Sxi1α and Sxi2a are predicted to form a heterodimer and homodimers

To study the impact of the homeodomain proteins Sxi1α and Sxi2a on unisexual reproduction in Cryptococcus, we selected to study XL280α and XL280a, a pair of congenic C. deneoformans strains that are well-known for their ability to undergo unisexual reproduction involving robust self-filamentation and sporulation [30,37]. Sxi1α and Sxi2a are conserved between C. neoformans and C. deneoformans, and homeodomain (Pfam:00046) or homeobox KN domains (from KNOX family transcription factors, Pfam:PF05920) were found within the sequences via an NCBI conserved domain search (Fig 1A). Both domains are structurally related, sharing DNA-binding functions through a helix-turn-helix motif, consistent with potential functional relevance in governing gene expression. Consistent with our previous report with yeast two-hybrid analysis [2,22], a heterodimer between Sxi1α and Sxi2a was predicted by AlphaFold3 protein-protein complex prediction (Fig 1B). Pairwise structural alignments between the predicted Sxi1α-Sxi2a heterodimer and the crystal structure of the S. cerevisiae a1-α2 heterodimer [40] revealed robust structural similarity between the two homeodomain complexes with an RMSD value of 0.561 angstroms between 49 pruned atom pairs (Fig 1C). Similarly, the Sxi1α-Sxi2a heterodimer exhibited high structural similarity to the AlphaFold3-predicted bW-bE heterodimer from the corn smut fungus U.maydis, with an RMSD of 1.069 angstroms across 40 pruned atom pairs (Fig 1C). Interestingly, the HD2 proteins [2], represented by Sxi2a, a1, and bW, superimpose closely, reflecting strong evolutionary conservation. In contrast, greater structural divergence was observed among the HD1 proteins, including Sxi1α, α2, and bE, consistent with their roles in conferring regulatory specificity (Fig 1C). Together, these comparisons support a conserved functional framework in which HD2 provides a structurally stable DNA-binding platform, while HD1 contributes specificity to transcriptional regulation, with heterodimer formation serving as a central regulatory module in mating control [41].

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. AlphaFold3 predicts a heterodimer and homodimers of Sxi1α and Sxi2a.

(A) Predicted domain structures of Sxi1α and Sxi2a in C. neoformans and C. deneoformans. (B) AlphaFold3 (AF3)-predicted heterodimer structure of Sxi1α and Sxi2a from C. deneoformans. C-terminal disordered regions were hidden for visualization. Only regions with a predicted local distance difference test (pLDDT) score >60 were colored and surfaced. The predicted alignment error (PAE) plot of the predicted Sxi1α-Sxi2a heterodimer is displayed in the right panel. (C) Overlay of the AlphaFold3-predicted Sxi1α-Sxi2a heterodimer with the crystal structure of a1-α2 (from S. cerevisiae, PDB:1AKH) and the predicted bW-bE heterodimer (from U. maydis, with high confidence ipTM = 0.74 pTM = 0.64). Pairing was performed with the “Matchmaker” function in Chimera X. (D) AlphaFold3-predicted homodimer structures with PAE plots for Sxi1α-Sxi1α (left panel) and Sxi2a-Sxi2a (right panel). Only regions with a pLDDT score >50 were colored and surfaced. (E) Yeast two-hybrid analysis of interactions among Sxi1α, Sxi2a, and the N-terminal region of Sxi2a (Sxi2aN). Schematics of Sxi2a and Sxi2aN constructs are shown on the left, with the homeodomain indicated. S. cerevisiae strains expressing Gal4 DNA-binding domain (BD) or Gal4 activation domain (AD) fusions were mated, and diploids were selected on synthetic dextrose medium. Protein-protein interactions between bait and prey fusions drive Gal4-dependent activation of ADE2 and HIS3 reporter genes, allowing for growth on medium lacking adenine or histidine. Activation of ADE2 in the ade2 mutant reporter strain background also reduces red pigment accumulation, resulting in lighter colony coloration on nonselective medium. Quantification of lacZ reporter activation was performed using a chlorophenol red-β-D-galactopyranoside (CPRG) liquid assay, and β-galactosidase activity is shown on the right. Error bars indicate mean ± standard deviation from three technical replicates.

https://doi.org/10.1371/journal.pgen.1012084.g001

Previous studies showed that, in addition to forming a heterodimer with a1, S. cerevisiae α2 forms a homodimer to repress a-specific genes [9]. To benchmark the reliability of our structural predictions, we applied AlphaFold3 to the well-characterized yeast a1 and α2. Consistent with experimental evidence, AlphaFold3 predicted robust homodimer formation for α2 but not for a1, supporting the validity of this approach for distinguishing biologically relevant homodimerization events (S1 Fig). Using this validated framework, we next asked whether Sxi1α and Sxi2a might also form homodimers, and we utilized AlphaFold3 to model Sxi1α-Sxi1α and Sxi2a-Sxi2a interactions. The predictions indicated potential low-confidence homodimer formation for Sxi1α-Sxi1α (ipTM = 0.14, pTM = 0.23) and Sxi2a-Sxi2a (ipTM = 0.20, pTM = 0.24) (Fig 1D). These low interface and predicted TM-scores suggest weak or unstable interactions. It is also possible that any homodimerization could involve additional interacting partner(s) that are not yet known, or that Sxi1α and Sxi2a could have independent roles analogous in function but distinct in structure from those observed in ascomycetous yeast.

To determine whether Sxi1α or Sxi2a could form homodimers, we performed yeast two-hybrid (Y2H) analyses with Gal4 DNA-binding domain (GBD) and transcriptional activation domain (GAD) fusion constructs (Fig 1E). Because homeodomain transcription factors frequently function as dimers, we specifically tested whether either protein could self-associate in this heterologous system. We also included the putative C. deneoformans ortholog of Mcm1, a transcription factor involved in the pheromone response pathway in other fungi including S. cerevisiae, where it interacts with the homeodomain protein α2 and transcription factor Ste12 to regulate mating-specific gene expression [42,43]. We identified C. deneoformans Mcm1 (CNN00060) and C. neoformans Mcm1 (CNAG_07924) as orthologs of S. cerevisiae Mcm1 through reciprocal BLASTp searches. Our AlphaFold3 prediction and alignment suggest a conserved structural organization between the S. cerevisiae α2-Mcm1 [44] and the C. deneoformans Sxi1α-Mcm1(CNN00060) (S2 Fig), and we hypothesize that Mcm1 could similarly interact with Sxi1α or Sxi2a in Cryptococcus. Thus, we tested all possible pairwise interactions between Sxi1α, Sxi2a, and Mcm1.

Plasmids encoding the fusion proteins were transformed into the Y2H reporter strains Y2HGold and Y187. Following mating of the two strains, diploids were selected and assayed for Gal4-dependent activation of ADE2, HIS3, and lacZ reporter genes. In a previous study, full-length Sxi2a fused to GBD was shown to auto-activate reporter gene expression [2]. Consistent with this observation, auto-activation was also detected in our assays for both Sxi2a and Mcm1 (S3 Fig). To avoid false-positive interactions resulting from auto-activation, we instead analyzed a truncated form of Sxi2a (Sxi2aN), containing only the N-terminal region and the homeodomain (HD), for interaction testing. A robust protein-protein interaction was detected between Sxi2aN and full-length Sxi2a, supporting the AlphaFold-predicted model of a Sxi2a homodimer (Fig 1E). In contrast, we did not detect evidence for Sxi1α homodimerization, or between Mcm1 and Sxi1α or Sxi2a in this assay (S4 Fig). While these results suggest that such interactions are either weak or absent in this context, we cannot exclude the possibility that dimerization or complex formation may occur in vivo and require post-translational modifications or additional binding partners not recapitulated in the yeast two-hybrid system.

