Sex-Determination System in the Diploid Yeast Zygosaccharomyces sapae

Sexual reproduction and breeding systems are driving forces for genetic diversity. The mating-type (MAT) locus represents a mutation and chromosome rearrangement hotspot in yeasts. Zygosaccharomyces rouxii complex yeasts are naturally faced with hostile low water activity (aw) environments and are characterized by gene copy number variation, genome instability, and aneuploidy/allodiploidy. Here, we investigated sex-determination system in Zygosaccharomyces sapae diploid strain ABT301T, a member of the Z. rouxii complex. We cloned three divergent mating type-like (MTL) α-idiomorph sequences and designated them as ZsMTLα copies 1, 2, and 3. They encode homologs of Z. rouxii CBS 732T MATα2 (amino acid sequence identities spanning from 67.0 to 99.5%) and MATα1 (identity range 81.5–99.5%). ABT301T possesses two divergent HO genes encoding distinct endonucleases 100% and 92.3% identical to Z. rouxii HO. Cloning of MATa-idiomorph resulted in a single ZsMTLa locus encoding two Z. rouxii-like proteins MATa1 and MATa2. To assign the cloned ZsMTLα and ZsMTLa idiomorphs as MAT, HML, and HMR cassettes, we analyzed their flanking regions. Three ZsMTLα loci exhibited the DIC1-MAT-SLA2 gene order canonical for MAT expression loci. Furthermore, four putative HML cassettes were identified, two containing the ZsMTLα copy 1 and the remaining harboring ZsMTLα copies 2 and 3. Finally, the ZsMTLa locus was 3′-flanked by SLA2, suggesting the status of MAT expression locus. In conclusion, Z. sapae ABT301T displays an aααα genotype missing of the HMR silent cassette. Our results demonstrated that mating-type switching is a hypermutagenic process in Z. rouxii complex that generates genetic diversity de novo. This error-prone mechanism could be suitable to generate progenies more rapidly adaptable to hostile environments.

homothallism mating-type evolution chromosomal rearrangement HO endonuclease genetics of sex sex chromosome Sexual reproduction is ubiquitous in eukaryotic organisms, from yeasts to human (Hadany and Comeron 2008). Hemiascomycetes in particular have evolved homothallic and heterothallic repertoires of bipolar mating strategies orchestrated by a single MAT locus, encoding key transcription factors that govern sexual identity and compatibility (Fraser and Heitman 2003). In contrast, other yeasts, such as Candida albicans, have developed alternative cryptic sexual cycle governed by a same-sex mating. The variability in mating system and sex chromosome may drastically affect population genetic structure, pathogen evolution, and ecological processes of survival and adaptation (Fraser and Heitman 2003;Bubnick and Smulian 2007;Hsueh et al. 2008), offering an in-deep understanding of factors that shape sex evolution, one of the major challenges in biology (Billiard et al. 2012).
In the haplo-diplontic yeast Saccharomyces cerevisiae, the MAT locus is located in centromeric region of chromosome III (CEN-MAT linkage) in two versions (idiomorphs), either MATa or MATa genes, enabling yeast to specify three cell types: haploid a, haploid a, and diploid a/a. In heterothallic strains of S. cerevisiae, mating takes place between cells bearing complementary MAT idiomorphs. However, S. cerevisiae exists in nature mainly as homothallic diploid strains (Mortimer 2000; reviewed in Greig and Leu 2009), and sexually reproduces in clonal cell populations by meiosis followed by motherdaughter mating (also referred to as haplo-selfing) (Knop 2006). A cassette model for mating-type switching has been proposed and further experimentally verified to explain haplo-selfing in S. cerevisiae (Hicks et al. 1977;Herskowitz 1988;Herskowitz et al. 1992). Matingtype switching is a programmed DNA rearrangement process that occurs in haploid budded cells and converts MATa into MATa, or vice versa (Strathern et al. 1982;Haber 1998). During switching, DNA at the MAT locus is removed and replaced with DNA copied from the heterochromatic silent cassettes near the telomeres of the chromosome III, either HML or HMR. The gene conversion is mediated by a LAGLIDADG homing endonuclease (HO), which catalyzes a sitespecific double-strand break (DSB) at the boundary between the Y sequences unique to the MATa or MATa alleles and the shared flanking Z sequences (reviewed in Haber 2012).
Based on comparative genomic analyses, the HO-catalyzed homothallic switching in the family Saccharomycetaceae arose from an obligate heterothallic ancestor system via a two-step process: (i) the origin of the silent cassettes (after the divergence of family Saccharomycetaceae from other families such as Debaryomycetaceae and the Candida albicans clade); (ii) the recruitment of HO gene, after the occurrence of a whole-genome duplication (WGD) event that split off the Saccharomycetaceae into the pre-WGD and post-WGD species, respectively (Wong et al. 2002;Butler et al. 2004). Despite the conservation of HML and MAT in cis, and of the a genotype at HML, the family Saccharomycetaceae displays consistent variability in idiomorph content and chromosomal organization at the MAT locus (Tsong et al. 2003;Butler et al. 2004;Fabre et al. 2005;Gordon et al. 2011). Unlike S. cerevisiae and closest relatives, other yeasts have no constrained HMR linked to MAT and HML loci on sex chromosome (Fabre et al. 2005). Moreover, the S. cerevisiae MAT loci code for only three proteins (the homeodomain proteins a1 and a2 and the "a-domain" protein a1), whereas an additional gene (MATa2) coding for an HMG domain DNA-binding protein is present in the MATa idiomorph of several species (Butler et al. 2004). Almost all the pre-WGD species retain a stable chromosomal organization with a restricted set of ancestrally conserved genes flanking the MAT locus. On the contrary, in post-WGD species the MAT locus is subjected to a continual process of erosion, leading different genes incorporated into the Z and X regions, making the sex chromosome a hotpsot for deletion and transposition (Martin et al. 2010;Gordon et al. 2011).
