Cucumber Team members:
- Yiqun Weng (YW), USDA-ARS, University of Wisconsin Madison,
- Rebecca Grumet (RG), Michigan State University
- Kai-Shu Ling (KL), USDA-ARS Charleston, SC
- Anthony Keinath (AK), Clemson University, SC
CucCAP Affiliated postdocs and graduate Students
- Dr. Junyi Tan, Postdoc research associate, University of Wisconsin Madison (Weng)
- Ying-Chen Lin, graduate students, Michigan State Univ (Grumet)
- Stephanie Rett-Cadman, graduate student, Michigan State Univ (Grumet)
- Daoliang Yu, graduate student, University of Wisconsin Madison (Weng)
- Sierra Zardus, technical support, Clemson University (Keinath)
- Anna Mothersbaugh, technical support, Clemson University (Keinath)
- Bazgha Zia, postdoc research associate, USDA-ARS, Charleston, SC (Ling)
Objectives and timeline
- Develop genomic, bioingormatic, mapping approacches and tools for cucurbits (2021,2022, 2023)
- Map and develop markers for disease resistance
- QTL introgression into breeding advanced lines and release to breeders
- Economic impact analyses and disease control information
Obj. 1.2 Seed multiplication of cucumber core collection
(YW & industry collaborators)
Seed increase of 388 accessions from diverse taxonomic groups, geographic origins, and market groups was continued in 2023 by industry collaborators. As of February 27, 2024, seed increase of 372 accessions has completed with 310 having >1000 seeds each and 62 with 100-1000 seeds per line. The remaining 16 failed for seed increase.
Obj. 2. Map and develop markers for disease resistance
2.1 QTL mapping of resistances
2.1.1 Downy mildew (YW & AK)
We collected phenotypic data for downy mildew (DM) responses to natural infection in field conditions in North and South Carolinas (multiple years), as we all as Hancock, Wisconsin fields (2022 only). We are performing association analysis using the data. Here is an example of GWAS with data from Hancock field in 2022 when an epidemic of DM occurred unexpectedly. In total 207 lines were grown in the trial with two replications. DM symptoms were scored with necrosis (Nec), chlorosis or yellowing (Yel), and general impression (GI). Frequency distribution of mean GI scores is shown in Figure 1. Nearly half of the lines are highly susceptible. The majority of lines, and less than 1% (17 of 207) showed moderately or highly resistant.
Figure 1. Frequency distribution of mean DM disease scores (GI) of 207 cucumber accessions upon natural infection of P. cubensis pathogen from 2022 field trial at Hancock, WI.
GWAS was conducted with mean disease scores and nearly half million high quality SNPs among 197 lines. No significant association was detected when GI or Nec were used. However, when Yel (chlorosis) ratings were employed, a SNP 2,0328,829 position (Gy14v2.1) on Chromosome 5 was detected to be significantly association with Yel in this panel. (Gy14v2.1) (Figure 2). This is only ~150 kb away from the dm1/CsSGR (staygreen, CsGy5G003280; 2,148,563-2,150,155) locus that confers broad-spectrum disease resistances including pre-20024 P. cubensis field strain(s). In the same filed, DM data from a population of Gy14 x 9930 cross with 112 RILs were also collected including Yel, Nec and GI. QTL analysis in this population revealed a major-effect QTL that is consistent with CsSGR/dm1 for DM resistance in Gy14 (Figure 3). These observations suggest CsSGR still plays a critical role in DM resistance, especially anti-chlorosis in the field.
Figure 2. A significant association of DM resistance was detected near the dm1/CsSGR locus on cucumber chromosome 5 using 2022 Hancock mean Yel data and kinship + PC1.
Figure 3. LOD curve from QTL analysis of DM disease scores in the Gy14 x 9930 RIL population grown in 2022 Hancock field.
2.1.2 Phytophthora fruit rot
(Rebecca Grumet and Ying-Chen Lin)
GWAS
The cucumber core collection (370 accessions for which data were collected) was screened from 2019 to 2021; 70% of the accessions had at least two years of data. The score for each accession is the mean of 30-50 fruits/year. Disease scores for the population were normally distributed. The correlation between years ranged from 0.48-0.80.
