High-resolution mapping of Rym14Hb, a wild relative resistance gene to barley yellow mosaic disease

Key message We mapped the Rym14Hb resistance locus to barley yellow mosaic disease in a 2Mbp interval. The co-segregating markers will be instrumental for marker-assisted selection in barley breeding. Abstract Barley yellow mosaic disease is caused by Barley yellow mosaic virus and Barley mild mosaic virus and leads to severe yield losses in barley (Hordeum vulgare) in Central Europe and East-Asia. Several resistance loci are used in barley breeding. However, cases of resistance-breaking viral strains are known, raising concerns about the durability of those genes. Rym14Hb is a dominant major resistance gene on chromosome 6HS, originating from barley’s secondary genepool wild relative Hordeum bulbosum. As such, the resistance mechanism may represent a case of non-host resistance, which could enhance its durability. A susceptible barley variety and a resistant H. bulbosum introgression line were crossed to produce a large F2 mapping population (n = 7500), to compensate for a ten-fold reduction in recombination rate compared to intraspecific barley crosses. After high-throughput genotyping, the Rym14Hb locus was assigned to a 2Mbp telomeric interval on chromosome 6HS. The co-segregating markers developed in this study can be used for marker-assisted introgression of this locus into barley elite germplasm with a minimum of linkage drag. Electronic supplementary material The online version of this article (10.1007/s00122-020-03733-7) contains supplementary material, which is available to authorized users.


Introduction
Viruses are an increasing threat to crops worldwide. The soil-borne barley yellow mosaic disease, caused by a complex of two Bymoviruses (Barley yellow mosaic virus (BaYMV) and Barley mild mosaic virus (BaMMV)) is one of the most important diseases of winter barley. Widespread in central Europe and East-Asia, it causes severe yield losses up to even total crop failure (Plumb et al. 1986;Jianping 2005;Kühne 2009). As chemical control of those viruses, transmitted by the plasmodiophorid Polymyxa graminis (Kanyuka et al. 2003), is not possible, only the use of resistant varieties can preserve yield in infected fields.
To date, 20 barley resistance genes have been identified, almost exclusively conferring recessive resistance (Jiang et al. 2020). Two of these loci have been cloned: the EUKARYOTIC TRANSLATION INITIATION FACTOR Communicated by Kevin Smith.

Electronic supplementary material
The online version of this article (https ://doi.org/10.1007/s0012 2-020-03733 -7) contains supplementary material, which is available to authorized users. 4E gene (eIF4E), (Stein et al. 2005) of which several allelic forms providing resistance are described, including rym4 and rym5, (Hofinger et al. 2011;Perovic et al. 2014;Yang et al. 2017;Shi et al. 2019), and the PROTEIN DISULFIDE ISOMERASE LIKE 5-1 (PDI5-1) gene which is also represented by a handful of alleles providing resistance, including rym1 and rym11 (Yang et al. 2017). The rym4 allele provides a recessive resistance to BaMMV and to the common BaYMV pathotype BaYMV-1, but not to pathotype BaYMV-2, which emerged in Europe at the end of the 1980s (Adams et al. 1987;Huth 1989;Adams 1991;Graner and Bauer 1993;Steyer et al. 1995). The spectrum of rym5 covers also BaYMV-2, however, resistance-breaking isolates of BaMMV and BaYMV have emerged (Kanyuka et al. 2004;Habekuß et al. 2008;Li et al. 2016). Facing the prospect of boom-and-bust cycles for known resistance genes (Brown and Tellier 2011), it is critical to continue searching for alternative resistance loci to underpin resistance breeding and to allow pyramiding of disease resistance loci. In particular, sources of non-host resistance, e.g. resistance exhibited from a plant species against all isolates of a pathogen which is not coevolutionary adapted, are particularly promising as they are thought to cover a larger resistance spectrum and to be more durable (Ayliffe and Sørensen 2019). Bulbous barley (Hordeum bulbosum L.), a perennial wild relative and representative of the secondary gene pool of cultivated barley (Hordeum vulgare L.), has been described as source of resistance to numerous barley pathogens, including barley leaf rust (Johnston et al. 2013;Yu et al. 2018) and barley powdery mildew (Xu and Kasha 1992;Pickering et al. 1995;Shtaya et al. 2007). So far, all H. bulbosum accessions investigated exhibited resistance to BaMMV and BaYMV (Ruge et al. 2003), suggesting that the species is probably a non-host to those viruses. Two major dominant resistance genes from H. bulbosum to both BaMMV and BaYMV have been described: Rym14 Hb (Ruge et al. 2003) and Rym16 Hb (Ruge-Wehling et al. 2006). Rym14 Hb was introgressed to barley by translocation of a H. bulbosum segment to barley chromosome 6HS (Ruge et al. 2003). In the past, a lack of suitable markers, alongside severely reduced recombination in the target region between the barley and H. bulbosum fragments, rendered precise mapping of Rym14 Hb elusive. Thanks to the development of genetic and genomic resources for H. bulbosum (Wendler et al. 2014(Wendler et al. , 2015, it is now possible to fine-map loci from this species in a H. vulgare background.
We aimed to map Rym14 Hb at high resolution, and to provide markers for its introgression into elite barley, ideally without linkage drag, using large populations and high-throughput genotyping to overcome the lack of recombination.

