Deadly fungus can multiply by having sex, which could produce more drug-resistant, virulent strains

Yue Wang, Jianping Xu. Population genomic analyses reveal evidence for limited recombination in the superbug Candida auris in nature. Computational and Structural Biotechnology Journal, 2022; 20: 3030 DOI: 10.1016/j.csbj.2022.06.030

Abstract: Candida auris is a recently emerged, multidrug-resistant pathogenic yeast capable of causing a diversity of human infections worldwide. Genetic analyses based on whole-genome sequences have clustered strains in this species into five divergent clades, with each clade containing limited genetic variation and one of two mating types, MTLa or MTLα. The patterns of genetic variations suggest simultaneous emergence and clonal expansion of multiple clades of this pathogen across the world. At present, it is unclear whether recombination has played any role during the evolution of C. auris. In this study, we analyzed patterns of associations among single nucleotide polymorphisms in both the nuclear and the mitochondrial genomes of 1,285 strains to investigate potential signatures of recombination in natural C. auris populations. Overall, we found that polymorphisms in the nuclear and mitochondrial genomes clustered the strains similarly into the five clades, consistent with a lack of evidence for recombination among the clades after their divergence. However, variable percentages of SNP pairs showed evidence of phylogenetic incompatibility and linkage equilibrium among samples in both the nuclear and the mitochondrial genomes, with the percentages higher in the total population than those within individual clades. Our results are consistent with limited but greater frequency of recombination before the divergence of the clades than afterwards. SNPs at loci related to antifungal resistance showed frequencies of recombination similar to or lower than those observed for SNPs in other parts of the genome. Together, though very limited, evidence for the observed recombination for both before and after the divergence of the clades suggests the possibility for continuous genetic exchange in natural populations of this important yeast pathogen.

4. Discussions

In this study, we analyzed the whole genome SNPs of 1,286 C. auris strains collected from across the world over the past 20+ years to investigate the potential signatures of recombination in this species. SNPs in both the nuclear and mitochondrial genomes were analyzed, both in the total sample as well as for each of the four clades where multiple strains have been sequenced. Our analyses revealed signatures of infrequent recombination in both the total sample as well as within each of the four individual clades. In addition, specific groups of SNPs, including those in genes involved in antifungal drug resistance as well as those that are shared among all four clades, were separately analyzed to help identify the potential contributors to the observed signatures of recombination. Different patterns of allelic associations were found among the sample types and between the nuclear and mitochondrial genomes. Below we discuss the main findings of our analyses and the major implications of our results.

4.1. Comparison between nuclear and mitochondrial genomes

The 1,285 genomes analyzed here represented all the strains of C. auris that have been sequenced and deposited in GenBank by researchers, up to May 2022. Multiple studies have analyzed variable numbers of strains, with the largest number of strains analyzed by Muñoz et al. [20] where 304 strains from many geographic regions were included. Most studies have focused on nuclear genomes. Analyses of nuclear genomes in those studies revealed that the global population of C. auris could be grouped into five distinct clades, with Clades I to IV represented by multiple strains in each while Clade V was represented by only one whole-genome sequenced strain (so far). However, one previous study analyzed mitogenome variations. Based on mitogenome SNPs of 130 C. auris strains, Misas et al. [69] showed that the mitogenome and nuclear genome SNPs clustered the strains into four similar clades. However, their analyses included only one Clade II strain and their results based on 10 Clade III strains from South Africa revealed no mitogenome sequence variation within Clade III. Our analyses here significantly expanded the sample sizes of all four clades with a total of 1,285 strains. While a similar pattern of sequence divergence within C. auris into five clades for both the nuclear genome and the mitochondrial genome was observed as previously reported (20,70), our analyses also revealed several notable features. Specifically, first, despite having more than twice as many strains as the earlier study (210 vs 86), we found no unambiguous SNP within the mitogenome of Clade IV, similar to that found by Misas et al. [69]. Second, the inclusion of 23 additional Clade II strains (versis one strain in the Misas et al. study) revealed no mitogenome SNP within Clade II. Third, the inclusion of 504 additional strains from more diverse geographic sources in Clade III (i.e., 514 in this study vs 10 in the Misas et al study) revealed abundant mitogenome SNPs within Clade III. Together, these analyses revealed that the amounts of sequence variations between the nuclear and mitochondrial genomes differed at both the whole species as well as within individual clades. At the whole species level, the SNP frequency in the nuclear genome was 1.876% (232,179/12.37×106), about six times of that of the mitochondrial genome (0.315%; 89/28212).

