Genetic diversity of PM resistance in the CG population
The 109 CG lines were evaluated for PM resistance three times, in spring 2013 (pm_2013S), spring 2014 (pm_2014S) and fall 2014 (pm_2014A). For each line, the DI of each experiment was calculated and shown in Supplementary Table 1. The mean DI of the three experiments varies among the CG lines between 25.9 and 29.4 (Fig. 2a). The DI correlation among the three experiments was significant with a R2 = 0.7 ~ 0.81 (Fig. 2b). Based on the SNP in the three loci and the DIs of three seasons, we detected the genotypic effect at each of the two seasons (Supplementary Fig. 1) . Results showed that the three loci were independent in the three seasons, which indicated that these three loci work independently for powdery mildew.
The mean DI per line of the CG population was used for phylogenic analysis to study the PM resistance diversity, resulting in three clusters based on the complete method (1.5) of SAS (Supplementary Fig. 2). The first cluster contains 34 lines with a DI < 20. Among these lines, 23 lines showed a DI < 10 and classified as highly resistant (HR) lines. Of these HR lines, nine (CG7, CG9, CG32, CG35, CG44, CG63, CG70, CG82 and CG101) showed a consistent DI score in all three experiments. The second cluster contains 33 lines with a DI > 35 and regraded as highly susceptible (HS) lines. Five HS lines, CG3, CG23, CG31, CG55, CG209 showed a comparable DI among the three tests. Interestingly, seven tested Japanese type (CG3, CG4, CG28, CG29, CG31, CG103 and CG104) showed high sensitivity. The third cluster includes 30 sensitive lines with a DI between 20 and 35.
Genome-associated analysis of powdery mildew resistance
In order to identify genetic loci that are associated with PM resistance in the CG population, DIs of each experiments was used for GWAS. Total twelve loci (pmG1.1, pmG2.1, pmG2.2, pmG3.1, pmG4.1, pmG4.2, pmG5.1, pmG5.2, pmG5.3, pmG5.4, pmG6.1 and pmG6.2) were detected, which are scattered on six chromosomes (Fig. 3). Three loci, pmG2.1, pmG5.2 and pmG5.3 were detected in all the three experiments with a higher p-value than others, indicating that these loci were stable and strongly related to PM resistance. Four loci, pmG1.1, pmG4.1, pmG4.2 and pmG5.4, were detected in two of the three experiments. Five loci, pmG2.2, pmG3.1, pmG5.1, pmG6.1 and pmG6.2, were detected in one of three experiments. Among the 12 loci, 9 loci were located in chromosomal regions where QTLs for PM resistance were reported in previous studies [6, 10, 11, 29]. While, the other three (pmG2.1, pmG4.1 and 6.1) have not been reported yet (Fig. 1) and thus representing novel cucumber loci for PM resistance.
Candidate genes for the PM resistance
Aiming at identifying potential candidate genes of the stable loci (pmG2.1, pmG5.2 and pmG5.3) and novel loci (pmG4.1 and pmG6.1), chromosomal region of 50 kb around the peak SNPs were further analyzed.
Candidate gene analysis for novel loci ( pmG4.1 and pmG6.1)
Physical distance of the peak SNP in pmG4.1 was at Chr.4:2,436,922 bp. Based on the cucurbit genomics (http://cucurbitgenomics.org/), 14 genes were predicted in the pmG4.1 region (Supplement table S2). In this region, Csa4G022270 encodes NPR1-like protein that regulates salicylic acid (SA)-mediated plant defense . Physical distance of the peak SNP in pmG6.1 was at Chr.6:2,291,66 bp. Ten genes were predicted in the pmG6.1 candidate region (Supplement table S3). Csa6G022350 encodes splicing factor U2AF subunit and its homologous gene in Arabidopsis is involved in pathogen response via the salicylic acid pathway . Csa6G022370 encoded MAC/Perforin domain containing protein and its homologous gene in Arabidopsis is related to response during germinivirus infection . These two genes might be the candidate genes of pmG6.1.
