SNPs associated with barley resistance to isolates of Pyrenophora teres f. teres

Net blotch caused by Pyrenophra teres f. teres is a major foliar disease of barley. Infection can result in significant yield losses of susceptible cultivars of up to 40%. Of the two forms of net blotch (P. teres f. teres and P. teres f. maculata), P. teres f. teres (net form of net blotch) is the dominant one in Russia. The goal of the current study was to identify genomic regions associated with seedling resistance to several pathotypes of the net form of net blotch in Siberian spring barley genotypes. For this, a genome-wide association study of a Siberian barley collection, genotyped with 50 K Illumina SNP-chip, was carried out. Seedling resistance of 94 spring barley cultivars and lines to four Pyrenophora teres f. teres isolates (S10.2, K5.1, P3.4.0, and A2.6.0) was investigated. According to the Tekauz rating scale, 25, 21, 14, and 14% of genotypes were highly resistant, and 19, 8, 9, and 16% of genotypes were moderate-resistant to the isolates S10.2, K5.1, P3.4.0, and A2.6.0, respectively. Eleven genotypes (Alag-Erdene, Alan-Bulag, L-259/528, Kedr, Krymchak 55, Omsky golozyorny 2, Omsky 13709, Narymchanin, Pallidum 394, Severny and Viner) were resistant to all studied isolates. Nine additional cultivars (Aley, Barkhatny, Belogorsky, Bezenchuksky 2, Emelya, G-19980, Merit 57, Mestny Primorsky, Slavaynsky) were resistant to 3 of the 4 isolates. The phenotyping and genotyping data were analysed using several statistical models: GLM + Q, GLM + PCA, GLM + PCA + Q, and the MLM + kinship matrix. In total, 40 SNPs in seven genomic regions associated with net blotch resistance were revealed: the region on chromosome 1H between 57.3 and 62.8 cM associated with resistance to 2 isolates (to P3.4.0 at the significant and K5.1 at the suggestive levels), the region on chromosome 6H between 52.6 and 55.4 cM associated with resistance to 3 isolates (to P3.4.0 at the significant and K5.1 and S10.2 at the suggestive levels), three isolate-specific significant regions (P3.4.0-specific regions on chromosome 2H between 71.0 and 74.1 cM and on chromosome 3H between 12.1 and 17.4 cM, and the A2.6.0-specific region on chromosome 3H between 50.9 and 54.8 cM), as well as two additional regions on chromosomes 2H (between 23.2 and 23.8 cM, resistant to S10.2) and 3 (between 135.6 and 137.5 cM resistant to K5.1) with suggestive SNPs, coinciding, however, with known net blotch resistance quantitative trait loci (QTLs) at the same regions. Seven genomic regions on chromosomes 1H, 2H, 3H, and 6H associated with the resistance to four Pyrenophora teres f. teres isolates were identified in a genome-wide association study of a Siberian spring barley panel. One novel isolate-specific locus on chromosome 3 between 12.1 and 17.4 cM was revealed. Other regions identified in the current study coincided with previously known loci conferring resistance to net blotch. The significant SNPs revealed in the current study can be converted to convenient PCR markers for accelerated breeding of resistant barley cultivars.


(Continued from previous page)
Conclusions: Seven genomic regions on chromosomes 1H, 2H, 3H, and 6H associated with the resistance to four Pyrenophora teres f. teres isolates were identified in a genome-wide association study of a Siberian spring barley panel. One novel isolate-specific locus on chromosome 3 between 12.1 and 17.4 cM was revealed. Other regions identified in the current study coincided with previously known loci conferring resistance to net blotch. The significant SNPs revealed in the current study can be converted to convenient PCR markers for accelerated breeding of resistant barley cultivars.
Keywords: Association mapping, Barley, GWAS, Hordeum vulgare, Net blotch, Resistance, SNP Background Net blotch, caused by Pyrenophora teres (anamorph: Drechslera teres [Sacc.] Shoem.), is a major foliar disease of barley worldwide and in Russia. The pathogen exists in two forms based on the symptoms they cause: the net form of net blotch (P. teres f. teres) and the spot form of net blotch (P. teres f. maculata). The net form of net blotch (NFNB) is the dominant form in different regions of Russia; the spot form of net blotch (SFNB) was found only in the southern part of European Russia [1]. NFNB epidemics in Northwest Russia appear with a frequency of 5 times every 10 years [2]. Infection can result in significant yield losses of up to 40% on susceptible cultivars under favourable environmental conditions [3]; also, the disease can cause reductions in the quality of barley [4].
The most cost-effective and environmentally friendly way to control the disease is the development of resistant cultivars. The success of resistance breeding relies on the genetic diversity of resistance and the availability of resistance genes in locally adapted germplasm.
The genomic regions associated with resistance of barley to P. teres f. teres have been found on all barley chromosomes [3,[5][6][7][8][9][10][11][12][13][14][15][16][17][18] using both linkage mapping in biparental mapping populations and association mapping (AM). Some QTLs provide resistance during whole ontogenesis, such as QRpt6 on chromosome 6H, determining both seedling and adult resistance to the net form of net blotch [5]. Other QTLs appear to be either seedling-or adult-specific [19]. Among the genomic regions associated with P. teres f. teres resistance, the region on chromosome 6H is the most well studied. It is supposed that either 3 different alleles of a single locus or three closely linked resistant genes exist in this region [9].
The goal of the current study was to identify genomic regions associated with seedling resistance to several pathotypes of the net form of net blotch. For this, a genome-wide association study of a Siberian barley collection, genotyped with 50 K Illumina SNP-chip, was carried out.

