The In-Silico Development of DNA Markers for Breeding of Spring Barley Varieties That Are Resistant to Spot Blotch in Russia

: The fungal pathogen Cochliobolus sativus Drechs. Ex Dastur, anamorph Bipolaris sorokiniana (Sacc.) Shoemaker is one of the most common barley pathogens worldwide and causes spot blotch and root rot in barley. Spot blotch is considered to be the major biotic stress hampering the commercial production of barley. During high disease severity, which occurs in the northwestern region of Russia once every three to four years, yield losses for barley may reach 40%. An increase in common root rot severity results in yield losses that can reach 80%. The goal of the current study was to identify signiﬁcant markers that can be employed as diagnostic DNA markers to breed C. sativus pathogen-resistant varieties of barley. In 94 spring barley cultivars and lines, the resistance of seedlings and adult plants to the impact of C. sativus on their leaves and roots was investigated. Five genomic regions associated with resistance to Spot blotch were identiﬁed (on chromosome 1H (50–61.2 cM), 2H (68.7–69.68 cM), 3H (18.72–26.18 cM), 7H (7.52–15.44 cM)). No signiﬁcant loci were determined to be associated with root rot. According to obtained data, 11 signiﬁcant SNPs were converted into KASP markers and 6 markers located on chromosome 3H were determined to possess good accuracy and the potential to be employed in marker-assisted selection.

Spot blotch is considered to be the major biotic stress hampering the commercial production of barley. Spot blotch is distributed across all barley-growing areas. During high disease severity, which occurs in the northwestern region of Russia once every three to four years, yield losses for barley may reach 40% [4]. Severe spot blotch epidemics lasting one to two weeks prior to maturity may reduce barley yields by 10-20%, while epidemics lasting three to four weeks may reduce yields by 20-30%, including a 10-15% reduction in kernel weight [5].

Plant Material and Genotyping Data
The research was based on a study of the Siberian spring barley collection, which was described previously [11], as well as on independent barley accessions described in Table S1 [4]. Additional information on 50K Illumina SNP-array loci was extracted from Cantalapiedra et al., 2015 [30] and the BARLEYMAP resource available online: http://floresta.eead.csic.es/barleymap (accessed on 6 April 2020). The independent barley accession set resistance data were available from Novakazi et al., 2019 [4]. DNA of accessions presented in the independent sample set was extracted using a Qiagen kit (Qiagen GmbH, Germany, DNeasy Plant Mini Kit 250) according to the instructions of the manufacturer.

Pathogen Isolates and Culture Conditions
For spot blotch (at the adult stage) and root rot (seedling stage) evaluation, C. sativus isolate O18.2, which originated from Omsk (Siberia region of Russia), was employed. Artificial infection in the field for evaluating adult plants was developed by using a single conidial isolate Ch3 (northwestern Russia, Leningrad region). The data obtained in this study were compared with data from a previous research, where the isolates Ch3 and Kr2 (South of European part of Russia, Krasnodar region) were used for seedling evaluation [11]. Methods of cultivation and isolate maintenance were described in a previous report [11].

Resistance to Spot Blotch
• Experimental design Plant growing and disease assessment of seedling resistance to spot blotch was described in our previous paper [11]. Plants were inoculated by spraying with conidial suspension (10,000 conidia/mL) 12-14 days after planting (two-or three-leaf stage).
A study of adult plant resistance of 96 barley genotypes was conducted on the experimental plots of VIR (St. Petersburg, Pushkin). The rows were 1-metre-long, with 20 seeds being planted per row. The distance between rows was 30 cm. The susceptible variety Harrington was planted at every 10 varieties as a control. Plants in stage BBCH 30 (BBCH is a scale for determining the developmental stages of cereals) were inoculated by spraying with suspension at a concentration of 20,000-25,000 conidia per mL. To improve the contact of conidia with the leaf surface, 100 µL/L of surfactant Tween 20 was added to the suspension.