Candidate gene analysis for pmG2.1
For the pmG2.1 locus, SNPs of the chromosome region Chr. 2: 2,710-2,949 kb were analyzed by pairwise LD correlations with a focus on candidate genes in the interval from 2.715 to 2.768 Mb (Fig. 4a). Based on the Cucumber Genome Browser (http://www.icugi.org/cgi-bin/ICuGI/index.cgi), 11 candidate genes were located in this interval. Three genes, Csa2G022770, Csa2G022780 and Csa2G022790 were selected, since they all encode a disease resistance-like protein although the DNA sequence similarity among them is very lower than 25%. The relative expression of the three genes were analyzed by qRT-PCR (Fig. 4d-f) with gene specific primers (Supplement table S4) in the selected HR and HS lines. Results showed that Csa2G022790 was significantly up-regulated in three of the four HR lines at 12 h after PM inoculation. In one of the three HR lines, CG44, the induction was also seen at 8 d after PM inoculation. A single nucleotide polymorphisms (SNP) (G to T) in the promoter of Csa2G022790 was located on the TCA-element (salicylic acid responsiveness) with the G presents in HR lines (Fig. 4c). Salicylic acid is a key signaling component involved in the activation of certain plant defense responses . In the HS lines the expression of Csa2G022790 was down-regulated in the CG3 line and no difference was found in the other lines (Fig. 4g). These results indicated that Csa2G022790 might be the candidate gene of the pmG2.1 locus.
Candidate gene analysis for pmG5.2
For the pmG5.2 locus, SNPs in the chromosomal region of Chr.5: 16,650 − 17,150 kb were analyzed (Fig. 5a). We focused on the interval from 16,878,456 bp to 16,879,990 bp based on pairwise LD correlations (r2 ≥ 0.6). Csa5G484620 was identified in this region, encoding an NBS resistance protein. Two SNPs (SNP2431819 and SNP2431821) were found in this gene (Fig. 5c). Twenty nine out of 30 HR lines were G/G and the other one is A/C, whereas 9 of 29 S lines were A/C and the others were G/G. Based on the qRT-PCR, the Csa5G484620 was significantly up-regulated in CG44 (HR) at the 12 h and down-regulated at 24 h after PM inoculation (Fig. 5d). Then, four HR lines and four HS lines were applied to qRT-PCR (Fig. 5e). In the HS lines, the expression of Csa5G484620 was down-regulated in CG23 while up-regulated in CG55, and no difference in the other two lines. Whereas the expression of all the HR lines were significantly up-regulated at 12 h after PM inoculation. These results indicated that Csa5G484620 might be the candidate gene of the pmG5.2 locus.
Candidate gene analysis for pmG5.3
For the pmG5.3 locus, SNPs in the chromosomal region of Chr.5: 22,230 − 22,610 kb was analyzed with the focus on the interval from 22,506 kb to 22,518 kb (~ 12 kb) using pairwise LD correlations (r2 ≥ 0.6) (Fig. 6a). Two annotated genes were identified; Csa5G604330 encoding alpha-amylase and Csa5G604340 encoding Pectate lyase. The latter is homologous to the Arabidopsis PMR6 gene . SNP_2526314 was on the CDS of Csa5G604340. For this site, 29 of 30 HR lines was A, the other HR lines was G, whereas that 7 of 29 HS lines was G, the other HS lines was A (Fig. 6b). Then, four HR lines and four HS lines were applied to analyze the expression of Csa5G604340 using qRT-PCR. Two of the HS lines (CG23 and CG31) and three of the HR lines showed up-regulated expression at 12 h after PM inoculaton (Fig. 6c). We focus on all the 18 European type lines, all the G genotype in this site was HS lines, and 55.6% of A genotype were HR lines, which showed that the gene Csa5G604340 could be related to the PM resistance of European type (Fig. 6d). Csa5G604340 might be the candidate gene of the pmG5.3 locus.