Plant material
A first round of low-resolution genetic mapping was performed using four F 6 families derived from F 5 plants heterozygous at the Rym14 Hb locus from the BAZ-4006 family of the population described in Ruge et al. (2003) and obtained from the cross between the susceptible winter barley cv 'Borwina' and the resistant H. bulbosum accession 'A42'.
To achieve a population size suitable for fine mapping, an additional eight F 2 families were generated by crossing an Rym14 Hb /Rym14 Hb F 6 plant (derived from F 5 4006/337) to either (i) var. 'KWS Orbit' or (ii) var. 'KWS Higgins', both missing the Rym14 Hb resistance locus (-/-). In the purpose of instant pyramiding of disease resistance loci, both cultivars carry rym4-based resistance (rym4/rym4) to BaMMV and BaYMV.

DNA extraction
Genomic DNA of plants from the low-resolution mapping population was isolated as described by Stein et al. (2001). Genomic DNA of plants from the fine mapping population was extracted according to the guanidine isothiocyanatebased protocol described by Milner et al. (2019).

Genotyping-by-sequencing and data analysis
Genotyping-by-sequencing (GBS) libraries for the low-resolution mapping were prepared from genomic DNA digested with PstI and MspI (New England Biolabs) as described by Wendler et al. (2015). Between 93 and 153 barcoded samples were pooled in an equimolar manner per lane and sequenced on the Illumina HiSeq 2500 for 107 cycles, single-end reads, using a custom sequencing primer.
The GBS reads were processed, aligned, and used to generate variant calls as described by Milner et al. (2019). Alignment was performed against the TRITEX genome assembly of barley cultivar 'Morex' (Monat et al. 2019). Individual variant calls were accepted wherever the read depth exceeded four. Variant sites were retained if they presented a minimum mapping quality score (based on read depth ratios calculated from the total read depth and depth of the alternative allele) of 20, a maximum fraction of 40% of missing data, a fraction of heterozygous calls between 30 and 70%, and between 10 to 40% of each homozygous call. Individuals with more than 40% missing data were excluded.

