Efficient Mating-Type Switching in Candida glabrata Induces Cell Death

Candida glabrata is an apparently asexual haploid yeast that is phylogenetically closer to Saccharomyces cerevisiae than to Candida albicans. Its genome contains three MAT-like cassettes, MAT, which encodes either MATa or MATalpha information in different strains, and the additional loci, HML and HMR. The genome also contains an HO gene homolog, but this yeast has never been shown to switch mating-types spontaneously, as S. cerevisiae does. We have recently sequenced the genomes of the five species that, together with C. glabrata, make up the Nakaseomyces clade. All contain MAT-like cassettes and an HO gene homolog. In this work, we express the HO gene of all Nakaseomyces and of S. cerevisiae in C. glabrata. All can induce mating-type switching, but, despite the larger phylogenetic distance, the most efficient endonuclease is the one from S. cerevisiae. Efficient mating-type switching in C. glabrata is accompanied by a high cell mortality, and sometimes results in conversion of the additional cassette HML. Mortality probably results from the cutting of the HO recognition sites that are present, in HML and possibly HMR, contrary to what happens naturally in S. cerevisiae. This has implications in the life-cycle of C. glabrata, as we show that efficient MAT switching is lethal for most cells, induces chromosomal rearrangements in survivors, and that the endogenous HO is probably rarely active indeed.


Introduction
Sexual reproduction in fungi takes diverse forms [1], some species preferring out-breeding and others inbreeding, either because spores resulting from meiosis conjugate inside the ascus, before germination, or because they mate with daughter cells, through homothallism [2]. Homothallism is formally defined by the fact that a single spore isolated after meiosis is able to undergo a full sexual cycle, without needing to meet a spore from the opposite mating-type. This is in opposition to heterothallism, where conjugation of cells from strains of opposite mating-types is required. Different underlying mechanisms for homothallism exist [1,3,4], MAT are on chromosome II, and HMRa is on chromosome V [43]. This configuration is common to four species in the Nakaseomyces, while one, N. delphensis, has the three loci on the same chromosome like S. cerevisiae. As for the final species, N. bacillisporus, the sequence in its present state does not contain the HMR locus.
C. glabrata has an HO gene homolog, where known domains for nuclear localization, site recognition and endonuclease activity are conserved [44]. This is also true in all Nakaseomyces [42]. Mating-type switching has been suggested by PCR experiments in C. glabrata [45], but no living cell where spontaneous switching has occurred and been followed experimentally has been isolated, such as is possible in S. cerevisiae, where HO haploid cells switch every generation, and pedigrees of cells can be performed to follow this. Recently, induction of HO expression inducing mating-type switching in C. glabrata has been reported, using a constitutive promoter and the endogenous gene, resulting in switching of a MATalpha cell to a MATa cell [46]. Populations of C. glabrata contain strains of both mating-types with some collections exhibiting bias [35,47], and it has been reported that some strains exhibit "abnormal" cassette configurations, i.e, differing from the canonical HMLalpha and HMRa configuration [36]. This is also known in S. cerevisiae [48,49].
It has been proposed that triplicated MAT-like cassettes originated before the HO gene, and that these were used to allow rare mating-type switching events, and that the HO gene, a "selfish" self-transposing gene, was then "domesticated" [50][51][52]. Indeed, K. lactis switches matingtypes inefficiently through mechanisms independent of Ho but dependent on triplicated cassettes [53,54]. The HO gene is part of the family of Homing Endonuclease Genes (HEGs), selfish genetic elements that can propagate through populations. HO is closely related to VDEI, the intein in S. cerevisiae's genome [55]. Many HEGs propagate as self-splicing introns in organelle genomes; the intron encodes an endonuclease that can cut its site in a genome without the intron, upon mating of an intron-containing and an intron-less strain. The cut is then repaired by homologous recombination (HR), using the intron-containing gene as template, resulting in intron propagation [56]. Ho is an intein whose sole known activity in the cell is to cut its recognition site at MAT and that is encoded by a free-standing gene, dissociated from its own recognition site on chromosome III. Endonucleases of the Ho family have rather low activity, recognize large sites but tolerate degenerate bases within the site, as shown by mutagenesis of the Ho and I-Sce I recognition sites [20,21,57].
We have previously published the composition of MAT-like cassettes and comparison of the HO genes in the Nakaseomyces [42]. We now report on heterologous expression of HO from different Nakaseomyces species and from S. cerevisiae in C. glabrata, and the resulting switching events in both directions, with abnormal structures resulting from gene conversion events. We observe that a high frequency of switching is associated with cell mortality.

