia) [33], [55]. When pairs of compatible K. mangroviensis strains are co-cultured on CMA medium, dikaryotic hyphae with clamp connections and basidia are produced [55]. Of 19 K. mangroviensis strains isolated from the Bahamas, 7 were fertile (37%) and 12 were sterile (63%). Among the seven fertile isolates, six were assigned to the α mating type and one to the A mating type. From these results, the authors proposed that the mating system appeared to be bipolar and biallelic. The nature of the K. mangroviensis mating system will be further clarified by analysis of additional fertile strains, progeny from genetic crosses, and the MAT locus.

In this study, we analyzed the mating behavior and MAT locus structure of C. heveanensis. First, we discovered an extant sexual cycle for C. heveanensis involving dikaryotic hyphae with clamp connections and basidia with cruciate septa and termed this teleomorph Kwoniella heveanensis. Second, we determined the structure of the MAT locus involving two probably unlinked gene clusters corresponding to the homeodomain locus and the pheromone/receptor locus. The presence of two divergently transcribed homeodomain genes in the homeodomain locus suggests this MAT locus arrangement is ancestral to the pathogenic Cryptococcus species, and one or the other homeodomain gene has been lost recently from MAT in C. neoformans and C. gattii. In addition, the presence of numerous other genes around the C. heveanensis MAT locus that correspond to genes located on the same chromosome but far away from MAT in C. neoformans suggests intrachromosomal rearrangements occurred during the evolution of the C. neoformans MAT locus and flanking regions. Characterizing MAT alleles from other C. heveanensis isolates suggests that the homeodomain locus is multiallelic and the pheromone/receptor locus is at least biallelic. Additionally, the MAT-specific region is likely to be restricted to the SXI1 and SXI2 genes in the homeodomain locus and spans at least the STE3, STE12, MFA1/2, CNB00600, and CNG04540 genes in the pheromone/receptor locus, consistent with a tetrapolar mating system. This analysis reveals an evolutionary intermediate with two apparently unlinked MAT loci into which genes have been acquired, lost, and translocated, providing insights into the events that punctuated the evolution of a contiguous sex determining gene cluster in C. neoformans and C. gattii. These studies serve as an evolutionary window both for the evolution of mating type loci in fungi, and also into similar events that have driven the evolution of sex chromosomes in fungi, plants, and animals including insects, fish, and mammals [18], [56]–[58].

Results

Discovery of the sexual cycle of C. heveanensis

To study mating behavior and the MAT locus of C. heveanensis, the isolates (Table 1) were first analyzed to validate their species assignment. When the D1/D2 and ITS regions were sequenced and compared to the type strain CBS 569, all 9 strains were found to have identical D1/D2 sequences (Table 1, Figure S1). Seven isolates also had identical ITS sequences whereas two isolates (BCC 8396, BCC 11754) had one nucleotide difference in the ITS region (Table 1, Figure S1). These findings are in accord with the assignment of these isolates as C. heveanensis. Two other isolates (CBS 9459,CBS 6097) appear to represent related, but distinct species based on a higher level of divergence in the D1/D2 and ITS sequences (see Figure S1). All strains appear to be haploid based on fluorescence-activated cell sorting (FACS) analysis by comparison to C. neoformans haploid and diploid reference isolates (Table 1, Figure S2).

To test if this species has an extant sexual cycle, isolates were co-cultured and examined morphologically for production of mating specific structures. Strains were inoculated on V8 (pH 5 or 7) or MS media alone and in every pair-wise combination and incubated at room temperature in the dark or light. Five strains were self-filamentous when incubated alone while the other strains were not (Table 1). The filaments produced during monoculture were highly branched, irregular in shape, lacked clamp connections, and monokaryotic (nuclei stained with Sytox green) (data not shown). In contrast, dikaryotic hyphae with clamp connections and basidia fruiting bodies indicative of sexual reproduction were produced by three mating combinations (BCC 8398 × CBS 569, BCC 8398 × BCC 8396, and BCC 8398 × BCC 11754). Optimal mating conditions were on V8 medium (pH 5) in the dark. By 10 days incubation, compatible mating pairs had begun to produce dikaryotic filaments with clamp connections (Figure 1A and 1B), and basidia embedded in the agar were first apparent at 2 to 3 weeks of incubation (Figure 1C), and later developed into clusters in a matrix reaching the agar surface and visible by eye (Figure 1D–1G). The basidia are globose with a diameter of 7.5–10.5 µm and cruciate septa, similar to the related species K. mangroviensis and Tremella mesenterica (Figure 1H–1K, [55]). Basidiospores (subglobose, diameter 2.3–3.3 µm) were associated with the basidial clusters (Figure 1J and 1K) and were clearly distinguishable from the larger elliptical budding yeast cells (3.5–4.9 µm×5.0–7.2 µm) (Figure 1L). Fecundity was associated with self-filamentation (4/5 self-filamentous strains were fertile; 4 non-filamentous isolates were sterile (Table 1)). These morphological observations define an extant sexual cycle for this species. The dikaryotic sexual state that formed when compatible pairs of C. heveanensis were co-cultured was named Kwoniella heveanensis because the basidial morphology of this species is quite similar to that of K. mangroviensis (for Standard and Latin descriptions, see Materials and methods). Both species are phylogenetically related in a sister species clade to the pathogenic Cryptococcus species complex [54].

