Limits...
Diversifying selection in the wheat stem rust fungus acts predominantly on pathogen-associated gene families and reveals candidate effectors.

Sperschneider J, Ying H, Dodds PN, Gardiner DM, Upadhyaya NM, Singh KB, Manners JM, Taylor JM - Front Plant Sci (2014)

Bottom Line: We demonstrate that in planta up-regulated genes that are rapidly evolving are found almost exclusively in pathogen-associated gene families, affirming the impact of host-pathogen co-evolution on genome structure and the adaptive diversification of specialized gene families.In particular, we predict 42 effector candidates that are conserved only across pathogens, induced during infection and rapidly evolving.This shows that comparative genomics incorporating the evolutionary signal of adaptation is powerful for predicting effector candidates for laboratory verification.

View Article: PubMed Central - PubMed

Affiliation: Plant Industry, Centre for Environment and Life Sciences, Commonwealth Scientific and Industrial Research Organisation Perth, WA, Australia.

ABSTRACT
Plant pathogens cause severe losses to crop plants and threaten global food production. One striking example is the wheat stem rust fungus, Puccinia graminis f. sp. tritici, which can rapidly evolve new virulent pathotypes in response to resistant host lines. Like several other filamentous fungal and oomycete plant pathogens, its genome features expanded gene families that have been implicated in host-pathogen interactions, possibly encoding effector proteins that interact directly with target host defense proteins. Previous efforts to understand virulence largely relied on the prediction of secreted, small and cysteine-rich proteins as candidate effectors and thus delivered an overwhelming number of candidates. Here, we implement an alternative analysis strategy that uses the signal of adaptive evolution as a line of evidence for effector function, combined with comparative information and expression data. We demonstrate that in planta up-regulated genes that are rapidly evolving are found almost exclusively in pathogen-associated gene families, affirming the impact of host-pathogen co-evolution on genome structure and the adaptive diversification of specialized gene families. In particular, we predict 42 effector candidates that are conserved only across pathogens, induced during infection and rapidly evolving. One of our top candidates has recently been shown to induce genotype-specific hypersensitive cell death in wheat. This shows that comparative genomics incorporating the evolutionary signal of adaptation is powerful for predicting effector candidates for laboratory verification. Our system can be applied to a wide range of pathogens and will give insight into host-pathogen dynamics, ultimately leading to progress in strategies for disease control.

No MeSH data available.


Related in: MedlinePlus

A pipeline for predicting effector candidates in expanded pathogen genomes incorporating comparative analysis, expression data and evolutionary support. The stem rust genes are grouped into gene families using TRIBE-MCL (Enright et al., 2002) and then split into families associated with pathogenicity (including those that are specific to P. graminis f. sp. tritici) and into fungal families, using comparative information from 78 fungal genomes. To limit the number of putative effector candidates, we use two additional lines of evidence: up-regulation during in planta infection and signatures of diversifying selection. Genes that are rapidly evolving, induced during infection and conserved predominantly across pathogenic fungi are returned as prime effector candidates.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4150398&req=5

Figure 1: A pipeline for predicting effector candidates in expanded pathogen genomes incorporating comparative analysis, expression data and evolutionary support. The stem rust genes are grouped into gene families using TRIBE-MCL (Enright et al., 2002) and then split into families associated with pathogenicity (including those that are specific to P. graminis f. sp. tritici) and into fungal families, using comparative information from 78 fungal genomes. To limit the number of putative effector candidates, we use two additional lines of evidence: up-regulation during in planta infection and signatures of diversifying selection. Genes that are rapidly evolving, induced during infection and conserved predominantly across pathogenic fungi are returned as prime effector candidates.

Mentions: We prioritize effector candidates in the stem rust P. graminis f. sp. tritici by linking comparative data, patterns of diversifying selection, and expression data during infection to a gene family's likelihood of participation in host-pathogen interactions (Figure 1). Using comparative information from 78 fungal genomes (pathogens and non-pathogens, Table S1), gene families were divided into those that can be associated with pathogenicity and those that are not related to pathogenicity (with the latter referred to as fungal genes). To dissect the nature of the gene families on a protein similarity level, a gene family representative was randomly chosen and the averages of 35 sequence-derived features for the gene family (e.g., average signal peptide prediction score, average molecular weight, average amino acid composition, see Materials and Methods) were assigned to each representative. This analysis then allowed us to use a previously developed unsupervised k-means clustering technique of sequence-derived features (Sperschneider et al., 2013). Enrichment and depletion analysis of the predicted protein clusters could then show whether putative effector clusters with certain characteristic features (e.g. secretion signals, small amino acids and cysteines) were present in an unbiased way, and whether commonalities or differences across pathogen-associated and fungal gene families existed. To include further evidence for pathogenicity function, ortholog sets were analyzed using prediction of their likelihood of undergoing diversifying selection. Pairwise dN/dS ratios from entire genes can only indicate diversifying selection if it has acted on all or the majority of the codons, which is rarely the case. Diversifying selection prediction is more powerful using site-specific or branch-specific models, which allow varying dN/dS ratios across codons or lineages. Proteins that were up-regulated in planta and were observed as part of pathogen-associated gene families predicted to undergo diversifying selection were selected as strong effector candidates for P. graminis f. sp. tritici.