Strains, cultures and transformation
Strains used for in vivo experiments and for amplification of HO genes are listed in Table 1. Yeast strains are grown in broth or on plates at 28°C, in YDP (non-selective, 1% Yeast Extract, 1% Peptone, 2% glucose), Synthetic Complete medium lacking uracil (SC-Ura, 0.67% Yeast Nitrogen Base without amino acids, 2% glucose, supplemented with all amino acids and adenine) for HO induction experiments. For selection of transformants and maintenance in repressive conditions, strains are grown in SC-Ura added with 2mM each of methionine and cysteine. Stability of plasmids in C. glabrata and S. cerevisiae was confirmed by comparing the number of colonies on SC-Ura and SC+Ura. Transformation of S. cerevisiae and C. glabrata was done according to the lithium acetate transformation protocol from Gietz et al [58].
It must be noted that the S. cerevisiae strain used for HO induction has a mutant endogenous HO gene; while the C. glabrata strains still possess their endogenous HO gene.

Cloning of HO genes into expression plasmids for C. glabrata
The different HO genes were amplified by PCR from genomic DNA on type strains for the Nakaseomyces and on an HO + strain in the case of S. cerevisiae (Tables 1 and 2). SpeI/ SalI  sites were added to primers used for HO gene amplification. Appropriate PCR fragments were cloned at the SpeI/SalI sites into the pYR32 plasmid. The pYR32 plasmid, kindly provided by B. Cormack [63], is a replicative plasmid for C. glabrata, and was also used for transformation of S. cerevisiae. Absence of mutation in all the HO genes (amplified and cloned) was verified by sequencing.

PCR, Southern Blot and Sequencing
To test mating-type switching, we performed PCR on colonies before and after induction, using specific primers, which can discriminate the mating-type at the MAT locus, by amplifying the upstream part of the locus (Tables 3 and 4). In some cases, we also typed HML and HMR (Tables 3 and 4, and see results). Cells from a fresh colony were incubated in 5 μL of NaOH 20 mM 5 min at 95°C in the PCR tubes before placing at 4°C, adding the buffer, primers, dNTPs and Taq polymerase and proceeding with the PCR amplification. For Southern blots, genomic DNAs were prepared using the Qiagen genomic DNA kit, according to manufacturer's instructions. 5 μg DNA was subjected to enzymatic digestions by EcoRV or HindIII/PstI. Gels (0.8% agarose in 0.5x TBE) were transferred on positively charged Nylon membranes from Roche, using an Appligene vacuum blotter. Membranes underwent pre-hybridization for 4 hrs and overnight hybridization at 65°C in Church's buffer [64], with a DIG-labeled PCR probe, and were submitted to high stringency washes at 65°C in 0.1% SDS, 1X and then 0.1X SSC buffer. DIG labeling and detection was done using a Roche kit, according to manufacturer's instructions. Primers used for probe PCR amplifications are given in Table 2. For the three cassettes MAT, HML and HMR, the whole locus was amplified using primers indicated on Tables 3 and 4, and purified PCR fragments were sequenced by Beckman Coulter Inc.