The protoploid yeast Zygosaccharomyces rouxii is one of the few pre-WGD species that split off from post-WGD species after the gain of HO gene (Butler et al. 2004). Z. rouxii strains commonly inhabit low a w environments and have been used for centuries as fermented food starters for the production of sugary and salty food, but they can also determine food spoilage, which accounts for huge economical loss to food industry (Solieri and Giudici 2008;Dakal et al. 2014). Z. rouxii traditionally has been considered as a predominantly haploid yeast with a bipolar mating system (Wickerham and Burton 1960). Because sporulation requires a diploid DNA content, the species with a haploid lifestyle, such as Z. rouxii, must first undergo mating between heterothallic a and a cells in response to osmostress. The resulting transient a/a diploid zygote usually enters in meiosis, producing from two to four haploid gametes. Syngamy of homothallic strains is also possible between genetically identical haploid cells by mating-type switching, followed by meiosis to restore the haploid status. Remarkably, alternative modes of reproduction have been observed but poorly investigated. For example, cell fusion could be not followed by nuclear fusion, resulting in a dikaryon that produces haploid buds (Mori 1973). In addition, zygote may lose the meiotic ability and begins clonal euploid/aneuploid lineages (Solieri et al. 2013a). Indeed, Z. rouxii in yeast culture collections have been demonstrated considerable variation in ploidy and karyotype (James et al. 2005;Gordon and Wolfe 2008;Solieri et al. 2008Solieri et al. , 2013a, that corresponds to phenotypic variability in survival under stress cues . Based on these evidences, at least three groups have been delineated and globally referred to as Z. rouxii complex: the group of haploid Z. rouxii, including the strain CBS 732 T , an allopolyploid group composed of strain ATCC 42981 and aneuploid relatives, and the novel diploid species Zygosaccharomyces sapae, which display mainly a clonal reproduction and rarely goes through meiosis resulting in ascospores (Gordon and Wolfe 2008;Solieri et al. 2013a,b). The coexistence in the same phylogenetic group of very closely related species of sexual and putative asexual taxa with similar ecological and physiologic properties raises several questions: (i) is Z. sapae truly asexual, having thus no traces of MAT genes in their genomes? (ii) Alternatively, has asexual species formed recently and, therefore, it still exhibits unfunctional sex related genes? (iii) Is mating-type imbalance possibly responsible for asexual lineages? Recently, the analysis of the MAT structure in haploid Z. rouxii strains revealed a remarkable rearrangement of sex chromosome by ectopic recombination, leading to strains with unusual genetic make-up aaa and aaaa (Watanabe et al. 2013). These evidences support that sex chromosome is prone to nonhomologous recombination in Z. rouxii species complex. However, no evidences about the MAT loci organization have been reported in diploid lineages. In this study, we surveyed the presence and integrity of MAT and HO genes in Z. sapae diploid type strain ABT301 T .

Strains and mating tests
The Z. sapae ABT301 T strain was retrieved from high sugary traditional balsamic vinegar (Solieri et al. 2006(Solieri et al. , 2013b and deposited to the Yeast Collection of the Centraalbureau voor Schimmelcultures (CBS; Utrecht, The Netherlands) and to the Mycothéque de l'Université Catholique de Louvain (MUCL; Louvain-la-Neuve, Belgium) under the codes CBS 12607 T and MUCL 54092 T , respectively. Zygosaccharomyces rouxii strains CBS 732 T , CBS 4837 (mating-type a) and CBS 4838 (mating-type a) were achieved from CBS collection. Strains were cultured and maintained in the yeast extract-peptone-glucose medium (1.0% yeast extract, 1.0% peptone, and 2.0% glucose, w/v). To study sexual compatibility, 2-to 4-d-old cultures of ABT301 T were incubated alone or in mixture to Z. rouxii CBS 4837 or CBS 4838 both on malt extract agar (MEA; Difco) and MEA supplemented with 6% (w/v) NaCl (6%NaCl-MEA), at 27°for 223 wk and examined microscopically using phase-contrast optics for production of conjugated asci.

Standard DNA manipulation
Genomic DNA (gDNA) was extracted from early stationary cultures via the phenol/chloroform method (Hoffman and Winston 1987). The restriction enzymes were purchased from Fermentas (Burlington, ON, Canada); rTAQ DNA polymerase and high-fidelity Phusion DNA polymerase from Takara (Takara Bio Inc., Shiga, Japan) and Thermo-Fisher (ThermoFisher Scientific, Waltham, MA), respectively; and the DNA ligation kit from Promega (Madison, WI). Plasmid preparations, polymerase chain reactions (PCRs), and other standard molecular biology techniques were performed as described elsewhere (Sambrook et al. 1989) or as instructed by suppliers. In particular, standard PCR mixtures (25-50 mL) contained 10 mM Tris-HCl (pH 8.3), 50 mM KCl, 1.5 mM MgCl 2 , 200 mM of each deoxynucleotide triphosphate, 0.4 mM each primer, 0.02 U/mL of rTaq DNA polymerase, and 1002200 ng of template DNA. The thermal program consisted of one cycle of 5 min at 94°followed by 35240 cycles of 94°for 45 sec, 58°for 1 min, and 72°for 2 min. For amplification of DNA fragments .2 kb, PCR mixtures (20 mL) contained 1X Phusion HF Buffer, 200 mM each deoxynucleotide triphosphate, 0.5 mM each primer, 0.02 U/mL Phusion DNA polymerase, and 1002200 ng of template DNA. The thermal program consisted of 1 cycle of 98°for 1 min, 25235 cycles of 98°for 10 sec, 60268°for 30 sec, 72°for 30 sec/kb, followed by 1 cycle of 72°for 10 min. All the PCRs were performed with BioRad T100 Thermalcycler (Bio-Rad Laboratories, Hercules, CA). Primer design was performed using the Primer3 software (Untergasser et al. 2012). Screening of cloning libraries containing PCR products from degenerate primers were performed by sequencing at least three plasmids. All the sequencing reactions were carried out by a custom sequencing service provider (BMR Genomics, Padova, Italy) Cloning of MAT loci Schematic strategy of MAT idiomorphs cloning is outlined in Figure 1. To summarize, degenerate primers were designed based on a set of amino acid sequences that represent highly conserved regions of homologous proteins MATa1, MATa1, and MATa2 from the species S. cerevisiae and Z. rouxii (Supporting Information, Table S1). These degenerate primer pairs were used to amplify via PCR similar conserved regions in Z. sapae gDNA. Individual gel bands from amplified MATa1 and MATa2 PCR products showing predicted sizes of 495 and 578 bp, respectively, were gel-extracted by using the Qiaquik column method (Qiagen) and cloned into pGEM-T Easy vector (Promega). Inserts from recombinant plasmids pAlpha2.2, pAlpha2.8, and pAlpha1.6 were submitted to sequencing in both directions with vector primers T7 and SP6. Similarly, MATa1-targeting degenerated primers were used to amplify via PCR a fragment of expected size of 153 bp, which was gel extracted and cloned as reported previously, resulting in a plasmid pA12 submitted to sequencing, as described above.
The MAT sequences were further extended by inverse PCR and PCR walking using plasmid partial sequences pAlpha2.2, pAlpha2.8, pAlpha1.6, and pA12 as starting points. To summarize, to extend MATa and MATa sequences, gDNA (200 ng) was digested with HhaI and MspI, respectively, and the resulting DNA digests were ligated with T4 DNA ligase (Promega). The digestion/ligation products were 10-fold diluted, 1 mL was used for 25 mL of PCR using rTaq polymerase (Takara), and the primers listed in Table S2. To complete MATa2 sequence, the primer 301_MATA2F1, spanning the 59UTR region of MATa2 open reading frames (ORFs) in Z. rouxii ZYR-O0C18326g locus was used together with an internal MATa2-specific primer (301_MATA2R1) in PCR amplification (Table S2).