The lower observed genetic variation in the mitochondrial genome than in the nuclear genome has been reported in several other fungal species, including the human pathogen Cryptococcus gattii species complex and the ectomycorrhizal mushroom Tricholoma matsutake species complex [71], [72], [73]. However within individual clades, while two clades (Clades II and IV) showed limited to no mitochondrial SNPs (consistent with the overall pattern within the species), the remaining two clades (Clades I and III) showed greater mitochondrial SNP frequencies than their respective nuclear genomes. At present, the mechanisms for the different amounts of sequence diversity between the two genomes among the clades are unknown. The small sample size and limited ecological niches (mostly from ear discharges) might have contributed to no mitochondrial sequence variation in Clade II. However, this explanation cannot hold for Clade IV where 210 strains from four continents and a variety of human body sites were examined, similar to those of Clades I and III strains in this study (Table S1). Geographically, strains of Clades II and IV are predominantly found in East Asia and the Americas respectively while Clades I and III are predominantly from South Asia and Africa respectively. It is possible that the higher temperature and other potential environmental factors in South Asia and Africa may have contributed to the higher mutation rates in mitochondrial genomes than in nuclear genomes in Clades I and III. The mechanisms for the observed divergent mitochondrial vs nuclear genetic variations among the four clades within C. auris remain to be elucidated.

4.2. Evidence of recombination

At both the species and individual clade levels, though the frequencies were generally low, evidence for PI was observed in both the nuclear and the mitochondrial genomes, with a significant proportion of those PI SNP pairs also in linkage equilibrium. However, the frequencies of PI SNP pairs differed among the samples. Overall, the frequency of nuclear PI SNP pairs at the species level was from twice to over 30 times of those within individual clades (Table 3, Table 4). A similar pattern was also observed for the mitochondrial genome SNPs. Together, these results suggested that there was more frequent recombination before the divergence of the clades than after individual clades were established. Specifically, though signatures of recombination were also detected within each of the four clades after their respective divergence, clonal reproduction and expansion seemed more dominant in natural populations of C. auris after the divergence of clades than before their divergence. Our observed pattern is largely consistent with the expectations of each clade having only one mating type and therefore less likely to mate and recombine among strains of the same clade.

Relative to the frequent reports of recombination in the nuclear genomes of fungal populations, reports of mitochondrial genome recombination are still rare. However, the list of fungal species and populations showing evidence of mitogenome recombination is growing. For example, since 1998, mitochondrial DNA recombination has been reported for the honey mushroom Armillaria gallica [74], the commercial button mushroom Agaricus bisporus [75], the wild ectomycorrhizal mushroom Russula virescens species complex [76], and the opportunistic human fungal pathogen Cryptococcus gattii species complex [71]. In the commercial mushroom A. bisporus, the observed frequency of mitochondrial loci with PI was correlated with the life cycles of two varieties within the species, with the outcrossing heterothallic population showing more evidence of mitochondrial genome recombination than the secondarily homothallic populations [75].

Because the ancestral population of C. auris contained strains of both mating types, evidence for recombination in the total sample was expected. The higher rate of SNP pairs that showed evidence for PI than those within individual clades is consistent with sexual recombination in the ancient population of this species. The absence of incongruent relationships among clades between nuclear and mitochondrial genome phylogenies is consistent with the absence of mating and recombination among the four clades after their divergence from each other. However, the observed PIs among SNP pairs within individual clades after their divergence are puzzling. Specifically, each clade is known to contain strains of only one mating type, MTLa for Clades I and IV, and MTLα for Clades II and III. In addition, we found limited evidence of parallel mutations in the genes that are most likely under parallel selective pressure, the antifungal drug resistance-related genes. While we cannot completely exclude the possibility that convergent mutations might have contributed to some of the observed PIs, our analyses revealed that even if they existed, such an effect would likely be minimal. However, evidence for recombination have been found in natural fungal populations known to contain only a single mating type. For example, same-sex mating has been reported in the human fungal pathogen Cryptococcus neoformans species complex and such mating can generate genetic recombinants, similar to what have been reported for opposite-sex mating and to natural populations containing strains of both mating types [77]. It is possible that low-frequency same-sex mating could have similarly happened to the individual C. auris clades to generate the observed PIs and linkage equilibrium. Alternatively, low frequency strains of the alternative mating type may exist within each of the four clades in nature and mating between strains of opposite mating types could have generated the observed signatures of recombination. Indeed, these two possibilities are not mutually exclusive, and both could have contributed to the observed signatures of recombination. Broader and more intensive sampling as well as experimental investigations of genetic crosses are needed in order to test these two possibilities.