Induction of HO expression and quantification of the efficiency of matingtype switching
Experimental procedure is shown on Fig 1. Transformants with the different HO plasmids were streaked onto repressive medium and directly tested by PCR (see above) to check that they had not switched before induction. Cells from a starter overnight culture in repressive conditions (SC-URA +2mM Met/Cys) were then counted, and around 200 cells were usually plated on inductive (SC-URA) and on repressive medium as control, and incubated at 28°C, until colonies appeared (one to two days). Determination of mating-type was done on individual colonies (a minimum of 14 colonies were tested per transformant plated, and at least 2 transformants were tested per experiment), by PCR amplification of both MATa and MATalpha.
At this step, we obtain "negative" colonies (those that had not switched and yielded a PCR product only with primers amplifying the original mating-type), and "mixed" colonies (colonies yielding a PCR product with both pairs of primers, in which some cells have switched, and others have maintained their original mating-type).
At least two "mixed" colonies were then restreaked on repressive medium, cells allowed to grown into individual colonies, and the PCR experiment performed again on at least 20 colonies per mixed colony. Colonies become "pure" as to their mating-type at this stage (i.e, no switching back and forth between mating-types occurs and all cells respond identically to the PCR determination, yielding a single positive response, either MATa or MATalpha). The percentage of "pure" switched colonies out of the number of colonies sub-cloned was then used as an estimate of the efficiency of mating-type switching (see results).
In the case of the expression of the HO gene from S. cerevisiae, we observed a high cellular mortality, so we decided to analyze this phenomenon more precisely: we plated dilutions from 10 6 to 10 2 cells on inductive medium and on repressive medium and compared the number of colonies growing. This was done on three and five independent transformants for strains BG87 and HM100 respectively.
For the induction of HO expression in BG87s, (BG87 switched from MATa to MATalpha) a colony which had lost the original plasmid was isolated by growing the strain on SC+URA, and transformed anew with the plasmid expressing the HO gene from S. cerevisiae (see Results).

Results
Mating-type switching in a C. glabrata MATa strain HO is particularly well-conserved in the Nakaseomyces [42], and we cloned all HO genes from the Nakaseomyces and from S. cerevisiae into a URA3 selectable plasmid that allows controlled expression of the cloned gene under the MET3 promoter [63]. Expression is repressed in conditions of high methionine and cysteine concentrations, and induced in the absence of these two amino acids in the medium (see Materials and Methods).
We first tested the expression of all HO genes in C. glabrata, using the MATa strain BG87, isogenic to the BG2 strain, commonly used in laboratories [38,62] (Table 1). Colonies were tested by specific PCR in order to discriminate the mating-type (specific primers and corresponding size expected are given in Tables 3 and 4). Results obtained after induction are shown in Table 5 (experiment A and 1 to 7). All HO genes, from the Nakaseomyces and from S. cerevisiae, are able to induce mating-type switching, i.e, induction of HO expression yields "mixed colonies", as evidenced by the obtainment of both MATa-and MATalpha-specific amplifications (see Materials and Methods).  From these mixed MATa/MATalpha colonies, we isolated pure switched cells (see Materials and Methods, and Fig 1). When S. cerevisiae's HO gene is expressed, more than 80% of the isolated cells present the opposite sexual type ("frequent" in experiment 1, Table 5) whereas less than 10% of the isolated cells had switched when the different Nakaseomyces HO gene were expressed ("rare" in experiments 2, 3, 4, 6, and 7, Table 5). In the case of N. castelli's HO, we could not isolate pure switched cells from a mixed colony exhibiting both MATa/MATalpha amplification ("absent" in experiment 5, Table 5).
As a control (experiment A, Table 5), we performed the same experiment with the pYR32 plasmid which does not contain an HO gene and we never observed mixed MATa/MATalpha colonies, but only non-switched parental colonies.
Southern blot analysis (Fig 3) of pure switched strains shows that the structure of the cassettes is as expected after mating-type switching i.e, HMLalpha, MATalpha, HMRa. The MAT locus from several pure switched clones (originating from expression of both S. cerevisiae HO and some Nakaseomyces HOs) was sequenced. This confirmed that the C. glabrata MAT locus undergoes switching from MATa to MATalpha. Sequence analyses show that the molecular structure of MAT locus is normal, suggesting a correct double strand break at the specific MATa HO cutting site and normal repair by homologous recombination with HMLalpha. Sequencing also confirmed that the switched clones obtained, truly originated from the BG87 strain, since several polymorphic sites at the MAT locus (discriminating the two C. glabrata background strains HM100 and BG87) were detected (Fig 4). HMLalpha and HMRa sequencing have shown that these loci were normal. Since switching is observed at a high frequency only with S. cerevisiae's HO, and that all other give switching rates similar to C. glabrata's HO, we focused on S. cerevisiae HO gene expression for further analyses.

