Evolution of loss of heterozygosity patterns in hybrid genomes of Candida yeast pathogens

Abstract

Background: Hybrids are chimeric organisms with highly plastic heterozygous genomes that may confer unique traits enabling the adaptation to new environments. However, most evolutionary theory frameworks predict that the high levels of genetic heterozygosity present in hybrids from divergent parents are likely to result in numerous deleterious epistatic interactions. Under this scenario, selection is expected to favor recombination events resulting in loss of heterozygosity (LOH) affecting genes involved in such negative interactions. Nevertheless, it is so far unknown whether this phenomenon actually drives genomic evolution in natural populations of hybrids. To determine the balance between selection and drift in the evolution of LOH patterns in natural yeast hybrids, we analyzed the genomic sequences from fifty-five hybrid strains of the pathogenic yeasts Candida orthopsilosis and Candida metapsilosis, which derived from at least six distinct natural hybridization events.

Results: We found that, although LOH patterns in independent hybrid clades share some level of convergence that would not be expected from random occurrence, there is an apparent lack of strong functional selection. Moreover, while mitosis is associated with a limited number of inter-homeologous chromosome recombinations in these genomes, induced DNA breaks seem to increase the LOH rate. We also found that LOH does not accumulate linearly with time in these hybrids. Furthermore, some C. orthopsilosis hybrids present LOH patterns compatible with footprints of meiotic recombination. These meiotic-like patterns are at odds with a lack of evidence of sexual recombination and with our inability to experimentally induce sporulation in these hybrids.

Conclusions: Our results suggest that genetic drift is the prevailing force shaping LOH patterns in these hybrid genomes. Moreover, the observed LOH patterns suggest that these are likely not the result of continuous accumulation of sporadic events-as expected by mitotic repair of rare chromosomal breaks-but rather of acute episodes involving many LOH events in a short period of time.

Keywords: Candida metapsilosis; Candida orthopsilosis; Comparative genomics; Hybrids; Loss of heterozygosity.

Fig. 1

Phylogenetic position of C. parapsilosis species complex. a Maximum likelihood tree reconstruction of the concatenated amino-acid multiple sequence alignment of the phylome 866 [22] available at PhylomeDB database (http://phylomedb.org/phylome_866) [24]. The light blue background highlights the species of the CUG clade. The dark blue background highlights the C. parapsilosis species complex. b Schematic representation of the evolutionary trajectory of the three opportunistic pathogens of the C. parapsilosis species complex, where A, B, and C in the pink background denote the different C. metapsilosis parental lineages and A and B in the green background denote the two C. orthopsilosis parental lineages

Fig. 2

C. orthopsilosis and C. metapsilosis isolates. a Geographical distribution of C. orthopsilosis and C. metapsilosis isolates. Circles, with colors representing different clades, are placed on the country where at least one sample of a given clade was isolated (Additional file 24: Table S1). b Distribution of loss of heterozygosity (LOH, dark gray) and heterozygosity (light gray) regions in chromosome 1 of four randomly selected strains of each C. orthopsilosis hybrid strain. C. orthopsilosis clade 1 is marked in blue, clade 2 in yellow, clade 3 in green, and clade 4 in orange. c Distribution of loss of heterozygosity (LOH, dark gray) and heterozygosity (light gray) regions in scaffold 1 of four randomly selected strains of C. metapsilosis clade 1.1, the two isolates of clade 1.2, and the only strain isolated thus far of clade 2. C. metapsilosis clade 1.1 is marked in pink, clade 1.2 in purple, and clade 2 in dark blue

Fig. 3

Maximum likelihood phylogenetic trees. a C. orthopsilosis tree with clade 1 highlighted with a blue background, clade 2 with yellow, clade 3 with green, and clade 4 with orange. Due to its dubious clade adscription, s498 has a red background. Isolates with no background correspond to parent A. b On the top right, C. metapsilosis main tree with clade 1 indicated with a square and clade 2 indicated with blue background. An additional zoom-in tree with strains of clade 1.1 with pink background and strains of clade 1.2 with purple background is presented. Newly sequenced strains are written in purple, and the environmental isolate is also in bold. Both reference strains are in blue

Fig. 4

Analysis of the LOH patterns in C. orthopsilosis supports the occurrence of meiotic recombination. a Non-linear accumulation of LOH blocks with time (SNPs/kb). b The average percentage of LOH and number of recombination events in 1-kb windows of C. orthopsilosis chromosome 2, only considering the 16 randomly selected strains represented in Fig. 2b to avoid bias of different clade sampling size. c Negative correlation between the average LOH length and the chromosome size in C. orthopsilosis clade 1 and s498. d Positive correlation between the number of LOH blocks and the chromosome size in C. orthopsilosis clade 1 and s498

Fig. 5

Schematic representation of the expected levels of LOH acquisition through a mitotic recombination or b stress-induced DNA breaks or meiotic recombination. The color grade scale indicates the color variation between low levels of LOH (light blue) and high levels of LOH (dark blue). The dashed line marks the common ancestor of strains of the same clade, after which different strains of a clade start diverging. In a scenario where only mitotic recombination takes place, the acquisition of LOH blocks is expected to proceed slowly and be proportional to time. In addition, SNPs and LOH levels are expected to increase in parallel as both processes would linearly accumulate with time. In a scenario of stress-induced DNA breaks or meiotic recombination, multiple LOH events are created in a single event, and strains progress much faster to high levels of LOH, as compared to mitotic recombination. In addition, LOH and SNP accumulation are uncoupled, and thus, levels of LOH and SNPs are not expected to correlate