BED domain‐containing NLR from wild barley confers resistance to leaf rust

Summary Leaf rust, caused by Puccinia hordei, is a devastating fungal disease affecting barley (Hordeum vulgare subsp. vulgare) production globally. Despite the effectiveness of genetic resistance, the deployment of single genes often compromises durability due to the emergence of virulent P. hordei races, prompting the search for new sources of resistance. Here we report on the cloning of Rph15, a resistance gene derived from barley’s wild progenitor H. vulgare subsp. spontaneum. We demonstrate using introgression mapping, mutation and complementation that the Rph15 gene from the near‐isogenic line (NIL) Bowman + Rph15 (referred to as BW719) encodes a coiled‐coil nucleotide‐binding leucine‐rich repeat (NLR) protein with an integrated Zinc finger BED (ZF‐BED) domain. A predicted KASP marker was developed and validated across a collection of Australian cultivars and a series of introgression lines in the Bowman background known to carry the Rph15 resistance. Rph16 from HS‐680, another wild barley derived leaf rust resistance gene, was previously mapped to the same genomic region on chromosome 2H and was assumed to be allelic with Rph15 based on genetic studies. Both sequence analysis, race specificity and the identification of a knockout mutant in the HS‐680 background suggest that Rph15‐ and Rph16‐mediated resistances are in fact the same and not allelic as previously thought. The cloning of Rph15 now permits efficient gene deployment and the production of resistance gene cassettes for sustained leaf rust control.


Introduction
In 2018, barley (Hordeum vulgare subsp. vulgare) was ranked fourth amongst grain crops in production (141 million tonnes) behind maize, rice and wheat (FAOSTAT, 2018). It is used primarily for animal feed and malt production, but also serves as a major food staple in the mountainous areas of Central Asia, Southwest Asia, the Andes of South America and North Africa. Leaf rust, caused by the fungus Puccinia hordei Otth, is the most damaging and widespread rust disease of barley (Park et al., 2015). Leaf rust epidemics can occur in most if not all barley growing regions and have been reported to cause significant reductions in grain quality and yield. Yield losses up to 62% have been reported in highly susceptible barley cultivars (Cotterill et al., 1992). Resistance to leaf rust in barley is conferred by either qualitative type Rph (Reaction to Puccinia hordei), all-stage resistances (ASR) that are typically race-specific or by quantitative trait loci (QTL) conferring partial adult plant resistance (APR) that is generally not race-specific (Niks et al., 2015). Due to P. hordei evolution and the subsequent emergence of new pathogenic variants, ASR resistance to leaf rust is often transiently effective. Of the 22 catalogued ASR genes, five (Rph10, Rph11, Rph13, Rph15 and Rph16) originate from H. v. ssp. spontaneum (Park et al., 2015).
Leaf rust resistance gene Rph15 was originally sourced from PI 355447, an accession of wild barley collected from Israel (Chicaiza, 1996;Jin and Steffenson, 1994). The reported widespread effectiveness of the Rph15 resistance to >350 P. hordei isolates likely reflects its limited deployment in agriculture (Martin et al., 2020a). Virulence for Rph15 does, however, exist in nature and was identified in an Israeli isolate (90-3) from a global survey, presaging the possible breakdown of Rph15 should it be deployed singly in cultivars (Martin et al., 2020a). Given the apparent rare occurrence of virulence for Rph15, it represents a valuable gene for leaf rust control in cultivated barley when deployed in combination with other effective resistance genes especially given the limited number of Rph genes that have been isolated to date. Chicaiza (1996) first determined that the Rph15 resistance was inherited as a single dominant gene that mapped to the centromeric region of the short arm of chromosome 2H in close linkage to the RFLP marker MWG2133. Martin et al. (2020a) recently developed a series of introgression lines (Rph1-Rph15) in the genetic background of cultivar Bowman and mapped Rph15 to a physical interval (44-57Mb) on chromosome 2H in the Morex reference genome (Figure 1). Genetic mapping of resistance genes Rph15 and Rph16 on chromosome 2H was originally performed using the Bowman + Rph15 9 Bowman (Weerasena et al., 2004) and the HS-680 9 L94 (Perovic et al., 2004) mapping populations, respectively. Further genetic studies determined that Rph15 and Rph16 from wild barley were likely allelic, and both were closely linked or colocated with RFLP marker MWG2133 (Weerasena et al., 2004). Other studies have repeatedly mapped leaf rust resistance at the Rph15 locus, suggesting the possible presence of an allelic series or complex resistance locus on chromosome 2H (Derevnina et al., 2014;Ivandic et al., 1998;Perovic et al., 2004;Weerasena et al., 2004). In this study, our aim was to isolate the Rph15 resistance gene and assess the molecular basis of the postulated allelic relationship between Rph15 and Rph16 on the short arm of chromosome 2H.