Characterization of the C. heveanensis MAT locus

To elucidate the structure of the C. heveanensis MAT locus, probes to MAT-specific genes of C. neoformans and C. gattii were generated by degenerate PCR. These probes (LPD1, RPO41, CAP1, ZNF1, STE20, IKS1, FAO1, and NOG2) were used to screen a fosmid library for the type strain C. heveanensis CBS 569. Sequencing of positive fosmids and assembly of the sequences resulted in two large gene clusters: a region containing the homeodomain locus and flanking regions spanning ∼80 kb and a region containing the pheromone/receptor locus and flanking regions spanning ∼180 kb (Figure 2). The homeodomain locus of C. heveanensis contains two divergent homeodomain genes (SXI1 and SXI2) encoding an HD1 and an HD2 class product, similar to other basidiomycete MAT loci [59]–[61]. Because the MAT locus alleles in C. neoformans and C. gattii contain only SXI1α or SXI2a but not both, this suggests that one or the other homeodomain gene has been lost recently in these pathogenic species.

The genomic region containing the homeodomain (HD) locus of C. heveanensis also contains the FAO1, FCY1, and UAP1 genes, which form the left boundary of the pathogenic Cryptococcus species MAT locus (Figure 3). Also included are the RPL22, SPO14, and CAP1 genes in addition to the homeodomain genes, which are part of the MAT locus in C. neoformans var. neoformans, C. neoformans var. grubii, and C. gattii [46], [47], [49]. These three genes are also linked to the homeodomain genes of Tremella mesenterica (Figure 3).

The remaining genes found in the MAT locus of pathogenic Cryptococcus species are present in the region containing the pheromone/receptor (PR) locus of C. heveanensis (Figure 3). These include the pheromone receptor gene, STE3, the pheromone gene MFA1, the pheromone sensing MAPK cascade components STE11, STE20 and STE12, and other genes whose functions are either unknown, essential, or may not be related to mating (including BSP1, ETF1, ZNF1, PRT1, NCP1, MYO2, IKS1, RPL39, and BSP3). This region also contains the LPD1, GEF1, CID1, RPO41, and BSP2 genes, which were hypothesized to be present at the boundaries of the ancestral sex determining regions in the proposed evolutionary model (RPO41 and BSP2 in the ancestral homeodomain locus and LPD1, GEF1 and CID1 in the ancestral pheromone receptor locus) and to have been entrapped in MAT when the HD and PR loci were brought together via a chromosomal translocation [47]. The presence of these genes in the PR locus region of both C. heveanensis and T. mesenterica implies that these genes were likely to have all been linked to the ancestral PR locus prior to fusion to a contiguous bipolar sex-determinant.

The HD and PR gene clusters of C. heveanensis have numerous other linked genes that are not present in the MAT locus of C. neoformans or C. gattii. Most of these genes (such as CND01630, CND01640, CND05260, CND05390, CNN00870, CNE02670, CNI00160, CND01540, CNG02430, CNE02190, CND01240, CND01560, and CND01570) are also present in the T. mesenterica PR locus region suggesting these genes have recently been lost from the MAT locus and relocated to other genomic locations in C. neoformans and C. gattii (Figure 4). Some of these genes do not have apparent homologs in C. neoformans or C. gattii, suggesting these genes may have been lost during the evolution of the MAT locus. It is also interesting that while a majority of these genes are on the 4^th chromosome of C. neoformans strains JEC21/B3501A, where MAT is also located, they correspond to the opposite telomeric region, suggesting intrachromosomal rearrangements may have occurred during MAT locus evolution (Figure 4, [62]).

Homeodomain locus of C. heveanensis is multiallelic, and the MAT–specific region is likely to be restricted to the SXI1 and SXI2 genes

Southern blot analysis of the 8 additional C. heveanensis isolates and the type strain with SXI1 and SXI2 gene probes showed a highly divergent profile, suggesting the homeodomain locus structure of C. heveanensis might be multiallelic (Figure 5A). Strains BCC 8305 and BCC 8313 showed the same hybridization profile and sequence analysis also showed that the SXI1, SXI2, LPD1, RPL22, ZNF1, and STE11 genes in these strains are 100% identical for the regions sequenced (data not shown). Strains BCC 8396 and BCC 11754 also appeared identical. Therefore, in these cases, one isolate from each strain pair was selected for further studies. Sequencing of an ∼12 kb region, spanning SXI1 and SXI2, from seven strains provided evidence that the locus is multiallelic with 6 different alleles of SXI1 and SXI2 (Figure S3). Percent sequence identity plots comparing the homeodomain region of CBS 569 with the corresponding region from other isolates showed that there is a highly dissimilar core corresponding to the 5′ ends of the SXI1 and SXI2 genes and spanning a fairly restricted region of ∼2 kb (indicated by a blue box in Figure 5B) followed by a more similar region (∼80%), while the 3′ ends of the genes are almost identical. A representative dot plot comparing the homeodomain region of CBS 569 and BCC 11757 also shows the divergence at the 5′ ends of both genes (Figure 5C). Percent identity plots comparing the homeodomain region from each strain with the others are shown in Figure S4. The homeodomain genes of BCC 8384 and BCC 8398 were found to be very similar (Figure S3, Figure S4). When pairwise comparisons were conducted, percent identity of predicted amino acid sequences between BCC 8384 and BCC 8398 was found to be 92% for Sxi1 and 94% for Sxi2, while percent identity varied between 71–77% for Sxi1, 75–82% for Sxi2 for the other pairwise combinations. Considering the absence of mating between these two isolates, although they are both self-filamentous, these strains may have the same allelic versions of SXI1 and SXI2, preventing mating.

Homeodomain proteins have distinct domains for different functions [63]. The homeodomain is the DNA binding region consisting of 3 alpha helices. Fungal homeodomain proteins involved in mating are either HD1 or HD2 type according to the length of this homeodomain region [14]. In HD2 type homeodomain proteins, the distance between the first and second helix is three amino acids shorter than HD1 homeodomain proteins. The C. heveanensis Sxi1 and Sxi2 proteins are HD1 and HD2-type homeodomain proteins, respectively, like their C. neoformans counterparts (Figure 6A, [50], [51]).