Cassette system determination
To verify whether the gene organization around Z. sapae MTL loci resembles those described in other yeast species (Butler et al. 2004;Watanabe et al. 2013), PCR amplification of gDNA was carried out by using primer sets spanning putative MTL-flanking genes (Table S3). To summarize, the first round of long-range PCR was done with highfidelity DNA polymerase (Phusion, Thermofisher) and the external primers 1, 2, 3, A, B, B9, C (Watanabe et al. 2013), and DownMATa1R1 (this study) in 20 mL of reaction volume, following the manufacturer's instructions. Subsequently, a seminested PCR amplification was done using a 1:20 dilution of the previous PCR and internal MTL locusspecific primers. In case of negative results in first round of PCRs, we tested alternative combinations of MTL-flanking genes by direct PCR amplifications from gDNA with the following primers sets: 1, 2, 3/ reverse nested MTL-specific primer (for 59 end flanking genes) and forward nested MTL-specific primers/A, B, B9, C, and DownMATa1R1 (for 39 end flanking genes). Amplified products were purified using the DNA Clean & Concentrator-5 kit (Zymo Research, Irvine, CA) and sequenced with the same primers used in PCRs.
Cloning of HO genes Schematic outline of cloning strategy was reported in Figure S1. To identify highly conserved amino acid sequences, homology comparison among the HO proteins from the species S. cerevisiae (AAA34683; NP_010054) and Z. rouxii (ZYRO0C10428p), as well as the S. cerevisiae VMA intein (AAL18609) was performed by ClustalW2 alignment (Larkin et al. 2007). Relied on the resulting conserved motifs, two degenerate primer pairs, ZrHOF2/ZrHO_R2 and ZrHOF3/ZrHO_R3, were designed and used to amplify the N-and C-terminal coding regions of the putative Z. sapae HO gene, respectively (Table S1). PCR fragments of expected length were gel extracted and cloned as previously reported. The plasmids pHO2.3 and pHO2.8 bearing two inserts coding for putative HO N-terminal portions and pHO3.5 containing an insert covering the HO C-terminal portion were identified by sequencing in both directions. Genomic portions cloned in pHO2.3 and pHO2.8 were joined to the insert cloned in pHO3.5 by PCR Figure 1 Outline of the cloning strategy and the resulting Z. sapae mating-type loci. Small horizontal arrows indicate degenerate primers (dotted) or genespecific primers (solid). Dotted horizontal lines represent unknown genomic sequences. The discovered ZsMTLa and ZsMTLa loci are summarized in colored red and blue boxes, respectively. Divergent copies of ZsMTLa are surrounded by blue (copy 1), green (copy 2), and gray (copy 3), respectively. Abbreviation: iPCR, inverse PCR.
amplifications with primers pairs 301_59HOF1/301_59HOR1 and 301_59HOF3/301_59HOR1, respectively (Table S2). The resulting two partial HO contigs were referred to as copy 1 and copy 2. Subsequently, the full-length ORF sequences of HO copies 1 and 2 were achieved by PCR-based walking. For upstream walking, a forward primer targeting the 59 UTR of Z. rouxii CBS 732 T HO gene (ZYR-O0C10428g) was combined with two HO copy-specific reverse primers (Table S2). The sequences flanking the 39 ends of both copies were covered through a two-steps PCR walking strategy. In the first step, HO copy-specific forward primers were combined with degenerate reverse primer targeting the HO conserved domain FYRDWSG. In the second one, forward HO copy-specific primers were exploited together with a downstream reverse primer, targeting the 39 UTR of Z. rouxii HO gene (Table S2).
gDNA-and PFGE-based Southern blot assays Southern blot assays were performed according to standard procedures described by Sambrook et al. (1989). gDNA (7 mg) was digested with the restriction enzymes listed in Table S4 following the manufacturer's instructions and resolved on 0.8% (w/v) agarose gel in 0.5X TBE buffer. Chromosomal DNA preparation in plug, gel preparation, and pulsed-field gel electrophoresis (PFGE) were performed as previously reported (Solieri et al. 2008). Digested gDNA and chromosomal DNAs separated by PFGE were transferred onto a Hybond-N+ membrane (GE Healthcare, Buckinghamshire, UK) by upward capillary transfer. In both experiments, probe synthesis was performed using a PCR DIG probe synthesis kit (Roche Applied Science, Basel, Switzerland) and detection was carried out by chemiluminescence, using an antidigoxigenin antibody and CDP-star (Roche Applied Science) according to the manufacturer's instructions. Primers engaged in probe synthesis and restriction enzymes for Southern blot assays were listed in Table S4.
Sequence analysis, phylogenetic construction, and protein domain identification Database searches were run with the BLAST server at the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov/ BLAST). Multiple sequence alignments were performed with the ClustalW program at the European Molecular Biology Laboratory (http:// www.ebi.ac.uk/clustalw) and manually refined. Phylogenetic trees were constructed by the neighbor-joining (NJ) method using MEGA version 5.0 from ClustalW alignment (Tamura et al. 2011). Bootstrap support was estimated using 1000 pseudoreplicates for distance analysis. Statistics relating to the identification of Pfam domains of predicted proteins were obtained from PFAM protein family database, version 27.0 (Punta et al. 2012). Structure predictions were obtained with Jpred3 (Cole et al. 2008). Sequence data from this article have been submitted with the EMBL/GenBank Data libraries under accession numbers HG9317122HG931721.

Mating test
We first assessed the mating behavior of ABT301 T strain in pure and mixed cultures with the Z. rouxii mating partners CBS 4837 (matingtype a) and CBS 4838 (mating-type a), respectively. Our previous observations show that ABT301 T rarely formed asci in pure culture on MEA medium, which involved mother and daughter cells that remained attached to each other (Solieri et al. 2013b). No conjugated asci were observed on 6%NaCl-MEA after 14 d of incubation. Furthermore, strain ABT301 T showed no mating reaction with Z. rouxii CBS 4837 or CBS 4838 tester strains, even after 3 wk of incubation both on MEA and 6%NaCl-MEA media (data not shown), suggesting the homothallic state for ABT301 T or that ABT301 T did not respond to Z. rouxii pheromone signaling or that its pheromone expression might be repressed or defective.