Sxi1α and Sxi2a are dispensable for in vitro growth and virulence-related traits

To further characterize the phenotypic impact of Sxi1α and Sxi2a, we generated sxi1αΔ and sxi2aΔ mutants in XL280α and XL280a, respectively, with a CRISPR/Cas9-based strategy following the previously-established TRACE electroporation protocol [38,39]. Multiple independent drug-resistant transformants (four for XL280α sxi1αΔ::NAT and two for XL280a sxi2aΔ::NEO) were constructed and confirmed to have correct genotypes as deletion mutants based on 5’ junction, 3’ junction, and internal ORF-specific PCR analysis (Fig 2A). Mapping of reads generated from Illumina whole-genome sequencing further confirmed the complete deletion of the genes in all four sxi1αΔ mutants and one of the two sxi2aΔ mutants, and a partial gene deletion in the other sxi2aΔ mutant (Figs 2B and S5). In some of the mutants, elevated read-depth was observed in the flanking regions employed for homologous recombination during gene deletion (Figs 2B and S5). This could indicate potential concatemeric or ectopic insertion [45] of the resistance cassette (Figs 2B and S5). For the partial deletion mutant XL280a sxi2aΔ#6, the deletion occurred between the 5’ microhomology sequence and the second gRNA targeting site, instead of at the 3’ microhomology sequence (S5 Fig). This pattern is consistent with a hybrid integration event in which NHEJ and HR [46] occurred on opposite sides of the construct. Importantly, no sequencing reads mapped to the N-terminal region of the gene that encompasses the conserved homeodomain (S5 Fig). Therefore, Sxi2a is functionally disrupted in this mutant, and the strain was retained for subsequent analyses.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Sxi1α and Sxi2a are not involved in growth, capsule formation, or melanin production.

(A) Generation of independent sxi1αΔ and sxi2aΔ mutants in XL280α and XL280a, respectively. Transformants were verified through internal PCR targeting the ORF, along with 5’ and 3’ junction PCRs specific to the drug resistance marker to confirm proper integration at the genomic locus. (B) Whole-genome sequencing (WGS) reads mapping confirming the complete or partial deletion of SXI1α (red block arrow) and SXI2a (blue block arrow). Genes located within and outside of the mating-type locus are highlighted in green and gray, respectively. (C) In vitro competition assays of the sxi1αΔ and sxi2aΔ mutants, each carrying either the NAT or NEO resistance markers, and their corresponding XL280 wild-type strains. Each competition was assayed after 0, 24, and 48 hours of incubation, with four independent replicates. Error bars indicate standard deviation. (D) Serial dilutions of sxi1αΔ and sxi2aΔ mutants, along with the corresponding XL280 wild-type strains, were spotted on solid YPD media and incubated at 30°C and 37°C. (E) Saturated cultures of the indicated strains, grown in RPMI medium, were stained with India ink to visualize capsule formation. The H99 strain and isogenic cap59Δ mutant served as positive and negative controls, respectively. (F) Overnight cultures of the indicated strains were spotted on Niger seed medium to monitor melanin formation. The wild-type H99 strain and isogenic lac1Δ mutant served as positive and negative controls, respectively.

https://doi.org/10.1371/journal.pgen.1012084.g002

We next performed competition assays to assess the impact of Sxi1α and Sxi2a on fungal fitness during vegetative growth. Equal amounts of mutant and wild-type cells were inoculated in liquid YPD medium and cultured for 48 hours at 30°C. Samples were collected at 0h, 24h, and 48h to assess population dynamics. In all cases, the sxi1αΔ and sxi2aΔ strains maintained similar proportions to wild type during co-culture, indicating that the mutations confer no detectable fitness defect (Fig 2C). Additionally, we characterized the contributions of Sxi1α and Sxi2a to several major virulence traits: thermotolerance at 37°C, capsule formation, and melanin production. Consistent with our previous reports in C. neoformans [23], none of the sxi1αΔ or sxi2aΔ C. deneoformans deletion strains displayed fitness differences compared to the XL280 wild-type strains in these assays (Fig 2D-2F). Taken together, these results indicate that the loss of SXI1α or SXI2a does not affect the fitness of the XL280 strain during vegetative growth and has no detectable impact on virulence-related traits under the conditions tested.

Sxi1α and Sxi2a are dispensable for morphology, sporulation, and cell viability during unisexual reproduction

We next investigated whether Sxi1α and Sxi2a are involved in unisexual reproduction in the XL280 background. When grown on mating- and selfing-inducing MS solid medium [47] as solo-cultures, all of the tested sxi1αΔ and sxi2aΔ mutants produce elongated hyphae, basidia, and basidiospore chains at levels similar to the XL280 wild-type controls (Fig 3A and 3B). No discernable differences in the basidia or basidiospores were observed by scanning electron microscopy (SEM) (Fig 3C). Thus, the deletion of SXI1α or SXI2a did not affect unisexual reproduction morphologically in the XL280 background. Additionally, we collected random meiotic basidiospores by microdissection to test the effect of Sxi1α and Sxi2a on spore germination rates and cell viability and found no significant difference in germination rates between spores produced by the sxi1αΔ mutants and XL280α (p-values: 0.682 (sxi1αΔ#4 vs XL280α) and 0.125 (sxi1αΔ#5 vs XL280α), Welch’s t-test), as well as between spores produced by the sxi2aΔ mutants and XL280a (p-values: 0.591 (sxi2aΔ#6 vs XL280a:) and 0.750 (sxi2aΔ #7 vs XL280a), Welch’s t-test) (Fig 3D). No significant difference in spore germination rate was observed between the congenic wild-type strains XL280α and XL280a (p-value: 0.0856, Welch’s t-test) (Fig 3D).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Sxi1α and Sxi2a are not required for hyphal growth or basidium and spore production.

