Genome analyses of the wheat yellow (stripe) rust pathogen Puccinia striiformis f. sp. tritici reveal polymorphic and haustorial expressed secreted proteins as candidate effectors

Abstract

Background: Wheat yellow (stripe) rust caused by Puccinia striiformis f. sp. tritici (PST) is one of the most devastating diseases of wheat worldwide. To design effective breeding strategies that maximize the potential for durable disease resistance it is important to understand the molecular basis of PST pathogenicity. In particular, the characterisation of the structure, function and evolutionary dynamics of secreted effector proteins that are detected by host immune receptors can help guide and prioritize breeding efforts. However, to date, our knowledge of the effector repertoire of cereal rust pathogens is limited.

Results: We re-sequenced genomes of four PST isolates from the US and UK to identify effector candidates and relate them to their distinct virulence profiles. First, we assessed SNP frequencies between all isolates, with heterokaryotic SNPs being over tenfold more frequent (5.29 ± 2.23 SNPs/kb) than homokaryotic SNPs (0.41 ± 0.28 SNPs/kb). Next, we implemented a bioinformatics pipeline to integrate genomics, transcriptomics, and effector-focused annotations to identify and classify effector candidates in PST. RNAseq analysis highlighted transcripts encoding secreted proteins that were significantly enriched in haustoria compared to infected tissue. The expression of 22 candidate effector genes was characterised using qRT-PCR, revealing distinct temporal expression patterns during infection in wheat. Lastly, we identified proteins that displayed non-synonymous substitutions specifically between the two UK isolates PST-87/7 and PST-08/21, which differ in virulence to two wheat varieties. By focusing on polymorphic variants enriched in haustoria, we identified five polymorphic effector candidates between PST-87/7 and PST-08/21 among 2,999 secreted proteins. These allelic variants are now a priority for functional validation as virulence/avirulence effectors in the corresponding wheat varieties.

Conclusions: Integration of genomics, transcriptomics, and effector-directed annotation of PST isolates has enabled us to move beyond the single isolate-directed catalogues of effector proteins and develop a framework for mining effector proteins in closely related isolates and relate these back to their defined virulence profiles. This should ultimately lead to more comprehensive understanding of the PST pathogenesis system, an important first step towards developing more effective surveillance and management strategies for one of the most devastating pathogens of wheat.

Figure 1

US PST-43 may belong to the same clonal lineage as the UK isolates PST-87/7 and PST-08/21. A. Dendrograms illustrating that US isolate PST-43 clustered with the two UK isolates PST-87/7 and PST-08/21. Dendrograms were constructed using the homokaryotic SNP information either in the coding or non-coding regions of the genome from all pair-wise alignments. B. Pair-wise comparison of non-synonymous mutations in synthetic gene sets illustrated that polymorphisms were more apparent when corresponding proteins representing PST-21 and PST-130 were compared against other isolates. Each gene for a given reference was taken in turn and any homokaryotic SNPs incorporated for each isolates mapped. The five genes (one reference gene and four synthetic genes) were then subjected to pair-wise polymorphism and positive selection analysis using Yn00. Circle sizes represent the number of proteins with at least one non-synonymous mutation (green circles) or under positive selection (purple circles). Pair-wise comparisons that showed enrichment in non-synonymous mutations in secreted proteins are illustrated with an orange background. C. The total number of genes determined as absent by mapping the sequence reads from each isolate in turn against every other isolate as a reference was greater for alignments against PST-21 and PST-130 when compared to alignments against PST-43, PST-87/7 and PST-08/21. A similar pattern was also observed when genes encoding predicted secreted proteins were assessed. Total number of genes absent from white to blue (0-482) for full gene complement and white to red (0-14) for secretome.