Isolation and characterization of Z. sapae MTLa loci
To determine how the mating-type information is retained in Z. sapae genome, we cloned the MATa loci from ABT301 T strain. Two degenerate primer pairs built on highly conserved regions of MATa1 and MATa2 were employed for cloning MATa1 and MATa2 ORFs, respectively. One putative MATa1-coding and two MATa2-coding partial sequences were obtained ( Figure 1). Chromosome walking by inverse PCR and PCR was used to further extent these sequences, resulting in three divergent Z. sapae mating type-like loci a, referred to as ZsMTLa locus copy 1, copy 2, and copy 3 ( Figure 1). Based on a BLAST-type search, two ORFs, namely ZsMATa1 and ZsMATa2, were predicted in each ZsMTLa locus, encoding proteins of 200 and 225 amino acid homologous to Z. rouxii MATa1 and MATa2, respectively, and separated by an intervening 343-bp sequence ( Figure  1). All three ZsMTLa loci displayed an identical organization, with the ZsMATa1 and ZsMATa2 genes orientated in opposite direction on complementary DNA strands, suggesting a configuration similar to those found in S. cerevisiae and other hemiascomycetes (Butler et al. 2004). To establish the genomic location of ZsMTLa loci, we combined Southern blot analysis and PFGE-karyotyping. As previously reported (Solieri et al. 2008), PFGE-Southern blotting failed to clearly resolve the highest molecular weight chromosomes spanning from 1.6 to 2.2 Mbp, and labeled as I, L and L9, respectively ( Figure S2). Hybridization of PFGE-Southern blot with an a-idiomorph specific probe resulted in a double band spanning from chromosome I and L, suggesting that ZsMTLa loci reside on at least two similar high molecular weight chromosomes. The coding regions of ZsMTLa loci were compared each others and with known sequences from Z. rouxii and S. cerevisiae orthologs ( Table 1). The ZsMATa1 genes from ZsMTLa copies 2 and 3 diverged from the ZsMATa1 gene harbored in ZsMTLa copy 1 for 68 and 82 nt substitutions, respectively. The deduced proteins ZsMATa1 copies 2 and 3 were 200-amino acid long and showed less percentage identities with the Z. rouxii counterpart compared to ZsMATa1 copy 1 ( Table 1). The NJ-tree was constructed using a selection of MATa1 sequences from representative taxa of post and pre-WGD species. As expected, ZsMATa1 copies 2 and 3 did not group to Z. rouxii MATa1, but instead clustered separately n  (bootstrapping values of 91 and 99%, respectively), with copy 2 being closer to Z. rouxii MATa1 than copy 3 ( Figure 2A). The alignment of ZsMATa1 copies with Z. rouxii and S. cerevisiae MATa1 proteins revealed the regions of highest similarity inside the MATa-HMG domain (Martin et al. 2010) and the acidic carboxyl terminal end ( Figure 2B), whose integrity is required for DNA binding and vegetative incompatibility, respectively (Philley and Staben 1994). Z. rouxii MATa1 and all ZsMATa1 variants conserved the first three a helices predicted in S. cerevisiae (Martin et al. 2010), whereas they lacked the fourth alpha helix predicted at the C-terminus of S. cerevisiae MATalpha1. Searching in PFam-A database, we found that MATa-HMG domain from ZsMATa1 copy 2 adhered a little better to the consensus profile (PF04769; E-values 6.6e-09) than the homologous regions in copy 1 and 3 (E-value 1.8e-08 and 1.7e-08, respectively). For example, MATa-HMG domain in ZsMATa1 copy 2 had an E159 residue as in S. cerevisiae when in the same position this amino acid was replaced by alanine in ZsMATa1 copy 1 and 3, as well as in Z. rouxii MATa1. Although the consensus profile does not consider this substitution as conservative, it is still detectable in MATa-HMG domain of related species such as Torulaspora delbrueckii, Vanderwaltozyma polyspora, and Candida glabrata. In addition, the MATa-HMG domain of ZsMATa1 copy 3 displayed a H163N substitution compared with ZsMATa1 copies 1 and 2 and Z. rouxii MATa1. However, this position is poorly conserved in the consensus profile for MATa-HMG domain even inside Saccharomycetes. The amino acid substitutions among ZsMATa1 copies occurred mainly at their amino terminal ends, with the most divergent copy 3 displaying 17 unique residues, as well as 12 and 7 common substitutions with copy 2 and copy 1, respectively.
The MATa2 coding sequences from ZsMTLa loci copies 1, 2, and 3 showed 99.9, 77.6, and 67.7% of identities with Z. rouxii MATa2 ortholog, respectively (Table 1). Phylogeny inferred from the MATa2 amino acid sequences of post and pre-WGD species showed a tree topology congruent with the species relationships established by using the MATa1 sequences. ABT301 T genome harbors three MATa2 variants, one (copy 1) clustered with Z. rouxii MATa2, whereas the others (copies 2 and 3) were related but phylogenetically distinct because of a high level of amino acid divergence ( Figure 3A). All three copies contained a conserved HD1 class homeodomain (HD; Pfam PF00046; E-values 4.1e-7, 2.0e-7, and 9.5e-8, for ZsMATa2 copy 1, copy 2, and copy 3, respectively), consisting in a three-helix globular domain which contacts both major groove bases and the DNA backbone (Wolberger et al. 1991;Kues and Casselton 1992) (Figure 3B). Seven residues in helix 3 that contact the backbone with their side chains in S. cerevisiae MATa2 homeodomains also were conserved in Z. rouxii, Z. sapae along with the tyrosine residue (Y10 in Z. rouxii MATa2 just upstream at N-terminal of helix 1; Figure 3B). A further key tyrosine residue with the same structural role in S. cererevisiae MATa2 was indeed replaced by lysine in Z. rouxii (Y150L). The three residues of S. cerevisiae MATa2, which form additional interactions with the DNA minor groove, were conserved both in Z. sapae and Z. rouxii (R146, G147 and R149) (Ke et al. 2002). However, portions of the protein outside the homeodomain which mediate interactions with accessory proteins had a different degree of conservation. The unstructured carboxy-terminal tail of a2 is required for formation of a stable al/a2-operator complex in S. cerevisiae and, thus, for the heterodimer-mediated repression of transcription. This domain is fully conserved in Z. sapae and Z. rouxii MATa2 and largely resembled that found in S. cerevisiae (Mak and Johnson 1993). The intervening flexible hinge that connects the amino-terminal domain and the homeodomain of S. cerevisiae MATa2 mediates the interaction of MATa2a2 homodimer with two subunits of MCM1 and hence its operator binding capacity (Vershon and Johnson 1993). This sequence is more divergent in ZsMATa2 copy 3 compared with ZsMATa2 copies 1 and 2, and between MATa2 proteins in Z. rouxii and S. cerevisiae. The ability of MATa2 to form both homodimers (a2/a2) and heterodimers (a2/a) mainly relies on the integrity of the N-terminal portion (Ho et al. 1994(Ho et al. , 2002. N-terminal homology between MATa2 in Zygosaccharomyces species and S. cerevisiae is less than that found for the homodomains, probably revealing a speciesspecific coevolution of the dimerization binding motifs. ZsMATa2 copy 3 was the most divergent from copies 1 and 2 ( Figure 3A), owing to unique amino acid replacements even if, in a few positions, the residue was different in all three copies ( Figure 3B), suggesting that these amino acid substitutions were less affected by functional constrains.
Isolation and characterization of the Z. sapae MTLa locus A strategy similar to that used for cloning ZsMTLa loci, was carried out to isolate the MATa-like locus from Z. sapae genome (Figure 1). We obtained one single 1641-bp ZsMTLa locus, which included two ORFs encoding putative MATa1 and MATa2 proteins, respectively, separated by a 279-bp intergenic sequence. The 474 bp MATa1-coding ORF, namely ZsMATa1, displayed a putative 51-bp intron and resulted in a deduced ZsMATa1 140-aa sequence 100% identical to Z. rouxii MATa1 ( Figure 4A). With respect to genomic location, PFGE-Southern blotting showed that ZsMTLa locus resides on the single high molecular weight chromosome L poorly resolved from chromosome L9 ( Figure S2B). The MATa1 harbored a conserved HD2 class homeodomain (Pfam E-value, 8.1-e10, PF00046), consisting of an unstructured N-terminal arm and three helices linked by two loops ( Figure 4A) (Kues and Casselton 1992;Anderson et al. 2000).