4.3. Genes adjacent to clade-shared SNPs

Our analyses revealed three clade-shared SNP regions, with SNPs in two of these regions showing high frequency of PI with other SNPs in the genome. Interestingly, all three clade-shared SNP regions are in intergenic regions between genes coding for hydrolases, oxidoreductases, and transcription factors with potential impacts on cell growth and lifespan (Table 7). For example, the ortholog of B9J08_000508, the downstream gene of clade-shared SNP region 1, is known to regulate the target of rapamycin complex 1 (TORC1) signaling. TORC1 is a multiprotein signaling complex functions as the organizer that incorporate internal and external cues to regulate cell growth and cell cycle progression [78]. A recent study demonstrated that TORC1 signaling plays an important role in controlling NaCl resistance through Sir2 in Saccharomyces cerevisiae [79]. The clade-shared SNPs within the upstream region of TORC1 gene may be involved in regulating the expression levels of TORC1.

Interestingly, the downstream gene of clade-shared SNP region 2, B9J08_002254, codes for a protein containing a putative FMN-binding domain which is known to be most frequently found in bacteria. It has been hypothesized that proteins containing such a domain in fungi may have been horizontally transferred from bacterial to fungal genomes [80]. Indeed, multiple independent transfers of such genes and the associated upstream sequences from bacteria to strains of C. auris in different clades could have contributed to the observed distributions of clade-shared polymorphisms and PIs. Our BLAST searches revealed that based on the amino acid sequence, the closest match to B9J08_002254 was in the bacterial genus Achromobacter, with a 99% query coverage and an E value of 1e-54.

The two genes located upstream and downstream of the clade-shared SNP region #3 were B9J08_003771 and B9J08_003772. Gene B9J08_003771 has a predicted unfolded protein-binding activity, while B9J08_003772 has a predicted DNA-binding transcription factor activity, zinc ion binding activity, and transcriptional regulation activity. The unfolded protein response is known to help human fungal pathogens survive in the host through balancing the load of proteins entering the endoplasmic reticulum and the protein-folding capacity of the organelle [81]. For example, in C. albicans, the zinc finger protein CZF1 is one of the DNA-binding proteins of the Cys6Zn2 class of transcriptional regulators with a multitude of functions such as biofilm induction, hyphal growth regulation, white-opaque switch, and yeast cell adherence [82], [83], [84], [85], [86]. While functionally likely important, how the polymorphisms in the intergenic regions of these two genes contribute to strain and population fitness remains to be investigated.

4.4. Conclusions and perspectives

This study identified limited but unambiguous evidence of recombination in both the total sample and within individual clades. In addition, evidence of recombination was found in both the nuclear and mitochondrial genomes, as well as between the nuclear and mitochondrial genomes. Overall, signatures of recombination were more prominent in the total sample than within individual clades, consistent with greater frequencies of recombination before the divergence of the four clades than after their divergence. At present, while several possibilities were suggested, the mechanism(s) for the observed recombination is not known. Nevertheless, the signatures of recombination identified here suggested a number of avenues from which further investigations could be conducted, including more extensive sampling for alternative mating types within each clade, laboratory attempts of both same-sex and opposite-sex mating, and identifying the adaptive significance of clade-shared SNPs. Such investigations should allow us to better understand the genetic architecture of virulence and drug resistance evolution within and among the divergent clades of this pathogen in natural and clinical environments.