GWAS was performed using BLUE values calculated from disease scores from 2019-2021 using MLM, FarmCPU, BLINK and MLMM models (Figure 4, Table 1). SNP data for the core collection was downloaded from CucGenDBv2 (Yu et al., 2023). SNPs were filtered using BCFtools (Danecek et al., 2021) and GATK (Van der Auwera et al., 2013) with the following criteria: bi-allelic, GQ scores >20, maximum read depth within two standard deviations of the mean read depth, minor allele frequency > 0.1, missing rate <20%, resulting in 1,168,270 SNPs for analysis. Eleven significant SNPs (Bonferroni-corrected genome-wide significance threshold at α = 0.05) were identified from the different models. The phenotype variance explained (PVE) ranged from 0.38-24.49%.
Table 1. Significant SNPs identified in multiple GWAS models (FarmCPU, Blink, MLMM, and MLM) for young fruit resistance in the cucumber core collection.
Figure 4. (A) Disease score distribution for young fruit resistance to Phytophthora capsici from the cucumber core collection, and BLUE distribution of combined data 2019-2021. The score for each accession is the mean of 30 50 fruits/year. (B) Manhattan plots and quantile–quantile plots of the genome-wide association study analyses for young fruit resistance in the cucumber core population. The horizontal blue and red lines represent significance thresholds of p = 0.05 and p = 0.01 based on Bonferroni correction, respectively. The dotted vertical lines show the locations of SNPs that were significant in at least two models.
Phenotypes were significantly different between accessions carrying homozygous reference vs. alternate alleles for all significant SNPs except S3_18480786. Of the nine SNPs, five alternate alleles led to increased resistance (lower disease scores) and four to increased susceptibility. Of the alternate alleles conferring increased resistance, only two were present in PI 109483-derived breeding line ‘A4-3’ (SNP S1_21117743 and S3_37752706), suggesting that the other alleles identified by GWAS may provide additional sources of resistance. When the alternate alleles associated with lower disease scores were rare in the core collection (< 10%, i.e., < 38 accessions), the majority of accessions (64%-81%) carrying the alternate allele originated from the India/South Asia region (Table 2). Conversely, four of the five SNPs associated with increased resistance were very uncommon in the East Asian accessions (0-3%). For S2_10226744, where the rare alternate allele was associated with increased susceptibility, 77% of the accessions were from East Asia. When the alternate alleles occurred frequently in the germplasm (>50%) (e.g., S3_37752706 and S7_3391182), the origins were widely distributed across regions.
Table 2. Geographical origin of accessions carrying the alternate alleles for the significant SNPs for young fruit resistance to P. capsici as identified by GWAS.
XP-GWAS
Extreme phenotype (XP) GWAS was performed to enable additional replication of phenotypic data. Weighted disease scores from 2019-2021 data were used to select the 29 most resistant and susceptible accessions that were then tested again in 2022. The clear difference in disease scores between the bulks was reproduced in the replicated trial in 2022 (Figure 5A,B), verifying accuracy of the bulk selection for XP-GWAS analysis. Correlations for the selected resistant and susceptible bulks among 2019, 2021, 2022 were 0.755-0.912. SNP data from the selected accessions were
combined via in-silico bulking. XP-GWAS analysis identified significant SNPs distributed across the seven chromosomes (Bonferroni corrected p=0.05 threshold) including 39 significant SNPs located on chromosomes 1 and 5 (Figure 5C). The XP-GWAS SNP identified on chromosome 5 overlapped with the QTL previously identified by QTL-seq.
The QTL peaks detected on chromosome 1 were located at 21.17 Mb by GWAS and 24.75 Mb by XP-GWAS. The QTL on chromosomes 5 and 6 were consistently identified by QTL-seq, GWAS and XP-GWAS methods (on chromosome 5 all were located within 7 Mb, and on chromosome 6 all were within 3 Mb). The signal on chromosome 5 was stronger in QTL-seq and XP-GWAS compared to GWAS, possibly due to the use of the resistant line ‘A4-3’ in the bi
parental QTL-seq analysis, and the enrichment of rare alleles in the resistant bulk for XP-GWAS analysis. On chromosome 3, the peak SNP was located at 37.75 Mb in GWAS and at 39.29 Mb in XP-GWAS. Both fall within the QTL region previously identified from QTL-seq analysis for age-related resistance.
