Genomic Signatures of a Major Adaptive Event in the Pathogenic Fungus Melampsora larici-populina

Abstract

The recent availability of genome-wide sequencing techniques has allowed systematic screening for molecular signatures of adaptation, including in nonmodel organisms. Host-pathogen interactions constitute good models due to the strong selective pressures that they entail. We focused on an adaptive event which affected the poplar rust fungus Melampsora larici-populina when it overcame a resistance gene borne by its host, cultivated poplar. Based on 76 virulent and avirulent isolates framing narrowly the estimated date of the adaptive event, we examined the molecular signatures of selection. Using an array of genome scan methods based on different features of nucleotide diversity, we detected a single locus exhibiting a consistent pattern suggestive of a selective sweep in virulent individuals (excess of differentiation between virulent and avirulent samples, linkage disequilibrium, genotype-phenotype statistical association, and long-range haplotypes). Our study pinpoints a single gene and further a single amino acid replacement which may have allowed the adaptive event. Although our samples are nearly contemporary to the selective sweep, it does not seem to have affected genome diversity further than the immediate vicinity of the causal locus, which can be explained by a soft selective sweep (where selection acts on standing variation) and by the impact of recombination in mitigating the impact of selection. Therefore, it seems that properties of the life cycle of M. larici-populina, which entails both high genetic diversity and outbreeding, has facilitated its adaptation.

Keywords: coevolution; genome scan; genome-wide association studies; plant–pathogen interactions; population genomics.

Fig. 1.

Unrooted neighbor-joining tree of all isolates based on genomic distances. The distances are expressed in number of differences per compared position (due to missing data, the number of compared positions is less than the number of polymorphic sites and varies between comparisons). Colored disks indicate sample membership. Isolates of the Vir1994 sample that are assigned to a specific group by the DAPC analysis are indicated by an additional red circle.

Fig. 2.

Selective sweep signatures in the Melampsora larici-populina genome. π, nucleotide diversity; $D_{\mathrm{a}}$, net pairwise population divergence; ${\widehat{\theta }}_2$, Weir and Cockerham’s cluster fixation index; ln(Rsb), excess of long-distance haplotype homozygosity. For genome scan methods, the test P value is presented. The average sequencing depth over isolates is given for all considered sites. π and D_a are computed over 1,000-bp overlapping windows (step: 250 bp). Each value is placed at the window midpoint. For selective sweep detection methods, the test statistic is given for all SNPs. Significant or outlier values are indicated by larger red disks and their position is outlined by a blue vertical line. For the GWAS results, the SNPs which are significant as cofactors are indicated by a star and the SNPs only passing the threshold in the model without cofactors by red disks.

Fig. 3.

Focus on the candidate region. iEG, integrated EHH. Other statistics are like in figure 2. The average sequencing depth over isolates is given for all polymorphic sites. For all statistics, all genes (full genic region) are indicated by gray frames. The significance threshold is indicated by a dotted line for BayPass and the GWAS. GWAS P values from step 1 (without cofactor) are given. Significant SNPs are denoted by red circles (for EHH, significance is assessed by the ln(Rsb) ratio of iEG Vir1994 to iEG Avr1993) and a red star for the top GWAS SNP. The bottom panel replicates the gene localization (upper frames are forward genes and lower frames are reverse genes). Bars show nonsynonymous mutations found in the whole data set. The one nonsynonymous mutation found to be significant in any test is denoted by a purple triangle on all panels.