Plant material and genotyping data
The study was based on a Siberian barley panel, consisting of 94 spring cultivars and breeding lines from the ICG GenAgro collection (Novosibirsk, Russia). Half of this panel was represented by cultivars and lines developed in breeding centres located in Siberia, whereas the other half consisted of cultivars and lines maintained in the Siberian spring barley collection, but originating from other regions and countries. Genotyping data for these 94 cultivars and lines were available from our previous study [20]. Additional information on 50 K Illumina SNP-chip loci was extracted from [21] and the BARLEYMAP resource (http://floresta.eead.csic.es/barleymap).
Propagation of the P. teres isolates was conducted on Czapek's modified medium containing 0.5 g/L KH 2 PO 4 , 0.5 g/L MgSO 4 , 0.5 g/L KCl, 1.2 g/L urea, 20 g/L lactose, and 20 g/L agar. To produce inoculum, single spore cultures were grown under near ultraviolet (UV) light with a 12 h photoperiod at 18-20°C for 14 days. Conidia were harvested by adding distilled water to the plate and scraping the agar surface with a spatula. The suspension was filtered through two layers of cheesecloth to remove fragments of mycelia. The concentration of the inoculum was adjusted to 5000 conidia per ml. The surfactant Tween 20 was added (100 μl per litre) to facilitate dispersion of the inoculum over the leaf surfaces. Inoculation was completed by spraying at a rate of approximately 0.2 ml per plant.

Plant growing and disease assessment
Seedling resistance was evaluated in controlled conditions in a climate chamber in the All-Russian Research Institute for Plant Protection (St. Petersburg, Russia). Three seeds of each barley cultivar were sown per pot containing nutrient-supplemented peat and cultivated for 2 weeks at 20-22°C with a photoperiod 16 h light (exposure 5000 lx)/8 h darkness in a split-plot design with three replicates. After inoculation, plants were covered with plastic bags and placed for 48 h at 20-22°C without light. After 2 days, inoculated plants were placed at 20-22°C with a photoperiod 16 h light (exposure 5000 lx)/8 h darkness and air humidity of 60-70% and were grown till the disease assessment. Seedling infection responses (IRs) were assessed on the second leaf 10-12 days after inoculation. P. teres resistance was scored by using the 10-point scale of Tekauz [22] 1-3 = highly resistant (HR); 3.1-5.0 = moderately resistant (MR); 5.1-6.9 = moderately susceptible (MS); 7.0-10.0 = high susceptible (HS).

Population structure
The population structure was analysed using STRUC-TURE v 2.3.4 [23] based on the genotypic data of a subset of 13,659 markers. Each second marker of a set of 27,319 markers previously selected by quality control [20] was taken to reduce the computing time. The number of subpopulation (k) in the panel was inferred using an admixture model with correlated allele frequencies, a burn-in period length of 5000 and 5000 Markov chain Monte Carlo (MCMC) repetitions. Independent analyses were run for each k between 1 and 32. The estimated likelihood values [LnP(D)] were compared with k using a graph to determine the optimal k.