Studies involving S. cerevisiae a1 and α2, Coprinopsis cinerea HD1 and HD2, and U. maydis bE and bW homeodomain proteins showed that following cell-cell fusion the two homeodomain proteins from compatible partners dimerize via their N-terminal domains [59], [64], [65]. These studies suggest that the interaction is through coiled-coils formed by alpha helices aligned such that the hydrophobic residues are adjacent. Although, we couldn't detect extensive coiled-coil formation in the C. heveanensis homeodomain proteins using the COILS software, there are 3–4 alpha helical regions in the N-terminal domains of both Sxi1 and Sxi2, which might form coiled-coil interactions (Figure S3).

In C. cinerea, two bipartite nuclear localization signals (NLSs) were found in HD1 and this region targeted a reporter protein to the nucleus [66]. In contrast, no NLS was apparent in the C. cinerea HD2 protein and HD2-reporter fusion proteins were not nuclear localized. Similarly in U. maydis, a bipartite NLS in bE (HD1) was predicted [67]. We found one bipartite and one 7-residue pattern (pat7) NLSs in Sxi1 (HD1), while only one pat7 type NLS in Sxi2 (HD2) using the PSORT II software (Figure 6B).

The RPL22 and CAP1 genes are part of the MAT locus in C. neoformans and C. gattii [47], [49]. The RPL22 gene shares 88–90% identity among mating types while the 5′ and 3′ regions of the CAP1 gene show different profiles. For C. neoformans var. neoformans and C. neoformans var. grubii, the identity is 90–92% for the 5′ region of CAP1 and 84–85% for the 3′ region. For C. gattii, the identity is 88% for the 5′ region of CAP1 and 93% for the 3′ region. Sequences from RPL22 and CAP1 were analyzed from C. heveanensis to determine if the HD locus encompasses these genes. When 530 bp of RPL22 gene sequence from all 7 unique strains was analyzed, 97 to 99% sequence identity was observed. Similarly, comparison of 212 bp of CAP1 gene sequence from strains CBS 569, BCC 8313, and BCC 15000 resulted in 99.5–100% identity. Based on this analysis revealing the RPL22 and CAP1 genes are more similar between C. heveanensis isolates than the a and α alleles of C. neoformans, it is likely that the HD locus of C. hevenensis only includes the homeodomain genes, SXI1 and SXI2, although other linked genes should also be examined to further support this conclusion.

These results show that the HD locus of C. heveanensis is multiallelic with at least 6 different alleles determined among 7 strains. Thus, additional extant HD locus alleles likely remain to be discovered. In addition, the MAT-specific region is likely to be limited most prominently to the 5′ end of the SXI1 and SXI2 genes coding for the N-terminal regions of the proteins known to be involved in the specificity of heterodimer formation.

Pheromone/receptor locus of C. heveanensis is at least biallelic

When the pheromone and the pheromone receptor gene sequences were amplified by PCR with CBS 569 specific primers, PCR fragments of the expected sizes were obtained for all strains except BCC 8398. Sequencing of these PCR products showed that the pheromone and receptor gene sequences were identical for all of the strains that yielded PCR products (data not shown). For the atypical strain BCC 8398, degenerate primers were used to amplify the STE3, STE12, CNB00600, and CNG04540 genes and the gaps between these genes were closed by primer walking. This sequence analysis resulted in an ∼9 kb sequence spanning the 5′ end of STE3, STE12, a pheromone gene, which was named MFA2, and the CNG04540 and CNB00600 genes. All of these genes were divergent between BCC 8398 and CBS 569 except CNB00600. Sequence identity values are 54% for STE3, 53% for STE12, 75% for MFA1/2, and 63% for CNG04540. In addition, the gene order also differed as indicated in the comparison and the dot plot (Figure 7A and 7B). Interestingly, the divergent region between BCC 8398 and CBS 569 starts at the 3′ end of the CNB00600 gene. The majority of this gene corresponding to the 5′ portion (1285 bases) is 98% identical, while for a small portion at the 3′ end (139 bases for CBS 569 and 124 bases for BCC 8398), the sequence identity is 53% suggesting this gene may have represented the border of MAT or have been subject to more recent gene conversion.

The pheromone genes of C. heveanensis, MFA1 (from CBS 569) and MFA2 (from BCC 8398), encode 39 amino acid-long pheromone precursors (Figure 7C). Both were identified by the characteristic C-terminal motif of lipopeptide pheromones: CAAX, where A is an aliphatic amino acid [68]. While Mfa1 ends with the motif CVIA like pheromone precursors from C. neoformans, C. grubii, C. gattii and T. mesenterica, Mfa2 has CIIA at its C-terminus. Tremerogen a-13, a T. mesenterica pheromone, was isolated and its structure determined [69]. According to this study, after proteolytic cleavage, the mature pheromone starts with a glutamic acid (E) residue indicated with an arrow in Figure 7C. Since this residue is conserved in all Cryptococcus pheromones, the proteolytic cleavage might occur at the same site in other pheromones as well.

No or minimal sequence differences were observed for the LPD1 (100% sequence identity in 530 bases), STE11 (99% in 470 bases), ZNF1 (100% in 410 bases) or IKS1 (99% in 1740 bases) genes. Also, Southern blot analysis using STE11 and MYO2 gene fragments as probes showed the same hybridization profile among all strains tested (Figure 7D). This shows that the divergent region is likely to be more restricted than that of C. neoformans and C. gattii and contains at least the receptor gene STE3, the pheromone gene MFA1/2, STE12, and CNG04540.