Mating-type switching in C. glabrata MATalpha strains
Contrary to S. cerevisiae where the three cassettes HML, HMR and MAT are on the same chromosome (ChIII), in C. glabrata, the HMR locus is located on a different chromosome from the HML and MAT loci. In S. cerevisiae, it is known that the relocation of HMR or HML lowers the efficiency of repair of the Ho cut [65,66]. We wanted to know whether switching occurred in the same way in both directions in C. glabrata. Our first experiments described above involved repair of the MAT locus with the HML locus on the same chromosome. In a MATalpha strain, such as HM100, switching involves repair with HMRa on a distinct chromosome. Switching from MATalpha to MATa has been performed previously, using the HO gene from C. glabrata expressed from a constitutive promoter [46], but was not examined in detail. Therefore, expression of the HO gene from S. cerevisiae was induced in the MATalpha strain, HM100, isogenic to the sequenced strain, CBS138 (experiment 8, Table 5). Like previously, colonies grown on inductive medium were tested by specific PCR at the MAT locus ( Fig  5). The expression of S. cerevisiae's HO allows the obtainment of several pure switched subclones. Sequence analyses of these clones show that the C. glabrata MAT locus can switch from MATalpha to MATa, and suggest again a correct double strand break at the MATalpha Ho cutting site and normal repair with HMRa as donor. Switching from MATalpha to MATa reveals unexpected cutting at HML in C. glabrata Sequencing of HML and HMR loci showed that half of the clones display a normal structure of these loci ( Fig 6A); but others present switching at HML, i.e, HMLalpha has been switched to HMLa, leading to a "triple a" genotype: HMLa, MATa, HMRa (Fig 6B). Normal and "triple a" genotypes were confirmed by sequencing and by Southern blot experiments (data not shown). The finding of these "triple a" strains is evidence of cutting at the HML locus by S. cerevisiae's Ho endonuclease.
Since the original MATa and MATalpha strains at our disposal were non isogenic, we decided to use a switched BG87s MATalpha strain to test whether "triple a" strains could be obtained in a different strain background (experiment 1-b, Table 5). We isolated two MATa BG87s clones, upon S. cerevisiae's HO expression, confirming that the MATalpha switched locus can switch back to MATa. Sequencing shows that the first clone displays a normal structure at the MATa, HMLalpha, and HMRa loci, but that the second one exhibits a "triple a" structure: MATa, HMLa, HMRa, similar to the one obtained in the HM100 strain (data not shown). The HML locus can act first as a recipient and then as a donor during mating-type switching Sequence comparison showed the presence of a polymorphic site inside the X region between the three loci, (MAT, HML and HMR) in the HM100 strain (Fig 6). The MAT and HMR loci possess a "C" whereas the HML locus presents an "A" at the same position (+1269 from the Ho cut site). In S. cerevisiae, we know that copying of MAT information during switching, includes not only the Y region but extends in the X region with variable lengths [67]. We took advantage of this polymorphic site to understand the chronological steps of the establishment of the "triple a" strains. Sequencing of one "triple a" strain shows that the MAT locus contains an "A" at the polymorphic site. This result proves that the MAT locus has been cut and necessarily repaired using HML as a template after it has switched itself, using HMRa as template. This is unexpected because it suggests that HML is preferentially cut over MAT.

