BackgroundCandida glabrata follows C. albicans as the second or third most prevalent cause of candidemia worldwide. These two pathogenic yeasts are distantly related, C. glabrata being part of the Nakaseomyces, a group more closely related to Saccharomyces cerevisiae. Although C. glabrata was thought to be the only pathogenic Nakaseomyces, two new pathogens have recently been described within this group: C. nivariensis and C. bracarensis. To gain insight into the genomic changes underlying the emergence of virulence, we sequenced the genomes of these two, and three other non-pathogenic Nakaseomyces, and compared them to other sequenced yeasts.ResultsOur results indicate that the two new pathogens are more closely related to the non-pathogenic N. delphensis than to C. glabrata. We uncover duplications and accelerated evolution that specifically affected genes in the lineage preceding the group containing N. delphensis and the three pathogens, which may provide clues to the higher propensity of this group to infect humans. Finally, the number of Epa-like adhesins is specifically enriched in the pathogens, particularly in C. glabrata.ConclusionsRemarkably, some features thought to be the result of adaptation of C. glabrata to a pathogenic lifestyle, are present throughout the Nakaseomyces, indicating these are rather ancient adaptations to other environments. Phylogeny suggests that human pathogenesis evolved several times, independently within the clade. The expansion of the EPA gene family in pathogens establishes an evolutionary link between adhesion and virulence phenotypes. Our analyses thus shed light onto the relationships between virulence and the recent genomic changes that occurred within the Nakaseomyces.Sequence Accession NumbersNakaseomyces delphensis: CAPT01000001 to CAPT01000179Candida bracarensis: CAPU01000001 to CAPU01000251Candida nivariensis: CAPV01000001 to CAPV01000123Candida castellii: CAPW01000001 to CAPW01000101Nakaseomyces bacillisporus: CAPX01000001 to CAPX01000186

Superintegrons (SIs) and multiresistant integrons (MRIs) have two main structural differences: (i) the SI platform is sedentary, while the MRI platform is commonly associated with mobile DNA elements and (ii) the recombination sites (attC) of SI gene cassette clusters are highly homogeneous, while those of MRI cassette arrays are highly variable in length and sequence. In order to determine if the latter difference was correlated with a dissimilarity in the recombination activities, we conducted a comparative study of the integron integrases of the class 1 MRI (IntI1) and the Vibrio cholerae SI (VchIntIA). We developed two assays that allowed us to independently measure the frequencies of cassette deletion and integration at the cognate attI sites. We demonstrated that the range of attC sites efficiently recombined by VchIntIA is narrower than the range of attC sites efficiently recombined by IntI1. Introduction of mutations into the V. cholerae repeats (VCRs), the attC sites of the V. cholerae SI cassettes, allowed us to map positions that affected the VchIntIA and IntI1 activities to different extents. Using a cointegration assay, we established that in E. coli, attI1-؋-VCR recombination catalyzed by IntI1 was 2,600-fold more efficient than attIVch-؋-VCR recombination catalyzed by VchIntIA. We performed the same experiments in V. cholerae and established that the attIVch-؋-VCR recombination catalyzed by VchIntIA was 2,000-fold greater than the recombination measured in E. coli. Taken together, our results indicate that in the V. cholerae SI, the substrate recognition and recombination reactions mediated by VchIntIA might differ from the class 1 MRI paradigm.