Results and discussion
The approach used in this study for the cloning of Rph15 was based on two hypotheses. The first hypothesis was that based on the susceptibility of cultivar Morex to races of P. hordei that are avirulent for Rph15, its sequenced genome (Mascher et al., 2017) may be useful for identifying the rph15 susceptibility allele within the physical interval reported by Martin et al. (2020a) (Figure 1). The second hypothesis was that due to the characteristic hypersensitive response associated with Rph15-mediated resistance, we anticipated the involvement of immune receptors encoded by R genes such as nucleotide-binding site leucine-rich repeats (NLRs). The 12.5 Mb introgressed segment defined for the near-isogenic line (NIL) Bowman + Rph15 (referred to as BW719) was analysed for the presence of NLR genes using the NLR annotator tool, which is an automatic high-throughput analysis for the NLR-specific conserved motif combinations (Steuernagel et al., 2020). A single NLR was identified represented by the gene model HORVU2Hr1G019120.5 from the Morex reference (Mascher et al., 2017) and was deemed the best candidate for further investigations based on a loss of function mutant population derived from the BW719 background.
A mutagenesis-based approach was used to verify the HORVU2Hr1G019120.5 candidate as the gene of interest and chemically treated 1500 seed sourced from the same BW719 stocks reported in Martin et al. (2020a). To confirm the BW719 stock was the same as that used to map the introgression in Martin et al. (2020a) to use for mutagenesis, we reconfirmed the previously reported monogenic inheritance and linkage of Rph15 with RFLP marker MWG2133 on chromosome 2H using a Bowman + Rph15 9 Gus F 2:3 population ( Figure S1). Rust testing of progeny seeds from 4320 M 2 spikes with the Rph15-avirulent P. hordei isolate 5457 Pj + identified eight independent mutants (rph15) that were confirmed at the M 3 generation. A 9442 base pair (bp) full-length genomic sequence that included 1.3 kb and 2.5 kb of 5 0 and 3 0 sequence, respectively, was amplified from Bowman + Rph15 using primers designed to the candidate gene model (HORVU2Hr1G019120.5) based on the Morex (rph15) sequence. The NLR motif prediction was used to develop a fulllength NLR candidate gene model annotation ( Figure 2). We then sequenced the full-length gene from the eight mutants and identified non-synonymous mutations in six out of the eight lines including: a premature stop codon (M1695), serine to proline substitutions or vice versa (M1371-3, M1727 and M4022) and substitutions from glycine to either glutamic or aspartic acid (M4022 and M4321). All non-synonymous mutations were identified in the LRR region of the Rph15 gene ( Figure 2; Figure S2). The remaining two susceptible mutants did not contain sequence changes in the candidate gene, suggesting they likely carry non-synonymous mutations within downstream signalling components essential for Rph15-mediated resistance. Similar findings were also identified during the recent cloning of the leaf rust resistance gene Rph1 from cultivated barley (Dracatos et al., 2019). A KASP marker was developed by interrogating a G/C SNP within the 3rd intron of the Rph15 gene that was further validated as predictive for the Rph15 resistance. We assessed the utility of the KASP marker developed within the Rph15 gene across 61 near-isogenic lines in the Bowman background (BW lines -Martin et al., 2020a) and 80 Australian cultivars that lack the Rph15 resistance based on extensive leaf rust phenotyping. All 80 Australian cultivars lacked the Rph15 marker allele corroborating their phenotypic response. Based on data reported in Martin et al. (2020a), from the 61 BW lines tested for rust resistance, 50 were postulated to carry Rph15 either singly or in combination with another resistance gene based on the presence of the Rph15 KASP marker allele, whilst the remaining 11 were postulated to carry unknown resistances (Table S1). Interestingly, six of the BW lines postulated to carry Rph15 and a further six BW lines with unknown leaf rust resistance failed to amplify (same signal profile as water blank) and likely carry a SNP haplotype at the marker different to either Bowman or BW719. Further work is underway to determine whether these BW lines carry uncharacterized resistance or allelic variants of the Rph15 gene (Table S1).
The full-length genomic sequence of Rph15 from BW719 was cloned into a binary vector and transformed into the leaf rust susceptible barley cultivar Golden Promise. Six independent hemizygous T 0 lines were produced that contained the wild-type Rph15 candidate in addition to two control lines that contained an empty vector only ( Figure S3). The progeny from four T 1 Golden Promise + Rph15 transgenic families was evaluated at the seedling stage in Australia with Rph15-avirulent pathotype 5457 P+ and in North America with the Rph15-avirulent P. hordei pathotype 92-7 and the only known Rph15-virulent P. hordei isolate 90-3 collected from Israel ( Figure S4; Table S2; Table S3). All four Golden Promise + Rph15 T 1 generation transgenic lines showed the expected segregation for low infection types (0;CN-;1CN) characteristic of the Rph15 resistance observed on BW719 seedlings in response to both Rph15-avirulent P. hordei isolates ( Figure 3). In response to the Rph15-virulent isolate (90-3), nearly all plants assessed from the four transgenic T 1 families were susceptible; however, variation was observed between the two experimental replicates due to the likely presence of inoculation escapes in replicate 1 [evidenced by the presence of a few plants that were immune (00;) with no necrosis that was uncharacteristic of the Rph15 resistance] (Table S3). Based on qRT-PCR data, the relative expression of Rph15 was variable both within and between T 1 families possibly reflecting differences in copy number or the site of T-DNA integration in the genome of Golden Promise and the assessment of siblings that were either heterozygous or homozygous for the Rph15 transgene/s (Figure S5).