The MATa2-coding ORF, namely ZsMATa2, was shorter in length that the Z. rouxii ortholog (ZYRO0C18326g) due to a 26-bp deletion. Thus, the deduced ZsMATa2 amino acid sequence is 9 amino acids shorter than Z. rouxii MATa2 and lacks the domain 219 (QAQA-QAANA) 227 (Figure 4, B and C). MATa2 was provided with single MATA_HMG-box, class I member of the HMG-box superfamily of DNA-binding proteins (NCBI's Conserved Domain Database code: cd01389; residues 72-145; E-value 4.31e-06; Figure 4B), coding by a sequence spanning across Ya and X regions. Beyond this putative functional domain, there were a very few spotted similarities with MATa2 annotated in close related species. The inferred joint point responsible for peptide removal from MATa2 in Z. sapae laid on X region and went through an imperfect tandem sequence (CAAGCA/ C) 3 at the nucleotide position 653 ( Figure 4C).

System cassette analysis
In S. cerevisiae, the functional MATa locus is flanked by BUD5 at the 59 end of MATa2 and by TAF2 at the 39 end of MATa1, whereas the silent HMR and HML loci are flanked by YCRWDDta12/YCR097Wa and YCL068C/HCl065W, respectively. In Z. rouxii, several chromosomal arrangements have been revealed in different strains or in different collection cultures of the same strain (Watanabe et al. 2013), suggesting that the MAT locus is an ectopic recombination hotspot. The analysis of 39 end flanking genes showed that SLA2 gene is frequently linked both to MAT and HML cassettes in all the chromosomal rearrangements described in Z. rouxii (Watanabe et al. 2013) and in other hemiascomycetes (Gordon et al. 2011). To assign chromosomal positions and establish neighboring genes of ZsMTL idiomorphs, PCR amplifications across the whole cloned cassettes sapae (ZsMATa2 copy 1, 2, and 3; GenBank: CDM87332, CDM87335, and CDM87338) and orthologous MATa2 annotated in Z. rouxii (ZrMATa2; GenBank: XP_0024978881) and S. cerevisiae genomes (ScMATa2; GenBank: NP_009866). The S. cerevisiae DNA binding homeodomain of MATa2 (Pfam PF00046) consisting in three three-helix globular domains that contact major groove bases and the DNA backbone are indicated by horizontal black bars (Wolberger et al. 1991). Evolutionary conserved key residues involved in DNA binding are highlighted with black asterisks. Green asterisks denote amino acids that take additional interactions with the DNA minor groove in S. cerevisiae MATa2, present in the unstructured tail at the N-terminal side of homeodomain (light blue bar). The unstructured carboxy-terminal tail of S. cerevisiae MATa2 required for formation of a stable a1/a2-operator complex is also shown (red bar). were performed by employing primer sets designed on genes flanking all MAT, HML, and HMR cassettes observed in other Z. rouxii strains ( Figure 5A). To capture possible divergent sequences of 39 end flanking regions of cloned ZsMTL idiomorphs, we designed a further degenerate primer DownMATa1R, spanning the motif FEFYADC of Z. rouxii SLA2 gene (ZYRO0F15862g) (SLA2_D). Positive PCR products were obtained with the primer pairs 2/A, 3/C, 2/DownMATa1R and further screened via seminested approach using primers specific for ZsMTLa, ZsMTLa copies 1, 2, and 3, respectively ( Figure 5A).
No PCR products were gained with the primer 1 on CHA1 (ZYRO0F15774g) gene sequence at MAT. To exclude alternative combinations of flanking genes other than those described by Watanabe et al. (2013), direct PCR was performed combining ZsMTL-copy specific primers and primers laying on other potential flanking genes (Figure 5A). The results of both approaches are reported in Figure 5B. A total of seven mating-type a cassettes were detected. Four were arranged in the following gene order: CHA1_L (ZYRO0F18524g)-ZsMTLa copy 1-SLA2 (ZYRO0F18364g); DIC1-ZsMTLa copy 1-SLA2 (ZYRO0F15862g); DIC1-ZsMTLa copy 2-SLA2 (ZYRO0F15862g); and DIC1-ZsMTLa copy 3-SLA2 (ZYRO0F15862g). The arrangement CHA1_L (ZYRO0F18524g)-ZsMTLa copy 1-SLA2 (ZYRO0F18364g) is consistent with the designation of this locus as silent HML cassette (ZsHML copy 1) (Watanabe et al. 2013). The arrangement DIC1-MAT-SLA2 indicates that three a-idiomorph ZsMTL cassettes are orthologous to MAT expression loci in other pre-WGD species and thus they are labeled as ZsMATa copies from 1 to 3. Furthermore, the 2/DownMATa1R PCR amplicons were positive to all three ZsMTLa copy-specific internal primers, resulting in three additional a-idiomorphs cassettes. These cassettes had a Z. rouxii DIC1-like upstream region and a downstream region (SLA2_D) divergent from those found at 39 ends of Z. rouxii MAT and HML loci (SLA2 gene and ZYRO0F18524g locus, respectively). Based on the syntenic pattern DIC1-ZsMTLa-SLA2_D, these ZsMTLa cassettes were referred to as ZsHML_D copies 1, 2, and 3. Southern blot hybridization on digested gDNA with a ZsMATa1 probe confirmed that at least seven mating-type a cassettes are present in ABT301 T ( Figure S3A). Finally, the downstream region of the ZsMTLa locus resulted to be orthologous to the Z. rouxii SLA2 gene (ZYRO0F15862g). However, the gene at its 59 end was still unknown, since all the PCR amplifications failed. This result suggested that the ZsMTLa could be a MATa expression locus in ABT301 T with an upstream genomic region not conserved between Z. rouxii and Z. sapae. The presence of a single mating-type a-idiomorph locus was also confirmed by gDNA-based Southern blotting ( Figure S3B).