These results suggest that the PR locus of C. heveanensis is at least biallelic similar to U. maydis and Tremella species, and it contains not only STE3 and MFA1/2, but also other genes including at least STE12, CNB00600, and CNG04540. Sequence analysis of the genomic regions flanking this region from the isolate BCC 8398 will be necessary to establish whether any additional regions of the PR MAT locus remain to be identified.

Mating type genes exhibit three different phylogenetic patterns

Phylogenetic analyses were conducted on selected mating type genes (CID1, GEF1, LPD1, BSP1, SPO14, ETF1, STE20, STE11, STE3 and STE12) from C. heveanensis, T. mesenterica, C. neoformans var. neoformans, C. neoformans var. grubii and C. gattii (Figure S5). These genes were classified into three groups based on their evolutionary history and two representative examples from each group are demonstrated in Figure 8. These analyses show that, STE3 and STE12 exhibit a mating type specific pattern, where C. heveanensis CBS 569 and T. mesenterica show a closer relationship to the α mating type group, while C. heveanensis BCC 8398 is closer to the a mating type group. BSP1, SPO14, ETF1, STE20, and STE11 are mating type-specific in C. neoformans and C. gattii, however, none of these genes from either C. heveanensis or T. mesenterica followed a mating type-specific pattern. The third gene group consists of CID1, GEF1 and LPD1 that do not show a mating type-specific pattern, but demonstrate a species-specific pattern for all species.

Discussion

We report here two advances with respect to understanding the evolution of sexual reproduction and mating type determination in fungi. First, we discovered and characterized the sexual cycle for C. heveanensis. Second, we cloned and sequenced the mating type locus for this organism. Unlike related species that commonly infect animals, C. heveanensis is not a pathogen and is instead associated with insects and flowers. This species represents an evolutionary window that provides novel insights into understanding transitions that occur between tetrapolar and bipolar mating type systems, and in model and pathogenic species. In particular, this analysis reveals how a tetrapolar ancestral species with two unlinked sex determining regions of the genome gave rise to bipolar extant species in which the sex determinants are linked, providing direct experimental validation of the central tenants of sex chromosome evolution originally proposed by Ohno [56]. These models involve the emergence of sex determinants on autosomes, which are then captured into a sex determining region or chromosome by sequential gene acquisition into strata of different evolutionary ages involving genes of coherent function. The basidiomycete mating type locus represents an exceptional experimental paradigm to test, and validate experimentally, key features of this model. That similar events have transpired repeatedly, and independently, in plants and animals and other fungi, illustrates common underlying principles that are involved in the emergence, evolution, and plasticity of genomic sex determining regions as diverse as fungal mating type loci and sex chromosomes.

Recent multilocus studies support the assignment of C. heveanensis to a clade of species related to but distinct from the pathogenic species C. neoformans and C. gattii [54]. Importantly, this analysis revealed that C. heveanensis occurs in a sensu lato group of species, including K. mangroviensis, which was recently discovered and found to have an extant sexual cycle that was defined as bipolar [55]. A critical aspect in finding the sexual cycle for C. heveanensis was the recent discovery of several new isolates for the species, which were made readily available by their deposition in a public strain collection [70]. Of three isolates previously reported to be C. heveanensis, our D1/D2 and ITS sequence analysis provides evidence that only the type strain CBS 569 is in fact C. heveanensis, whereas two other isolates (CBS 9459 and CBS 6097) appear to represent isolates of closely related cryptic species (Figure S1). Neither of these isolates was fertile with the C. heveanensis type strain. All eight of the recently reported isolates are, based on D1/D2 and ITS sequence analysis, bona fide C. heveanensis isolates. Pairwise incubation of all nine available C. heveanensis isolates under a broad set of conditions and media resulted in mating for three of the 36 possible mating combinations. Thus, at least four of the nine C. heveanensis isolates are fertile. Mating resulted in the production of dikaryotic hyphae with fused clamp connections, basidia with cruciate septa, and basidiospores. Notably, in solo incubations all four of the fertile strains were self-filamentous and produced monokaryotic hyphae, whereas only one of the sterile isolates was self-filamentous. Thus, the formation of monokaryotic hyphae by mating partners appears to be associated with and may precede successful mating, and these may facilitate location of mating partners or fusion of isolates. Based on the similarities of the sexual cycle between C. heveanensis and K. mangroviensis, we have assigned the teleomorphic designation Kwoniella heveanensis. In addition to providing insights into the evolution of mating and MAT, the discovery of the sexual cycle for C. heveanensis will also facilitate further laboratory investigations and strain construction.

How might the self-filamentous phenotype that may be necessary for sexual reproduction be controlled? Five of the nine C. heveanensis isolates are self-filamentous, and four of these are fertile in genetic crosses. The filaments produced may represent something similar to a conjugation tube, however as these occur during mono-culture, they are not induced by pheromones produced by a mating partner of compatible mating type. Possibilities include that elements of the pheromone response pathway have been constitutively activated by mutations, or in response to nutritional cues or limitation alone. Alternatively, it may be that these isolates are self-responsive to their own pheromone as a partial agonist of their Ste3 receptor homolog. In such a model, co-culture leads to more robust pheromone signaling necessary to enable cell-cell fusion and generation of a heterodimeric homeodomain complex that enables completion of sexual reproduction. Future studies to address these and other models will be necessary to test the activity and specificity of the mating pheromones and the Ste3 homologs involved in sensing them, and to isolate mutations that abrogate the self-filamentous phenotype testing if, for example, the pheromone and pheromone receptor genes are involved and whether this is necessary for mating.

