Unbiased discovery of natural sequence variants that influence fungal virulence

Abstract

Isolates of Cryptococcus neoformans, a fungal pathogen that kills over 120,000 people each year, differ from a 19-megabase reference genome at a few thousand up to almost a million DNA sequence positions. We used bulked segregant analysis and association analysis, genetic methods that require no prior knowledge of sequence function, to address the key question of which naturally occurring sequence variants influence fungal virulence. We identified a region containing such variants, prioritized them, and engineered strains to test our findings in a mouse model of infection. At one locus we identified a 4-nt variant in the PDE2 gene, which severely truncates its phosphodiesterase product and significantly alters virulence. Our studies demonstrate a powerful and unbiased strategy for identifying key genomic regions in the absence of prior information, suggest revisions to current assumptions about cAMP levels and about common laboratory strains, and provide significant sequence and strain resources to the community.