Diversifying selection in the wheat stem rust fungus acts predominantly on pathogen-associated gene families and reveals candidate effectors.

Sperschneider J, Ying H, Dodds PN, Gardiner DM, Upadhyaya NM, Singh KB, Manners JM, Taylor JM - Front Plant Sci (2014)

A pipeline for predicting effector candidates in expanded pathogen genomes incorporating comparative analysis, expression data and evolutionary support. The stem rust genes are grouped into gene families using TRIBE-MCL (Enright et al., 2002) and then split into families associated with pathogenicity (including those that are specific to P. graminis f. sp. tritici) and into fungal families, using comparative information from 78 fungal genomes. To limit the number of putative effector candidates, we use two additional lines of evidence: up-regulation during in planta infection and signatures of diversifying selection. Genes that are rapidly evolving, induced during infection and conserved predominantly across pathogenic fungi are returned as prime effector candidates.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4150398&req=5

Figure 1: A pipeline for predicting effector candidates in expanded pathogen genomes incorporating comparative analysis, expression data and evolutionary support. The stem rust genes are grouped into gene families using TRIBE-MCL (Enright et al., 2002) and then split into families associated with pathogenicity (including those that are specific to P. graminis f. sp. tritici) and into fungal families, using comparative information from 78 fungal genomes. To limit the number of putative effector candidates, we use two additional lines of evidence: up-regulation during in planta infection and signatures of diversifying selection. Genes that are rapidly evolving, induced during infection and conserved predominantly across pathogenic fungi are returned as prime effector candidates.
Mentions: We prioritize effector candidates in the stem rust P. graminis f. sp. tritici by linking comparative data, patterns of diversifying selection, and expression data during infection to a gene family's likelihood of participation in host-pathogen interactions (Figure 1). Using comparative information from 78 fungal genomes (pathogens and non-pathogens, Table S1), gene families were divided into those that can be associated with pathogenicity and those that are not related to pathogenicity (with the latter referred to as fungal genes). To dissect the nature of the gene families on a protein similarity level, a gene family representative was randomly chosen and the averages of 35 sequence-derived features for the gene family (e.g., average signal peptide prediction score, average molecular weight, average amino acid composition, see Materials and Methods) were assigned to each representative. This analysis then allowed us to use a previously developed unsupervised k-means clustering technique of sequence-derived features (Sperschneider et al., 2013). Enrichment and depletion analysis of the predicted protein clusters could then show whether putative effector clusters with certain characteristic features (e.g. secretion signals, small amino acids and cysteines) were present in an unbiased way, and whether commonalities or differences across pathogen-associated and fungal gene families existed. To include further evidence for pathogenicity function, ortholog sets were analyzed using prediction of their likelihood of undergoing diversifying selection. Pairwise dN/dS ratios from entire genes can only indicate diversifying selection if it has acted on all or the majority of the codons, which is rarely the case. Diversifying selection prediction is more powerful using site-specific or branch-specific models, which allow varying dN/dS ratios across codons or lineages. Proteins that were up-regulated in planta and were observed as part of pathogen-associated gene families predicted to undergo diversifying selection were selected as strong effector candidates for P. graminis f. sp. tritici.

Bottom Line: We demonstrate that in planta up-regulated genes that are rapidly evolving are found almost exclusively in pathogen-associated gene families, affirming the impact of host-pathogen co-evolution on genome structure and the adaptive diversification of specialized gene families.In particular, we predict 42 effector candidates that are conserved only across pathogens, induced during infection and rapidly evolving.This shows that comparative genomics incorporating the evolutionary signal of adaptation is powerful for predicting effector candidates for laboratory verification.

View Article: PubMed Central - PubMed

Affiliation: Plant Industry, Centre for Environment and Life Sciences, Commonwealth Scientific and Industrial Research Organisation Perth, WA, Australia.

ABSTRACT
Plant pathogens cause severe losses to crop plants and threaten global food production. One striking example is the wheat stem rust fungus, Puccinia graminis f. sp. tritici, which can rapidly evolve new virulent pathotypes in response to resistant host lines. Like several other filamentous fungal and oomycete plant pathogens, its genome features expanded gene families that have been implicated in host-pathogen interactions, possibly encoding effector proteins that interact directly with target host defense proteins. Previous efforts to understand virulence largely relied on the prediction of secreted, small and cysteine-rich proteins as candidate effectors and thus delivered an overwhelming number of candidates. Here, we implement an alternative analysis strategy that uses the signal of adaptive evolution as a line of evidence for effector function, combined with comparative information and expression data. We demonstrate that in planta up-regulated genes that are rapidly evolving are found almost exclusively in pathogen-associated gene families, affirming the impact of host-pathogen co-evolution on genome structure and the adaptive diversification of specialized gene families. In particular, we predict 42 effector candidates that are conserved only across pathogens, induced during infection and rapidly evolving. One of our top candidates has recently been shown to induce genotype-specific hypersensitive cell death in wheat. This shows that comparative genomics incorporating the evolutionary signal of adaptation is powerful for predicting effector candidates for laboratory verification. Our system can be applied to a wide range of pathogens and will give insight into host-pathogen dynamics, ultimately leading to progress in strategies for disease control.

No MeSH data available.


Related in: MedlinePlus