(A and B) Colony morphology (top), self-filamentation (middle), and sporulation (bottom) phenotypes of independent sxi1αΔ (A) and sxi2aΔ (B) mutants when grown under mating-inducing conditions. Scale bars, 1 cm (upper panel), 100 μm (middle panel) and 20 μm (bottom panel). (C) Scanning electron microscopy (SEM) images of basidia and basidiospores from the indicated solo-culture. Samples were grown on MS media for 1 to 2 weeks before microscopy. Scale bar: 5 μm. (D) Germination rate of random spores dissected from the indicated solo-cultures. At least 3 replicates were performed for each strain (indicated by dots), with error bars depicting the standard deviation among replicates. (E) Competition assays of sxi1αΔ and sxi2aΔ mutants against their corresponding XL280 wild-type strain. The co-cultures were grown on MS medium under mating-inducing conditions. Spores from the periphery, and yeast cells from the center, of the co-culture spots were dissected and analyzed to determine their origin (i.e., wild type or mutant), which were then used to infer the relative competitiveness of the mutant strains. Five different co-culture spots were analyzed for each competition.

https://doi.org/10.1371/journal.pgen.1012084.g003

We previously observed that strains capable of undergoing unisexual reproduction outcompete mutants with unisexual reproduction defects under mating-inducing conditions, attributable to enhanced hyphal growth and sporulation [48]. To assess the effects of SXI1α and SXI2a deletion on this metric of unisexual reproduction efficiency, for each deletion strain, we mixed equal numbers of cells from the mutant (containing a dominant drug-resistant marker) and its wild-type control strain (sensitive to drug), and spotted the mixture on MS medium to induce unisexual reproduction. After robust sporulation was observed (10 days post plating), we dissected two populations: 1) random basidiospores from the periphery where sporulation occurs, and 2) the center of the spot where cells remain mostly as yeast (Fig 3E, left panel). Germinated dissected spores/yeast cells were transferred onto YPD plates containing selective drugs (NAT or NEO) to determine which strain in the mixture they originated from. For both sxi1αΔ and sxi2aΔ, the ratios between mutant and wild type among the basidiospores at the periphery were similar to those among the yeast cells at the center (spores/yeast, p-values: 0.5398275 (sxi1αΔ) and 0.379667 (sxi2aΔ), paired t-test), indicating the loss of Sxi1α or Sxi2a does not affect competitive fitness during unisexual reproduction (Fig 3E, right panel).

The DNA-binding protein Mcm1 functions as a central transcriptional co-regulator of mating type homeodomain proteins during sexual development in budding yeast [43,44]. Because Cryptococcus Sxi1α and Sxi2a have structural and functional commonalities with S. cerevisiae α2 and a1, we questioned whether an analogous co-factor may exist in Cryptococcus. We first examined whether deletion of MCM1 affects unisexual reproduction by generating mcm1Δ mutants in the XL280 background (S6 Fig) and tested their self-fertility under mating-inducing conditions. Deletion of MCM1 did not result in obvious defects in unisexual mating morphology in either XL280α or XL280a (S7 Fig). To test for potential synthetic effects between MCM1 and SXI1α or SXI2a, we constructed sxi1αΔ mcm1Δ and sxi2aΔ mcm1Δ double mutants (S6 Fig). No clear unisexual mating morphology defects were observed in these strains (S7 Fig). Interestingly, it appears that all of the engineered strains containing mcm1Δ deletion displayed an increased abundance of yeast cells in the center of mating spots, along with a reduced frequency of basidia bearing robust basidiospore chains (S7 Fig). Additionally, we tested the role of Mcm1 in α x a sexual reproduction with an mcm1Δ strain derived from the C. neoformans H99 transcription factor deletion collection [49]. Through mating, we generated the congenic strain KN99a mcm1Δ. Crosses were performed on MS and V8 (pH = 5) media by mating mcm1Δ with either wild-type, sxi1αΔ, or sxi2aΔ strains [23]. Consistent with previous observations [2,22], crosses involving sxi1αΔ or sxi2aΔ exhibited clear mating defects in hyphal development and sporulation (S8 Fig). In contrast, no detectable mating defects were observed in either unilateral or bilateral α x a crosses involving mcm1Δ (S8 Fig). Collectively, these results indicate that, unlike its essential co-regulatory role in budding yeast, Mcm1 is largely dispensable for both unisexual and α x a sexual reproduction in C. deneoformans and C. neoformans.

Sxi1α is critical for unisexual cell-cell fusion

In the presence of a helper strain serving as a pheromone donor, α x α and a x a outcrossing can be facilitated in C. deneoformans in so-called Ménage à trois matings [22,30] (Fig 4A). Previously, our results from α x a sexual reproduction suggested that Sxi1α and Sxi2a function after cell-cell fusion [2]. Although cell-cell fusion is dispensable for solo unisexual reproduction, it may still occur under certain conditions, such as during α x α outcrossing [22,30]. Given the limited understanding of the role of Sxi1α in this context, we sought to determine whether Sxi1α is involved in cell-cell fusion during α x α outcrossing in the presence of a donor strain (JEC20, MATa NAT^S NEO^S). Strains in the XL280α SXI1α and sxi1αΔ backgrounds that were differentially resistant to NAT and NEO were used to set up three types of mating: SXI1α x SXI1α, SXI1α x sxi1αΔ, and sxi1αΔ x sxi1αΔ. Equal number of cells of the two strains in each mating were mixed and spotted on V8 agar (pH = 7) to induce sexual reproduction. After 72 hours of incubation, cells were collected. The fusion frequency was fewer than 5.04x10^-8 in two independent sets of the XL280α NAT x XL280α NEO crosses, with fusion products identified by screening colonies for NAT and NEO resistance. In contrast, significantly higher fusion frequencies were found in sxi1αΔ::NAT x sxi1αΔ::NEO (22.5-fold higher than wild type, frequency: 1.13x10^-6, p-value: 0.0013, Welch’s t-test) and sxi1αΔ::NAT x XL280α NEO (62.9-fold higher than wild type, frequency: 3.17x10^-6, p-value: 0.0288, Welch’s t-test) (Fig 4B). Taken together, these findings demonstrate that cell-cell fusion between MATα cells is significantly increased when SXI1α is absent in at least one of the two strains.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Fusion frequency and ploidy analysis of products from α x α and a x a Ménage à trois matings.

(A) Schematic of the Ménage à trois mating assay. In each mating-type background, two strains carrying different selectable markers (NAT or NEO) were co-cultured with a pheromone donor of the opposite mating type to induce cell-cell fusion. NAT and NEO resistant colonies were recovered on selective medium. (B-C) Quantification of α x α (B) and a x a (C) fusion frequencies for the indicated bilateral and unilateral crosses. Cells from the indicated mating assay on V8 agar (pH = 7.0) for 72 h were plated on YPD + NAT + NEO and YPD to determine the ratio of cell-cell fusion products in the population. Data are presented as mean ± SD from four to five independent biological replicates; each point represents an individual replicate. ns, not significant; * (p-value < 0.05), ** (p-value < 0.01), *** (p-value < 0.001) indicate statistical significance compared to wild-type control. (D) Fluorescence-activated cell sorting (FACS) analysis of DNA content in parental strains and fusion products. H99 (1N control) and CnLC6683 (2N control) originate from the same experiment; their profiles are duplicated in both the left and right panels to facilitate comparison. Fusion products (highlighted by dashed boxes) are annotated with their mating type(s) as determined by genotyping PCR (right). 1N and 2N indicate haploid and diploid DNA content, respectively. (E) Proportion of mating type of fusion products from the indicated crosses. (F) Principal component analysis (PCA) of 3’ enriched RNA sequencing data from WT (XL280α NAT x XL280α NEO x JEC20) and sxi1αΔ bilateral (sxi1αΔ::NAT x sxi1αΔ::NEO x JEC20) Ménage à trois mating harvested after 72 h on V8 (pH = 7) medium. Each point represents an independent biological replicate. The percentage of variance explained by PC1 and PC2 is indicated. (G) Volcano plot showing differential gene expression between sxi1αΔ bilateral and WT ménage à trois mating. Genes classified as differentially expressed are highlighted in red based on | log₂ fold change| ≥1 and adjusted p-value ≤ 0.05 thresholds, while non-differentially expressed genes are shown in gray. Selected genes of interest are labeled. (H) Bar plots showing normalized CPM expression levels for representative differentially expressed genes and non-differentially expressed genes across WT and sxi1αΔ bilateral ménage à trois mating. Data are presented as mean ± standard deviation across biological replicates. Asterisks indicate statistically significant differences. Statistical significance was assessed using two sample Welch t tests on TMM normalized log₂ CPM values for each gene, with p values adjusted for multiple testing with the Benjamini–Hochberg method. Genes labeled as DEG correspond to those classified as differentially expressed in the DEG analysis.

