Background

Cucumber (Cucumis sativus) cultivated worldwide is an economically important vegetable crop, and its fruit is consumed fresh or pickled [1]. Cucumber is understood to have originated in India and was domesticated in Asia approximately 3,000 years ago [2]. The global production of cucumber was 93.5 million tonnes in 2021, and the largest cucumber producer is China followed by Turkey, Russia, Ukraine, and Mexico [3].

The genus Begomovirus (family Geminiviridae) contains 445 virus species [4], each of which carries a circular, monopartite or bipartite, single-stranded DNA genome [5]. Most begomoviruses have bipartite genomes organized of two circular DNA components (DNAs A and B), each approximately 2,800 nucleotides (nt) in length [6]. A bipartite begomovirus tomato leaf curl New Delhi virus (ToLCNDV) that infects Solanaceae crops (tomato [Solanum lycopersicum], pepper [Capsicum spp.], eggplant [Solanum melongena], potato [Solanum tuberosum]) and cucurbit crops (cucumber, melon [Cucumis melo], zucchini squash [Cucurbita pepo], luffa [Luffa spp.]) [7, 8], is the second most important begomovirus after the monopartite tomato yellow leaf curl virus (TYLCV) and has a high economic impact on horticultural production. ToLCNDV was first detected in India in 1995, and its isolation has subsequently been reported across the Indian subcontinent (India, Pakistan, and Bangladesh) and Southeast Asia (Thailand, Laos, Indonesia, and Taiwan) [9,10,11,12,13,14,15]. ToLCNDV mainly threatens the production of cucurbit crops such as cucumber, melon, squash (Cucurbita spp.), bottle gourd (Lagenaria siceraria), Luffa spp., and Sechium edule in Asian countries [9, 11, 16,17,18]. Based on reports from Iran and Turkey in the Middle East and from Spain, Portugal, Italy, Greece, Morocco, Tunisia, and Algeria in the Mediterranean Basin, from 2012 onward, ToLCNDV has spread westward, affecting cucurbit crops such as zucchini squash, melon, and cucumber [19,20,21,22,23,24,25,26,27,28].

The insect vector, whitefly Bemisia tabaci (Hemiptera: Aleyrodidae), has driven the remarkable emergence of begomoviruses [29]. In tropical and subtropical regions, the distribution and polyphagous feeding habits of B. tabaci critically affect the prevalence and economic importance of begomoviral diseases. Generally, insecticides that target B. tabaci populations are used to control begomoviruses caused diseases; however, insecticide resistance in B. tabaci has emerged by intensive and unregulated use of insecticides [30, 31]. An integrated pest management approach which uses begomovirus-resistant cultivars can be an effective alternative to control begomoviruses. Several begomovirus resistance genes have been cloned from Solanaceae crops using elaborate forward and reverse genetic methods [32,33,34,35,36]. Among the identified resistance genes, Ty-1/Ty-3 in tomato and Pepy-2 in capsicum encode DFDGD-class RNA-dependent RNA polymerases (RDRs) and are dominant resistance genes to multiple begomoviruses [33, 35]. Furthermore, it was demonstrated that the replication of begomoviral DNA is restricted by RDR via a transcriptional gene silencing pathway [37]. Breeding for begomovirus-resistant tomato plants has advanced, and Ty-1 has been introgressed into most of the TYLCV-resistant F1 hybrid cultivars bred worldwide [38, 39]. Several studies have reported ToLCNDV resistance in cucurbits, such as melon, C. moschata, and L. cylindrica accessions [40,41,42,43]. In cucumber, ToLCNDV-resistant materials and the detection of quantitative trait loci (QTLs) on chromosomes 1 and 2 have been described in international patents from Vilmorin & Cie, Rijk Zwaan, and Nunhems [44,45,46] More recently, additional ToLCNDV-resistant cucumber accessions were reported and a QTL for resistance was mapped on chromosome 2 [47]. However, no resistance genes have been identified.