Disease evaluation
Infection responses (IRs) of seedlings were measured 10 days after inoculation at the two-to three-leaf stage using the one-to-nine rating scale of Fetch and Steffenson [31]. The lesion size and the degree of associated chlorosis were the basis of this scale, which was described in our previous paper [11]. A susceptible cultivar, Harrington, was selected as a high-IR control.
The infection responses to spot blotch exhibited by adult plants were reported in the phase of milk-wax ripeness (BBCH 73-83) as a percentage of leaf area infected with disease on the upper and lower leaves. The average disease severity was determined for ten plants of each barley accession. A four-class rating scale for assessing disease severity was employed: R-highly resistant (1-10.5%); MR-moderately resistant (11-20%); MS-moderately susceptible (21-39%); and S-susceptible (40-80%).

Experimental design
Laboratory experiments were performed in a climate-controlled room at a temperature of 23 • C and a light period (9000 lux) of 16 h. The seeds of the tested accessions were germinated on moistened filter paper in Petri dishes in the dark for two to three days. Sprouted grains were placed on sterile sand moistened with distilled water in 200 mL plastic pots (100 g of sand in each) and inoculated with 10 mL per pot of suspension at a concentration of 50,000 conidia/mL (5000 spores per 1 g of sand). As a control, the same number of plants of each accession was employed without inoculation. Harrington cultivar susceptible to root rot were utilized as infection controls. The experiment was conducted in three replicates with 15 plants of each accession in one replicate (total of 45 plants for each accession).

•
Disease evaluation At 14 days after inoculation, when the plants were in the phase of two to three leaves, the damage of barley samples by root rot was measured. The plants were carefully removed from the sand, the roots were thoroughly washed with water, and the degree of damage to roots and coleoptiles was visually assessed using a five-point scale.
This scale of assessment of seedling root rot caused by C. sativus is based on the brown lesion size and color of coleoptile. Low IRs of 0-1 (small light brown lesions not measuring more than 0.5 cm on the roots and single strokes on coleoptile) correspond to resistance (R), a score of 2 (light brown lesions on the roots up to 1 cm long, strokes on coleoptile strongly marked) indicates moderate resistance (MR), a score of 3 (light brown/black lesions on the roots up to 1.5 cm long and light brown color of coleoptile) demonstrates moderate susceptibility (MS), and scores of 4 (strong brown/black lesions on the roots up to 2 cm long; strong brown colour of coleoptile; brown/black lesions) and 5 (strong brown/black on the roots longer than 2.5 cm in length, black lesions on coleoptile, seedling dies), correspond to accessions that are susceptible (S) and highly susceptible (HS), respectively, to root rot. Types of reactions were reported only if the susceptible cultivar Harrington exhibited high IRs.

Association Analysis
Association analysis was performed using the TASSEL 5 package [32]. In total, four statistical tests were employed for this study, including a general linear model (GLM) without population structure and accounting for population structure across the Q-matrix and principal component analysis (PCA) and a mixed linear model with kinship matrix (MLM + K). In this study, the newly obtained phenotypic data were utilized along data obtained previously [11], and new data were analyzed with the same models as the old data to compare them. Genotyping data were analysed using the barley 50K Illumina iSelect SNP array at the Traitgenetics GmbH (Gatersleben, Germany), and a set of 27 319 markers were selected for the next analysis after quality control [11].
To detect significant SNPs, two corrections were employed: (i) the 5% Bonferroni threshold was set at 1.8309 × 10 -6 , which meant that the significance level (0.05) was divided by the total number of tests (27,319), and (ii) the false discovery rate (FDR) was calculated for each isolate in each model. The suggestive level corresponded to p < 10 −4 , which implied that the significance of an association was sufficiently high but did not exceed the threshold value.

Conversion to KASP Markers
When the candidate SNPs were detected, their reference sequence was obtained using the http: //plants.ensembl.org/index.html database. Next, sequences were analyzed with the UGENE program. Polymorphic DNA alleles with flanking sequences measuring 101 bp were presented using the format in which two allele states at investigated SNPs were divided by the symbol "/" and enclosed in square brackets. The known polymorphic base pairs were identified using standard nomenclature. The 11 markers, which were converted into the KASP markers, were chosen based on the obtained data. KASP genotyping was conducted by LGC Genomics (UK). Primers sequences are presented in the Table S5.