Association analysis
Different statistical models were tested on disease resistance scores (separately for each of four isolates) with the help of the TASSEL 5 package [24] to detect significant marker associations: (1) generalized liner model (GLM) without correction for population structure; (2) GLM + Q: GLM + Q-matrix to account for population structure; (3) GLM + PCA: GLM with a principal component analysis (PCA) to account for population structure, (4) GLM + PCA + Q; and (5) MLM + K: MLM with kinship matrix. Genotyping data for a set of 27,319 markers previously selected by quality control [20] were used in the association analysis.
To identify significant single nucleotide polymorphisms (SNPs), two corrections were used: (i) the Bonferroni correction, where the significant threshold (0.05) is divided by the total number of tests, in this case, the total number of markers (27,319), giving the threshold 1.8302*10 − 6 , and (ii) the false discovered rate (FDR) that was calculated for each isolate in each model. The suggestive level corresponded to p < 10 -4th and was considered as suggestive evidence of an association if SNPs in the model of an isolate did not exceed the threshold value.

Population structure
The most likely number of subpopulations was k = 4 as determined by STRUCTURE v 2.3.4 (Fig. 2). The set of barley genotypes was divided into 4 groups (Fig. 3) consisting of 17, 29, 20, and 34% of genotypes. Group III contained the highest percentage of Siberian accessions (67%). Groups I, II and IV contained 31, 30 and 39% of Siberian accessions, respectively ( Table 1). Percentages of highly resistant (HR), moderately resistant (MR), moderately susceptible (MS), and highly susceptible (HS) genotypes in each group is given in Table 1.

Genome-wide association study (GWAS) analysis
The results of all statistical models were first compared in quantile-quantile (QQ) plot to find proper models for each dataset. QQ-plots for the models used are presented for each P. teres f. teres isolate in Additional file 2. The GLM analysis without correction for population structure showed a great number of false positive SNPs in a QQ-plot (Additional file 2). A QQ-plot using the GLM model accounting for population structure (GLM + Q) appeared to be more proper; in the case of the P.3.4.0 isolate, a very good match with expected values was observed (Additional file 2). Similarly, the GLM + PCA and GLM + PCA + Q models appeared to be more proper than GLM. Association mapping results using different models are presented in Additional file 3. With the help of the GLM + Q model, two significant SNPs on chromosome 6H were revealed for the isolate P3.4.0, one significant SNP on chromosome 3H for the isolate A2.6.0, one suggestive SNP on chromosome 1H for the isolate K5.1 and two suggestive SNPs (1 SNP on chromosome 2H and 1 SNP on chromosome 5H) for isolate S10.2 (Additional file 3; Table 2).
The GLM analysis with PCA accounting for the population structure (GLM + PCA) revealed 2 significant SNPs on chromosome 6H and 2 significant SNPs on chromosome 2H associated with resistance to the isolate P.3.4.0. Additionally, 2 suggestive SNPs (1 SNP on chromosome 3H and 1 SNP on chromosome 6H close to the region revealed for P3.4.0 isolate) were associated with resistance to the K5.1 isolate, and 3 suggestive SNPs on chromosome 6H were associated with resistance to S10.2 (Additional file 3; Table 3).
The GLM analysis with a combination of two corrections GLM + PCA + Q revealed 7 significant SNPs on chromosome 6H, 7 significant SNPs on chromosome 2H, 8 SNP on chromosome 1H and 6 significant SNPs on chromosome 3H (Additional file 3; Table 4).
No significant SNP was revealed using the MLM analysis with the kinship matrix (MLM + K model).

Discussion
The analysis of the population structure of the Siberian barley panel revealed 4 clusters. We noticed that among 14 accessions susceptible to all four net blotch isolates, 8 originated from Siberia (57.1%), and among 23 accessions susceptible to three isolates, 5 (22%) originated in Siberia.
The GWAS performed using five statistical models revealed seven genomic loci associated with resistance to one to three net blotch isolates. The comparison of these regions with locations of previously known P. teres resistance QTLs is presented in Table 5.    (Table 5).

Chromosome 2H
The resistance to different isolates P3.4.0 and S10.2 was associated with two different loci. The locus associated with resistance to the S10.2 isolate included one suggestive SNP (JHI-Hv50k-2016-74407) mapped in the interval 23.3-23.8 cM. We suggest that seedling resistance to the S10.     (Table 5).

Chromosome 6H
The region on chromosome 6H between 52.6 and 55.4 cM was associated with resistance to three isolates (to P3.4.0 at the significant and K5.1 and S10.2 at the suggestive levels). Among the 12 SNPs revealed in the current study is SCRI_RS_239642 and SCRI_RS_224389,

Conclusions
Seven genomic regions on chromosomes 1H, 2H, 3H, and 6H associated with the resistance to four