In our preliminary study, a total of 575 cucumber accessions were screened for ToLCNDV resistance by agroinfiltration of a Spanish isolate of ToLCNDV (ToLCNDV-ES), and No.44 was selected as one of the resistance sources. In this study, ToLCNDV resistance of a cucumber accession No.44 were evaluated. Reliable QTLs were detected on chromosomes 1 and 2 by two rounds of phenotyping with two independent F2 populations generated by crossing the begomovirus-susceptible cultivar SHF and No.44. Further fine-mapping and functional analysis of a candidate gene on chromosome 1 identified CsRDR3 as the gene responsible for ToLCNDV resistance at the QTL.

Results

No.44 is resistant to ToLCNDV isolates from the Mediterranean Basin and Southeast Asia

The resistance of cucumber accession No.44 to two ToLCNDV isolates was evaluated by agroinfiltration. In the first experiment, agroinfiltration of ToLCNDV-ES resulted in 100% infection in the susceptible cultivar ‘Sagami Hanjiro Fushinari’ (SHF), and the plants consistently exhibited severe symptoms, with an average disease severity index (DSI) score of 4 (Fig. 1A, Table 1). Moreover, only 26% of the inoculated No.44 plants were virus-infected, and the infected plants showed no symptoms, with an average DSI score of 0 at 30 days post inoculation (dpi). F1 showed moderate resistance to ToLCNDV-ES, with an average DSI score of 2.8 at 30 dpi (Table 1). In the second experiment, ToLCNDV-ES was agroinfiltrated into No.44 plants and another susceptible cultivar ‘Natsu Suzumi’ (NS) (Table 1). Again, No.44 was resistant to ToLCNDV-ES, while NS was susceptible to virus infection. Moreover, the accumulation of ToLCNDV-ES viral DNA was significantly lower in No.44 plants than in NS plants (Fig. 2A). In the third experiment, agroinfiltration of ToLCNDV-IDN [15] with higher pathogenicity induced severe symptoms in susceptible SHF plants (Fig. 1B, Table 1). Meanwhile, No.44 was resistant to ToLCNDV-IDN with an average DSI score of 1.4 at 30 dpi. Moreover, the accumulation of ToLCNDV-IDN viral DNA was significantly lower in No.44 plants than in SHF plants (Fig. 2B). From these results, we concluded that No.44 is resistant to ToLCNDV isolates from multiple cucumber-producing regions, including the Mediterranean Basin and Southeast Asia.

Fig. 1
figure 1

No.44 (resistant) and ‘Sagami Hanjiro Fushinari’ (SHF; susceptible) plants infected with ToLCNDV. A ToLCNDV-ES-agroinfiltrated SHF and No.44 plants at 30 days post inoculation (dpi). B ToLCNDV-IDN-agroinfiltrated SHF and No.44 plants at 30 dpi. C ToLCNDV-ES-infected SHF and No.44 plants at 30 days after graft inoculation. D Ninth and 20th leaves of SHF and No.44 plants at 60 days after graft inoculation

Full size image
Table 1 Evaluation of tomato leaf curl New Delhi virus (ToLCNDV) resistance in No.44 and F1 population obtained by crossing with SHF
Full size table
Fig. 2
figure 2

Quantification of ToLCNDV viral DNA in the leaves of cucumber No.44 and susceptible cultivars. Relative amount of viral DNA accumulated in A ToLCNDV-ES-agroinfiltrated NS and No.44 plants, B ToLCNDV-IDN-agroinfiltrated SHF and No.44 plants, C ToLCNDV-ES-infected SHF and No.44 plants after 30 days from graft inoculation, and D plants after 60 days from graft inoculation. Only ToLCNDV-infected plants were used for this analysis. NS and SHF indicate the begomovirus-susceptible cultivars ‘Natsu Suzumi’ and ‘Sagami Hanjiro Fushinari’. Young upper leaves of ToLCNDV-infected plants were collected at 15 and 30 days after agroinfiltration and 30 and 60 days after grafting for analysis of viral DNA accumulation by qPCR. Normalized relative viral DNA values were calculated using the 25S rRNA gene. The data are presented as the mean ± standard error (SE). Significant differences between means (Student’s t test; p < 0.05) are indicated with asterisks