Marker development
Exome capture data of the introgression line '4006/163', described in Wendler et al. (2014) (accession number ERP004445), were mapped to the TRITEX genome assembly of barley cultivar 'Morex' (Monat et al. 2019) together with the exome capture data of the H. bulbosum genotype 'A42' and of eight barley varieties: 'Bonus', 'Borwina', 'Bowman', 'Foma', 'Gull', 'Morex', 'Steptoe', and 'Vogelsanger Gold', described in Mascher et al. (2013b) (accession number PRJEB1810). Read mapping and variant calling were performed as described by Milner et al. (2019). The variant matrix was filtered for the following criteria: heterozygous and homozygous calls had to be covered by a minimum depth of three and five reads, respectively, and have a minimum quality score of 20. Single nucleotide polymorphism (SNP) sites were retained if they had less than 20% missing data and less than 20% heterozygous calls. SNPs that were carrying the reference call in all eight barleys and the alternate call in 'A42' and '4006/163' was selected as candidates to design Kompetitive Allele Specific PCR (KASP) markers, either using KASP-by-design (LGC Genomics, Berlin, Germany) or 3CR Bioscience (Essex, UK) free assay design service. Those markers are latter designated as KASP and PCR Allele Competitive Extension (PACE) markers, respectively. Since no suitable SNPs were identified in the first 500 kbp of chromosome 6HS on the 'Morex' reference genome, the exome capture data were additionally mapped to the genome assembly of cultivar 'Barke' (Jayakodi et al. in press). The SNP at coordinate 241,723 bp on chromosome 6H of the 'Barke' genome assembly was retrieved and used to design the telomeric marker Rym14_Bar241723. Furthermore, in order to control the genetic state at the segregating rym4 resistance locus, the diagnostic SNP for the resistance conferring allele (Stein et al. 2005) was also used to design a KASP marker. Further information on KASP and PACE markers is provided in supplementary tables 1 and 2, respectively.

Genotyping
Genotyping assays with KASP markers were carried out in a final volume of 5 μl consisting of 0.7 μl genomic DNA (50-100 ng/µL), 2.5 μl of KASP V4.0 2X Master Mix High Rox (LGC Genomics, Berlin), 0.07 μl KASP assay mix (KASP-by-design, LGC Genomics, Berlin) containing the primers, and 2.5 μl of sterile water. PCR amplifications were performed using the Hydrocycler 16 (LGC Genomics, Berlin) with cycling conditions as follows: 94 °C for 15 min, followed by a touchdown profile of 10 cycles at 94 °C for 20 s and 61 °C for 1 min with a 0.6 °C reduction per cycle, followed by 26 cycles at 94 °C for 20 s and 55 °C for 1 min. Genotyping assays with PACE markers were carried out in a final volume of 5 μl consisting of 0.7 μl genomic DNA (50-100 ng/µL), 2.5 μl of PACE Master Mix High Rox (3cr Bioscience, Essex, United Kingdom), 0.07 μl primer mix containing the primers (12 µM of each allele specific primers and 30 µM of the common reverse primer), and 2.5 μl of sterile water. PCR amplifications were performed using the Hydrocycler 16 (LGC Genomics, Berlin) with cycling conditions as follows: 94 °C for 15 min, followed by a touchdown profile of 10 cycles at 94 °C for 20 s and 65 °C for 1 min with a 0.8 °C reduction per cycle, followed by 30 cycles at 94 °C for 20 s and 57 °C for 1 min.
For both marker types, the genotyping results were read out using the ABI 7900HT (Applied Biosystems) using an allelic discrimination file. Readings were made before and after PCR, and the data were analyzed using SDS 2.4 Software (Applied Biosystems).

Phenotyping
Resistance to BaMMV was tested under greenhouse conditions as described by Habekuß et al. (2008). After sowing, the plants were grown in a greenhouse (16 h day/8 h night, 12 °C). The susceptible barley variety 'Maris Otter' was systematically included to monitor success of infection. At the 3-leaf stage (around 2 weeks after sowing), the plants were mechanically inoculated twice at an interval of 5-7 days with the isolate BaMMV-ASL1 (Timpe and Kühne 1994) using the leaf-sap of BaMMV-infected leaves of susceptible cv. 'Maris Otter', mixed in K 2 HPO 4 buffer (1:10; 0.1 M; pH 9.1) containing silicon carbide (caborundum, mesh 400, 0.5 g/25 ml sap). Five weeks after the first inoculation, the number of infected plants with mosaic symptoms were scored, and double antibody sandwich enzyme-linked immunosorbent assay (DAS-ELISA) with polyclonal BaMMVspecific antibodies produced by the Serum Bank of the Institute of Epidemiology and Pathogen Diagnostics (JKI Quedlinburg, Germany) was carried out in parallel according to published protocols (Clark and Adams 1977). Virus particles were estimated via extinction at 405 nm using a Dynatech MR 5000 microtiter-plate reader. Plants with an extinction E 405 > 0.1 were qualitatively scored as susceptible.