To confirm the correlation between the presence of the transgene and the observed phenotypic response in the T 1 transgenic families, 12 sib plants per T 1 family were genotyped with the predictive KASP marker. Genotypic analysis confirmed, without exception, cosegregation between the presence of the heterozygous SNP genotype (Golden Promise susceptibility allele in combination with the T-DNA from BW719 in all resistant sibs, and only the susceptibility allele found within all susceptible sib lines). Thus, the exploitation of previous mapping data, combined with both mutation and complementation experiments in the present study, facilitated confirmation of the candidate as the causal gene for Rph15-mediated resistance.
Rare sub-families of NLR immune receptors in plants have evolved to encode additional integrated domains that act as decoys to recognize and interact with pathogen effectors that would normally target host transcription factors containing zinc finger BED (ZF-BED) domains (Grund et al., 2019;Kroj et al., 2016). The ZF-BED domain was originally characterized in Drosophila through mutagenesis studies and was subsequently shown to be essential in rice for Xa1-mediated resistance to bacterial blight (Yoshimura et al., 1998). More recently, three BED domain NLRs (Yr5, Yr7 and YrSP on the long arm of chromosome 2B) with unique pathogen specificity conferred resistance to stripe rust in bread wheat, comprising a complex resistance NLR cluster (Marchal et al., 2018). Sequencing of the cDNA from BW719 indicated that the Rph15 gene consists of four (three small and one large) exons and encodes an immune receptor with a single predicted BED domain followed by the canonical NLR (1696 residues) (Figures 2 and 4). The presence of four exons and only a single BED domain differs from the three exons structure identified for BED domain-containing NLRs (Yr5/Yr7/YrSP) in bread wheat (Marchal et al., 2018). The identification of a BED domain within the encoded Rph15 immune receptor reveals that different rust pathogen species adapted to distinct hosts may have effectors targeting similar transcription factor domains.
Phylogenetic analysis using numerous recently cloned NLR immune receptors from the Triticeae determined that Rph15 was most closely related to the susceptible version of the Rph15 protein from Morex followed by other BED domain carrying NLR proteins, including both Yr5 and Yr7 homologues from wheat and Xa1 from rice on chromosome 4 (corresponding to 2H in barley) ( Figure 5). Interestingly, the phylogenetic relatedness between NLRs was not based on the presence or absence of a BED domain as we performed sequence alignments using fulllength NLRs both with and without the BED domains ( Figure 5).  Figure 2 Model of the Rph15 gene amplified from BW719 near-isogenic line using primers designed to candidate gene HORVU2Hr1G019120.5. The gene consists of four exons (blue boxes) and three introns (black lines). The CC and NB-ARC domains were predicted using the NLR annotator software developed by Steuernagel et al., (2020) and the LRR motifs were predicted as described by Martin et al., (2020b). Red arrows indicate the positions of susceptible mutants identified within the BW719 (Rph15) background and the orange arrow was a single mutant identified in the HS-680 (Rph16) background. Areas spanning conserved domains are highlighted in green. Rph15 was closely related to the stem rust resistance protein sequence of Sr21 from Triticum monococcum and other NLRs conferring rust and mildew resistance from bread wheat and its relatives whilst the MLA clade and the recently cloned leaf rust resistance gene encoding Rph1 from cultivated barley clustered separately from Rph15 (Dracatos et al., 2019; Figure 5). Further examination of the BED domains between Rph15, Yr5/Yr7 and Xa1 suggests that the BED domain from Rph15 was more closely related to the BED II clade consensus and yet only carried six out of the nine conserved residues, suggesting the possible presence of a functionally diverse BED domain distinct from BED I and II ( Figure S6). According to Marchal et al. (2018), Yr5 and Yr7 carried BED domains that clustered with other clade I BED domain carrying NLRs from the Triticeae, whilst Xa1 from rice clustered separately from both clades I and II yet still carried eight out of the nine conserved residues between BED I and II clades. Further experiments assessing both the function and involvement of the BED domain in the Rph15-mediated resistance are required to test this hypothesis.