Switching-induced cell death in C. glabrata
Comparison of colony numbers between inductive and repressive conditions shows that S. cerevisiae HO gene expression induces a very strong lethality in C. glabrata cells (Table 6). This lethality is not observed during the Nakaseomyces HO gene expression. We observed the same rate of lethality between the two strains BG87 MATa and BG87s MATalpha (experiments 1 and 1-b respectively, Table 6) suggesting that the switching direction (MATa versus MATalpha) does not influence cell death. We noticed that, in the HM100 strain background, lethality is ten times higher ( Table 6, experiment 8).
The high lethality observed could be due to cutting of an Ho site outside of the triplicated cassettes, which could not be repaired by homologous recombination. We unsuccessfully searched the entire genome in silico, for additional Ho sites, even though we cannot exclude the presence of degenerate sites. Since we observe unexpected switching of HML (triple "a" genotype, see above), we hypothesize that the high lethality is linked to high efficiency of cutting at the Ho sites, including the HML site which can be cut in C. glabrata (see discussion).

S. cerevisiae and C. glabrata HO gene expression in S. cerevisiae
We were intrigued by the strong lethality observed in C. glabrata cells, when the HO gene from S. cerevisiae was expressed. Such lethality has not been reported, to our knowledge, when HO is expressed from a galactose-inducible promoter in a wild-type strain of S. cerevisiae. We therefore wanted to test whether our construction, with the HO gene under the control of the MET3 In the case of lethality (*), dilutions from 10 6 to 10 2 cells were plated on inductive medium and on repressive medium, and numbers of colonies compared.
For experiments 1 and 1-b, number of cells counted, is between approximately 4 and 300 clones for dilutions ranging from 10 3 to 10 5 cells per plate. For experiment 8, we counted between 1 and 50 colonies for dilution ranging from 10 4 up to 10 6 cells per plate. This was done on three and five independent transformants for strains BG87 and HM100 respectively. promoter from C. glabrata could induce lethality in S. cerevisiae, taking advantage of the fact that ARS sequences can function as origins in both species [68]. A MATa ho S. cerevisiae strain (FY1679-18D) was transformed with the two constructs used previously, carrying either the C. glabrata or S. cerevisiae HO gene (experiments 21 and  22, Tables 5 and 6). Several independent transformants underwent the same treatment as C. glabrata cells, i.e they were grown in broth in repressive conditions, and plated on either inductive or repressive medium at various concentrations. We confirmed the stability of the plasmid in S. cerevisiae by comparing the number of colonies on medium with or without uracil. In S. cerevisiae, switching events were searched for, by a mating assay with a complementing strain possessing the same MAT type, so that mating is possible only if switching has occurred. We observed that expression of S. cerevisiae's HO leads to switching of the MATa locus of S. cerevisiae, whereas no switch is observed when the C. glabrata HO gene is expressed (experiments 21 and 22, Table 5).
No lethality is observed upon expression of either HO genes in S. cerevisiae (experiments 21 and 22, Table 6). Thus, the strong lethality observed in C. glabrata is not due to a toxic effect of our S. cerevisiae HO construct, but seems to be specific to C. glabrata cells.