https://doi.org/10.1371/journal.pgen.1012084.g004

We then asked whether Sxi2a has a similar effect on cell-cell fusion during a x a unisexual mating (Fig 4A). Interestingly, with the presence of a donor strain (JEC21, MATα, NAT^S NEO^S), the fusion frequency between XL280a NAT and XL280a NEO was 2.43x10^-6, while no significant difference in fusion frequency was found in sxi2aΔ::NAT x sxi2aΔ::NEO (20% of the wild type, frequency 5.43x10^-7, p-value: 0.1204, Welch’s t-test). In contrast, the fusion frequencies between sxi2aΔ::NAT x XL280a NEO (7.2-fold higher than wild type, frequency 1.79x10^-5, p-value: 0.0086, Welch’s t-test), and XL280a NAT x sxi2aΔ::NEO (4.6-fold higher than wild type, frequency 1.13x10^-5, p-value: 0.0002, Welch’s t-test) were significantly higher than the wild-type cross (Fig 4C).

We hypothesized that NAT and NEO resistant colonies, carrying both selectable markers after cell-cell fusion, would exhibit diploid DNA content, or triploid if the fusion involved a non-selective donor strain (Fig 4A). To verify the ploidy of fusion products from α x α and a x a outcrosses, flow cytometry analysis was performed. With C. neoformans haploid strain H99 and diploid strain CnLC6683 as controls [50], all NAT- or NEO- containing outcrossing parental strains were confirmed to possess haploid DNA content (Fig 4D). Two independent NAT and NEO resistant colonies from Ménage à trois matings were randomly selected for analysis. As expected, in the α x α fusion assay, diploid DNA content was observed in bilateral crosses between sxi1αΔ deletion mutants. In contrast, haploid DNA content was detected in XL280α NAT x XL280α NEO and sxi1αΔ::NAT x XL280α NEO crosses. The presence of haploid NAT and NEO resistant colonies suggests that our assay may slightly overestimate the true frequency of cell-cell fusion events, likely due to genome recombination during unisexual reproduction. Similarly, in a x a crosses, FACS analysis revealed diploid DNA content in fusion products from XL280a NAT x XL280a NEO, sxi2aΔ::NAT x sxi2aΔ::NEO, and XL280a NAT x sxi2aΔ::NEO, whereas haploid DNA content was found in the sxi2aΔ::NAT x XL280a NEO cross (Fig 4D). To further confirm whether the donor cell directly participated in fusion events, genotyping PCR was performed to amplify STE20a and STE20α, allelic variants of the same mating-type-specific genes located with the MAT locus. None of the independent fusion products contained the STE20 allele from the pheromone donor’s mating type, indicating that triploid products involving the donor strain are rare and were not observed in our analysis (Fig 4D and 4E).

To dissect the transcriptional differences underlying increased cell fusion frequency, we performed 3’ enriched RNA-sequencing for wild-type and sxi1αΔ bilateral Ménage à trois matings conducted on V8 (pH = 7) for 72 hours, the identical conditions used for fusion frequency assays. Wild-type samples consisted of XL280α NAT x XL280α NEO x JEC20, while sxi1αΔ bilateral samples consisted of sxi1αΔ::NAT x sxi1αΔ::NEO x JEC20. Libraries were prepared with a 3’ enriched RNA-seq protocol incorporating unique molecular identifiers to enable deduplication and accurate transcript quantification, yielding an average of 11.75 million aligned reads per sample (S9 Fig). Principal component analysis revealed clear separation between wild-type and sxi1αΔ samples along PC1, which explained 63.2 percent of the variance, while PC2 accounted for 19 percent of the variance and showed partial overlap between groups (Fig 4F). Differential expression analysis with an adjusted p-value threshold of 0.05 or less (adjusted p-value≤0.05) and an absolute fold change cutoff of at least two (Fold Change ≥ 2) identified 134 differentially expressed genes, including 18 upregulated and 116 downregulated genes in sxi1αΔ bilateral matings relative to wild type (S1 Table). Consistent with previous findings, the known Sxi1α target CLP1 was significantly downregulated in sxi1αΔ matings. Notably, the pheromone receptor gene STE3α and pheromone genes MFα1–3 were significantly upregulated in sxi1αΔ matings (Fig 4G). Because genome-wide DEG analysis applies stringent statistical thresholds that may exclude biologically relevant mating and fusion genes, we next examined normalized expression levels using log2 counts per million (CPM) to directly compare transcript abundance across conditions. Examination of log2 CPM for mating and cell fusion related genes confirmed elevated expression of MFα1–3 and STE3α, but not for the key transcription factor gene STE12α, in sxi1αΔ matings. Pheromone genes at the MATa locus, MFa1–3, were also upregulated, whereas STE3a and STE12a showed no significant changes (Fig 4H). In addition, significant expression differences were observed for additional Sxi1α targets such as CPR2 and for key mating regulators ZNF2 [51] and CFS1 [52], while no significant differences were detected for SXI2a, MAT2, CVA1 or the fusion associated genes PRM1 and KAR5 [53] (Fig 4H).

Sxi1α and Sxi2a regulate expression of unique gene subsets in unisexual crosses

To explore potential transcriptional regulatory roles for Sxi1α and Sxi2a in unisexual and α x a sexual reproduction, we performed RNA-seq of solo-cultures (XL280α, XL280a, sxi1αΔ, and sxi2aΔ), and co-cultures including a wild-type cross (XL280α x XL280a), unilateral crosses (sxi1αΔ x XL280a and XL280α x sxi2aΔ), and a bilateral mutant cross (sxi1αΔ x sxi2aΔ). Reads were mapped to a concatenated assembly consisting of a chromosome-level XL280α genome and the XL280a MATa locus [54], with MATa- or MATα- masked versions used for solo-cultures and the unmasked assembly for co-cultures. We acquired a mean of 22.4 million reads per sample (SD = 7.6 M) and achieved a mean alignment rate of 95.2% (SD = 0.8%). Genes with at least 1 transcript per million (TPM) in at least one sample were retained for downstream analysis. In both α and a solo-cultures, principal component analysis (PCA) of the 500 most variable transcripts showed that the presence/absence of SXI1α or SXI2a explained the majority of the variance (S10A and S10B Fig). One replicate from the sxi1αΔ solo culture clustered separately from the others in both principal component space and in hierarchical clustering of the relative (scaled variance stabilizing transformation (vst) normalized) expression of the top 500 most variable genes (S10C Fig). Therefore, this replicate was excluded from downstream expression analysis. Re-analysis without this replicate resulted in reduced variance between replicates and a larger proportion of variance explained by PC1 (S10D Fig). All four co-culture crosses showed clear clustering of technical replicates and separation by genotype, indicating that transcriptional differences in these experiments are explained primarily by the independent variable with minimal contributions from confounding variables (S1E Fig). PC1 (62% variance) appears to be explained by the presence of a functional copy of both SXI1α and SXI2a, with the samples from the wild-type cross clustering separately from the crosses where one or both genes were deleted. Interestingly, PC2 (25%) shows strong, discrete clustering by genotype, indicating transcriptional differences between crosses lacking SXI1α, SXI2a, or both SXI1α and SXI2a (S10E Fig).

