Comparative genomics of downy mildews reveals potential adaptations to biotrophy

Abstract

Background: Spinach downy mildew caused by the oomycete Peronospora effusa is a significant burden on the expanding spinach production industry, especially for organic farms where synthetic fungicides cannot be deployed to control the pathogen. P. effusa is highly variable and 15 new races have been recognized in the past 30 years.

Results: We virulence phenotyped, sequenced, and assembled two isolates of P. effusa from the Salinas Valley, California, U.S.A. that were identified as race 13 and 14. These assemblies are high quality in comparison to assemblies of other downy mildews having low total scaffold count (784 & 880), high contig N₅₀s (48 kb & 52 kb), high BUSCO completion and low BUSCO duplication scores and share many syntenic blocks with Phytophthora species. Comparative analysis of four downy mildew and three Phytophthora species revealed parallel absences of genes encoding conserved domains linked to transporters, pathogenesis, and carbohydrate activity in the biotrophic species. Downy mildews surveyed that have lost the ability to produce zoospores have a common loss of flagella/motor and calcium domain encoding genes. Our phylogenomic data support multiple origins of downy mildews from hemibiotrophic progenitors and suggest that common gene losses in these downy mildews may be of genes involved in the necrotrophic stages of Phytophthora spp.

Conclusions: We present a high-quality draft genome of Peronospora effusa that will serve as a reference for Peronospora spp. We identified several Pfam domains as under-represented in the downy mildews consistent with the loss of zoosporegenesis and necrotrophy. Phylogenomics provides further support for a polyphyletic origin of downy mildews.

Keywords: Biotrophy; Gene loss; Genomics; Oomycete; Peronospora effusa; Peronospora farinosa; Peronospora lineage; Spinach downy mildew.

Fig. 1

SyMap plots of P. effusa R13 aligned against P. effusa R14, P. sojae v3.0 against P. effusa R13 and P. sojae v3.0 against P. effusa R14. The plots aligned against P. sojae are scaled so the area of the plot occupied by P. sojae scaffolds is scaled to the size of the P. effusa sequences. No scaling is applied to the P. effusa cross isolate comparison

Fig. 2

Orthology analysis of oomycete gene models. a Venn diagram depicting the number of orthogroups shared between two P. effusa isolates and unique gene counts. b Venn diagram depicting the number of orthogroups shared between two P. tabacina and unique gene counts. c UpsetR plot demonstrating the number of orthogroups shared by sets of oomycete species. The intersection size is the number of orthogroups and the black dots on the x-axis represent whether the orthogroups are present or absent in that set. For instance, the first bar demonstrates that 4893 orthogroups are shared among all oomycetes, while the second demonstrated that 1145 are shared among Phytophthora spp. and are absent in the downy mildews. Only intersections over 60 orthogroups are depicted. The table inset shows how many gene models are unique to each species and include models not assigned to orthogroups and models included in orthogroups made up from a single species

Fig. 3

Heat map showing the distribution of orthogroups where a Pfam domain, found to be depleted in downy mildews, is detected as encoded by at least one gene model of that orthogroup. Red indicates the orthogroup contains at least one model from the species, blue indicates that no model is detected from the species. Green tabs on the right of the row indicates that the orthogroup contains a maximum of one gene from each species except for P. effusa and P. tabacina where two models were permitted as two isolates were combined in the analysis. P. halstedii is bordered to highlight that it does not have the same loss pattern exhibited by other downy mildew species across all categories. The inlayed table indicates the number of gene models that encode an under-represented Pfam domain, but were not assigned to an orthogroup

Fig. 4

K-mer and read mapping overview. a KAT density plots demonstrating the 21-mer multiplicity of read 1 (x axis) against read 2 (y axis) for P. effusa R13, P. effusa R14, P. tabacina J2 and P. tabacina S26. Approximate locations of homozygous and heterozygous 21-mer are labeled on the x-axis. b KAT spectra-cn plots of the 21-mer multiplicity of the read set (x axis) against their respective assembly (y axis) for P. effusa R13, P. effusa R14, P. tabacina J2 and P. tabacina S26. Black areas under the peaks represent 21-mers present in the reads, absent in the assembly, red indicates 21-mers are present once in the assembly, purple 21-mers are present twice, green thrice. c Frequency of reads supporting the alternative allele over the entire assembly and gene space for P. effusa R13 and P. effusa R14. The allele frequency is cut-off at 0.2 and 0.8 on the x-axis of all plots. d Normalized read depth of every gene predicted in both P. effusa R13 and P. effusa R14. The height of each plot indicates the number of gene models at the normalized coverage displayed on the x axis

Fig. 5

Maximum likelihood phylogeny of 49 BUSCO gene models, identified as conserved as a single copy gene across all assemblies surveyed. Node labels indicate the score after 1000 bootstraps. The scale bar indicates the nucleotide divergence per site