Analysis of the mating type locus for C. heveanensis revealed two apparently unlinked genomic regions, one corresponding to the homeodomain gene locus (B) and the second to the pheromone/pheromone receptor gene locus (A) (Figure 9). Notably, the B locus (HD) includes a pair of divergently oriented genes encoding homeodomain proteins related to Sxi1/Sxi2 from C. neoformans and bE/bW from U. maydis. The A locus (PR) includes genes encoding a Ste3 pheromone receptor homolog and a mating pheromone. Both loci are embedded in larger gene clusters, including some previously identified in the C. neoformans MAT locus and novel ones. To establish which genes in the ∼80 and ∼180 kb assemblies are mating type specific, representative regions were isolated by PCR and sequenced from fertile and sterile isolates. This analysis revealed striking diversity in the intergenic region and 5′ coding regions of the SXI1/SXI2 genes, flanked by a less divergent region (∼80% identity) and followed by nearly identical 3′ coding regions (Figure 5). The divergent sequence region spans ∼2 kb. Importantly, this locus appears to be multi-allelic, based on sequence divergence, and at least 6 alleles are present in the 7 isolates examined, suggesting additional alleles may remain to be discovered. This MAT organization is quite similar to the U. maydis b MAT locus [59], [71], [72] corresponding to the diverse N-terminal coiled-coil dimerization domain that dictates self- vs. non-self recognition. The RPL22 and CAP1 genes linked to SXI1/SXI2 in C. heveanensis had very similar sequences from the strains analyzed (97 to 100% identity), suggesting these genes are outside the mating-type specific region. Similar to other basidiomycetes, including U. maydis, S. commune, and C. cinerea, we suggest dimerization occurs between HD1 and an HD2 class homeodomain proteins encoded by different mating type locus alleles [65], [73], [74].

The A locus (PR) of C. heveanensis encodes a Ste3 pheromone receptor homolog, a mating pheromone, and other genes involved in pheromone sensing (Ste11, Ste12, and Ste20) are present or linked. Thus far, our sequence analysis for this large gene cluster spans ∼180 kb for the CBS 569 type strain and representative sequence from other isolates. This analysis allows us to conclude that the gene sequences, and the gene organization, differ substantially between two different alleles present in the population (Figure 7). The conclusion that at least two alleles are present is also supported by consideration of which strain combinations are fertile and which are infertile (Table 1). The finding that the C. heveanensis A MAT locus might be biallelic is similar to the tetrapolar species U. maydis, in which the a mating type locus is also biallelic [75]. The C. heveanensis A locus is distinguished from that of U. maydis in that additional genes are present within the locus, and the linked or flanking genes are also quite distinct. While further analysis of additional genomic regions from other isolates will be required to define the full extent of these MAT locus alleles, it is apparent that highly conserved sequences are embedded within more divergent ones (see Figure 7B). This may illustrate that gene conversion occurs within the locus, as is apparent in the MAT locus of C. neoformans, or that this MAT locus allele represents a transitional form into which related genes have been recruited but have not yet diverged into mating type specific allelic forms. Recent studies of the ectomycorrhizal fungus Laccaria bicolor, a tetrapolar species, have illustrated differences in the evolutionary trajectory for the two mating type loci [76]. Whereas the homeodomain locus appears to be in a region of the genome where syntenic gene organization is conserved, the pheromone and receptor gene locus is more dynamic and subject to different selective pressures, leading to dramatic changes in gene order and organization and insertion of many transposable elements. Similar properties may hold for the two mating type locus alleles of C. heveanensis and may be responsible for some of the unusual features of MAT in this and related species.

The molecular organization of the C. heveanensis MAT locus allows us to propose that this species has a tetrapolar mating type determination system. This conclusion is based on the finding of two apparently unlinked loci that correspond to the A and B mating type locus alleles of other tetrapolar fungal species [16], [77], [78]. Second, a high level of genetic sequence divergence is found for genes within each of the two loci, consistent with a role in mating type determination. Third, the B locus is multiallelic, as is the case of other tetrapolar fungi but not in bipolar species. Fourth, the specific alleles found for the isolates are well correlated with those that are in fact fertile in that all three successful combinations differ at both MAT loci, whereas for at least some of the infertile strain combinations, they share a closely related or identical allele at one or both MAT loci (see Table 1). Fifth, the finding that in the population that the A1 locus allele occurs in isolates with at least six different highly diverged B locus alleles is consistent with two unlinked loci in linkage equilibrium defining mating type. Given the dramatic sequence divergence between the different B alleles, models in which these arose by divergence while linked to the A1 locus seem less plausible. Put another way, if the A and B loci were to be linked, we would expect a higher proportion of the population would consist of A1 linked to, for example, B1, than is observed.

The conclusion that C. heveanensis has a tetrapolar mating type determination system can be further tested by two approaches. First, additional physical evidence that the two MAT loci are unlinked in the genome, based on electrophoretic separation and demonstration by chromoblot that the two lie on different chromosomes, whole genome sequence analysis, or independent segregation following meiosis, would all provide evidence the two are unlinked. Thus far it has not been possible to resolve chromosomes of this species by pulsed field gel electrophoresis, even employing conditions that work effectively for the closely related species C. neoformans, C. gattii, and C. amylolentus. Whole genome analysis may prove to be a more facile approach. The evidence that the two MAT loci are unlinked is that we have been unable, thus far, to find fosmids that physically span both loci or provide evidence the two are linked. It is formally possible that the PR locus could be linked to the right most end of the HD locus, which would place the two ∼20 kb apart. We view this as unlikely as we found no evidence for fosmids in the library that would span such a theoretical region. It is formally possible that the two loci are on the same chromosome but far enough apart to be unlinked genetically, or that the two loci are partially linked. Future studies will be necessary to examine this in further detail.