Low-resolution mapping
A population of 427 F 6 from the cross 'Borwina' x 'A42' was genotyped by GBS and phenotyped for resistance to BaMMV. Data for 389 plants and 77 SNPs passed the quality filters (supplementary table 3). On chromosome 6H, 73 plants were homozygous for the 'Borwina' allele, 92 were homozygous for the 'A42' allele, 220 were heterozygous, 1 3 and four recombined. The infection rate was low with only 10% of plants infected, compared to an expected 25% when resistance is controlled by a single dominant gene. Two susceptible barley cultivars were tested: 92.5% and 72.2% of Maris Otter and Igri plants, respectively, were infected. It is known that the penetrance of infection in such experiments is never complete, and that the genetic background of the plant plays a role in this phenomenon, Maris Otter being the most susceptible cultivar tested (Adams et al. 1986(Adams et al. , 1993. Despite that, among the 39 plants phenotyped as susceptible to BaMMV, 38 were homozygous for the 'Borwina' allele and one recombined on chromosome 6H, indicating a strong association of phenotype and genotype. To further confirm this association, 26 lines were phenotyped on progenies of 12-20 plants (Fig. 1a, Table 1, supplementary table 4). These included (i) 17 lines with the susceptible genotype on chromosome 6H but scored as resistant, (ii) five heterozygous lines, and (iii) the four recombinant lines. Progenies of lines presenting the susceptible genotype displayed infection rates between 50 and 95%, while those of heterozygous lines displayed rates between 5 and 20%. The progeny of line 5204-58 displayed an intermediate level of susceptibility, with 35% of infected plants. However, this line had been phenotyped as susceptible in F 2 generation, and was therefore classified as susceptible.
These results support the low penetrance of the infection in this experiment, with only half of the expected susceptible plants successfully infected, as well as the association of the chromosome 6H locus with resistance to BaMMV. Moreover, the phenotypes of the four recombinant progenies defined Rym14 Hb interval between the telomere of chromosome 6HS and the marker position at base pair 4,553,134.

Fine mapping
The population of 7500 F 2 was genotyped at the Rym14 Hb locus with four KASP markers (Rym14_Bar241723, Rym14_2370223, Rym14_3087282, and Rym14_5003183, supplementary table 1). The recessive resistance gene rym4 to BaMMV and BaYMV-1, located on chromosome 3HL, also segregated in the population. Therefore, in order to properly assign the resistance to the control of Rym14 Hb , the segregation of rym4 was monitored with the rym4_SNP KASP marker (supplementary table1). We identified 28 recombination events, corresponding to a genetic distance of ~ 0.2 cM, between the markers Rym14_Bar241723 and Rym14_5003183. These results confirmed the strongly reduced recombination rate between the H. bulbosum and the H. vulgare fragments on chromosome 6HS. In cultivated barley, the syntenic 5 Mbp Rym14 Hb interval on chromosome 6HS corresponds to a genetic distance of 4 cM (Mascher et al. 2013a), implying a 20-fold reduction in recombination frequency between the H. bulbosum and the H. vulgare fragment.
All recombinants were genotyped with seven PACE markers (Fig. 1b, supplementary tables 2 and 4). Among the recombinants, ten plants were homozygous for the rym4 allele, nine were heterozygous, and the remaining nine were homozygous wildtype at the rym4 locus (supplementary table 5). As plants homozygous for the rym4 allele would be resistant to BaMMV, irrespective to their genotype at Rym14 Hb , only F 3 families derived from the 18 Rym14-recombinants heterozygous or homozygous for the susceptible allele at rym4 were phenotyped using 30 and 20 F 3 siblings, respectively ( Table 2). The infection rate during  Table 2). The cosegregation of the phenotype with the genotype at the three markers in the F 3 progenies largely confirmed the analysis (supplementary table 6). The only discrepancy was the significant χ 2 result for line Rym14_49/7_149, that was explained by a distortion of segregation of the susceptible allele in the small number of plants tested: a single plant was homozygous for H. vulgare allele in Rym14 Hb interval.
Based on this analysis, the Rym14 Hb target region was reduced to a 2 Mbp interval on the 'Morex' reference genome, between the telomere of chromosome 6HS and Rym14_2066975 (Fig. 1b, Table 2).

