Stable predictive markers for Phytophthora sojae avirulence genes that impair infection of soybean uncovered by whole genome sequencing of 31 isolates

BACKGROUND: The interaction between oomycete plant pathogen Phytophthora sojae and soybean is characterized by the presence of avirulence (Avr) genes in P. sojae, which encode for effectors that trigger immune responses and resistance in soybean via corresponding resistance genes (Rps). A recent survey highlighted a rapid diversification of P. sojae Avr genes in soybean fields and the need to deploy new Rps genes. However, the full genetic diversity of P. sojae isolates remains complex and dynamic and is mostly characterized on the basis of phenotypic associations with differential soybean lines. RESULTS: We sequenced the genomes of 31 isolates of P. sojae, representing a large spectrum of the pathotypes found in soybean fields, and compared all the genetic variations associated with seven Avr genes (1a, 1b, 1c, 1d, 1k, 3a, 6) and how the derived haplotypes matched reported phenotypes in 217 interactions. We discovered new variants, copy number variations and some discrepancies with the virulence of previously described isolates with Avr genes, notably with Avr1b and Avr1c. In addition, genomic signatures revealed 11.5% potentially erroneous phenotypes. When these interactions were re-phenotyped, and the Avr genes re-sequenced over time and analyzed for expression, our results showed that genomic signatures alone accurately predicted 99.5% of the interactions. CONCLUSIONS: This comprehensive genomic analysis of seven Avr genes of P. sojae in a population of 31 isolates highlights that genomic signatures can be used as accurate predictors of phenotypes for compatibility with Rps genes in soybean. Our findings also show that spontaneous mutations, often speculated as a source of aberrant phenotypes, did not occur within the confines of our experiments and further suggest that epigenesis or gene silencing do not account alone for previous discordance between genotypes and phenotypes. Furthermore, on the basis of newly identified virulence patterns within Avr1c, our results offer an explanation why Rps1c has failed more rapidly in the field than the reported information on virulence pathotypes. ELECTRONIC SUPPLEMENTARY MATERIAL: The online version of this article (10.1186/s12915-018-0549-9) contains supplementary material, which is available to authorized users.