ATG CC CC ? ZnF-BED NB-ARC LRR
To determine the functional SNP variant/s required for Rph15mediated resistance, we aligned Rph15 protein sequence with homologues from four different cultivated barley genotypes that lack the Rph15 resistance, viz. Morex, Barke, Gus and Golden Promise (Mascher et al., 2017;Schreiber et al. 2020). Despite all barley genotypes assessed having distinct haplotypes, three main diverse haplogroups were identified amongst the six sequences: the Rph15 resistance haplotype and two distinct susceptibility haplotypes rph15_1 and rph15_2 (Figure 4). The Gus and Morex rph15 proteins differed by a single amino acid substitution, and their rph15_1 haplotype was characterized by numerous nonsynonymous SNPs mainly in the LRR domain. One critical SNP was identified that caused a substitution from a cysteine to a premature STOP codon (Figure 4). The other susceptible haplotype rph15_2 was identified in both Barke and Golden Promise and differed by >45 residues largely due to a deletion event of 30 residues relative to the other haplotypes also in the LRR domain.
Based on the alignment, we resolved the difference between resistance and susceptibility to six functional SNP variants. Further sequencing efforts of additional barley accessions is likely to determine how many susceptibility haplotypes exist within the Hordeum gene pool and permit the identification of the causal polymorphisms differentiating susceptibility vs Rph15-mediated resistance. Many of these variants can be further interrogated for diagnostic marker design and future gene-editing applications.
The discovery of two other widely resistant H. vulgare sp. spontaneum accessions (HS078 and HS-680) led to the identification of another leaf rust resistance allele at the Rph15 locus (Ivandic et al., 1998). The development of segregating doubled haploid (DH) populations using the susceptible line, L94 and subsequent genetic map construction led to the mapping of the resistance loci of both HS078 and HS-680 in the short arm of chromosome 2H via linkage with RFLP markers MWG784 and MWG2133 (Ivandic et al., 1998). An early study previously mapped Rph15 to the long arm of chromosome 2H; therefore, the gene in both HS078 and HS-680 was predicted as independent and was catalogued as Rph16 (Ivandic et al., 1998). Weerasena et al. (2004) subsequently concluded that based on the lack of segregation observed in 1027 F 2 individuals from the cross BW719 9 HS-680, Rph15 and Rph16 are likely alleles of each other. Furthermore, closely linked AFLP markers led to a change in map position from the long arm to the short arm of chromosome 2H. Hence Rph15 was in close proximity to Rph16 based on cosegregation with RFLP marker MGW2133. To further unravel the genetic relationships between Rph15 and Rph16, in this study, we used the same primer pairs used to PCR-amplify the Rph15 gene from BW719 to determine the Rph15 haplotype in HS-680. Both BW719 and HS-680 had identical (100% nucleotide identity over 9.5kb) Rph15 haplotypes; therefore, we propose an alternative hypothesis, that the Rph16 resistance in HS-680 is due to the same BED domain-containing NLR conferring Rph15-mediated resistance originally sourced from PI 355447 in BW719.