Discussion
In S. cerevisiae, mating-type switching is a complex system relying on the existence of: 1) three cassettes (donors and recipient), 2) the presence of the Ho endonuclease 3) repair mechanisms, 4) silencing mechanisms at HMLalpha and HMRa. These silencing mechanisms allow the switching system to work with the MAT locus as the sole recipient and HMLalpha and HMRa as donors.
C. glabrata has never been shown to mate nor to switch mating-types naturally. Its genome has retained the three cassettes and the HO gene, but misses the SIR1 gene, involved in silencing of HML and HMR in S. cerevisiae. In this work, we address the functionality of different Ho proteins and of the switching system in C. glabrata by expressing all HO genes from the Nakaseomyces and S. cerevisiae. Our induction experiments are done on solid medium, so that the HO genes are expressed continuously. Studies in S. cerevisiae were performed in the same way (for review, [5]) and allowed the dissection of the MAT-switching mechanisms. This is very different from the natural situation in which HO gene expression is tightly regulated, and could lead to continuous cleavage during cell growth on plates. Nonetheless, in our experiments, we obtained a very reproducible proportion of switched colonies. This seems to show that the Ho proteins do not cut repeatedly and do not induce switching back and forth between MATa and MATalpha in our conditions.
HO genes from the Nakaseomyces and S. cerevisiae allow MAT switching in a MATa C. glabrata strain (BG87). We observe that the Nakaseomyces' Hos are poorly efficient, even those of the two mating-competent species of the clade, N. delphensis and N. bacillisporus. Unexpectedly, S. cerevisiae's Ho is the most efficient to induce MAT switching in C. glabrata and we thus focused the rest of our work on this endonuclease.
S. cerevisiae's Ho is efficient in both directions and in both genetic backgrounds (strains BG87 and HM100). Thus, the localization of HMR on another chromosome from MAT does not prevent its role as donor during double-strand break repair in C. glabrata. In addition, inducing the switch in the two backgrounds (BG87 and HM100) allows us to obtain MATa and MATalpha isogenic strains, which could be useful for mating assays.
In addition to its high efficiency of switch, S. cerevisiae's Ho induces a very high cell lethality in C. glabrata, which is not the case for the Nakaseomyces HO genes. Moreover, this strong lethality seems to be specific to C. glabrata since 100% of S. cerevisiae's cells survive during S. cerevisiae's HO gene expression in the same conditions. Thus the lethality is not due to any construct toxicity or HO gene expression conditions. The same survival rate is observed in both sexual types BG87 (MATa) and BG87s (MATalpha), so that using the HMR locus as template does not seem to be involved in the observed lethality.
Switching in a MATalpha strain results in numerous clones which have also switched at HML, proof that Ho also cuts at this locus. On the contrary, we do not observe switching at HMR in our conditions. We cannot exclude that HMR is atypically cleaved by Ho, like HML, and that this leads to repair without switch, so that we cannot detect it. The fact that we observe cutting at HML and not at HMR is unexpected since HMR is transcriptionally leaky, contrary to HML [38]. Therefore, accessibility to transcription factors and accessibility to cleavage by Ho do not necessarily go hand in hand. In S. cerevisiae, extensive studies have revealed that HML and HMR are silenced by several proteins (mainly Sir proteins) and usually protected from Ho cleavage (for review, [5]). In sir-mutants, the two HML and HMR loci are unsilenced and can act as recipients and thus, switch [27][28][29]. Furthermore, cleavage at HML has been previously reported at a very low rate during constitutive HO expression in wild-type cells while none was observed at HMR [69], in accordance to what we observed in C. glabrata.
Natural variants carrying opposite configuration at the HML locus have been described previously in C. glabrata [36]. We screened 100 different strains of our C. glabrata collection [35] by typing the Ya or Yalpha information at the three loci MAT, HMR and HML by specific PCR amplifications (data not shown). We found 10% of the strains revealing rearrangements, with at least one strain harboring the HMLa genotype. These data confirm that cuts outside of MAT can occur in natural conditions.
Our results show, that in a wild-type strain of C. glabrata, both MAT and HML are cut at high levels by Ho during switching. We believe that this "illegal" cleavage is the result of a silencing system which is less efficient than S. cerevisiae's, possibly due to the absence of SIR1 and that this may be the cause of the mortality. C. glabrata does not follow the rule of "one recipient for two opposite donors", essential for the proper functioning of the switching system in S. cerevisiae. The high mortality associated to efficient switching, explains perhaps, why C. glabrata is not observed to switch regularly. Our work, relying on deregulation of the genes, shows that the low efficiency of the endogenous system is probably not due to a weak endogenous promoter, but may be linked to the properties of the protein itself, such as stability, nuclear localization and/or cleavage activity. Nonetheless, the fact that the genome of C. glabrata has retained the elements necessary for switching as well as the genes involved in mating, point to the possibility that undiscovered conditions may induce switching and/or mating.
Further experiments are under way to understand the correlation between switching and cell death in C. glabrata. Especially, mutations in the Ho recognition sites should allow us to understand better the switching system and the rules for cleavage of recipients and for donor preference in this pathogen.