Phenotyping
The results of the investigation of the seedling and adult resistance to spot blotch isolates are presented in Table S2. According to the Fetch and Steffenson rating scale, 19% of genotypes were resistant, 34% of genotypes were moderately resistant to isolate O18.2 at the seedling stage, 5% of genotypes were resistant, and 18% were moderately resistant to isolate Ch3 at the adult plant stage. The obtained results were compared with previous data obtained for seedling resistance for two isolates: Ch3 and Kh2 [11]. Three genotypes (B-1, Kolchan, Svetik) were resistant to all isolates at both the seedling and adult plant stages. Seven cultivars, Aley, Biom, G-21219, Mutant 68, Omsky golozyorny 2, Severny, and Signal, were resistant to three of four isolates. Twenty per cent of genotypes were resistant, and 11.7% of genotypes were moderately resistant to isolate O18.2 when it was used for seedling evaluation of resistance to root rot (Table S3). One cultivar, Kolchan, was resistant to all isolates and exhibited seedling and adult spot blotch resistance and root rot resistance. One cultivar, Aley, was resistant to common root rot, and one cultivar was resistant to spot blotch at the seedling stage. The results of correlation analysis of data on seedling and adult resistance to different isolates are given in Table 1. Table 1. Pearson correlation analysis between resistance to spot blotch and common root rot, as well as between resistance to spot blotch between seedlings and adult plants.  The correlation is shown in the low triangle; the correlation significance is shown in the upper triangle. The significant p-value in all cases was <0.05.

GWAS Analysis
• Isolate O18.2. Seedling resistance To estimate the significance of the observed associations, the results of every model were first considered in the QQ plot. Ideally, the line representing the observed p-values should match that for the expected p-values. For the O18.2 experiment performed at the seedling stage, the most accurate result was obtained using the GLM + Q, GLM + PCA or MLM + K models, while the GLM model without correction for population structure provided a large number of false-positive SNPs Figure A1. With the help of the GLM + Q model, in total, 47 SNPs were identified, namely, 2 SNPs on chromosome 1H, 17 SNPs on chromosome 3H, 1 SNP on chromosome 7H and 28 SNPs on chromosome 2H, but the locus region is too large, and we suggest that among SNPs of the second chromosome, false-positive SNPs are presented ( Figure A2a). With the help of the GLM + PCA model, in total, 14 SNPs were identified, namely, 13 SNPs on chromosome 3H and 1 SNP on chromosome 7H ( Figure A3a). No significant SNPs were identified with the MLM + K model ( Figure A4a). Results for every model were compared with previous researches (with data of Ch3 and Kr2 isolates) and it was showed that pathogen response locates in the same genomic regions ( Figure A2b Table S4. The comparison of seedling resistance loci obtained by testing the O18.2 isolate with previous data for other isolates [11] showed that significant genome loci were identified in the same chromosomal regions (Figure 1).

•
Isolate Ch3. Adult resistance According to the results of GLM + Q and GLM + PCA, only one significant SNP on chromosome 5H at the interval 1.12-1.22 cM was revealed (Figure 1). No significant SNPs were identified using MLM analysis with the kinship matrix (MLM + K model). •

Common root rot
No significant SNPs were revealed in these studies.

