Quantitative pathogenicity and host adaptation in a fungal plant pathogen revealed by whole-genome sequencing

Abstract

Knowledge of genetic determinism and evolutionary dynamics mediating host-pathogen interactions is essential to manage fungal plant diseases. Studies on the genetic architecture of fungal pathogenicity often focus on large-effect effector genes triggering strong, qualitative resistance. It is not clear how this translates to predominately quantitative interactions. Here, we use the Zymoseptoria tritici-wheat model to elucidate the genetic architecture of quantitative pathogenicity and mechanisms mediating host adaptation. With a multi-host genome-wide association study, we identify 19 high-confidence candidate genes associated with quantitative pathogenicity. Analysis of genetic diversity reveals that sequence polymorphism is the main evolutionary process mediating differences in quantitative pathogenicity, a process that is likely facilitated by genetic recombination and transposable element dynamics. Finally, we use functional approaches to confirm the role of an effector-like gene and a methyltransferase in phenotypic variation. This study highlights the complex genetic architecture of quantitative pathogenicity, extensive diversifying selection and plausible mechanisms facilitating pathogen adaptation.

Fig. 1. Population structure and linkage disequilibrium (LD) in the Z. tritici population (n = 103).

a Admixture plot of the fungal population. Each vertical bar represents a single isolate and is colored according to the membership coefficient (Q_i) to the three sub-population (K) clusters identified by STRUCTURE. b LD decay over physical distance. The r² values were calculated between pairs of SNPs up to a physical distance of 20 kb and fitted against the physical distance using a non-linear model. The dotted orange line shows the distance (0.5 kb) at which LD decays at the r² < 0.2 level.

Fig. 2. Overview of the identified candidate pathogenicity genes from the multi-host GWAS.

a Number of identified genes per cultivar. b Gene IDs and their significant associations. Black dots correspond to PLACP (percentage of leaf area covered by pycnidia) and red dots to PLACN (percentage of leaf area covered by necrosis) marker-trait associations. The larger the dot the stronger the association (p value). Radial lines connect dots to the cultivar where it was detected and to the gene ID. c Nucleotide diversity (π) estimates in a 50 kb sliding window with a step size of 25 kb. Genomic positions are displayed on the x-axis and π values are displayed on the y-axis. The y-axis ranges from 0 to 0.045. d Tajima’s D estimates in 50 kb sliding windows with a step size of 25 kb. Genomic positions are displayed on the x-axis. Tajima’s D values are displayed on the y-axis. The y-axis ranges from −2.5 to 2. e The number of associations for each gene (PLACN/PLACP per cultivar).

Fig. 3. The quantitative nature and host-specificity of Z. tritici pathogenicity.

a An upset plot of candidate pathogenicity genes across host cultivars. b Boxplots showing variation in quantitative pathogenicity (PLACP) of the isolates (n = 103 isolates) carrying different lead SNP alleles of the candidate pathogenicity genes (center line at the median, upper bound at 75th percentile, lower bound at 25th percentile, with whiskers chosen to show the 1.5 of the interquartile range). c Differential gene expression shown as log2-fold change compared to the culture medium. Red shades indicate up-regulated genes and blue shades indicate down-regulated genes during the infection cycle.

Fig. 4. Genetic diversity of quantitative pathogenicity genes in Z. tritici and factors contributing to gene diversification.

a Distribution of nucleotide diversity (π) among candidate pathogenicity genes (n = 58) and the rest of the protein-coding genes (n = 10,993). Statistical significance was determined using a permutation test with 1000 iterations; p = 0.001818**. b TE distance compared between pathogenicity genes (n = 58) and the rest of the protein-coding genes in Z. tritici (n = 12,012). Statistical significance was determined using a permutation test with 1000 iterations; p = 0.04683*. In (a) and (b) data are presented as box plots (center line at the median, upper bound at 75th percentile, lower bound at 25th percentile) with whiskers chosen to show the 1.5 of the interquartile range. c The relationship between TE distance and recombination events in candidate pathogenicity genes. The error band indicates the 95% confidence interval.

Fig. 5. An effector gene encoding a positively-selected small secreted protein is involved in quantitative pathogenicity of Z. tritici.

a Regional association plot of a candidate effector protein detected on cultivar “Arina”. X and Y axes indicate positions on chromosome 3 and −log10 (p value) for associations with PLACP, respectively. Annotations on the reference genome are shown at the bottom with the candidate gene represented by a green box, other genes by blue boxes and transposable elements by gray boxes. A linkage disequilibrium plot is shown below the annotation bar. b log2-fold expression changes of Zt_3_00467 based on RNAseq data collected throughout the time course of Z. tritici infection on wheat. c Percentage of leaf area covered by pycnidia (PLACP) produced by the wild-type strain 3D7 (n = 6 leaves) and the ectopic transformants expressing the virulent (3D7) and avirulent (IPO10273) alleles (n = 18 leaves each). Data are presented as box plots (center line at the median, upper bound at 75th percentile, lower bound at 25th percentile) with whiskers chosen to show the 1.5 of the interquartile range. Significant difference between group means are represented with different letters above the boxplots after a Tukey’s HSD test. d Alignment of the virulent and avirulent protein sequences. Residues in yellow constitute the predicted signal peptide, gray residues show protein mutations and red asterisks show positively selected sites with a Bayes Empirical Bayes posterior probability >0.95.

Fig. 6. A gene encoding a methyltransferase mediates quantitative pathogenicity in Z. tritici.

a Regional association plot of the pathogenicity gene detected on cultivar “Shafir”. X and Y axes indicate positions on chromosome 6 and −log10 (p value) for associations with PLACP, respectively. Annotations on the reference genome are shown at the bottom with the candidate gene represented by a green box, other genes by blue boxes and transposable elements by gray boxes. A linkage disequilibrium plot is shown below the annotation bar. b log2-fold expression changes of Zt_6_00682 based on RNAseq transcription data collected throughout the time course of Z. tritici infection on wheat. c Percentage of leaf area covered by pycnidia (PLACP) produced by the wild-type strain 3D7 (n = 6 leaves), knockout mutants (n = 18 leaves) and the complementation mutants carrying the virulent (3D7) and avirulent (3D1) alleles (n = 18 leaves each). Significant difference between group means are represented with different letters above the boxplots after a Tukey’s HSD test. Data are presented as box plots (center line at the median, upper bound at 75th percentile, lower bound at 25th percentile) with whiskers chosen to show the 1.5 of the interquartile range. d Alignment of the virulent and avirulent protein sequences. Residues in orange constitute the methyltransferase domain, gray residues show protein mutations and red asterisks show positively selected sites with a Bayes Empirical Bayes posterior probability >0.95.