We next performed differential gene expression analysis, treating genes with a false discovery rate (FDR)-adjusted p-value less than or equal to 0.05 and at least a two fold positive or negative expression change as significantly differentially expressed. In the sxi1αΔ comparison, SXI1α was identified as a DEG, and further analysis of the mapped RNA-seq reads showed that reads mapping to SXI1α were concentrated at the same regions with a high mapping rate of the Illumina DNA-seq reads (Figs S11 and 2B). No reads were detected in the open reading frame. This suggests that the 5’ UTR region used for homologous recombination may have integrated ectopically in the genome and/or increased in copy number. A MATα-adjacent gene, FAO1, was also identified as a DEG in the sxi1αΔ solo culture. Reads mapping to this gene were concentrated at the 3’ end that was also part of the homology region used for integration of the selectable marker during SXI1α gene deletion. It is therefore likely that this effect is also an artifact of ectopic integration of the resistance cassette. Based on these observations, SXI1α, FAO1, and SXI2a were excluded from consideration as DEGs.

In solo-cultures, sxi1αΔ had 19 DEGs compared to XL280α, while sxi2aΔ had 23 DEGs relative to XL280a (Fig 5A, 5B and S2 Table). There were notably no shared DEGs between sxi1αΔ and sxi2aΔ (Fig 5C). Amt2 (CNJ01880) [55], homologous to a well-characterized ammonium permease required for mating under low-nitrogen conditions in C. neoformans, was upregulated in the sxi1αΔ mutant. Gene ontology (GO) enrichment analysis of the identified DEGs in the sxi1αΔ mutant revealed two significantly enriched terms, nucleobase transport and transmembrane transport (S12A Fig), whereas no significant enrichment was observed for DEGs from the sxi2aΔ mutant.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. Sxi1α and Sxi2a regulate transcription of distinct subsets of genes during unisexual and α x a reproduction. (A and B) Differentially expressed genes in sxi1αΔ (A) and sxi2aΔ (B) relative to wildtype XL280α or XL280a, respectively, during solo cultures (72h at RT on V8 medium). Dashed lines indicate p_adj = 0.05 and log₂FC = ± 1. (C) Comparison of DEGs in (A) and (B). (D-F) DEGs relative to XL280α x XL280a in crosses between sxi1αΔ x XL280a (D), XL280α x sxi2aΔ (E), and sxi1αΔ x sxi2aΔ (F). (G and H) Comparisons of DEGs in sxi1αΔ (G) and sxi2aΔ (H) mutants between solo cultures and α x a crosses, with overlapping genes highlighted in the boxes.

https://doi.org/10.1371/journal.pgen.1012084.g005

SXI1α and SXI2a have distinct and overlapping transcriptional effects in unisexual and α x a sexual crosses

Given the apparent α- and a- specific transcriptional roles of SXI1α and SXI2a during unisexual reproduction, we next questioned whether these transcription factors may also function independently of one another, in addition to their known function as a heterodimer, during α x a sexual reproduction. We compared transcriptional changes in α x a crosses of opposite mating type XL280 parents deficient for one (unilateral cross) or both (bilateral cross) transcription factors. Comparing all mutant crosses to the wild-type cross XL280α x XL280a, we identified 336, 190, and 202 DEGs in sxi1αΔ x XL280a, XL280α x sxi2aΔ, and sxi1αΔ x sxi2aΔ, respectively (S2 Table). In contrast to the solo-cultures undergoing unisexual reproduction, where the sets of DEGs were markedly smaller and primarily upregulated in sxi1αΔ and sxi2aΔ relative to wild type, DEGs had both increased and decreased expression relative to the XL280α x XL280a co-culture cross (Fig 5D-5F). GO term enrichment analysis revealed that DEGs from the unilateral cross sxi1αΔ x XL280a were predominantly associated with carbohydrate transport and metabolism, including monosaccharide, hexose, and glucose transmembrane transport, as well as carbohydrate catabolic processes (S12B Fig). DEGs from the unilateral cross XL280α x sxi2aΔ showed enrichment for broader carbohydrate metabolic and catabolic processes (S12B Fig). In the bilateral mutant cross sxi1αΔ x sxi2aΔ, enriched terms were similar to those in the sxi1αΔ unilateral cross, with strong representation of carbohydrate transport, catabolism, and glucose-related processes (S12B Fig). These results suggest that Sxi1α and Sxi2a are involved in broad metabolic regulation during sexual reproduction, evidence of their functionality beyond processes directly implicated in sexual development. Further, observations that deletion of SXI1α has a more pronounced impact on carbohydrate-related pathways during mating than deletion of SXI2a provide additional support for the existence of distinct roles for these transcription factors.

There were eight DEGs in common between solo sxi1αΔ cultures and the sxi1αΔ x XL280a cross and five DEGs in common between solo sxi2aΔ and the XL280α x sxi2aΔ cross (Fig 5G and 5H). These genes are largely uncharacterized, with functions that remain unknown, except for DHA1 (CND04870) [56], a secreted protein whose expression is regulated by copper availability and which was identified as a DEG in both modes of sexual reproduction in the sxi2aΔ mutant. Comparison of the predicted protein sequences of these shared genes with S. cerevisiae proteins shows potential orthologs with roles in metabolic processes (S3 Table).

Given the enrichment for metabolic processes in the transcriptomic data, we used Biolog YT phenotypic plates to assess whether these changes translated into altered substrate utilization [57,58]. The Biolog YT panel is specifically designed for yeast metabolic phenotyping and contains 94 wells, each preloaded with a different carbon source, enabling comprehensive assessment of both respiratory and growth-based metabolic activity over time. PCA of metabolic activity at 72, 96, and 120 h showed partial separation between sxi1αΔ and XL280α (PC1 = 24.4% variance) (S13A Fig), with PERMANOVA confirming a significant difference (R² = 0.123, p = 0.026). sxi2aΔ and XL280a showed greater overlap (PC1 = 43%) (S13B Fig), with no significant difference (R² = 0.057, p = 0.389). To pinpoint the substrates contributing to these differences, pairwise t-tests were performed for each carbon source at each time point. In the sxi1αΔ mutant, significant differences (p < 0.05) compared to XL280α were observed for utilization of D-gluconic acid, D-sorbitol, L-malic acid, N-acetyl-L-glutamic acid plus D-xylose, and palatinose at one or more tested time points (S14 Fig). In the sxi2aΔ mutant, utilization of β-methyl-D-glucoside, gentibiose, quinic acid plus D-xylose, D-galactose, D-raffinose, D-sorbitol, and dextrin differed significantly from XL280a in at least one of the tested time points (S15 Fig). These phenotypic differences in carbon source utilization are consistent with the RNA-seq GO term enrichment results, which indicated that the Sxi1α and Sxi2a mutations affect genes involved in carbohydrate transport and metabolism.