Second, additional genetic evidence involving the isolation and analysis of meiotic progeny could further establish this species as tetrapolar. The sine qua non of a tetrapolar mating system is the production of four types of meiotic progeny (A1 B1 crossed with A2 B2 yielding four mating types: A1 B1, A2 B2, A1 B2, and A2 B1). Thus far we have not been successful in isolating basidiospores by micromanipulation, and this is a challenge as these are often embedded in a matrix together with basidia. Moreover, in our experience attempts to purify progeny from fungal mating mixtures often leads to a high rate of contamination from the parental isolates, confounding analysis. Future studies could approach this using genetically marked strains to enable selection of meiotic progeny to determine their mating type. A further implication of our findings and conclusions is that the mating type determination system for the closely related species K. mangroviensis, currently classified as a bipolar species with α and A mating types, may also turn out to be tetrapolar with further detailed analysis.

Our findings provide insight and direct experimental support for key features of an evolutionary model that was proposed earlier for the origins of the unusual bipolar mating type locus of the pathogenic Cryptococcus species complex based on a comparative genomic analysis of three closely related species (Figure 10) [47]. The proposed model hypothesized that the ancestral state was a tetrapolar organism with two unlinked loci, one encoding homeodomain genes and the other pheromones and the pheromone receptor that evolved in a stepwise fashion to the large bipolar mating type alleles in C. neoformans and C. gattii. In this model, two unlinked MAT loci first expanded by acquiring additional genes related to sexual reproduction, resulting in a series of gene strata of different evolutionary ages. Second, the two MAT loci fused via a translocation event in one of the two mating types, resulting in an intermediate “tripolar” state. Third, gene conversion between the linked and unlinked sex determinants resulted in fusion of the two loci in the other mating type and collapse to a bipolar state. Finally, more recent gene conversions and rearrangements occurred giving rise to the extant MAT alleles observed today.

Our analysis of the C. heveanensis mating type locus alleles reveals two large gene clusters, one corresponding to the left half of the Cryptococcus mating type locus encoding homeodomain genes and the other to the right half encoding pheromones, pheromone receptors, and signaling components. This organization serves as a molecular window on the organization of the MAT locus in the last shared common ancestor to the two lineages and represents key features of the hypothesized intermediate stages in the MAT evolutionary model (Figure 10). First, we see that the two ancestral MAT loci are probably unlinked and in some cases appear to have acquired additional related flanking genes. Second, the two large gene clusters have been formed, but have not yet undergone fusion to the tripolar or bipolar state. Given that two related large unlinked gene clusters are also present in the related species C. amylolentus (Findley and Heitman, unpublished results), the most parsimonious model is that the organization of MAT in C. heveanensis represents the ancestral state prior to fusion, rather than the result of a recent translocation that separated a contiguous MAT allele. Third, the strata of most recently acquired genes (RPO41, BSP2, GEF1, CID1, and LPD1) are all present in the right gene cluster linked to the pheromone and pheromone receptor genes, and thus were already present prior to linkage of the two regions. This differs in detail from the originally proposed model in which RPO41 and BSP2 had been depicted as linked to the homeodomain gene cluster, and GEF1, CID1, and LPD1 as linked to the pheromone/receptor locus [47]. It also raises the likelihood that these genes are linked because they are functionally related in some way to MAT function, rather than being simply entrapped when the two unlinked MAT loci were fused by translocation. We note that the organization of the two gene clusters in C. heveanensis represents a juxtapositioning of MAT-specific genes (SXI1, SXI2) with genes that are part of the MAT locus in C. neoformans/C. gattii but not yet MAT-specific in C. heveanensis. Similarly, for the pheromone/receptor gene cluster, sequences that are not mating type-specific are embedded between mating type-specific sequences. These gene clusters therefore represent transitional intermediates representing MAT loci to which genes of related function have been recruited, but not yet captured into the mating type-specific regions of the genome or begun to diverge into mating type-specific alleles.

The model presented in Figure 10 represents our model for the events that led to the formation of a large bipolar mating type locus from a tetrapolar ancestral system. The proposed expansion of the ancestral HD and PR loci would likely have involved rearrangements facilitated by repetitive sequences that accumulate in recombinationally suppressed regions of the genome. If these genes have a function in sexual reproduction, as many do, this could have sustained the organization of the locus until the two loci fused. Some features of the gene strata apparent in the extant C. neoformans and C. gattii MAT alleles reflect episodic gene conversion events that occurred after the two ancestral MAT loci fused [44], [47]. Thus, multiple mechanisms operated involving acquisition of linked genes of related function into two expanding gene clusters, as is apparent here for the STE12 gene linked to the PR locus, and gene conversion events between alleles both prior to and following linkage of the HD and PR loci that influenced the evolutionary trajectory and plasticity of this dynamic region of the genome.

A key finding of this comparative analysis is that two divergently oriented homeodomain genes are present in the MAT locus of C. heveanensis, whereas only one or the other homeodomain gene is present in the a or α MAT alleles of C. neoformans and C. gattii. This provides evidence that, like other tetrapolar basidiomycete fungi such as U. maydis, Schizophyllum commune, and C. cinerea, the ancestral Cryptococcal MAT locus contained two paired, divergent homeodomain genes and one or the other was lost recently in the C. neoformans/C. gattii lineage. Examples of homeodomain gene loss are also seen in MAT of pathogenic Candida species [79]–[81]. For example, the alpha2 homeodomain gene has been lost in two haploid emerging yeast pathogenic species, Candida lusitaniae and Candida guillieromondii, both of which have retained complete sexual cycles, and in addition C. guilliermondii has also lost the a1 homeodomain gene. As discussed elsewhere [17], the transition from paired to solo homeodomain genes further serves to restrict outbreeding potential. In species with paired homeodomain genes, there are two opportunities to form functional heterodimeric pairs following cell fusion, whereas species with only a single MAT encoded homeodomain protein have only one chance to form a productive heterodimer required for completion of sexual reproduction. A second key finding is that the C. heveanensis A locus (PR) is at least biallelic; when additional isolates become available it will be possible to test if other extant alleles are present (triallelic or multiallelic). Previous studies of U. maydis, which is tetrapolar with a biallelic pheromone/pheromone receptor locus [75], and the closely related species Sporisorium reilianum [77], which is tetrapolar and triallelic at the pheromone/pheromone receptor locus, provide evidence that transitions have occurred in other fungi from multiallelic/multiallelic to multiallelic/biallelic mating type determination systems. These transitions involving the loss of allelic diversity at one of the two unlinked mating type alleles result in a relative decrease in outcrossing frequency but do not alter the restriction to inbreeding engendered by the tetrapolar state [17]. Thus, a loss of allelic diversity at the A locus encoding pheromone and pheromone receptors may have similarly occurred in the lineages leading to C. heveanensis. The outcome for the species then in both cases is a restriction in outbreeding potential, which may serve to further promote inbreeding and clonality in the population, and this transition in sexuality may have contributed to the emergence of this clade of organisms as successful pathogens of animals, including humans.