Analysis of Z and X regions
In Saccharomyces species, the MAT, HMR, and HML cassettes share two homologous regions flanking the Y sequences, termed X and Z, which are regarded among the most slowly evolving sequences in the yeast genome (Kellis et al. 2003). Because HO creates a DSB within MAT locus at the junction between Y and Z sequences (Haber 1998), single base substitutions at the region near the Y/Z border are sufficient to inhibit HO-cut MAT switching (Weiffenbach et al. 1983;Nickoloff et al. 1986). To infer the functional state of Z sequences, we determined the extent of the sequence homology in the 39 flanking regions of the eight Z. sapae mating-type cassettes (three ZsHML_D, three ZsMATa, one ZsHML copy 1, and one ZsMATa, respectively). As expected in species with HO endonuclease, Y-Z junction was conserved in ZsMATa1 and all ZsMATa1 genes. The eight Z. sapae mating-type cassettes were always found with the HO site-consensus sequence CGCAGC at the first site of the Z regions. This sequence was also found in C. glabrata (Butler et al. 2004) and represents a variant of the canonical S. cerevisiae recognition sequence (CGCAAC) for the HO site-specific enzymatic cleavage of MAT during switching ( Figure  S4). Both HO site-specific sequences have been shown to be cleaved efficiently by the S. cerevisiae HO in vivo (Nickoloff et al. 1990). The high level of conservation at the Y/Z borders suggests that all the mating-type cassettes could be functional either as putative MAT or HML/HMR donor sequences. Otherwise, base substitutions were observed at the 39 end of Z region. In particular, four mating-type cassettes flanked by Z. rouxii-like 59 and 39 regions, namely ZsHML copy 1, ZsMATa copies from 1 to 3, and ZsMATa, displayed the Z regions 100% identical to those found in haploid Z. rouxii CBS 732 T , whereas the ZsHML_D copies 1, 2, and 3 differed for 8 SNPs from the canonical Z. rouxii Z sequences. Finally, X region analysis showed that ZsMATa2 extends into the X region, whereas the X/Ya junction is Polymerase chain reaction (PCR)-based strategies used for determining chromosomal three cassette system in Z. sapae. Forward and reverse ZsMTL-specific internal primers were used to screen PCR products obtained by using all possible combinations of primers spanning putative MTL-flanking genes (seminested PCR approach); in case of negative results, 59 and 39 PCR walking was done by using all possible combinations of ZsMTL-specific internal primers and MTL-flanking genes primers (direct PCR approach). Small arrows indicate gene-specific primers (solid lines) and degenerate primer (dotted lines). (B) PCR amplification results of the ZsMTL loci from ABT301 T . Left symbols indicate presence or absence of PCR products after seminested PCR; middle symbols indicate presence or absence of direct PCR products at 59 end; right symbols indicate presence or absence of direct PCR products at 39 end. SLA_D, divergent SLA2 gene partial sequence determined with the degenerate primer DownMATa1R; CHA1_L, CHA1 (ZYRO0F18524) located near to the silent HML cassette in CBS 732 T genome; cp, copy. All primer sequences are in Table S3.
upstream the codon stop of ZsMATa2 genes. Consistently to this organization, the X regions in six Z. sapae a and one a-idiomorphs loci differ from those found in Z. rouxii and ZsHML copy 1 for the same 26-bp indel previously described in ZsMATa2 gene ( Figure S5).

Cloning of HO genes
The occurrence of a HO-cleavable site in Z regions of all eight matingtype cassettes suggests that ABT301 T genome could harbor a HO endonuclease gene. To test this hypothesis, degenerate primer pairs were exploited to determine Z. sapae homologs of Z. rouxii HO gene (ZYRO0C10428g) ( Figure S1). Two putative full-leght ORFs, namely ZsHO copy 1 and copy 2, were identified with 100 and 86.2% identities to Z. rouxii HO gene, respectively. The predicted Z. sapae HO proteins have 100 and 92.3% sequence identities to Z. rouxii HO protein. NJ-based phylogeny inferred from amino acid HO sequences showed that ZsHO copy 2 is clearly distinct from ZsHO copy 1 and Z. rouxii HO ( Figure 6A). Southern blotting result on BanII-digested gDNA with a probe able to recognize both ZsHOs was congruent with the occurrence of two gene copies in ABT301 T genome ( Figure  S3C). To determine the chromosome location of HO copies, we performed a PFGE-Southern blotting with the same probe. The results showed that the chromosomal position of ZsHOs differed from that of Z. rouxii HO. In Z. rouxii CBS 732 T the single HO gene is located on the low molecular weight chromosome C (Souciet et al. 2009), whereas in ABT301 T both ZsHO genes appear to be on the same high molecular weight chromosome I, which harbors the ZsMTLa loci ( Figure S2C).
In S. cerevisiae homothallic strains, HO endonuclease is necessary to complete the sexual cycle by inducing the formation of cells with opposite mating-types within a clone. Because Z. sapae is unable to mate heterothallic sexual partner, its sexual reproduction may depend on the ability of some cells to switch mating-types and fuse with related cells. As shown in Figure 6B, the highest homology between both HO genes cloned in Z. sapae and the single HO genes found in Z. rouxii and S. cerevisiae corresponded to conserved motifs characteristic of intein-encoded LAGLIDADG endonucleases (Belfort and Roberts 1997;Stoddard 2005;Hafez and Hausner 2012). The two Z. sapae HO copies mostly differed in positions outside these functional domains ( Figure 6B). With a few exceptions, Z. sapae HOs shared high identity in eight intein motifs lying at their C-and N-terminals, which form the relic of the protein-splicing domain in HO proteins. The intervening sequences around the LAGLIDADG motif in both ZsHOs were conserved and organized in four amino acid domains responsible for HO endonuclease activity. The C-terminal end of S. cerevisiae HO harbors three zinc finger domains thought enhancing the specificity of HO binding (Bakhrat et al. 2004). In ZsHOs these finger domains had the same organization in the primary sequence, even the last HX2C residue was absent. However, this motif can also be deleted from S. cerevisiae HO, without affecting the mating-type switching activity (Bakhrat et al. 2004).
Structural and mutagenesis studies of LAGLIDADG endonucleases, such as HO and PI-SceI in S. cerevisiae, revealed that the region downstream the B motif, the DDR region, although not well conserved in its primary sequence, probably contacts the phosphate DNA backbones of target site through charged lateral chains of key amino acid residues (He et al. 1998;Moure et al. 2002). This hypothesis was supported by the effect of K99A substitution in S. cerevisiae HO that abrogated the mating-type switching activity (Bakhrat et al. 2004). Indeed, there was high identity in the primary sequence of putative DDR regions in both Z. sapae HOs, whereas there was poor similarity with S. cerevisiae HO. Noteworthy, another positive charged amino acid (N97) was found in Z. sapae HOs instead of K99 residue found in S. cerevisiae HO. Similarly, both Z. sapae HOs conserved a few amino acid residues (i.e., D222, G223, R286, K308, D333, K417), that, once replaced in S. cerevisiae HO, hampered the binding and/or endonuclease activities in vivo or in vitro, or are considered functionally relevant by homology modeling with PI-SceI (Meiron et al. 1995;Ekino et al. 1999;Bakhrat et al. 2004;Ezov et al. 2010). On the other hand, exceptions to this conservation were found. For example, residue H475 in S. cerevisiae HO, involved in DNA binding of endonuclease target sequence (Meiron et al. 1995;Ekino et al. 1999), was substituted by proline in both Z. sapae HOs.

