Distinct life histories impact dikaryotic genome evolution in the rust fungus Puccinia striiformis causing stripe rust in wheat

Abstract

<jats:title>Abstract</jats:title><jats:p>Stripe rust of wheat, caused by the obligate biotrophic fungus <jats:italic>Puccinia striiformis</jats:italic> f. sp. <jats:italic>tritici</jats:italic>, is a major threat to wheat production world-wide with an estimated yearly loss of US $1 billion. The recent advances in long-read sequencing technologies and tailored-assembly algorithms enabled us to disentangle the two haploid genomes of <jats:italic>Pst</jats:italic>. This provides us with haplotype-specific information at a whole-genome level. Exploiting this novel information, we perform whole genome comparative genomics of two <jats:italic>P</jats:italic>. <jats:italic>striiformis</jats:italic> f. sp<jats:italic>. tritici</jats:italic> isolates with contrasting life histories. We compare one isolate of the old European lineage (PstS0), which has been asexual for over 50 years, and a Warrior isolate (PstS7 lineage) from a novel incursion into Europe in 2011 from a sexual population in the Himalayan region. This comparison provides evidence that long-term asexual evolution leads to genome expansion, accumulation of transposable elements, and increased heterozygosity at the single nucleotide, structural and allele levels. At the whole genome level, candidate effectors are not compartmentalized and do not exhibit reduced levels of synteny. Yet we were able to identify two subsets of candidate effector populations. About 70% of candidate effectors are invariant between the two isolates while 30% are hypervariable. The latter might be involved in host adaptation on wheat and explain the different phenotypes of the two isolates. Overall this detailed comparative analysis of two haplotype-aware assemblies of <jats:italic>P</jats:italic>. <jats:italic>striiformis</jats:italic> f. sp<jats:italic>. tritici</jats:italic> are the first steps in understanding the evolution of dikaryotic rust fungi at a whole genome level.</jats:p>

A haplotype-phased genome of wheat stripe rust pathogen Puccinia striiformis f. sp. tritici, race PST-130 from the Western USA

“…Using the UNIQUE hits (excluding repetitive) from the Assemblytics table generated from the MUMmer alignments, we calculated that the two PST-130 genomes have an average of 5.22 SNP/kb between the aligned primary contigs (p) and haplotigs (h) and the data is summarized in Table 3, which include similar data for the phased genomes of PST-104E (5.59 SNP/kb) and PST-DK0911 (1.96 SNP/kb). These values correlate well (R = 0.98) with previously reported polymorphism levels between the two genomes of each race based on the mapping of Illumina reads to their respective genome references [13,17,18].…”

“…the more divergent haploid genomes in PST-104E and PST-130 (5.0 and 5.2 SNP/kb, respectively). The number of primary contigs and N50 values of PST-130 is similar to those reported for PST-104E [17] and PST-DK0911 [18] (Table 5). However, the number of haplotigs is much larger in PST-DK0911 (1,176) than in the other two races (458-475).…”

“…Even if we add 0.285 Mb from the collapsed orphan contigs with 0 SNP/kb to our calculation, the average divergences between the aligned primary contigs and haplotigs for PST-130 would be minimal (from 5.22 SNP/kb to 5.19 SNP/kb if collapsed orphans are included). However, the exclusion of collapsed orphans with 0 SNPs from the calculation of the average divergence values using the MUMmer/Assemblytics in PST-DK0911, where the collapsed orphans are abundant [18], may explain our higher divergence estimates (1.96 SNP/kb) relative to the values estimated by Illumina mapping (1.6 SNP/kb) for this race [18].…”

“…To visualize the proportion of the PST-130 genomes aligned at different levels of divergence, we generated a histogram using 0.5 SNP/kb intervals ( Fig 4A). We generated similar distributions for PST-DK0911 ( Fig 4B) and PST-104E ( Fig 4C) based on MUMmer / Assemblytics results recalculated from the published primary contigs and haplotigs [17,18]. PST-DK0911 ( Fig 4B) showed a rapid and continuous decrease of the percent identity at increasing levels of divergence, with 69.8% of the genome showing less than 2 SNP/kb and only 2.6% of the aligned genomes showing very high levels of divergences (>10 SNP/kb, Table 3).…”

Untitled

“…If we calculate the average haploid genome size for PST-DK0911 by adding its primary contigs (74.4 Mb) and haplotigs (52.1 Mb), and dividing the total by two (63.0 Mb), the size would be smaller than sizes of the other two phased genomes calculated in the same way (Table 5). Since in PST-DK0911 most orphan contigs displayed a 2x haploid coverage, the main contigs provide a better estimate of the genome size [18]. This is equivalent to assume that the 22.3 Mb difference between the primary contigs and haplotigs in PST-DK0911 are all collapsed orphans ((74.4 + 52.1 + 22.3)/2 = 74.4) and is close to the average haploid genome sizes of PST-130 and PST-104E ( Table 5).…”