Bowman + Rph15
Golden Promise B49-11a B49-2 B49-3 Vec8 B49-8a Figure 3 Complementation of the wild-type Rph15 gene from BW719 into Golden Promise using Agrobacterium-mediated transformation. Resistant sib plants from four different Golden Promise + Rph15 T 1 generation families showed the same resistance reaction in response to the Rph15-avirulent Puccinia hordei race 5457 P + 12 days after infection as the Bowman + Rph15 control relative to the susceptible cv. Golden Promise and the empty vector control Vec8. Examination of historical phenotypic data using five Rph15avirulent isolates from diverse origins and one Rph15-virulent isolate from Israel determined that Rph15 and Rph16 shared the same race specificity (Table S4). Furthermore, phenotypic screening of an EMS HS-680 mutant population using an avirulent P. hordei isolate (I-80) and subsequent Sanger sequencing identified a single non-synonymous G to A knockout rph15 mutation in two sib lines (Figure 2; Figure 4). Due to chance the possible involvement of second neighbouring gene also conferring Rph16-mediated resistance cannot be ruled out, however further comparative genomic analysis based on sequencing of knockout and/or the HS-680 wild type is required to test the twogene hypothesis.
The level of SNP marker diversity near the Rph15 allele is similar to that observed with barley genes that originated before the domestication of barley. Martin et al. (2020a) reported on the development and characterization of numerous Rph introgression lines in cultivar Bowman (BW lines). With 20 different haplotype patterns in the 50 BW lines likely to have the Rph15 resistance allele, the marker diversity near the Rph15 allele is greater than that observed for mutant genes that involved during or after domestication of barley such as non-brittle rachis (btr1 and btr2) (Casas et al., 2018) and six-rowed spike type (vrs1) (Pourkheirandish et al., 2015).
Three BW lines with the Rph15 resistance allele have the same T2H-4H reciprocal translocation with the breakpoint in 2HS distal to the Rph15 locus. The three BW lines (BW723 from PI 405179, BW724 from PI 405341 and BW725 from PI 391004) have identical marker haplotypes in the interstitial regions between the breakpoint and the centromere. This T2H-4H translocation is different from the T2H-4H translocation in H. v. subsp.
spontaneum described by Konishi and Linde-Laursen (1988) because the wild barley accessions originated from Israel instead of Russia and the breakpoint is in a sub-terminal position of 2HS, at about 50.5 Mb (53.3 cM), instead of near the centromere. Because survival of gametes with crossovers in the interstitial regions is extremely rare and there is no evidence that the critical allele for Rph15 originated more than once, Rph15 must have been present in H. v. subsp. spontaneum before the translocation event occurred. These observations suggest that the Rph15 resistance allele evolved prior to the domestication of barley. The emergence of the rph15_1 haplotype defined by the premature STOP codon variant may also explain the loss of Rph15-mediated resistance in the cultivated barley gene pool.
The wild barley Rph15 donor accession PI 355447 was originally collected in Israel and confers resistance to a wide range of P. hordei isolates (Chicaiza, 1996;Weerasena et al., 2004). We examined exome capture data developed from a previously characterized panel of 267 accessions based on the 2017 Morex assembly comprised of both wild and cultivated barley genotypes (Mascher et al., 2017;Russell et al., 2016). Due to the featured premature STOP codon in Morex (rph15_1), only a truncated rph15 sequence was available for comparison. Extensive functional diversity was identified amongst the 267 barleys; however, no wild accession carried the Rph15 allele sequence. Five nearest neighbours with 80-90% nucleotide similarity were identified in the Western Fertile Crescent region bordering Israel and Jordan ( Figure S7); however, further detailed SNP haplotyping is required to determine the precise origin and distribution of the Rph15 resistance.
A single virulent P. hordei isolate was collected in close geographic proximity to the predicted origin of Rph15. We superimposed the location of the collection site of isolate 90-3 and determined it colocated with the putative origin of Rph15 near Tel Aviv ( Figure S7). This highlights the ability of surrounding P. hordei pathogen populations to acquire virulence and overcome a resistance gene allele present in wild barley populations in Israel. Furthermore, the presence of the alternate host (star-of-Bethlehem -Ornithogalum umbellatum) in this region also may have been a contributing factor for the emergence of virulence for Rph15. Despite its wide effectiveness, to date Rph15 has only been deployed in one barley cultivar (ND Genesis, PI 677345) in the Upper Midwest region of the USA, and its durability in agriculture remains unknown. Although NLR resistance genes have often been overcome by matching pathogen virulence, there is evidence that they can be an important component of durable resistance when deployed in combination with resistance genes of different structure and function (Palloix et al., 2009). Given that the number of leaf rust resistance genes in barley is finite, it is crucial to use these genes with appropriate stewardship to ensure they remain effective for as long as possible. Experience has shown that the most effective strategy to protect resistance genes and achieve durable disease resistance in crop species such as wheat (Park, 2007) and barley (Park et al., 2015) is to deploy multiple genes in combination. Depending on public acceptance, Rph15 could also be pyramided in resistance gene cassettes in combination with other cloned resistance gene alleles such as Rphq2 (Wang et al., 2019) and Rph1 (Dracatos et al., 2019) using a cisgenic approach. Effective gene deployment is reliant on selection of suitable gene combinations and the presence of diagnostic molecular tools to reduce the possibility of hitchhiking effects of deleterious traits. From a breeding perspective, we propose a strategy of combining Rph15 with Rph20, a gene conferring partial adult plant resistance (Rphq4 in Qi et al., 1998). These genes are not located near any loci controlling undesirable traits, as evidenced by the deployment of Rph15 in ND Genesis and the widespread use of Rph20 for over 70 years during which time it has remained durable.
Although not yet cloned, the mode of action of Rph20 likely differs from Rph15 based on previously reported mechanisms of adult plant resistances derived from bread wheat (Krattinger et al., 2009;Moore et al., 2015). Therefore, to overcome this gene combination, the pathogen would need at least two mutations (either sequential or simultaneous) in contrasting response mechanisms.
In summary, we have cloned the leaf rust resistance gene Rph15 originating from wild barley accession PI 355447 collected in Israel and determined it encodes a BED domain-containing immune receptor. Sequence comparison between wild type and mutant in the donor accession for Rph16 (HS-680) suggests that Rph15 and Rph16 are neither allelic nor distinct closely linked genes as previously thought but in fact the same gene; however, the involvement of second neighbouring gene cannot be ruled out. A predictive KASP marker has also been developed and validated for efficient marker assisted selection, and an implementation strategy has been devised to ensure optimal gene stewardship. Figure 5 Phylogenetic analysis was performed by multiple alignment of Rph15 from BW719 with full-length cloned NLRs from the Triticeae and was visualized using a Neighbour-joining tree analysis. An alignment was performed both with and without the BED domain and no noticeable differences in node length or clustering was identified. Rph15 (denoted with star) clustered with other BED domain-containing NLRs from wheat and rice and was more distantly related to Rph1