DISCUSSION
Recently, nonconventional yeasts isolated from highly stress environments received enhancing attention both for biotechnological exploitation and genome evolution studies. Chronic osmotic stress triggers aneuploidy (Pfau and Amon 2012), increases the genome DNA content (Gerstein et al. 2006;Dhar et al. 2011), and favors chromosome instability (Aguilera and García-Muse 2013). The frequency of sex and the nature of breeding systems affect genome variation and adaptation to stress environments (Lee et al. 2010;Balloux et al. 2003). Although Z. rouxii and relatives are the most relevant osmo and halotolerant food yeasts, research into their mating systems is restricted to the haploid Z. rouxii strains (Butler et al. 2004;Gordon et al. 2011;Watanabe et al. 2013). Previous analysis demonstrated that Z. sapae diploid strains are genetically and phylogenetically distinct from Z. rouxii (Solieri et al. 2013a,b). Here, we examined mating-type system in Z. sapae strain ABT301 T and found that the pattern of ZsMTL loci is completely different from those described for haploid Z. rouxii strains. Based on genome project (Souciet et al. 2009 Here, we exploited three experimental approaches, i.e., MAT gene cloning, PCR MAT cassette placement, and PFGE-Southern blotting, to enroll the MAT loci cooccurring in ABT301 T genome and to inspect their genome configuration. First, we provided evidences that Z. sapae ABT301 T possesses four independent mating type-like loci, resulting in an unusual a,a,a,a genotype. In addition to one ZsMTLa locus harboring MATa2 and MATa1 genes, we identified three MTLa loci, each containing pairs of MATa1 and MATa2 genes. Remarkably, in two of three ZsMTLa loci, MATa1 and MATa2 genes were slightly divergent from those described in the canonical Z. rouxi MATa locus (ZYRO0F15840g and ZYRO0F15818g, respectively). A similar pattern of mating-type gene expansion has been recently found in Hortaea werneckii, a highly halotolerant and heterothallic black yeast, which possesses two divergent MAT1-1-1 genes (Lenassi et al. 2013).
We hypothesize that the presence of three divergent ZsMTLa loci variants could be arisen from two alternative events. One route may consist in the amplification of a chromosomal segment containing the ancestral linked MATa1 and MATa2 genes, leading to paralogs that progressively accumulate mutations in the postduplication period. Potentially, this duplication could also involve the entire sex chromosome due to a chromosome missegregation during mitosis, which provides a diploid progeny with three chromosomes harboring progressively divergent ZsMTLa loci. In the second route, the acquisition  (Bakhrat et al. 2004). In black shading are amino acid positions that are inferred to be critical for Sc-HO activity by analyzing naturally occurred or artificially induced HO mutants or by homology modeling with PI-SceI (Meiron et al. 1995;Ekino et al. 1999;Bakhrat et al. 2004;Ezov et al. 2010). In light gray, divergent positions between HO copies 1 and 2 in Z. sapae are shown. Amino acid identities are reported below the alignment following ClustalW rules: Ã , identity; :, conservative substitution; ., semiconservative substitution. of extra ZsMTL loci on homeologous sex chromosomes may take place after horizontal gene transfer (HGT) or interspecific introgression events. In fungi, interspecies MAT HGTs have been documented in clonal populations with increased adaptive phenotypes to new environments, but the underlying mechanisms are yet poorly understood (reviewed in Richards et al. 2011). Currently, there is no significant evidence to preferentially support one of the proposed alternatives about the generation of divergent ZsMTLa variants. Noteworthy, the amino acid sequence analysis reveals that substitutions among ZsMATa1 or ZsMATa2 copies are not randomly distributed. Accordingly, many residues crucial for transcriptional regulation activities of S. cerevisiae MATa1 and MATa2 are also conserved in the putative Z. sapae orthologs. These findings convey that divergent ZsMATa genes are under a selective driving force aimed to maintain functional integrity of the encoded transcription factors. The retention of three divergent and putatively functional ZsMTLa loci could be favored by the divergent transcription of MATa1 and MATa2 from the intervening promoter located on the intergenic region within each locus. Alternatively, the ZsMTLa extra loci could had been acquired from a close donor species through a very recent HGT event, limiting the sequence divergence among ZsMTLa copies.
The second goal was to establish whether Z. sapae has a HO/MAT cassette system like that in S. cerevisiae and Z. rouxii. Strain ABT301 T possesses two divergent HO genes, coding putatively functional endonucleases which share eight conserved intein motifs and the amino acid residues involved in DNA binding. Again these data hint that both the ZsHO genes are under the same selective pressure and that SNPs in ZsHO copies 1 and 2 are selectively neutral mutations, with negligible effects on gene function. However, in vitro switching tests are advisable to prove this hypothesis. Moreover, the high degree of divergence observed between ZsHO copies 1 and 2 suggests that these genes did not arise from a recent duplication event. Alternatively, they could result from a HGT event between two close yeast species, both bearing functional HO. All species that have HO genes have also silent cassettes (Butler et al. 2004). Although the post-WGD species contain highly variable organization of mating-type MAT locus and HMR/ HML silent loci, the pre-WGD species retain the ancestral gene arrangement DIC-MAT-SLA2 which distinguishes mating-type MAT locus from silent cassettes HML or HMR (Gordon et al. 2011).
To understand how the ZsMTL copies were organized in MAT, HML, and HMR cassettes, we explored the ZsMTL gene surroundings. One MTLa and three MTLa variants have been anchored to flanking regions by PCR amplification using one primer specific to Ya (copy 1, 2, and 3, respectively) or Ya together with a primer annealing on common neighboring sequences found in Z. rouxii genome. Three ZsMTLa loci resulted duplicated in two syntenic patterns. One set, namely ZsMATa copies 1, 2, and 3, exhibits the canonical synteny DIC1-MAT-SLA2. The other set includes three ZsMTLa loci with a gene layout DIC1-ZsMATa-SLA2_D, regarded as ZsHML_D. The ZsMTLa copy 1 locus also fits to CHA1-MAT-SLA2 gene organization (ZsHML copy 1). The SLA2 gene lays at the 39 end of ZsMTLa locus whereas the gene at its 59 end remained unknown. Because the position of SLA2 gene on the right side of MAT is conserved in a number of pre-WGD (Butler et al. 2004;Gordon et al. 2011), we considered the ZsMTLa locus as MATa expression locus. This hypothesis is supported by observing that in S. cerevisiae diploid cells, active MATa1-MATa2 repressor is necessary to turn off the transcription for a set of haploid-specific genes. As being a diploid strain (Solieri et al. 2008), ABT301 T should express MATa1 with the same functional role. Our preliminary expression analysis indicates that both MATa1 and MATa2 are transcribed in ABT301 T strain in standard as well as salt stressed conditions, excluding that the ZsMTLa locus is a silent cassette HMR (data not shown).