Figure 5. Disease score distribution and Manhattan plot of the XP-GWAS analysis to identify SNPs associated with young fruit resistance. (A) Disease score distribution of the resistant and susceptible bulks in different years. (B) Disease score values of the resistant (R), susceptible (S), and random bulks (**** indicates P<0.0001, Wilcoxon test). (C) Manhattan plot of the XP-GWAS analysis. The dashed line indicates the 5% FDR threshold; the solid line indicates significance threshold of p = 0.05 based on Bonferroni correction.
In most cases, the QTL identified for Phytophthora fruit rot co-localized with prior identified disease resistance hot spots (Wang et al. 2020) for resistances to downy mildew, powdery mildew, fusarium wilt, and gummy stem blight (Figure 6). The aggregation of these QTL suggests that these genomic regions play an important role in disease resistance to fungal and
oomycete pathogens.
Figure 6. Chromosomal locations of QTL identified for Phythophthora fruit rot of cucumber in relation to prior QTL identified for resistances to other cucumber diseases. The indicated PFR QTL were identified from multiple analyses: red bar – biparental QTL-seq; blue bar – fine mapping of biparental populations; red asterisk GWAS of cucumber core collection; blue asterisk XP-GWAS; purple asterisk – candidate gene identified by RNAseq analyses. Figure is adapted from Wang et al., 2020 Horticulture Research under Creative Commons license http://creativecommons.org/licenses/by/4.0/. Abbreviations: DM, downy mildew; PM, powdery mildew; ALS, angular leaf spot; Foc, fusarium wilt; GSB, gummy stem blight; MYSV, melon yellow spot virus; CYSDV, cucurbit yellow stunting disorder virus.
2.1.3 CGMMV (KL and YW)
Cucurbit green mottle mosaic virus (CGMMV) is an emerging seed-borne virus in North America. CGMMV causes serious disease symptoms and losses in cucurbits, particularly cucumber and watermelon. As an effort to combat CGMMV in cucumber and watermelon, genetic resources were explored to develop genetically resistant/tolerant cucumber and watermelon lines in this study. Initially a total of 50 cucumber lines were screened to assess phenotypic reactions to the CGMMV infection (Figure 7). As a result, three lines were identified as tolerant with no phenotypic symptoms but intermediate serological reactions. The three tolerant lines identified seems to have a common origin (‘Chinese Long’). The selected tolerant lines were crossed with susceptible cucumber lines to develop F1, F2 and RILs. The F2 populations developed from the tolerant and susceptible lines will be evaluated further for SNP genotyping.
In another experiment, a GWAS panel of 177 cucumber accessions were evaluated for their resistance to CGMMV. The GWAS panel showed diverse responses to CGMMV infection resulting in 12 lines in tolerant, 24 being intermediate and 137 with susceptibility (Figure 8). Phenotyping results are being validated by genotypic analysis to obtain the associated QTLs for resistance to CGMMV.
Disease Severity Index Classes
Figure 7. Rating classes of cucumber infected by CGMMV, rating 0: no symptom, rating 1: mild mosaic symptom, plant recovery; rating 2: severe mottle mosaic and rating 3: severe mottling and plant stunting.
Figure 8. Phenotypic distribution of accessions to CGMMV testing among the GWAS panel.
Phenotypic screening was conducted using 6 plants/line and symptom reading scoring of 0-3, with 0: no symptom, rating 1: mild mosaic symptom, plant recovery; rating 2: severe mottle mosaic and rating 3: severe mottling and plant stunting. Tolerant is lines with a disease severity index (DSI) = <25; Intermediate is lines with a DSI = >25 – <50, and Susceptible is lines with a DSI = >50.
2.2 Marker development and verification
We conducted fine mapping of the major-effect DM QTL, dm4.1, and dm5.3, and introgress them into different genetic backgrounds through marker-assisted QTL pyramiding. We found that there are actually four sub-QTL at the dm4.1 locus that are present in both WI7120 and PI 197088 including dm4.1.1, dm4.1.2A, dm4.1.2B, and dm4.1.3. The candidate genes for dm4.1.2A and dm4.1.3 in PI 197088 have been identified previously (Berg et al. 2020, 2021). So we focued on cloning of the dm4.12B QTLfrom WI7102 which was delimited into a 36.2 kb region on Chromosome 4. With multiple lines of evidence, we show that the gene for the L-2-hydroxyglutarate dehydrogenase (L-2HGDH) is a candidate for dm4.1.2B (Figure 9).
