Search

Mating-type switching is a complex mechanism that promotes sexual reproduction in Saccharomycotina. In the model species Saccharomyces cerevisiae, mating-type switching is initiated by the Ho endonuclease that performs a site-specific double-strand break (DSB) at MAT, repaired by homologous recombination (HR) using one of the two silent mating-type loci, HMLalpha and HMRa. The reasons why all the elements of the mating-type switching system have been conserved in some Saccharomycotina, that do not show a sexual cycle nor mating-type switching, remain unknown. To gain insight on this phenomenon, we used the yeast Candida glabrata, phylogenetically close to S. cerevisiae, and for which no spontaneous and efficient mating-type switching has been observed. We have previously shown that expression of S. cerevisiae's Ho (ScHo) gene triggers mating-type switching in C. glabrata, but this leads to massive cell death. In addition, we unexpectedly found, that not only MAT but also HML was cut in this species, suggesting the formation of multiple chromosomal DSBs upon HO induction. We now report that HMR is also cut by ScHo in wild-type strains of C. glabrata. To understand the link between mating-type switching and cell death in C. glabrata, we constructed strains mutated precisely at the Ho recognition sites. We find that even when HML and HMR are protected from the Ho-cut, introducing a DSB at MAT is sufficient to induce cell death, whereas one DSB at HML or HMR is not. We demonstrate that mating-type switching in C. glabrata can be triggered using CRISPR-Cas9, without high lethality. We also show that switching is Rad51-dependent, as in S. cerevisiae, but that donor preference is not conserved in C. glabrata. Altogether, these results suggest that a DSB at MAT can be repaired by HR in C. glabrata, but that repair is prevented by ScHo., Competing Interests: The authors have declared that no competing interests exist.

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.

Adhesion to digestive mucosa is considered a crucial first step in the pathogenicity of invasive Candida infections. Candida glabrata disseminated infections predominantly start from the gut. A mouse model of disseminated infection starting from the gut was set up. Hematogenous dissemination was obtained after a low-protein diet followed by a regimen of cyclophosphamide-methotrexate and an oral inoculation of the yeasts via the drinking water. The liver was the first organ infected (day 7 post-infection), and lethality was 100% at day 21 post-infection. This new mouse model was used to compare the mortality rate and fungal burden in deep organs induced by 5 strains exhibiting different levels of adhesion to enterocyte Caco-2 cells, as determined in a test on 36 C. glabrata strains. In this model, no statistical difference of lethality was demonstrated between the strains, and fungal burden varied in kidneys and lungs but without correlation with the level of adhesion to enterocytes. Further studies using the model developed here allow analysis of the crossing of the digestive mucosa by yeasts, and help relate this to yet-poorly understood adhesion phenotypes.