The identification of shared DEGs between unisexual and α x a sexual crosses indicates that Sxi1α, and potentially Sxi2a, could have distinct transcriptional effects from their heterodimer interacting partner. These observations, coupled with the computational prediction and yeast two-hybrid assays that Sxi2a may be capable of forming homodimers, led us to question the extent of Sxi1α- and Sxi2a-dependent transcriptional effects occurring independently of the presence of a functional heterodimer interacting partner during α x a crosses. To differentiate between transcriptional events dependent on the Sxi1α-Sxi2a heterodimer as opposed to only Sxi1α or only Sxi2a, we examined the direction of differential expression in DEGs identified in one or more crosses (Fig 6A). Compared to the XL280α x XL280a cross, candidate heterodimer-regulated genes should undergo either a positive (heterodimer-repressed) or negative (heterodimer-activated) change in expression in crosses lacking SXI1α, SXI2a, or both (Fig 6A, top panel). By these criteria, we identified 57 and 65 potential heterodimer-repressed and -activated genes, respectively (Fig 6A, top panel). Consistent with previous reports [26,29], CLP1 (CNB01190) was among the set of heterodimer-activated genes. We next reasoned that any genes whose expression depends only on Sxi1α would have either increased (Sxi1α-repressed) or decreased (Sxi1α-activated) expression relative to wild type in the sxi1αΔ x XL280a cross, but no expression change in the XL280α x sxi2aΔ cross because XL280α has a functional copy of Sxi1α (Fig 6A, middle panel). Strikingly, we identified a larger gene set that fit these criteria than potential heterodimer-regulated genes, with 78 and 109 Sxi1α-activated and -repressed DEGs, respectively (Fig 6A, middle panel). Lastly, we applied similar reasoning to identify DEGs whose expression depends only on a functional copy of SXI2a (Fig 6A, bottom panel). From this comparison, we identified 14 and 27 potential Sxi2a-activated and -repressed genes, respectively (Fig 6A, bottom panel). Surprisingly, CPR2 (CNB04250) [27], a gene thought to be activated by the Sxi1α-Sxi2a heterodimer, was present in the set of 27 genes with potential Sxi2a-dependent activation.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 6. Sxi1α and Sxi2a control mating genes in distinct and opposing manners. (A) Schematic of 3 hypotheses for Sxi1α and Sxi2a function during α x a sexual reproduction. Genes in the first scenario are either positively or negatively regulated by the Sxi1α-Sxi2a heterodimer and undergo similar expression changes in all crosses lacking functional Sxi1α and/or Sxi2a. Genes in the second scenario are regulated by Sxi1α and are differentially expressed in crosses lacking a functional copy of Sxi1α but not in the XL280α x sxi2aΔ cross because expression changes are independent of Sxi2a. Genes in the third scenario are differentially expressed in crosses lacking functional Sxi2a (XL280α x sxi2aΔ and sxi1αΔ x sxi2aΔ) but unchanged in the presence of a functional copy of Sxi2a (sxi1αΔ x XL280a). (B and C) log₂FC expression values of MATα (B) and MATa (C) genes in sxi1αΔ x XL280a, XL280α x sxi2aΔ, and sxi1αΔ x sxi2aΔ mutant crosses relative to the XL280α x XL280a wild-type cross.

https://doi.org/10.1371/journal.pgen.1012084.g006

SXI1α and SXI2a have distinct regulatory effects on the α and a MAT loci

We observed an enrichment of MATα and MATa genes in the Sxi1α- and Sxi2a-only dependent sets of DEGs compared to the heterodimer set, in which only two genes (MFa1 and MFa2) were predicted to be heterodimer repressed (Fig 6A, top panel). Plotting expression changes in all crosses for the MATα and MATa loci revealed an even more surprising pattern, where a majority of genes in each locus appear to be controlled by Sxi1α and Sxi2a in opposing directions. MATα genes tend to have increased expression in the absence of Sxi1α and decreased expression in the absence of Sxi2a, suggesting a negative regulatory role for Sxi1α and a positive regulatory role for Sxi2a (Fig 6B). Conversely, MATa genes generally have positive expression changes in the absence of Sxi2a and negative expression changes in the absence of Sxi1α (Fig 6C). In the case of MATa genes, expression differences in the sxi1αΔ x sxi2aΔ crosses more closely mirror those in the XL280α x sxi2aΔ than in sxi1αΔ x XL280a. Therefore, while sxi2aΔ may result in a smaller set of uniquely differentially expressed genes, Sxi2a may have an important role in driving MATa gene expression change in opposition to Sxi1α. Taken together, these results support distinct roles for SXI1α and SXI2a in both mating-type-specific and more general gene regulation.

Fungal strains and growth conditions

C. deneoformans and C. neoformans strains for this study are listed in S4 Table. The strains were stored in 15% glycerol at -80°C for long-term storage. Fresh strains were streaked from glycerol stocks and maintained on solid YPD medium (1% yeast extract, 2% Bacto Peptone, 2% dextrose) at 30°C.

Spot dilution assays were performed with overnight cultures of the tested strains grown in liquid YPD at 30°C. Strains were diluted to an OD₆₀₀ of 0.5 in sterile water. These cultures were further serially diluted 5-fold and spotted (3 μL for each dilution) onto YPD plates. The plates were incubated at 30°C and 37°C (testing thermotolerance) for 2 days before imaging.

Protein structural predictions

Protein structure predictions were made with AlphaFold3 [64] on the AlphaFold server (https://alphafoldserver.com/). The predicted models and error plots were further processed with Chimera X (v 1.9) [65] for predicted local distance difference test (pLDDT) value filtering, alignment, and visualization.

Yeast two-hybrid assay

Sxi1α, Sxi2a, and Sxi2aN constructs were generated previously [2]. Briefly, SXI1α and SXI2a coding sequences from XL280 were cloned from cDNA into Gal4 DNA-binding domain (GBD) and Gal4 activation domain (GAD) vectors, and truncated SXI2a fragments were generated by PCR amplification of the defined N-terminal region (encoding amino acids 1–217). For this study, the putative C. deneoformans MCM1 coding sequence was amplified from XL280 cDNA and cloned into GBD and GAD vectors to generate yeast two-hybrid constructs. All previously generated and newly constructed plasmids were sequence-verified prior to use (S4 Table). Reporter strains Y2HGold and Y187 were transformed with plasmids via the high-efficiency transformation method [66]. Diploids were generated by mating and selected for on synthetic dextrose (SD) dropout medium (0.67% yeast nitrogen base, 2% dextrose) minus leucine and tryptophan. Adenine sulfate was added to the medium at a concentration of 20 mg/L, and amino acids and uracil were included at standard concentrations to support auxotrophic growth requirements. Gal4-dependent activation of ADE2 and HIS3 reporter genes was assessed by growth on SD medium additionally lacking adenine or histidine, respectively. β-galactosidase activity was measured using a chlorophenol red-β-D-galactopyranoside (CPRG)–based liquid assay (OZ Biosciences) according to the manufacturer’s instructions. Absorbance was measured at 595 nm, and enzyme activity (milliunits) was calculated based on a linear standard curve.