Materials and Methods

Strains and media

Strains used in this study are listed in Table 1. C. heveanensis type strain CBS 569, CBS 9459, and CBS 6097 were obtained from the Centraalbureau voor Schimmelcultures (CBS), Netherlands. Other strains of C. heveanensis were from the BIOTEC Culture Collection (BCC), Thailand [70]. Strains were grown in YPD broth or on YPD solid medium. Mating assays were performed on V8 medium (pH 5 and pH 7) [82] and Murashige and Skoog (MS) medium minus sucrose (Sigma-Aldrich, [83]) by pairwise inoculation of strains and incubation at room temperature in the dark on dry upright media without parafilm.

Standard and Latin descriptions of the sexual stage

Standard description: Kwoniella heveanensis Metin, Findley, & Heitman sp. nov.

Etymol.: The epithet is chosen to be identical with that of Cryptococcus heveanensis (Groenewege) Baptist & Kurtzman comb. nov [84].

Heterothallic fungus. Hyphae dikaryotic, clamp connections fused. Basidia clustered, submerged initially, finally on aerial hyphae, globose, 7.5–10.5 µm, cruciately septate, sterigmata not observed. Basidiospores subglobose, 2.3–3.3 µm, germinating by conidia.

Latin description: Kwoniella heveanensis Metin, Findley, & Heitman sp. nov.

Species heterothallica. Hyphae dikaryoticae, fibulae fusae. Basidia aggregata, primum submersa, deinde superficialia, globosa, 7.5–10.5 µm diam, septis cruciatis divisa, sterigmatibus carentia, Basidiosporae subglobosae, 2.3–3.3 µm diam, cito conidia proferentia.

Fluorescence-activated cell-sorting (FACS) analysis

The protocol for the preparation of cells for FACS was based on a previous protocol [85]. Briefly, ∼50 µl of cells scraped from overnight grown cultures of YPD plates were washed with PBS and fixed in 1 ml of 70% ethanol overnight. Then the tubes were centrifuged and the supernatants were discarded. After washing the cells with 1 ml of NS buffer (10 mM Tris-HCl pH 7.2, 0.25 M sucrose, 1 mM EDTA, 1 mM MgCl₂, 0.1 mM CaCl₂, 0.1 mM ZnCl₂), they were resuspended in 180 µl NS buffer. Then, 20 µl RNase A (10 mg/ml, Qiagen) and 5 µl of propidium iodide (0.5 mg/ml, CALBIOCHEM) were added. After incubating at room temperature in the dark with mixing for at least 2 hours, 17.5 µl of the cells were mixed with 700 µl of 50 mM Tris-Cl pH 7.5 and the tubes were sonicated for 45 seconds. Flow cytometry was performed on 10,000 cells and analyzed on the FL1 channel with a Becton-Dickinson FACScan.

Microscopy

For microscopic examination of mating, either the mating mix was inoculated into V8 medium (pH 5) medium plates, and later a block of agar containing the hyphae and basidia was transferred to microscope slides, or the mating mix was directly inoculated onto slides containing a thin layer of mating medium.

For staining, a modified version of a previously published protocol was used [86]. Calcofluor white (0.05% solution, BD) and Sytox green (5 mM, Invitrogen) were used to stain the cell wall and nuclei, respectively. The slide containing the agar piece with hyphae or basidia was first washed with PBS, then, 5–10 µl of Calcofluor white was added and mixed gently with a cover slip. After incubating for 15 min, the excess dye was washed away with PBS. Then fixing was performed by incubating the slide in fixing solution (3.7% formaldehyde, 1% triton-X100 in PBS) for 15 min and the slide was washed with PBS. After that, 5 to 10 µl of 1∶10 diluted Sytox green was spotted on the slide, mixed with a cover slip, and incubated for 15 min. Then the slide was washed with PBS, mounted with a cover slip, and examined under the microscope. Microscopy was performed with an Axioskop 2 plus upright microscope (Zeiss). Images were captured using an AxioCam MRm camera and AxioVision application software (Zeiss). Scanning electron microscopy was performed with an XL30 environmental SEM (FEI).

Fosmid library preparation

For library production, the CopyControl Fosmid Library Production Kit (Epicentre) was used according to the manufacturer's instructions. In each library, a total of ∼16,000 clones were picked into 96 well plates. Purified genomic DNA (∼40 kb fragments) was first sheared, end-repaired, and separated using contour-clamped homogenous electric field (CHEF) on a CHEF DR-II apparatus (Bio-Rad). The following conditions were used: 1- to 6-sec switch time, 6 V/cm, 14°C for 14 to 15 hrs. The size-fractionated DNA was recovered by gel extraction and precipitated. The insert DNA was then ligated into the CopyControl pCC1FOS cloning-ready vector and incubated overnight at room temperature. The ligated DNA was packaged and plated on E. coli phage-resistant cells overnight at 37°C. Fosmid clones were then picked into 96 well plates and eventually transferred to 384 well plates, which were stored at −80°C.

