Pooled Enrichment Sequencing Identifies Diversity and Evolutionary Pressures at NLR Resistance Genes within a Wild Tomato Population

Abstract

Nod-like receptors (NLRs) are nucleotide-binding domain and leucine-rich repeats containing proteins that are important in plant resistance signaling. Many of the known pathogen resistance (R) genes in plants are NLRs and they can recognize pathogen molecules directly or indirectly. As such, divergence and copy number variants at these genes are found to be high between species. Within populations, positive and balancing selection are to be expected if plants coevolve with their pathogens. In order to understand the complexity of R-gene coevolution in wild nonmodel species, it is necessary to identify the full range of NLRs and infer their evolutionary history. Here we investigate and reveal polymorphism occurring at 220 NLR genes within one population of the partially selfing wild tomato species Solanum pennellii. We use a combination of enrichment sequencing and pooling ten individuals, to specifically sequence NLR genes in a resource and cost-effective manner. We focus on the effects which different mapping and single nucleotide polymorphism calling software and settings have on calling polymorphisms in customized pooled samples. Our results are accurately verified using Sanger sequencing of polymorphic gene fragments. Our results indicate that some NLRs, namely 13 out of 220, have maintained polymorphism within our S. pennellii population. These genes show a wide range of πN/πS ratios and differing site frequency spectra. We compare our observed rate of heterozygosity with expectations for this selfing and bottlenecked population. We conclude that our method enables us to pinpoint NLR genes which have experienced natural selection in their habitat.

Keywords: RENSeq; Solanum penellii; population genetics; resistance genes.

Fig. 1.—

NLR genes in Solanum pennellii. Phylogenetic tree for the identified S. pennellii NLR genes generated using PhyML (WAG) with 1,000 bootstraps after alignment of all NB-ARC using MUSCLE. TNLs are highlighted in yellow background. Collapsed triangles represent known NLR clusters with high bootstrap values (>75%, clade CNL-RPW8: 54%). NLR families are indicated above the different clades and several named resistance genes from other species have been included for references.

Fig. 2.—

Coverage of targeted region. The fraction of bases in the targeted area having a coverage of a certain depth (x-axis) or deeper. The lines represent the individual runs and the combined data, separated for the NLR regions and the set of control (ctl) genes. The plot represents the data after preprocessing.

Fig. 3.—

SNP calls from four different callers. (A) Overlap of called SNPs between different SNP callers. Popoolation and GATK share the most common SNPs. (B) SNPs called for a region of NLR Sopen11g028610. Top shows the coverage (gray) and SNPs that appear directly from the .bam file (including putative false positives). The blue lines in the lower parts of the figure show the SNPs as identified by Sanger sequencing and four SNP callers. Popoolation and GATK show the best performance judging by overlap.

Fig. 4.—

Site frequency spectra. Folded site frequency spectra for the SNPs detected in our NLR set. x-axis shows the number of variants per site, with ten equals a frequency of 0.5 in our population. (A) Absolute folded SFS; y-axis shows actual number of sites. (B) Relative folded SFS, y-axis shows the fraction of sites. (C) Absolute folded SFS per gene.