Genome-Wide Screen for Saccharomyces cerevisiae Genes Contributing to Opportunistic Pathogenicity in an Invertebrate Model Host

Abstract

Environmental opportunistic pathogens can exploit vulnerable hosts through expression of traits selected for in their natural environments. Pathogenicity is itself a complicated trait underpinned by multiple complex traits, such as thermotolerance, morphology, and stress response. The baker's yeast, Saccharomyces cerevisiae, is a species with broad environmental tolerance that has been increasingly reported as an opportunistic pathogen of humans. Here we leveraged the genetic resources available in yeast and a model insect species, the greater waxmoth Galleria mellonella, to provide a genome-wide analysis of pathogenicity factors. Using serial passaging experiments of genetically marked wild-type strains, a hybrid strain was identified as the most fit genotype across all replicates. To dissect the genetic basis for pathogenicity in the hybrid isolate, bulk segregant analysis was performed which revealed eight quantitative trait loci significantly differing between the two bulks with alleles from both parents contributing to pathogenicity. A second passaging experiment with a library of deletion mutants for most yeast genes identified a large number of mutations whose relative fitness differed in vivovs.in vitro, including mutations in genes controlling cell wall integrity, mitochondrial function, and tyrosine metabolism. Yeast is presumably subjected to a massive assault by the innate insect immune system that leads to melanization of the host and to a large bottleneck in yeast population size. Our data support that resistance to the innate immune response of the insect is key to survival in the host and identifies shared genetic mechanisms between S. cerevisiae and other opportunistic fungal pathogens.

Keywords: barseq; bulk segregant analysis; hybrid; pseudohyphae; virulence.

Figure 1

S. cerevisiae causes significant mortality to G. mellonella larvae. (A) At 4 hr after injection of 5 × 10⁶ cells of BY4722 in 10 μl PBS into 10 Galleria sixth instar larvae, animals show evidence of melanization and morbidity (right dish), whereas controls injected with 10 μl PBS appear healthy. (B) Survivorship for 72 hr is significantly different (P < 0.05) among treatments (PBS control, 2.5 × 10⁶ cells per animal, and 5 × 10⁶, Cox proportional hazards test)_, Error bars indicate SD across three replicate populations of 10 larvae per treatment.

Figure 2

Change in relative frequencies of the 43 genotypes in serial passaging experiment I. Each stacked bar plot represents the proportion of the barcode sequences recovered that could be assigned to one of the strains. In the top panel, the frequencies of the eight initial populations are shown. These samples were used to start both in vitro and in vivo passages.

Figure 3

Success in in vivo and in vitro habitats in passaging experiment I is not correlated with source of isolation. In the top and bottom panels are shown the fitness of the 43 strains measured after a single passage grouped by isolation source in the in vitro and in vivo environments, respectively. Fitness is measured as the fold change from the initial to after the first passage. Boxes indicate 25th–75th percentiles (IQR), and whiskers extend to the points within 1.5× the IQR. Outlier points are shown as separate dots.

Figure 4

The ability of strains to grow filamentously influences fitness in G. mellonella. (A) Representative photos of strains without or with evidence of filamentous growth on SLAD media after 5 d and photographed at low magnification. (B) Comparison of fitness measured as fold-change (F) in G. mellonella with the ability to grow filamentously. Horizontal lines show means and whiskers show SD. Differences between means were significant (P = 0.03) tested using Mann–Whitney U-test (used due to unequal variance between classes).

Figure 5

Bulk segregant analysis identifies eight QTL regions differentially affecting survivorship in vitro vs. in vivo. Each panel indicates a different chromosome with the top portion the frequency of the allele from the 322134S parent and the lower portion the LOD score. Black dashed line indicates the estimated allele frequency of the initial pool before passaging. Green dots indicate estimated SNP after three passages in vivo, and the green line indicates the estimated frequency of the bulk population. The red dots and lines indicate the corresponding values after three in vitro passages. Eight regions surpassing LOD 5.0 are labeled as QTL and discussed in the text.

Figure 6

Volcano plot of mean fold change between in vivo vs. in vitro passages of serial passaging experiment II. Green points indicate significant P-values corrected using a false-discovery rate of 0.05 (Benjamini–Hochberg), and red points are nonsignificant after correction.

Figure 7

Summary of GO terms enriched and depleted in sets of deletion mutants with significantly higher fitness in vivo (A) or in vitro (B). Size of circles is proportional to the number of genes with that GO term. Values on the x-axis indicate log odds ratio of terms being represented by genes in the selected set vs. the universe of possible genes. For complete data of significant terms see Table S3.