DNA extraction, degenerate PCR, and library screening

For DNA isolation, strains were inoculated in 50 ml YPD medium, grown for two days at room temperature with shaking and centrifuged. After freezing at −80°C, the pellet was freeze-dried overnight and DNA was isolated as previously described [87].

Degenerate primers were designed for LPD1, RPO41, CAP1, ZNF1, STE20, IKS1, FAO1 and NOG2 based on known gene sequences from C. neoformans var. neoformans, C. neoformans var. grubii, and C. gattii. These primers (Table S2) were used in PCR reactions with C. heveanensis genomic DNA using Ex Taq polymerase (Takara) and the products obtained were gel extracted with QIAquick Gel Extraction Kit (Qiagen). The PCR fragments were cloned into pCR 2.1-TOPO using the TOPO TA Cloning Kit (Invitrogen). From the resulting colonies, plasmid isolation was performed with QIAprep Spin Miniprep Kit (Qiagen) and the plasmids were sequenced with M13F and M13R vector specific primers. After obtaining MAT gene sequences, specific primers were designed using a web-based program, Primer3 (http://frodo.wi.mit.edu/). After PCR with specific primers, the resulting gene fragments were gel-extracted as previously described and used as probes to screen the library.

For library screening, colony lifts were performed on Hybond membranes (Amersham) using standard protocols [88]. The PCR products were labeled with Prime-It II Random Primer Labeling Kit (Stratagene) with dCTP. Hybridization was performed using ULTRAhyb hybridization buffer (Ambion) according to the manufacturer's instructions.

Sequencing and assembly

Positive fosmids were isolated from 200 ml of overnight grown cultures using the Qiafilter plasmid midi kit (Qiagen). The fosmid DNA obtained was sheared using a hydroshear apparatus (Genomic Solutions) to 1.6–3 kb fragments. The gel extracted sheared DNA was then ligated to adapters. To prepare the vector, pUC18 plasmid DNA was cut with EcoRI and HindIII removing the polylinker site, and ligated to single stranded oligos that contain one end complementary to the restriction enzyme site, and the other non-complementary to itself to prevent self-ligation. The sheared and adapter-ligated DNA was then ligated to the pUC18-derived vector and the reaction was used to transform DH5α competent cells (Invitrogen). White colonies were inoculated onto 96 well plates and plasmid isolation was performed with DirectPrep 96 MiniPrep kit (Qiagen). For each positive fosmid, four to five 96 well plates were used.

Sequencing reactions were performed with BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems) and analyzed on PE3700 96-capillary sequencer (Applied Biosystems) in the Sequencing Facility at Duke University. The sequence reads were assembled using Phred, Phrap, and Consed program packages [89]–[91]. The gaps were closed by designing primers from contig ends using Primer3, amplifying across gaps by PCR and sequencing both strands. Short gaps were closed using iProof polymerase (Biorad) and long gaps were closed using LA Taq polymerase (Takara).

Southern blot analysis

For Southern hybridization of C. heveanensis strains, 5 µg of genomic DNA from each strain was digested with EcoRV, electrophoresed on a 0.8% agarose gel, and blotted onto Hybond membranes (Amersham) using standard protocols [88]. The membrane was then hybridized to MAT gene probes generated by PCR as described above.

Annotations and bioinformatic analyses

Annotations of the genes were done manually by comparing the 3-frame translated C. heveanensis sequences with the most similar protein sequences identified by BLASTX. Genbank accession numbers are, GU129941 for CBS 569 homeodomain locus, GU205379 for CBS 569 pheromone/receptor locus, GU129940 for BCC 8398 pheromone/receptor locus, GU129942, GU129943, GU129944, GU129945, GU129946, and GU129947 for homeodomain loci of the strains BCC 8313, BCC 8384, BCC 8398, BCC 11754, BCC 11757, AND BCC 15000, respectively. Genbank accession numbers for D1/D2 and ITS regions are GU585738 and GU585749 for CBS 9459, GU585739 and GU585750 for CBS 6097, GU585740 and GU585751 for BCC 8305, GU585741 and GU585752 for BCC 8313, GU585742 and GU585753 for BCC 8384, GU585743 and GU585754 for BCC 8396, GU585744 and GU585755 for BCC 8398, GU585745 and GU585756 for BCC 11754, GU585746 and GU585757 for BCC 11757, and GU585747 and GU585758 for BCC 15000, respectively. The accession number of the ITS region of CBS 569 is GU585748.

Percent sequence identity plots and dot plots were generated using the MultiPipMaker [92]. Pairwise sequence identity values were calculated using Geneious 4.6.4.

The comparison plot between pheromone/receptor loci of strains CBS 569 and BCC 8398 was produced using Artemis Comparison Tool Release 8 [93]. The input comparison file was generated using WebACT with the tBlastx algorithm.

Homeodomain regions of the proteins Sxi1 and Sxi2 were predicted by comparison to the previously characterized homedomain proteins. Helical content of the N-terminal domains of Sxi1 and Sxi2 were predicted with Jpred 3 [94]. Nuclear localization signals were predicted with PSORT II [95].

Phylogenetic analysis

Phylogenetic analysis was conducted on coding sequences using MEGA4 [96]. The phylogenetic relationship was inferred using the Maximum Parsimony method.

Supporting Information

Attachment 1

Attachment 2

Attachment 3

Attachment 4

Attachment 5

Attachment 6

Attachment 7