To explain the peculiar genetic makeup of Z. sapae mating system, we inferred two nonexclusive scenarios of chromosomal arrangement (Figure 7), considering two assumptions: (i) MAT and HML loci are linked in hemiascomycetes (Gordon et al. 2011); and (ii) HMR and/or MAT loci are located on different chromosomes in Zygosaccharomyces species (Fabre et al. 2005;Souciet et al. 2009;Watanabe et al. 2013). Based on the first scenario, diploid ABT301 T genome bears two genetically distinct sets of sex chromosome pairs, both lacking HMR cassettes. One set contains MATa and MATa Z. rouxii-like sequences linked to ZsHML_D copy 1 and ZsHML copy 1, respectively. The other chromosome pair includes two slight divergent mating-type a loci, namely ZsMATa copies 2 and 3, linked with the homologous ZsHML_D copies 2 and 3, respectively ( Figure 7A). In the second scenario, the diploid ABT301 T strain has an aa genotype, homozygous for the MATa-HML loci (ZsMATa-HML_D copy 1) and heterozygous for the MATa-HML loci (ZsMATa copy 1-ZsHML copy 1 and ZsMATa copy 2-ZsHML copy 2, respectively). Furthermore, consistently with this model, ABT301 T strain displays an homeologous extra-copy of sex chromosome (trisomy) which hosts the most divergent cassettes ZsMATa copy 3-ZsHML_D copy 3 ( Figure 7B). This hypothesis implies that ABT301 T is not an euploid strain with a karyotype that is a multiple of the haploid complement, a status which partially disagrees with our previous data (Solieri et al. 2008). By combining FACS and PFGE, strain ABT301 T and its conspecific ABT601 resulted to be diploid yeasts bearing additional number of chromosomes compared to Z. rouxii. Unfortunately, loss or gain of individual chromosomes similar in size would be hardly detectable even by combining FACS and PFGE. Therefore, the occurrence of an additional sex chromosome hosting syntenic array of ZsMATa copy 3-HML_D copy 3 could be not excluded. However, in both scenarios the lack of HMR cassette implies that ABT301 T may be unable to reproduce by haplo-selfing. The loss of HMR cassette has been previously documented in S. cerevisiae haploid cells, where mutation or deletion of the MATa locus on chromosome III causes reversion to the default MATa mating-type, allowing these MATa cells, termed a-like fakers, to mate illegitimately with strains of the MATa matingtype (Strathern et al. 1982). This event involves mitotic crossover at a frequency of 3.1 · 10 26 (Hiraoka et al. 2000), leading to a deletion between MAT and HMR or a circular chromosome fusing MAT and HML (Hawthorne 1963;Strathern et al. 1979;Haber et al. 1980). a,a homozygous diploid strains have been found via same-mating sex in Cryptococcus neoformans (Lin et al. 2005) and via parasexual cycle in C. albicans (Magee and Magee 2000;Wu et al. 2005;Forche et al. 2008). Among species having the silent cassette system, a,a,a strains have been found in C. glabrata (Srikantha et al. 2003), whereas a,a,a and a,a,a,a strains in Z. rouxii (Watanabe et al. 2013). In Z. sapae, an interchromosomal recombination may lead to the loss of HMR and the subsequent translocation of ZsHOs to the same chromosomes harboring ZsMTLa loci, giving rise to a chromosomal configuration different from that of Z. rouxii CBS 732 T (Souciet et al. 2009). In ABT301 T the resulting aaaa genotype is likely to produce a matingtype imbalance, which determines the clonality as the main mode of reproduction observed in Z. sapae (Solieri et al. 2013a).
Our work provides a first insight to understand how the matingtype system is arranged in Z. sapae diploid genome. A question much harder to be addressed concerns why the Z. sapae genome is provided with a redundant number of divergent MTLa loci. Our hypothesis is that, although the MAT loci are typically nonrecombining genomic regions (Idnurm 2011), sex chromosome is a hotspot for DSBs, translocation, and mutation in Z. sapae. As in the relative Z. rouxii, in Z. sapae mating-type information is shared between two unlinked chromosomes, and this could favor outbreeding instead of inbreeding (Fraser and Heitman 2003). Illegitimate recombination at these "hot spots" can be induced by the exposure of Zygosaccharomyces yeasts to environmental stresses such as high osmotic conditions. This hypothesis is consistent with the results recently reported for haploid Z. rouxii strains (Watanabe et al. 2013) and with many reports that correlate increased DSBs frequencies to the upsurge of mutation rate and genome instability due to errors in DNA synthesis or microhomology-mediated jumps to ectopic templates (Hicks et al. 2010). DSBs that occur in MAT switching could trigger chromosomal rearrangements. When two specific DSBs are introduced simultaneously on separate chromosomes, DSBs-repair occurs via homologous recombination (with or without crossingover) (reviewed by Haber 2006) and in the absence of homology via nonhomologous end joining (Yu and Gabriel 2004), with reciprocal translocations and interchromosomal rearrangements. We speculate that under stress conditions imprecise mating-type switching and homeologous recombination between sex chromosomes further enrich the range of genetic diversity in Zygosaccharomyces species.
Furthermore, in S. cerevisiae the MAT-bearing chromosome III was found to be the most unstable chromosome (Kumaran et al. 2013) in haploinsufficiency (De Clare et al. 2011). Kumaran et al. (2013) suggest that chromosome III aberrant segregation during meiosis is mainly due to fast evolving centromeric sequences to which MAT and HML loci are tightly associated. This chromosome instability causes karyotype variability, giving rise to aneuploid descendants with diverse phenotypes. Chromosomal rearrangements, as well as sex chromosome instability, may result in a divergent adaptation with reproductive isolation and speciation (Dettman et al. 2007). Consistently with this thesis, S. cerevisiae (Magwene et al. 2011) and C. albicans (Forche et al. 2011) increase the number of recombination events in response to stress (fitness-associated recombination; Hadany and Beker 2003) to promote the evolution of complex traits and accelerate the adaptive rate. The present study provides a methodologic approach and sequence information to carry out a large-scale screening of mating-type loci organization in Z. sapae and Z. rouxii. This screening will be instrumental to confirm the role of genome plasticity and sex chromosome instability in stress adaptation. Figure 7 Inferred genomic organization of the three ZsMTLa copies (cp1, cp2, cp3) and one ZsMTLa in Z. sapae ABT301 T . (A) First hypothesis considers a diploid genome with two sex chromosome pairs, namely I and I' and L and L', bearing ZsMATa cassettes 1, 2, and 3, and ZsMATa cassette, which are linked to the putative silent ZsHMLa_D cassettes 1, 2, and 3 and to ZsHML copy 1 (cp1), respectively. (B) Second hypothesis considers an aneuploid number of chromosomes I. A set of three homeologous chromosomes, namely I, I', and I", harbor ZsMATa cassettes 1, 2, and 3, arranged with ZsHMLa_D cassettes 1, 2, and 3, respectively. Chromosomes L and L' are homozygous for the ZsMATa locus, which is linked to a silent ZsHML copy 1 cassette. In both chromosomal arrangements, ZsHO copies 1 and 2 are located on chromosome set I. Chromosomal letters are according to Figure S2. Dotted arrows indicate divergent ZsMTL sequences from Z. rouxii. Chromosomal organization of three cassette system in Z. rouxii haploid strains NBRC 1733 T and CBS 732 T were reported at the bottom for comparative purposes, according to Watanabe et al. (2013) and Souciet et al. (2009), respectively.