Figure 2

Comparison between infected tissue and isolated haustoria RNAseq libraries. Scatter plot of log₂ transformed sequencing counts generated by aligning RNAseq reads to all PST-08/21 genes (A), those that encode predicted secreted peptides (B), or encode non-secreted peptides (C). Red and green colored circles correspond respectively to transcripts that were identified as significantly enriched or depleted in isolated haustoria as determined by DESeq analysis (P ≤ 0.01). Red lines represent the locally weighted polynomial regression (LOWESS method).

Figure 3

Clustering of secreted proteins and annotation of protein tribes based on known effector features and PST-specific criteria. A bioinformatic pipeline was implemented to identify groups of secreted proteins with characteristic effector features. The proteomes of the five PST isolates were combined (totaling 100357 proteins), 5502 secreted proteins predicted, and redundancy reduced. The consolidated PST secretome (2999 proteins) was combined with predicted secretomes from P. graminis f. sp. tritici (PGT) and Melampsora larcia populina (Mellp) and proteins grouped based on sequence similarity (Markov clustering). Tribes containing at least one PST member (1037 tribes) were annotated with known effector features or PST-specific criteria. Finally, tribes were ranked and heirarchical clustering implemented based on their content of proteins with known effector features. Programs are indicated in red. NLS, nuclear localization signal.

Figure 4

The top 100 ranked protein tribes containing putative effector candidates. Clusters were determined using hierarchical clustering of the top 100 ranked tribes containing putative effector candidates. A. Combined score used to rank tribes based on their content of effector features. B. Score for similarity of tribe members to haustorial expressed secreted proteins (HESPs) or characterized fungal AVRs. C. Score for number of members encoded by genes with at least one flanking intergenic region >10 Kb. D. Score for number of members classified as repeat containing (RCPs). E. Score reflecting number of members classified as small and cysteine rich (SCRs). F. Score for number of members containing any characterized effector motifs or nuclear localization signals (NLS). G. Score for number of members not annotated by PFAM domain searches. H. Score for number of members in the top 100 expressed in infected material, determined by mRNA-seq analysis. I. Scatter plot indicating number of members in the top 1, 5 or 10% expressed in infected material. J. Score for number of members in the top 100 expressed in haustoria, determined by mRNA-seq analysis. K. Scatter plot indicating number of members in the top 1, 5 or 10% expressed in haustoria. L. Number of PST members showing sequence polymorphisms between isolates. Stars indicate tribes that contain members assessed for expression using qRT-PCR.

Figure 5

Secreted proteins with high likelihood as candidate effectors. A. Sequence alignment of PST21_18220, a member of tribe 238, and the corresponding alleles from the other four isolates illustrating sequence polymorphisms specifically between the US isolates, PST-21, PST-43 and PST-130. B. Sequence alignment of the second member of tribe 238, PST21_18221, and its alleles from other isolates illustrating that this protein was highly conserved across isolates. C. The two members of tribe 238, PST21_18220 and PST21_18221, are in close proximity within a single contig in the genome sequence. The corresponding genes were expressed during infection and were also highly expressed and enriched in haustorial samples as determined by mRNA-seq analysis. D. Features displayed by the 117 proteins that were identified as polymorphic between the two UK isolates PST-08/21 and PST-87/7. E. Sequence alignment of PST130_05023 and the synthetic genes that incorporate the SNP information from the other four isolates sequenced, illustrating sequence polymorphisms between isolates. Polymorphic residues are indicated below the sequence by red stars.

Figure 6

Quantitative RT-PCR revealed peaks of expression for the selected effector candidates during plant infection. A. Schematic representation of the stages of PST development during plant infection. B. Quantitative RT-PCR was undertaken at four stages of PST-08/21 infection for a subset of 22 effector candidates. Three peaks of expression were noted at 1 day post-inoculation (dpi), 6 dpi and 14 dpi. hpi, hours post-inoculation; S, uredinospore; SV, substomatal vesicle; IH, invasive hyphae; HM, haustorial mother cell; H, haustorium; P, pustule; G, guard cell.