The polygenic strategies of host-specific and general virulence of Botrytis cinerea across diverse eudicot hosts

Diverse qualitative and quantitative genetic architectures can successfully enable fungal virulence and host range. To model the quantitative genetic architecture of a generalist pathogen with an extensive host range, we conducted a genome-wide association study (GWAS) of the lesion area of Botrytis cinerea across eight hosts. This revealed that it was possible to partition the virulence, as defined by the lesion area, into common and host-specific components across all hosts from host-specific virulence. All traits showed that a large proportion of the Botrytis genome likely contributes to fungal lesion development on leaves with small effect sizes. The candidate genes are evenly spread across the core chromosomes with no indication of bipartite genomic architecture. The GWAS-identified polymorphisms and genes show that B. cinerea relies on genetic variants across hundreds of genes for growing on diverse hosts, with most genes influencing relatively few hosts. When pathogen genes were associated with multiple hosts, they were associated with unrelated rather than related host species. Comparative genomics further suggested that the GWAS-identified genes are largely syntenic with other specialist Botrytis species and not unique to B. cinerea. Overall, as shown in A. thaliana, B. cinerea’s generalist behavior is derived from the sum of the genome-wide genetic variation acting within gene networks that differentially coordinate the interaction with diverse hosts.

Methods
To begin querying how the host range and virulence are determined in Botrytis cinerea, we used previous measurements of disease caused by a population of 96 diverse Botrytis isolates across an array of eudicots including tomato, sunflower, lettuce, chicory, endive, turnip, Arabidopsis, and soybean (Caseys et al., 2021). In this work, we use these phenotypic measurements to identify the Botrytis genes that may shape host susceptibility. Combining the phenotypic measurements with the genomic variation in the Botrytis population, we mapped and analyzed the genetic architecture of host preferences across Botrytis isolates using genome-wide association study (GWAS).  Given the quantitative nature of Botrytis virulence, a Bayesian sparse linear mixed model (BSLMM) using the Markov chain Monte Carlo algorithm implemented in GEMMA was run for lesion area on each plant genotype. To incorporate the phenotypic information potentially provided by lesion area measured on up to 12 plant genotypes per host species, we implemented a multivariate approach to the BSLMM using multivariate adaptive shrinkage (MASH).To estimate the proportion of the candidate genes that might have recently evolved within B. cinerea, we performed comparative genomics analysis with 7 Botrytis species. B. cinerea B05.10 genes orthologous to B. fragariae, B. aclada, B. deweyae, B. porri, B. hyacinthi, and B. sinoallii were called by OrthoMCL.