Gene deletion and confirmation

Approximately 1.0 kb 5’ and 3’ flanking regions of the SXI1α gene and 60 bp 5’ and 3’ flanking regions of the SXI2a and MCM1 genes were constructed with primers listed in S5 Table to target their respective genes in XL280 for homologous recombination [46]. Overlap PCR with a selectable marker was used to construct the DNA donor with 1.0 kb homology, while 60 bp microhomology was included in the primers for donor amplification. C. neoformans codon optimized Cas9, gRNAs (a single gRNA targeting SXI1α and two gRNAs targeting SXI2a), and drug resistance cassettes (NAT from pAI3, NEO from pJAF1 or HYG from pJAF15) containing homologous flanking sequences were introduced into XL280 wild-type strains via the previously reported Transient CRISPR-Cas9 Coupled with Electroporation (TRACE) system [38,39]. Transformants were screened with PCR internal to the ORF, and 5′ and 3′ junction PCRs to the drug resistance marker integration sites to confirm integration at the genomic locus. Illumina whole-genome sequencing was performed on positive transformants with the Novaseq X Plus platform (2x150 bp). The sequencing reads were mapped to a modified version of the XL280α genome [54] with the Burrows-Wheeler Alignment tool (BWA) (v 0.7.17) [67] to further confirm the deletion mutant strains. The modification included adding the MATa locus sequence of JEC20 (GenBank: AF542530.2) as an additional contig.

H99 mcm1Δ was obtained from [49] and crossed with KN99a to obtain KN99a mcm1Δ. Both mutants were verified by internal PCR targeting the MCM1 ORF with the primers listed in S5 Table.

Capsule formation and melanin production assays

Tested strains, with an initial OD₆₀₀ of 0.1 in 4 mL of liquid RPMI medium (RPMI 1640 with 2% dextrose), were incubated on a roller drum at 70 rpm and 30°C for 3 days. After washing and resuspension in water, the cells were negatively stained with an equal volume of India ink to visualize the capsules.

To assess melanin production, overnight saturated cultures of test strains were spotted (10 μL) onto Niger seed agar plates (7% Niger seed, 0.1% dextrose). The Niger seed plates were incubated at 30°C or 37°C for 8 days before photographing.

Competition assays

Overnight cultures of tested strains grown in liquid YPD at 30°C were diluted to OD₆₀₀ of 0.2 and equally mixed in a 1:1 ratio in 5 mL of liquid YPD for co-culture. sxi1αΔ::NAT mutants (resistant to NAT) and sxi2aΔ::NEO mutants (resistant to NEO) were mixed with XL280α and XL280a, respectively. The mixture was plated on YPD plates, YPD + nourseothricin (NAT), and YPD + neomycin (NEO) to confirm the starting ratios. After 24 and 48 hours of co-culture, an aliquot of the mixture was sampled, and plating was repeated with the OD₆₀₀ diluted to 0.1 to track changes in relative abundance of mutants. Four independent replicates were performed for each competition.

For the competition assay performed on MS medium, overnight cultures were similarly diluted to OD₆₀₀ of 1. sxi1αΔ::NAT or sxi2aΔ::NEO mutants and the corresponding XL280 wild-type strain were equally mixed in a 1:1 ratio and spotted on MS medium. Spores were dissected 10–14 days post-spotting. Yeast cells from internal regions were also collected. Spores or yeast cells were further transferred to YPD + NAT, YPD + NEO and YPD to determine the ratio of mutants in the population. Five independent spots were investigated for each competition.

Mating and cell-cell fusion assay

Mating plates were set up as described previously [68]. α x α and a x a unisexual reproduction assays were performed on Murashige and Skoog (MS, Sigma, USA) plates for 1–2 weeks before imaging and dissection. Basidiospores generated during unisexual reproduction were randomly dissected and plated on YPD plates for 2–3 days to determine spore viability and germination rates.

Cell-cell fusion frequency during α x α unisexual outcrossing in XL280α NAT (XL561 and XL562) and XL280α NEO (XL563 and XL564) was compared to bilateral mutant crosses of sxi1αΔ::NAT and sxi1αΔ: NEO or unilateral crosses of sxi1αΔ::NAT and XL280α NEO. Similarly, a x a unisexual outcrossing in XL280a NAT (CF1816) and XL280a NEO (JHG223) was compared to bilateral mutant crosses of sxi2aΔ::NAT and sxi2aΔ::NEO or unilateral crosses of sxi2aΔ::NAT and XL280a NEO or sxi2aΔ::NEO and XL280a NAT.

Cells were diluted in sterile water to an OD₆₀₀ of 1. Two MATα strains (α x α outcrossing) or MATa strains (a x a outcrossing) containing either NAT or NEO resistance cassettes were mixed equally with JEC20 (MATa) or JEC21 (MATα) for Ménage à trois matings on V8 agar (pH = 7) and incubated for 72 hours in the dark at room temperature. The cells were then collected and plated on YPD + NAT (100 μg/mL) + NEO (200 μg/mL) and YPD alone to determine the ratio of colonies with dual resistance four to five days post-plating (i.e., fusion product) out of the total cells. Genotypes at the mating type locus of the fusion products were validated via STE20a or STE20α genotyping PCR.

Flow cytometry analysis

Ploidy of the fusion products from α x α or a x a cell fusion was determined by fluorescence-activated cell sorting (FACS) analysis as previously described [69]. Cells were grown overnight on solid YPD medium at 30 °C, harvested, and washed with 1 mL of PBS. The cells were fixed in 70% ethanol at 4 °C overnight, pelleted and washed with 1 x NS buffer (10 mM Tris-HCl pH = 7.5, 0.25 M sucrose, 1 mM EDTA, 1 mM MgCl₂, 0.13 mM CaCl₂, 0.1 mM ZnCl₂, 0.5 mM phenylmethylsulfonyl fluoride, 7 mM β-mercaptoethanol). Samples were treated with RNase A (1.05 mg/mL) and stained with propidium iodide (PI, 30 μg/mL) in a total volume of 200 μL of 1x NS solution. Stained cells were diluted in 1x Tris-PI buffer (0.964 M Tris-HCl pH = 7.5, 35 µg/mL PI) prior to analysis. FACS was performed at the Duke Cancer Institute Flow Cytometry Core Facility with approximately 10,000 cells analyzed per sample. Flow cytometry data were visualized using FlowJo (v.10.10.0).

Light and scanning electron microscopy (SEM)

Mating structures from MS plates were imaged with an AxioSkop 2 fluorescence microscope to visualize hyphae, basidia and basidiospores under brightfield. SEM samples were prepared following a previously described protocol [50] with minor modifications. Briefly, a ~ 2 cm x 2 cm agar plug containing a patch of mated cells from an MS plate was fixed in PBS containing 4% formaldehyde and 2% glutaraldehyde at 4°C for 2 hours. The fixed samples were then gradually dehydrated in a graded ethanol series (30%, 50%, 70%, 95%), with each step performed at 4°C for 15 min. This was followed by three washes in 100% ethanol, each at 4°C for 15 min. The samples were further dehydrated with a Ladd CPD3 Critical Point Dryer and subsequently coated with a thin layer of gold with a Denton Desk V Sputter Coater (Denton Vacuum, USA). Hyphae, basidia and basidiospores were observed with a scanning electron microscope equipped with an EDS detector (Apreo S, ThermoFisher, USA).