Plant materials and pathogen isolates
We used the near-isogenic line for Rph15, [BW719; PI 355447/ 7*Bowman - Martin et al., 2020a], as the resistance donor wild type for the development of mutants, population development and DNA template for complementation experiments. An F 2 -F 3 mapping population was developed by crossing BW719 with the leaf rust susceptible cv. Gus to re-confirm linkage with the chromosome arm location and the inheritance of the Rph15 resistance. Complementation analysis in barley was performed using barley cultivar Golden Promise. The donor accession for Rph16 (HS-680) was used for EMS mutation, detailed rust testing and for DNA extraction and Sanger sequencing. A detailed list of the rust pathogen isolates used in this study and their virulence phenotypes can be found in Table S2 and is also detailed further in Martin et al. (2020a).

Candidate gene identification in Morex reference and amplification from Bowman + Rph15
Based on the hypersensitive infection type of the Rph15 resistance, we used the physical interval of 44-57Mb reported in Martin et al. (2020a) for the introgression of the Rph15 resistance from PI 355447 into Bowman to search for predicted annotated NLR candidate resistance genes in the Morex reference assembly using the NLR annotator pipeline reported by Steuernagel et al. (2020). The sequence from 44 800 000 to 57 300 000 bp on chromosome 2H of the Morex reference genome (Mascher et al., 2017) was extracted by samtools version 1.9.0. The sequence was cut into fragments of 20 kb length with 5 kb overlap, and the NLR prediction was performed based as previously described (https://github.com/steuernb/NLR-Annotator Steuernagel et al., 2020).
We extracted flanking sequences from the Morex reference around the candidate gene locus and design a single primer pair (RGA4 F: and RGA4 R) based on the Morex rph15 candidate gene sequence (HORVU2Hr1G019120.5 GenBank accession AY641411.1) to amplify a 9442 base pair genomic fragment including approximately: 1.3 kb of sequence upstream from the translation Start codon and 2.5 kb of sequence downstream of the STOP codon. Polymerase chain reaction (PCR) was performed using Phusion polymerase as per the manufacturer's instructions (New England Biosciences). PCR was performed using 50 µL reactions including: 1 9 HF buffer, 100 ng of genomic DNA, 1 µM of primers, 3% of dimethyl sulfoxide (DMSO), 1 Unit of Phusion polymerase (New England Biosciences). PCR was performed at 1 cycle of 98°C for 3 min followed by 35 cycles of denaturing at 98°C for 20 s, annealing at 60°C for 20 s, extension at 72°C for 3 min and 30 s and then incubating at 72°C for 10 min. The PCR products were separated on a 1% agarose gel, and a single band of the expected size was excised and purified. The gel-purified PCR product was then cloned into the sub-cloning vector TOPO XL as per manufacturer's instructions (Invitrogen). We amplified and cloned genomic fragments from resistant donors for Rph15 (BW719) and Rph16 (HS-680) and the universally susceptible (Gus) barley accessions. The plasmid DNA of five positive clones from each amplicon was sent for Sanger sequencing using internal primers, and the sequences were compared with the template from Morex. All primers used in this study are detailed in Table S4.

cDNA analysis
Total RNA was extracted using the TRIzol method (Invitrogen) as per manufacturer's instructions. cDNA synthesis, 5 0 and 3 0 RACE (rapid amplification of cDNA ends) were performed using the manufacturer's instructions (Clontech).