Figure 9. Map-based cloning of the dm4.1.2B sub-QTL. A. Eight dm4.1.2B NIL-F2 recombinants defined by 11 marker loci delimit dm4.1.2B into a 36.3 kb interval on Chr 4. B. DM symptoms of two critical recombinants 4-94 and 4-97 scored at 7, 10 and 14 days post inoculation (dpi) at either adaxial (ada), abaxial (aba) or both sides. C. The 36.6kb region contains four predicted genes with gene #4 as the best candidate of dm4.1.2B, which encodes L 2HGDH. There are two non-synonymous SNPs in the coding region (red triangles) and a 53 bp insertion (blue triangle) in WI7120 that is located at -79 bp upstream of the translation start (TSS) of L-2HGDH. D. Alignment of deduced amino acid sequences (partial) of L-2HGDH among Gy14, 9930 and WI7120. The 1st and 2nd SNPs would result in M1K (change of start codon) and V18A amino acid substitutions, respectively.
There are multiple polymorphisms inside the promoter region of dm4.1.2B between PI 197088 and susceptible 9930 parental lines including a 52-bp insertion in PI 190788. DM resistance conferred by two other sub-QTL, dm4.1.2A and dm4.1.3 have been previously shown to be due to insertion of a 551bp and a 7,688 bp transposon insertion, respectively (Berg et al. 2020, 2021). We examined the allelic diversity of three the Sub-QTL among cucumber accessions with high resistance to DM (Figure 10; Table 3) which revealed commonalities and differences of alleles carried by these cucumber lines.
Figure 10. Sequence alignment of dm4.1.2B locus revealed association of the 52-bp deletion in the promoter region and DM resistance among cucumber lines with DM resistance.
Table 3. Correspondence of alleles at dm4.12A, dm4.1.2B and dm4.1.3 Sub-QTL with DM resistance among different cucumber lines.
Objective. 3. QTL introgression into breeding or advanced lines
(Yiquin Weng, Rebecca Grumet and Anthony Keinath)
One objective of this project is to develop inbred lines with both DM and PFR (Phytophthora fruit rot) resistances through marker-assisted QTL pyramiding. So far we have completed introgression of three DM QTL (dm4.1, dm5.2, dm5.3) into Gy14 (pickle), 9930 (Asian Long), and WI7204 (mini) backgrounds, which were named Gy14Q3, 9930Q3 and WI7204Q3, respectively. We further introduced the major-effect QTL for PFR resistance, qPFR5.1 into Gy14, which were in repulsive phase with dm5.2 and dm5.3. We took a revised marker-assisted backcrossing strategy and identified ideal recombinants carrying alleles for all four resistance loci (dm4.1, dm5.2, qPFR5.1, and dm5.3) in homozygous states.
Figure 9. DM resistance of Gy14Q3 (left and middle) and 9930Q3 under heavy P. cubensis infection in Hancock field in 022 summer season.
DM resistances of these introgression lines was observed in 2022 and 2023 filed trials. In 202 Hancock field with heavy DM infection, both 9930Q3 and Gy14Q3 showed much higher resistance than lines donor lines (Figure 9). However, in 2023 Spring field trial at Clemson, South Carolina, Gy14Q3 and 9930Q3 only shoed slightly high resistance than their respective donors that also depend on the time after infection (Table 4). Overall, compared with the original donor of these QTL (PI 197088), the resistance conferred by the three QTL does not seem to satisfactory). In particular, from multiple years’ observation, WI7024 was highly susceptible to DM. We also tested responses to infection by the PFR pathogen in Gy14Q4, which did not show significantly higher resistance than the donor (Gy14 or Gy14Q3). Whether this is due to wrong genotyping or negative genetic background effects (linkage drag) needs further investigation.
Table 4. Disease scores of DM of selected introgression lines and controls under natural infection (2023 Clemson)
For the fine mapped/cloned DM/PFR resistance QTL (dm1/CsSGR, dm4.1, dm5.2, dm5.3 and qPFR5.1), we are developing SNP assays adapted to high throughput SNP genotyping. Figure 10 is an example for dm5.3/CsSIB1 with the KASPar genotyping platform.