Candidate genes
In the absence of a genomic sequence for a Rym14 Hb plant, we cannot precisely define the genes present in the Rym14 Hb interval. However, as synteny between the two Hordeum species is high (Wendler et al. 2017), it is still relevant to assess the genes annotated in the orthologous interval of the H. vulgare reference genome as a proxy for suggesting Rym14 Hb candidate genes. In the respective interval of the 'Morex' V2 reference sequence, 30 high-confidence (HC) ( Table 3) and 17 low-confidence genes (Monat et al. 2019) are annotated. In addition, all HC gene models were checked for homology with other genes by a BLASTx (v2.9.0, default parameters) homology searches against the non-redundant protein sequence database (Camacho et al. 2009) and for presence of conserved domains in NCBI conserved domains (Lu et al. 2019). Among the HC genes, HORVU.MOREX. r2.6HG0448010 is annotated as a TIR-NBS-LRR gene in Monat et al. (2019), however, our analysis reveals that it does not contain any of the major domains of nucleotidebinding and leucine-rich repeat domain (NLR) genes (TIR or coiled-coil, NB-ARC and LRR), and monocotyledons so far have not been shown to contain TIR-NLR genes (Jacob et al. 2013). This gene is therefore interpreted as a pseudogene. HORVU.MOREX.r2.6HG0448100, annotated as a dirigent protein, is a jacalin-related lectin, while HORVU. MOREX.r2.6HG0448250, annotated as part of the protein kinase protein family, displays the highest homology with a wall-associated receptor kinase, and HORVU.MOREX. r2.6HG0448290 codes for a papain-like cysteine protease (PLCP). Interestingly, the interval also contains no less than 14 HC genes annotated as thionins, sharing with each other at least 88% of their coding sequence. In addition to these annotated genes in the 'Morex' genome, additional candidate genes could be unique to the resistant genotypes.

