In the present study, 56 Avena accessions were postulated to carry Pg6 and 20 Avena accessions were identified with potentially novel resistance (Table 4). Unique resistance was rare in the diploids and only three accessions, PI 158247, an A. strigosa accession from Portugal, PI 131695 an A. strigosa accession from Poland and PI 657297 an A. longiglumis accession from Morocco showed moderate resistance to race TQL. Additionally, PI 186614, from Rio Grande do Sul, Brazil had a unique ITs pattern across the Pga races and did not contain the allele associated with Pg6. All four of these diploid accessions warrant additional study to determine if their unique resistance is conferred by novel resistance genes.
Four of the accessions in this study were resistant to DBD but susceptible to all the other races tested. This type of race-specific resistance has not been previously documented within A-genome Avena species. This resistance could be conferred by either Pg2 or Pg4, as these genes are effective against race DBD and ineffective against race KBD, or some previously unreported resistance. However, this DBD-only resistance will be ineffective in fields where virulence to these genes is widespread.
Steinberg et al. (2005) identified 35 accessions of the 9,978 tested as having high levels of field resistance to oat stem rust. Of these, 33 were susceptible to race NA1 with virulence to Pg6, indicating that these accessions may carry Pg6. We were able to compare 22 of the accessions in their study using PI/CI accession numbers matching accessions with Pg6 postulations in the present study and determined that all of them likely contain Pg6 (Supplementary Table S1). In their study, only two A. barbata accessions, CN 23731 and CN 26171, were resistant to NA1 with IT of 0;1, which was more pronounced resistance than the IT of ‘2’ exhibited by the resistant tetraploid accessions in the present study, and may represent another novel source of resistance.
Five A. vaviloviana accessions from Oromīya, Ethiopia, had resistant ITs of ‘2’ across the oat stem rust races tested (Table 4). A. vaviloviana is an allotetraploid species with an AB-genome that is closely related to A. barbata Pott ex Link and A. abyssinica Hochst. (Chew et al. 2016; Yan et al. 2016). Intermediate levels of field resistance to oat stem rust were previously reported at low frequency among tested accessions in all three species (Steinberg et al. 2005). Given their similar origin, collection date, and IT,these five A. vaviloviana accessions likely contain a single novel source of resistance which warrants further investigation. A. barbata and A. abyssinica may harbor additional novel alleles, and accessions from these species and other tetraploids should be tested against important oat stem rust races DBD, TJS and TQL to identify additional resistant sources.
All 56 Pg6-carrying accessions from the diversity panel were A-genome diploids with As or Al genomes (Table 1). Maughan et al. (2019) demonstrated that A. atlantica, A. strigosa, and A. wiestii constitute a single species complex differentiated by seed dispersal mechanisms, whereas A. brevis could not be genetically differentiated from A. strigosa. Fifty-one of the postulated Pg6-carrying accessions were in this As clade and six accessions were in the distantly related A. longiglumis clade. The SNP marker most closely associated with the Pg6 phenotype in the diversity panel matched the Pg6 phenotype in all but one of the As-genome accessions, but did not align with the Pg6 phenotypes associated with A. longiglumis accessions (Table 3). A. longiglumis has a distinct morphology from other A-genome species, is distantly related to other A-genome diploids and is thought to be the progenitor of all extant Avena hexaploids (Yan et al. 2016). It would be interesting to understand whether the resistance in As and Al genome diploids is conferred by the same gene or different race-specific genes that have identical resistance patterns across stem rust races. Cloning Pg6 in A. strigosa and mapping the Pg6-like resistance in A. longiglumis would expand understanding of how resistance arose in A-genome Avena accessions and might provide valuable insights into race-specific resistance gene evolution.
Only 11 C-genome accessions were available for testing, and they were susceptible to all four oat stem rust races used in this study (Table 2). A more exhaustive investigation utilizing C-genome accessions from other collections would be required to conclude that oat stem rust resistance is not present in C-genome diploids. Recent genetic studies proposed that speciation between the A- and C-genome diploids occurred between 5.4 and 12.9 million years ago and subsequent tetraploidization and hexaploidization events likely occurred during the Miocene-Pliocene interval in northwest Africa (Chew et al. 2016; Liu et al. 2017; Maughan et al. 2019). If further testing verifies that the Pg6 phenotype is present in only A-genome diploids, then Pg6 may have arisen after the A-, C-genome diploid speciation event and was absent in the diploid progenitors of current tetraploid and hexaploid species.
Eight of the Pg6 carrying accessions showed mixed IT reactions (Table 3). Mixed IT reactions indicate the importance of deriving accessions from a single seed source and retesting the derived line to confirm the phenotype before proceeding with further genetic testing. Mixed accessions can also make it difficult to draw conclusive associations between previously genotyped or sequenced materials and current phenotyping efforts. Care was taken in this study to choose accessions with clear phenotypic responses for SNP development.
CIav 6956, the Pg6 carrier, showed strong seedling resistance and moderate field resistance to crown rust (T. Gordon, unpublished). Crown rust resistance in another A. strigosa accession PI 258731 is remarkably stable and has been introgressed into hexaploid oat (Rines et al. 2018). Another broadly effective source of oat crown rust resistance, Pc94, was introduced from the A. strigosa accession PI 186606, ‘Saia’ from Rio Grande do Sul, Brazil. The molecular markers that have been developed for the crown rust resistance loci in PI 258731 and Pc94 are on A. atlantica chromosome scaffolds ScoFOjO_350_483 and ScoFOjO_324_449, respectively, whereas Pg6 resistance was localized to ScoFOjO_1702_2338. These results support a hypothesis that resistance to these rusts is derived from different chromosomal regions, but the relationship between rust resistance loci within A. strigosa warrants further investigation.
