Diverse Lineages of Candida albicans Live on Old Oaks

Abstract

The human pathogen Candida albicans is considered an obligate commensal of animals, yet it is occasionally isolated from trees, shrubs, and grass. We generated genome sequence data for three strains of C. albicans that we isolated from oak trees in an ancient wood pasture, and compared these to the genomes of over 200 clinical strains. C. albicans strains from oak are similar to clinical C. albicans in that they are predominantly diploid and can become homozygous at the mating locus through whole-chromosome loss of heterozygosity. Oak strains differed from clinical strains in showing slightly higher levels of heterozygosity genome-wide. Using phylogenomic analyses and in silico chromosome painting, we show that each oak strain is more closely related to strains from humans and other animals than to strains from other oaks. The high genetic diversity of C. albicans from old oaks shows that they can live in this environment for extended periods of time.

Keywords: Population genomics; asexual; clonality; environmental reservoir; loss of heterozygosity (LOH); molecular evolution; yeast ecology.

Figure 1

C. albicans from oak are mostly diploid with some loss of heterozygosity (LOH). A. The proportion of base calls differing from the reference strain (allele ratios) are mostly 1.0 or 0.5 for oak strains (NCYC 4144-6), suggesting diploidy. Regions that recently homozygozed are shaded light blue. The points that occur in these LOH regions often correspond to the locations of known repeats where short reads are probably mismapped. The oak strain with $a/a$ at its mating locus (NCYC 4144) arrived at this state by LOH for the whole of chromosome 5. B. The proportion of heterozygous sites in 100 kb nonoverlapping windows are either high or low for oak strains (NCYC 4144, NCYC 4145, and NCYC 4146). For oak strains (shown here) and clinical strains (Figure S2) we see two modes; heterozygosity is either low (below the red line at 0.1%), or high (with a mean > 0.4%). Regions with >0.1% heterozygous sites in a 100 kb window were classed as LOH regions and are shown in light blue in A.

Figure 2

C. albicans from oak are more similar to clinical strains than to each other. Phylogenetic and pairwise sequence comparisons show that the oak strain NCYC 4146 is similar to clade 4 clinical strains, NCYC 4144 is similar to clade 18 clinical strains, and NCYC 4145 is diverged from the 17 clades sampled and 10 further strains of unknown clade. A. Genome-wide phylogeny. Maximum likelihood phylogenetic analysis of a whole-genome alignment (14,282,655 bp) shows that oak strains (green) are more closely related to clinical strains than they are too each other. Bootstrap support is estimated from 100 replicates and only values >70% are shown. The alignment contained 313,250 variable sites with at least one homozygous difference between strains. B. Painted chromosomes of oak strains. Most parts of the genomes of oak strains are more similar to clinical strains than to other oak strains (green). The genome of each oak strain was colored according to the clade assignment of the most similar strain for each 100 kb window in the genome using the colours for each clade shown in A. Regions are colored white if a strain sequence is diverged from all the other oak or clinical strains that we sampled (the proportion of sites differing is >0.028%). From left to right, NCYC 4146 in light blue is most similar to Clade 4 strains, NCYC 4144 in blue is most similar to Clade 18, and NCYC 4145 is mostly white and therefore diverged from all known clades.