PCA
The principal component analysis (PCA) was calculated for all phenotypic disease evaluation data regarding spot blotch. The principal component analysis for genotypic data was calculated through the distance matrix [33] using the software package JACOBI4 [34]. To calculate the distance matrix between barley varieties, we re-coded the barley genome from a two-letter code to a numerical code. After the re-coding, 0 was assigned to the effector allele, and 1 was assigned to the non-effector allele, and their intermediate forms were coded as 0.5. For example, the AA allele is reflected as 1, AG as 0.5, and GG as 0. Both sets of main components were taken as blocks in two-block partial least-squares (2B-PLS) analysis (first set is phenotypic block, second set is genotypic block).
"Phenotype-genotype" covariation was calculated as a set of linear bi-components. The first three linear bi-components capture 93% of the sum covariation. Therefore, three first components were analyzed.
Correlation coefficients are shown for all investigated isolates, both for seedlings and for adult plants. We consider bi-components simultaneously, both for phenotyping and for genotyping, to process the data for a more complete investigation. The phenotypic part reflects the totality of all phenotypic reactions in the infection response investigation. It was observed that seedling resistance to every isolate correlates with the first phenotypic bi-component, as well as the resistance of adult plants. Additionally, the resistance of adult plants correlates with the second phenotypic bi-component.
Therefore, using the phenotypic bi-component data as the phenotypic reaction in the GWAS analysis, we can obtain markers associated with resistance to all isolates within the investigated varieties. Therefore, the associated analysis of the first bi-component data was conducted. Model GLM showed that 41 SNPs are significant. Of these SNPs, 22 are on chromosome 1H, 17 SNPs are on chromosome 3H and 2 SNPs are on chromosome 4H. Moreover, according to the QQ plot, this model shows the presence of false-positive markers ( Figure A1).
A total of 35 significant SNPs were identified using model GLM + Q ( Figure A2e). Of these SNPs, 18 were on chromosome 2H, and 17 were on chromosome 3H. Model GLM + PCA identified 18 SNPs. Sixteen of these SNPs were on chromosome 3H, and 3 SNPs were on chromosome 7H. (Figure A3e). A total of 7 SNPs were identified with the MLM + K model ( Figure A4e). Locus markers on chromosome 3H and on chromosome 7H, which were detected using the first bi-component, coincide with the markers previously identified by the authors. The correlation of the first bi-component data with all isolates and with adult plants suggests that these markers may be determined to be diagnostic within the studied isolates.

PLS Analysis
As a result of the PLS analysis applied to the totality of phenotypic traits (resistance to spot blotch) and genotypic data, bi-components were obtained: phenotypic and genotypic, with most corresponding to each other. All phenotypic traits correlate with the first bi-component (Table 2), that is, the first bi-component reflects the overall stability of the studied traits. A correlation plot was constructed between the genotype and the phenotype data of the first bi-component ( Figure 2). Considering the resistance of varieties to various pathogen isolates at the stage of seedlings and adult plants, it was observed that varieties that showed primarily resistance are grouped at one end of the correlation line (the circle "R" at Figure 2). Additionally, varieties susceptible to the pathogen in all trials are on the other end of the correlation line (the circle "S" in Figure 2). As a result, two clusters were identified: resistant (eight varieties) and susceptible (30 varieties) varieties to spot blotch ( Figure 2). Accordingly, it can be concluded that these varieties possess the genetic background that responds to infection by the studied pathogens. The SNPs enabling the two clusters R and S to be distinguished ( Figure 2) were considered, and of these SNPs, 11 markers were selected for further conversion to KASP markers (Table S4): these SNPs are located on chromosome 1H (50-61.2 cM), chromosome 3H (18.72 cM and 24.63-26.18 cM), chromosome 5H (1.12-1.22 cM) and chromosome 7H (7.52-15.44). Ten out of the 11 SNPs were successfully developed at LGC, and 1 failed (** in Table 3). Despite the identification of significant SNPs on chromosome 2H, these SNPs did not help to clearly distinguish the two clusters R and S ( Figure 2).

KASP Genotyping Results
To validate that significant SNPs were converted to KASP markers, independent barley accessions (22 accessions, including 11 resistant and 11 susceptible to spot blotch) were chosen. The results were plotted on a Cartesian plot, where the x-axis shows the FAM signal fluorescence value for each sample associated with one allele and the y-axis shows the HEX signal fluorescence value associated with the second allele ( Figure 3).  Table 4.