Kebebe et al. (2020a) mapped the oat stem rust gene Pg13 between 67.7 and 68.5 cM on hexaploid linkage group Mrg 18. The diagnostic markers reported for Pg13 in their study were between 491,649,525 and 498,515,330 bp on the diploid chromosome AA2. They also found that the oat crown rust resistance gene Pc91 co-segregated with Pg13 on Mrg 18 at the 7C-17A translocation breakpoint. Pc91 was originally introgressed into A. sativa cultivars from the synthetic hexaploid, ‘Amagalon’, CIav 9364. This line was produced by crossing the tetraploid A. magna accession, CIav 8330, with the A. longiglumis line, ‘CW 57’, but it is not documented which species contributed this resistance (Rothman 1984). It is apparent that these three rust resistance genes, Pg6, Pg13, and Pc91, are very close to one another on the A-genome. However Pg6 and Pg13, show different race specificity (Supplementary Table S1) and the marker most closely associated with Pg6 in the present study, AA2_483439497, is at least 8 Mbp proximal to the markers closest to Pg13 and Pc91. Additional testing also indicated that Amagalon is susceptible to Pga race KBD (T. Gordon, unpublished). A comparative sequencing technique, like the one presented in the current study, could be used to elucidate the relationship between Pg6, Pg13, and Pc91.
Maughan et al. (2019) previously annotated 1,563 RGAs within the A. atlantica genome which typically clustered in sub-telomeric regions. In this study, three clusters of SNPs aligned perfectly with the Pg6 phenotype in the genomic region between 475 and 490 Mbp on AA2 (Fig. 2). The first cluster was composed of 1,138 SNPs, between 478.5 and 479.4 Mbp, the second was composed of 129 SNPs between 482.0 and 482.4 Mbp and the third was composed of 69 SNPs between 483.4 and 483.6 Mbp. Within the first large SNP cluster there was one RGA, a leucine-rich repeat receptor-like protein kinase (LRRK) in a 3 kb section beginning at 478,733,268 bp and annotated as ‘AA012417’ in the A. atlantica genome. Most SNPs with perfect association in this region were located slightly downstream from this LRRK gene. However, one SNP located at 478,733,705 bp was within this gene. In contrast, the assay that interrogated this SNP and other SNPs in the first cluster were several cM away from the resistance locus in the RIL population (Table 6).
Another RGA, a 5 kb resistance to Peronospora Parasitica protein 13 (RPP13) between 483,422,214 and 483,427,403 bp and annotated as ‘AA012586’ was located in the third SNP cluster. RPP13 is an NBS-LRR protein which initiates a race-specific hypersensitive response in Arabidopsis thaliana when challenged with the obligate biotrophic oomycete pathogen, Hyaloperonospora arabidopsidis (Rentel et al. 2008). The interaction between the cloned effector ATR13 and RPP13 elicits a common defense response that was effective against oomycete, viral, and bacterial pathogens (Rentel et al. 2008). Assays used to interrogate SNPs in the region close to the RPP13 analog were predictive of Pg6, specifically, marker AA2_483439497 was perfectly aligned with the Pg6 phenotype in the mapping populations and within the As genome accessions in the diversity panel (Table 6). This marker was flanked by two SNPs, AA2_483429191 and AA2_483451960, that were slightly less predictive of the Pg6 phenotype. Oddly, the SNP within the RPP13 gene sequence region AA2_483427147, and the SNP only 2 kb distal to the gene, AA2_483429191, were less predictive of the Pg6 phenotype than AA2_483439497 which was 12 kb distal indicating a slight rearrangement from the expected gene sequence. Nevertheless, since no other annotated RGA genes were found in this region, these results provide strong support for RPP13 as the candidate Pg6 resistant gene.
NBS-LRR type genes are effective at controlling biotrophic and hemibiotrophic pathogens but wide deployment of this type of gene has been problematic in the case of necrotrophic pathogens. Susceptibility to Victoria Blight caused by the necrotrophic fungal pathogen Bipolaris victoriae was shown to be conferred by the same NBS-LRR resistance gene that conferred resistance to crown rust caused by the biotrophic fungal pathogen Pca, and wide deployment of this type of resistance could induce susceptibility to necrotrophic pathogens (Lorang et al. 2007). Despite the close proximity of the most diagnostic SNPs to an NBS-LRR gene, a causal association has not been made, and further expression, annotation, and gene cloning studies will be required to elucidate a mechanism for Pg6 resistance.
In conclusion, Pg6 is a widely effective oat stem rust resistant gene, and SNP markers closely linked with this gene enabled identification of novel sources of oat stem rust resistance from within a diverse collection of Avena diploid germplasm. A comparative sequencing technique was used to quickly narrow a genomic region of interest and select a candidate RGA. The utility of the SNPmarker at 483439497 bp on AA2 was validated in diverse germplasm and can be used to screen additional germplasm collections and assist with introgression and gene pyramiding of Pg6.