Discussion
Resistance genes deployed in breeding and in the field are often overcome by new pathogen variants after only a few years (Brown and Tellier 2011). Pyramiding several resistance genes has proven to increase the resistance durability, however, this strategy requires the availability of several independent resistance loci Riedel et al. 2011;Kim et al. 2011). In light of these facts, non-adapted resistance genes from wild crop relatives are precious, since they are assumed to confer more durable resistance than genes originating from within the diversity of the cultivated species, owing to co-evolution between the cultivated host and pathogen genotypes (Fonseca and Mysore 2019). Until recently, the fine mapping of genes from crop wild relative species was impractical, owing to strong suppression of recombination with the cultivated species (Ruge et al. 2003;Kakeda et al. 2008;Wijnker and de Jong 2008;Prohens et al. 2017). The results of this study demonstrate that high-throughput genotyping coupled with large mapping populations can overcome this limitation, by constraining the interval of the Rym14 Hb viral resistance gene to the telomeric 2 Mbp of chromosome 6HS, and providing markers suitable for marker-assisted-selection. Rym14 Hb was described as providing resistance against both BaMMV and BaYMV (Ruge et al. 2003). However, phenotyping for resistance to BaYMV is only feasible in infested fields, and is not well adapted to gene mapping. Therefore, in this study, we only mapped BaMMV resistance. Among cloned by movirus resistance genes, the resistance alleles rym4 and rym5 of the eIF4E gene, and the alleles rym1 and rym11 of PDI5-1 gene provide resistance against isolates of both virus species (Kanyuka et al. 2004;Stein et al. 2005;Ordon et al. 2005;Habekuss et al. 2008). Thus, the two viruses are genetically similar enough for a gene to provide resistance against isolates of both viruses. But the possibility of the described Rym14 Hb BaYMV resistance being provided by a closely associated, but distinct, locus cannot be excluded at this point and will require further testing.
While genes coding for NLR are the usual suspects for dominant resistance to pathogens, including viruses (de Ronde et al. 2014;Boualem et al. 2016), only a pseudogene presenting similarities with this gene family is annotated in the Rym14 Hb interval on the barley reference genome. However, it is not rare that susceptible genotypes do not possess a functional copy of the resistance gene. NLRs are overrepresented in regions displaying presence/absence variation (Xu et al. 2012;Bush et al. 2013). Therefore, some NLR resistance genes, like RPM1 and RPS5, are only present in the resistant genotype (Grant et al. 1998;Henk et al. 1999). In the case of wheat leaf rust resistance gene Lr21, it was shown that the gene is a chimera of two non-functional alleles that probably evolved via a recombination event (Huang et al. 2009).
Among the other annotated genes at the Rym14 Hb locus, two are very good candidates. Wall-associated protein kinase-like HORVU.MOREX.r2.6HG0448250 is described resistance genes in plant-bacteria and plant-fungus pathosystems (Li et al. , 2020Dmochowska-Boguta et al. 2020). Their role in plant-virus pathosystems is less clear but it has been suggested that a cell wall-associated protein kinase was involved in the repression of plasmodesmal transport of the Tobacco mosaic virus by phosphorylating its movement protein (Citovsky et al. 1993;Waigmann et al. 2000). A second promising candidate is HORVU.MOREX. r2.6HG0448100. It codes for a jacalin-related lectin and is thus part of the family that includes the Arabidopsis thaliana genes RTM1 and JAX1 that provide dominant major resistance against poty-and potexviruses, respectively (Chisholm et al. 2000;Yamaji et al. 2012).
However, other genes in the Rym14 Hb interval, even if less likely candidates, might also play a role in resistance. For example, HORVU.MOREX.r2.6HG0448290 codes for a PLCP. PLCPs are known to play a major role in programmed cell death triggered by NLR genes. Interestingly, CYP1, a tomato PLCP, is targeted by the Tomato yellow leaf curl virus V2 protein, suggesting that V2 could downregulate CYP1 to counteract host defenses (Bar-Ziv et al. 2012). Rcr3, a tomato papain-like cysteine protease gene, is required for the function of the resistance gene Cf-2 to Cladosporium fulvum (Krüger et al. 2002), while NbCathB, from Nicotina benthamiana, is requested for the HR triggered by the non-host pathogens Erwinia amylovora and Pseudomonas syringae (Gilroy et al. 2007). The high level of thionin duplication at this locus also raised our attention. Thionins are part of common anti-bacterial and antifungal peptides (Bohlmann and Broekaert 1994), conferring enhanced resistance to several pathogens. Thionins were also found to exhibit increased expression in resistant compared to susceptible pepper genotypes during infection by the Chili leaf curl virus (Kushwaha et al. 2015), suggesting a possible role in basal defense. Additionally, the cytochrome P450 superfamily has been associated with resistance to the Soybean mosaic virus (Cheng et al. 2010;Yang et al. 2011). Some subtilisin proteases are induced by pathogens and involved in programmed cell death (Figueiredo et al. 2014), and GDSL lipases were found to be either negative or positive regulators of plant defense mechanisms (Hong et al. 2008;Kwon et al. 2009).
The feasibility of further reducing the target interval by recombination through additional fine mapping is low and would require the screening of tens of thousands of additional F 2 plants for the chance of finding one additional recombinant in the smallest target region. Therefore, a candidate gene approach may be a more fruitful strategy for continued progress. Despite the presence of promising candidate genes like HORVU.MOREX.r2.6HG0448250 and HORVU. MOREX.r2.6HG0448100 in the haplotype of the susceptible cultivar 'Morex', the resistance conferring gene may be present only in the haplotype of the resistant H. bulbosum. Therefore, deciphering the resistant haplotype, most likely though a high-quality chromosome-scale genome assembly of the interval in H. bulbosum, is an essential prerequisite to the prioritization of candidate genes for further functional testing.
The markers identified in this study are tightly linked to Rym14 Hb and therefore are of prime importance to barley breeding. These markers will allow the reliable introgression of this resistance into barley elite lines with a minimum of 1 3 linkage drag compared to the previously established markers (Ruge et al. 2003). This is essential for introducing this gene into new cultivars. As the prevalence of resistance-breaking isolates of rym4 and rym5 will increase in the barley growing area in Europe and Asia (Kühne 2009), introgression of Rym14 Hb into new elite varieties together with other resistance loci represents a critical opportunity to improve the durability and spectrum of barley resistance to BaMMV and BaYMV.