Discussion
“Omics” analyses such as comparative and evolutionary genomics, transcriptomics, epigenomics, or proteomics mainly rely on the accuracy of predicted gene models. The Zymoseptoria species complex provides a unique model system for the application of “omics” data to study the underlying genetics of pathogenicity and host specificity of plant pathogens (Stukenbrock et al. 2010, 2011, 2012). However, updated and comparable annotations of gene and TE content in species of Zymoseptoria have so far been missing. Here, we present a new gene and repeat annotation of the prominent fungal wheat pathogen Z. tritici and its three close relatives: Z. pseudotritici, Z. ardabiliae, and Z. brevis; therefore, we provide a valuable resource for future functional, experimental, and evolutionary studies.
We conducted the re-annotation of the IPO323 genome using a pipeline that identifies the most probable gene models based on information from transcript assemblies, homology searches, and ab initio gene predictors (Haas et al. 2011). This allowed us to identify 11,839 gene models, including 1200 models that were not predicted previously by the JGI. The support of RNA-seq data for most of these genes underlines the potential of our pipeline for identifying novel genes.
We applied the same annotation pipeline to re-sequenced genomes of the three close relatives of Z. tritici, Z. pseudotritici, Z. ardabiliae, and Z. brevis, and could call comparable numbers of genes in these species (∼10,500 models). Of these, 7786 represent core Zymoseptoria genes shared among the four species. Our exhaustive search for gene orthology also identified genes shared by two or three species and suggested recent gains or losses of genes among the Zymoseptoria species. Notably, we also identified 1798 orphan genes in Z. tritici enriched in putative pathogenicity-related factors. Species-specific orphan genes in Z. tritici and in the three other species may include key determinants of virulence in wheat and other grass hosts. Further functional studies will allow this hypothesis to be tested.
Recently, a new identification and classification of repeats and TEs in Z. tritici was conducted (Dhillon et al. 2014). In the present study we found very similar patterns in terms of repeat and TE distribution, number of families, and evidence of TE activity. However, we could further improve the assignment of repeat families to known TE families; 39.8% of the 93 repeat families identified by Dhillon et al. were not categorized into known TE classes. We were able to improve the TE classification, leaving only 9% of the 111 repeat families unclassified. The TE annotation pipeline used here contains a number of supplementary steps of clustering of the REPET outputs and allowed us to identify more complete elements. These were all manually checked to avoid the generation of false or chimera elements.
The gene and TE annotation of the four Zymoseptoria species genomes coupled with preliminary comparative analysis allowed us to highlight interesting features relevant to the biological life traits of these organisms. Establishment of homologous sequence families across the four species showed that a large part of genes are shared between two or more species and that the major difference of gene content relies on orphan genes. These orphan genes, in all four species, are significantly enriched in putative pathogenicity-related genes and potentially play a role in the determination of host specificity. Similar presence–absence patterns of pathogenicity-related genes were shown to be strong determinants of host range in plant pathogens, e.g., between lineages of the rice blast pathogen Magnaporthe oryzae, the wilt pathogen Verticillium dahliae, and lineages of the Leptosphaeria maculans–Leptosphaeria biglobosa species complex (Couch et al. 2005; de Jonge et al. 2012; Grandaubert et al. 2014). Similarly, comparison of TE families showed a majority of species-specific sequences; this underlies the fact that each species genome has been widely invaded after the species divergence. We speculate that TEs may have played a role in host specialization by the acquisition or modification of pathogenicity-related traits. Also, large TE-rich regions could have played a role in speciation through the suppression of homologous recombination in chromosomal regions with inversions or translocations as proposed for the Leptosphaeria maculans–Leptosphaeria biglobosa species complex (Grandaubert et al. 2014). In Dhillon et al. (2014), the authors stated that several genes with putative pathogenicity-related functions were found to be associated with TEs. Preliminary analyses of genes and TEs tend to show that orphan genes are located significantly closer to TEs than other nonorphan genes (J. Grandaubert, unpublished data). TEs thus may play a role in the origin and evolution of these genes.
The annotation data presented here will greatly improve our understanding of gene evolution, adaptation, and the role of TEs in pathogen evolution. The predicted gene models will furthermore greatly support future studies of gene expression in Z. tritici and its related species and provide a strong basis to understand the underlying genetics and molecular biology of host–pathogen interactions in Zymoseptoria spp.