RNA preparation and sequencing

For Ménage à trois matings, cells were diluted in sterile water to an OD₆₀₀ of 1. Two MATα strains (α x α outcrossing) containing either NAT or NEO resistance cassettes were mixed equally with JEC20 (MATa) on V8 agar (pH = 7) and incubated for 72 hours in the dark at room temperature before harvesting.

Three biological replicates were performed for each strain or mating pair. Harvested samples were lyophilized and processed for RNA extraction using the RNeasy Mini Kit (Qiagen), following the manufacturer’s instructions. Library preparation and sequencing was performed by Plasmidsaurus (USA). RNA libraries were generated from total RNA using a strand-specific 3’ end counting protocol with unique molecular identifiers (UMIs) to enable accurate transcript abundance quantification (https://plasmidsaurus.com/rna). Libraries were sequenced on the Illumina platform, yielding deduplicated 3’ end enriched reads for gene expression analysis.

For both unisexual and α x a sexual crosses, strains were collected from 2 to 3-day-old cultures grown on solid YPD medium and diluted to an OD₆₀₀ of 1. For unisexual crosses, 10 μL of the diluted culture (OD₆₀₀₌1) was spotted onto V8 agar plates (pH = 7), with 10 spots per plate and three technical replicate plates per biological replicate. For α x a crosses, equal volumes of MATα and MATa cell suspensions (each at OD₆₀₀₌1) were mixed prior to spotting under the same conditions. Three biological replicates were performed for each strain or mating pair. Plates were incubated at room temperature for 72 hours in the dark, after which cells were washed off and collected. Harvested samples were lyophilized and processed for RNA extraction using the RNeasy Mini Kit (Qiagen), following the manufacturer’s instructions.

RNA was sequenced by the Duke University Sequencing and Genomic Technologies Core Facility. Poly-A enriched libraries were prepared from samples with RIN ≥ 7 using the Kapa mRNA-seq HyperPrep kit (Kapa Bioscience). Pooled libraries were sequenced on an Illumina Novaseq X Plus instrument in 150 bp paired-end reads.

RNA-sequencing data analysis

For Plasmidsaurus RNA-seq, data was analyzed following Plasmidsaurus’s instructions. Briefly, raw reads were quality filtered by fastp (v0.24.0) [70], then aligned to the XL280α reference genome with an additional MATa sequence from JEC20 using STAR aligner (v.2.7.11b) [71], with non-canonical splice junction disallowed. PCR duplication was removed through UMI-based deduplication with UMIcollapse (v1.1.0) [72]. Alignment quality metrics, strand specificity, and read distribution across genomic features including enrichment toward transcript 3’ ends were assessed using RSeQC (v5.0.4) [73] and Qualimap (v2.3) [74], with results aggregated into a comprehensive quality control report using MultiQC (v1.32) [75]. Gene-level expression quantification was performed using featureCounts (v2.1.1) [76]. PCA were calculated on normalized counts (TMM, trimmed mean of M-values) using Pearson correlation. Differential expression was done with edgeR [77] (v4.0.16) using standard practice including filtering for low-expressed genes with edgeR::filterByExprwith default values. Differentially expressed genes (DEGs) were identified using a significance threshold of adjusted p-value ≤ 0.05 and an absolute log₂ fold change cutoff of 1. FeatureCounts gene level counts were filtered and TMM normalized using edgeR, and log₂ counts per million (CPM) values were calculated. Significance was determined using Welch two sample t-tests with Benjamini Hochberg false discovery rate correction.

For RNA-seq data from both unisexual and α x a sexual crosses, Illumina universal adaptors were trimmed from paired reads with CutAdapt (v4.9) [78]. Trimmed reads were aligned to the XL280α + MATa concatenated reference assembly. Modified assemblies with MATα (Chr4 coordinates 1522691–1656182) and MATa (MATa coordinates 0–148067) masked were generated using the bedtools (v2.31.1) maskfasta function on the concatenated assembly [79]. XL280α background unisexual crosses were mapped to the MATa-masked assembly, XL280a background unisexual crosses to the MATα-masked assembly, and α x a sexual crosses to the full, unmasked assembly with HISAT2 (v2.2.1) [80]. Counts per gene were quantified with featureCounts (v2.0.6) [76] with a gff annotation file generated by Liftoff (v1.6.3) [81] of annotated features from JEC20 and JEC21 [82].

Exploratory data analysis and differential expression analysis were performed with DESeq2 (v1.44.0) [83]. Prior to analysis, transcript counts were normalized by transcripts per million (TPM) within each replicate and genes with at least 1 TPM in 1 sample were retained. Clustering and principal component analyses were performed using vst normalized values for the top 500 most variable genes generated by DESeq2. Differential expression analysis was carried out in DESeq2 with adjusted p-value ≤ 0.05 and an absolute log₂ fold change cutoff of 1 was applied to identify DEGs from the resulting expression data. The R packages EnhancedVolcano (v1.22.0) (https://github.com/kevinblighe/EnhancedVolcano) and pheatmap (v1.0.13) (https://github.com/raivokolde/pheatmap) were used for data visualization. DEG sets were analyzed with the FungiDB (https://fungidb.org/fungidb/app) GO Enrichment tool using the JEC21 genome as the background. Significance was assessed using Fisher’s exact tests with Benjamini-Hochberg correction, retaining terms with FDR < 0.05. Enriched terms were visualized in RStudio as dot plots ranked by log₂(fold enrichment) or FDR, colored by −log₁₀(FDR), and annotated with gene counts.

Biolog phenotypic assays

Overnight cultures grown on solid YM agar (5 g peptone, 3 g yeast extract, 3 g malt extract, 10 g dextrose, 20g agar, pH = 6.2) were collected and diluted in sterile water to an OD₆₀₀ of 0.2. A volume of 100 μL of the diluted samples was inoculated into each well of the Biolog Yeast Identification Test Panel (YT Microplate) and cultured at 28 °C in the dark. Three biological replicates were performed for each strain.

Cell growth was monitored daily with a SpectraMax iD3 plate reader at OD₅₉₀ (oxidation test) and OD₆₀₀ (Assimilation test). Growth on each substrate was normalized using the formula (Abs_sample - Abs_water)/(Abs_glucose-Abs_water). Principal component analysis (PCA) was performed using scaled data from Biolog YT plate assays. Differences in global metabolic profiles between genotypes were assessed using PERMANOVA (adonis2 function, vegan R package), with 999 permutations and Euclidean distance as the dissimilarity metric.

Statistical analysis

Spore germination and unisexual competitiveness were analyzed with paired or unpaired student t-tests in Rstudio (version 2024.04.1) with the “t.test” function with argument “paired = TRUE” for paired tests or “paired = FALSE” for unpaired tests.

Fusion efficiency data were analyzed in RStudio (version 2024.04.1). Welch’s t-tests were performed using the t.test function with the argument “var.equal = FALSE” to compare fusion rates between selected sample groups. Mean values and standard deviations were plotted using ggplot2, with individual data points overlaid.