Generation of loss of function mutants for Rph15
We performed sodium azide mutagenesis on the wild-type Rph15 donor barley line BW719 (Martin et al., 2020a,) using the procedure described by Chandler and Harding (2013) with some modifications. Approximately 1500 seeds were immersed in water at 4°C overnight. The imbibed seeds were transferred to a 2-L measuring cylinder filled with water and aerated with pressurized air for 8 h, with one change of water after 4 h. The water was drained, and the seeds were incubated in a shaker for 2 h in freshly prepared 1 mM sodium azide dissolved in 0.1 M sodium citrate buffer (pH 3.0). Next, the seed was washed extensively in running water for at least 2 h and placed in a fume hood to dry overnight. Seeds were sown in the field, and single spikes from each plant were harvested separately from the remaining spikes, which were harvested in bulk.
The Bowman + Rph15 mutant M 2 spikes and M 2 -derived M 3 families were phenotypically assessed at the seedling stage as described by Dracatos et al. (2019). In all cases, at least two susceptible plants (putative rph15 knockouts) were transplanted for each candidate segregating M 2 family for subsequent progeny testing. Sequence confirmation for each mutant was performed through PCR amplification of M 3 derived susceptible progeny for each mutant family. Mutagenesis was also previously performed on HS-680 using the same approach, and susceptible knockouts were identified at the Julius K€ uhn-Institute (JKI), Quedlinburg, Germany using the P. hordei isolate I-80 (avirulent with respect to both Rph15 and Rph16) (Table S2).

Sequence confirmation of rph15 mutants
Sanger sequence confirmation of all mutants was performed at the M 3 stage using a three-step process. Firstly, only DNA from progeny-tested homozygous susceptible families was extracted using the CTAB method (Doyle and Doyle, 1987) for PCR amplification of the 9442 bp genomic fragment for the candidate Rph15 gene as described above. Secondly, the PCR products were cloned into the TOPO XL vector as described above and three positive clones for each amplicon were sent for Sanger sequencing on a fee-for-service basis to the Australian Genome Research Facility (AGRF) for comparison with the wild-type Rph15 candidate gene. Finally, only when all three clones carried the same non-synonymous sodium azide-induced (either G to A or C to T) mutation were they deemed confirmed mutants. including 4 µL of KASP master mix (LGC), KASP Assay mix (0.12 pmol two allele-specific primers and 0.3 pmol common reverse primer) and 25 to 50 ng of genomic DNA. The KASP marker reaction conditions were as follows: 1 cycle at 94°C for 15 min, 10 cycles of at 94°C for 20 s and 65°C for 1 min and then 32 cycles of at 94°C for 20 s and 57°C for 1 min, cooled down to 25°C for 5 min. Marker validation was performed using a panel of 80 Australian barley accessions all postulated to lack Rph15 and 60 leaf rust resistant Bowman near-isogenic lines of which many were postulated to carry the Rph15 resistance. All accessions are listed in Table S1.

Complementation using Agrobacterium-mediated transformation (ABMT) in barley
The 9442 bp fragment containing its native promoter and terminator was amplified by PCR with NotI at both sites, enzyme cutting and ligation into pWBVec8 . Subsequently, the vector containing the intact sequence-verified Rph15 candidate gene was transformed into Agrobacterium tumefaciens strain AgL0 using the freeze/thaw method (Chen et al., 1994). Barley transformation into Golden Promise was performed as described by Bartlett et al. (2008).