Discussion
According to previously published studies, no significant correlations were observed between root and leaf infection caused by B. sorokiniana [3,17].
As shown in Table 1, the correlation between seedling resistance to different isolates was high enough (from 0.61 to 0.73). The correlation between the seedling and adult stages was weaker. Adult plants were infected with the Ch3 isolate, and the correlation between the indicators for seedlings infected with Ch3 and similarly infected adult plants was 0.59. Table 1 shows that the correlation between root rot and spot blotch varies in the region from 0.22 to 0.28.
Several loci were identified in the significant interval of chromosome 1H (50-61.2 cM; Figure 1). In 1996, Steffenson et al. revealed QTLs for adult plant resistance (53.6-61.2 cM; [18]). The locus responsible for seedling resistance was determined to be at 59.7 cM by Roy et al., 2010 [9]. The locus between 40 and 50 cM was identified by Gutierrez et al., 2013 [35]. Later Afanasenko et al., 2015 [36] identified QTLs at the SNP locus BOPA_11_11015 (syn. BOPA1_946-2500; position in iSelect map-54.2 cM) by analysis of bi-parental mapping population Zernogradsky 85 (R)/Ranny 1 and Haas et al., 2016 [26] revealed the locus on 1H chromosome (42.2 cM) on population derived from the wild accession and the cultivar Rasmusson. However, the investigated marker JHI-Hv50k-2016-33568 from the current study showed only slight predictive power (55%), which does not enable it to be utilized in marker-assisted selection. It is possible that this marker could be used as race-specific to certain pathogen isolates.
On chromosome 3H, the QTL associated with resistance to spot blotch was found in a very similar position by Zhou and Steffenson, 2013 [20]. Additionally, the resistance locus was identified for seedlings at position 9.6 cM (Bopa_12_30818) and for adult plants at position 19.2 cM (BOPA_11_20742, BOPA_11_10565). Grewal et al., 2012 [37] detected the resistance locus for seedlings in the interval 24.9-31.1 cM and two QTLs of resistance for adult plants in the interval 23.0-24.9 cM. In the current study, the genomic locus with the most significant SNPs was determined to be on chromosome 3H in intervals of 18.72 and 24.6-26.18 cM. Three of the markers (JHI-Hv50k-2016-156842, JHI-Hv50k-2016-156833, JHI-Hv50k-2016-155569) in which the presence of one or another allele was associated with the resistance/susceptibility of varieties can be employed as diagnostic markers in breeding programs.
According to published data, there is not enough information indicating the locus of resistance on chromosome 5H. Bovill (2010) [38] detected minor QTLs on chromosome 5H in only one year from three years of research (marker P22M50-304). Later, an allele in position 110.25 cM (Rcs-qtl-5H-110.25 cM) for adult plants was described [23]. The locus on chromosome 5H identified in this study is in the interval of 1.12-1.22 cM (JHI-Hv50k-2016-277077). In this independent sampling, the marker showed 100% presence of the G allele in susceptible varieties. However, the A allele was not observed in resistant varieties. One of the reasons for this phenomenon may be that the resistance for this marker is not widespread enough, and the selected independent sampling turned out to be small.
The resistance locus was determined to be on chromosome 7H. Bilgic et al., 2005 [39] showed that seedling resistance is controlled monogenically by the Rcs5 gene on chromosome 7H, while adult plant resistance is controlled by two QTLs: the major locus on chromosome 1H and the minor locus on chromosome 7H. Among the three studied markers, one (JHI-Hv50k-2016-446766) failed to obtain results (N/A). The two remaining variables (JHI-Hv50k-2016-448898 and JHI-Hv50k-2016-451269) showed low predictability (27 and 36%, respectively).
The discovered KASP-markers can be used for selection in hybrid populations obtained using polymorphic parental forms. PCR markers that are easier to use in selection can be developed from the selected KASP-markers in the future.
Selection requires a quick assessment of the breeding material for the presence of the desired allelic resistance in elite lines, regardless of the degree of damage to plants in the field. Accelerating the breeding new varieties is one of the perspectives of this work.
Supplementary Materials: The following are available online at http://www.mdpi.com/2077-0472/10/11/505/s1. Table S1: Set of independent barley accessions for the validation of candidate diagnostic markers for the resistance to C. sativus, Table S2: Results of spot blotch resistance assessment within the Siberian spring barley cultivars collection, Table S3: Results of seedlings common root rot resistance assessment within the Siberian spring barley cultivars collection, Table S4: SNPs associated with resistance to O18.2 isolate, revealed by GLM + PCA and GLM + Q analysis, Table S5: Primers sequences.