Quantitative expression analysis of Rph15
The relative expression of the Rph15 resistance gene was assessed in two different mutant lines and the Golden Promise + Rph15 transgenic lines including the same four T 1 transgenic lines that were also rust tested against the Rph15virulent P. hordei isolate 90-3 from Israel. Barley genotypes Bowman, BW719, mutant lines M4022 and M1727 were grown in the glasshouse for 12 days at 20°C. Ten-day-old seedlings were inoculated with P. hordei pathotype 5457 P+, and both uninfected and infected leaf samples were harvested at two timepoints (0 and 24 h after inoculation, respectively). For the transgenics, DNA was extracted from six half seeds for each of the Golden Promise + Rph15 T1 transgenic lines (B49-1, B49-2, B49-3, B49-8a, B49-8c, B49-11a and B49-11c) and subsequently genotyped with the Rph15 KASP marker. Marker positive transgenic resistant sib lines, Bowman + Rph15, Golden Promise and Bowman were grown in the growth chamber, and tissue was sampled at the 2nd-3rd leaf stage for RNA extraction. RNA was extracted using Plant RNA Kit (Promega) on Maxwellâ RSC 48 Instrument (Promega) as per manufacturer's instructions. cDNA was synthesized from 1 µg of total RNA using SuperScript TM III Reverse Transcriptase (Invitrogen) as per manufacturers instructions. The total volume was brought to 100 µL with water after the reaction was finished. qPCR was performed by adding 5 µL of iTaq Universal SYBRâ Green Supermix (BIO-RAD), 1 µL of forward primer (10 µM), 1 µL of reverse primer (10 µM) and 3 µL of diluted cDNA, using the standard protocol. PCR reactions were performed at 1 cycle of 95°C for 3 min, 40 cycles of denaturing at 95°C for 10 s, annealing (extension) at 60°C for 30 s with plate reading at each cycle and melt curve from 65°C to 95°C with increment of 0.5°C for 5 s plus plate reading. Reactions for each cDNA sample contain two sets of primers: 1/reference gene primers Hv-ACT-F and Hv-ACT-R (Hordeum vulgare actin gene) and 2/Rph15 specific primers Rph15_1679-F and Rph15_1793-R (Table S5). The relative expression level is expressed as: RE = 2 DCt where DCt = Ct(actin) -Ct(Rph15). The relative expression levels of all samples were normalized as the relative expression of Bowman + Rph15 was set to 100.
Nearest neighbour analysis using the Hordeum exome capture The genomic sequence of Rph15 was aligned to the genome sequence assembly of barley cultivar Morex (Morex V1, Mascher et al., 2017) using the BWA-MEM algorithm (Li 2013) set to default parameters. Sequence variants between Morex and Rph15 were detected with BCFtools mpileup (Li 2011). The results were imported into R (R Core Team, 2016) and intersected with a SNP matrix derived from exome capture data of 267 wild and domesticated barley accessions (Russell et al., 2016). Nearest neighbours of Rph15 in a set of 91 wild barleys were found by a simple SNP matching distance. A map showing the collection sites of the 55 neighbours was plotted using the R package mapdata (https://cran.r-project.org/web/packages/mapdata/index.html) ( Figure 5). Collection sites were taken from Russell et al. (2016). The collection site of the Rph15-virulent isolate (90-3) in Sha'ar HaGai, Israel was superimposed on the same map to illustrate the colocation between the putative origin of the Rph15 resistance and matching virulence within P. hordei populations, respectively.

Supporting information
Additional supporting information may be found online in the Supporting Information section at the end of the article.

Figure S1
Confirmation of the BW719 genetic stock as the wild type source of Rph15 used in this study. Figure S2 Leaf rust infection phenotypes 11 days after inoculation of (L to R) wild type BW719 (Bowman+Rph15), Bowman, and the six sodium azide-induced non-synonymous rph15 knockout mutants: M1371-3, M1695, M1727, M4022, M4321 and M4651. Figure S3 Summary of phenotypic and molecular characterisation of the T 0 generation Golden Promise+Rph15 and control lines. Figure S4 Phenotypic assessment of seedlings from (L to R) Bowman+Rph15 (BW719), Golden Promise and individual sib plants from T 1 generation Golden Promise+Rph15 transgenic lines B49-2 and B49-11a inoculated with North American Puccinia hordei races that are virulent (90-3, A) and avirulent (92-7, B) with respect to Rph15. For a full description of infection types refer to Park et al. (2015). Figure S5 Relative expression analysis using qRT-PCR of the Rph15 resistance gene in the different T 1 generation Golden Promise+Rph15 transgenic lines generated in this study relative to the resistant wild type Bowman+Rph15 (BW719) and susceptible genotypes Golden Promise (GP) and Bowman (Bo).

Figure S6
Comparative amino acid sequence alignment of the predicted BED domains of the NLR proteins from Rph15 from BW719 identified in this study with Yr5 and Yr7 from bread wheat (Marchal et al. 2018) and Xa1 from rice (GenBank accession BAA25068.1). Figure S7 The geographic distribution of 91 wild barleys (Hordeum vulgare ssp. spontaneum) from the exome capture based on collection sites from the Fertile Crescent as given by Russell et al. (2016). Table S1 Genotypic results of 61 near-isogenic lines carrying leaf rust resistance in cultivar Bowman published in Martin et al. (2020) with the gene-based Rph15 KASP SNP marker developed in this study. Table S2 Summary table of leaf rust pathogen isolates used in this study and their virulence/avirulence spectra. Table S3 Summary of phenotypic data of four T 1 families and barley leaf rust differential genotypes infected with two Puccinia hordei isolates with contrasting virulence for Rph15. Table S4 Phenotypic response testing at the seedling stage of barley leaf rust differential host genotypes at the University of Minnesota, US using six pathogenically diverse Puccinia hordei isolates. The locations and details of these isolates are specified in Martin et al. (2020).