Full size image

ToLCNDV-ES was transmitted from a symptomatic NS scion to healthy No.44 or SHF rootstock and evaluated for resistance at 30 and 60 days after grafting. ToLCNDV-ES infection of the NS scion was confirmed by PCR before grafting. At 30 days after grafting, No.44 plants exhibited vigorous growth compared with the susceptible SHF plants (Fig. 1C). The average DSI values of the fifth and 10th true leaves of No.44 plants were significantly lower than those of their counterparts in SHF plants (Table 2). Moreover, at 60 days after grafting, intense yellowing was observed on the ninth true leaf of SHF plants, but only a few spots with mild yellowing were observed on the leaves of No.44, and the average DSI was significantly lower for No.44 than for SHF (Fig. 1D, Table 2). On the 20th true leaf, symptoms were relatively mild in both No.44 and SHF, but the average DSI was significantly lower in No.44 than in SHF (Table 2). Quantification of ToLCNDV-ES DNA in leaves revealed that significantly less ToLCNDV-ES DNA accumulated in the fifth and 10th true leaves of No.44 than in those of SHF at 30 days after grafting (Fig. 2C). At 60 days after grafting, there was no significant difference in the amount of viral DNA in the ninth true leaf between No.44 and SHF plants, but there was significantly less viral DNA in the 20th true leaf of No.44 than in its counterpart in SHF (Fig. 2D). Taken together, these results show that No.44 is resistant to ToLCNDV-ES not only at the young seedling stage but also at the fruit-setting stage.

Table 2 Evaluation of tomato leaf curl New Delhi virus (ToLCNDV) resistance in No.44 by viral graft transmission
Full size table

Genetic mapping of ToLCNDV-ES resistance in F2 populations

For the genetic mapping of ToLCNDV-ES resistance, two rounds of phenotyping were conducted with two independent F2 populations generated by crossing the begomovirus-susceptible cultivar SHF and No.44. In the first F2 population, 187 out of the 203 individuals agroinfiltrated with ToLCNDV-ES tested positive (92%) for virus infection at 30 dpi, as determined by PCR. These ToLCNDV-ES-positive plants were analyzed further. At 30 dpi, phenotypic segregations were as shown in Fig. S1A, and number of resistant individuals (DSI 0–1) to susceptible individuals (DSI 2–4) segregated into 1:3 for the F2 population (n = 187), as shown via the χ2 test, indicating that ToLCNDV-ES resistance was controlled by a single recessive gene (Table S1). Linkage analysis of ToLCNDV-ES resistance in the first round of the SHF × No.44 F2 population (n = 187) was performed using 309 SNPs discovered by restriction site-associated DNA sequencing (RAD-seq). The number of linkage groups was the same as the chromosome number of cucumber (C. sativus), and the total size of the linkage map was 647.8 cM (average marker distance = 2.1 cM). Two significant QTLs, one on chromosome 1 and the other on chromosome 2, were detected by a composite interval mapping (CIM) analysis of the SHF × No.44 F2 population; these QTLs were denoted as cucumber yellow leaf curl disease resistance-1 (Cy-1) and Cy-2, respectively (Fig. 3A, Table 3). The QTLs explaining less than 10% of the phenotypic variation were defined as minor QTLs, whereas the QTLs explaining more than 10% of the phenotypic variation were defined as major QTLs according to previous report [48]. The highest peaks and narrowest intervals of QTLs were detected for the DSI score at 30 dpi. A major QTL Cy-1 with logarithm of odds (LOD) score of 15.1 was detected at the physical position of 24,837,399 on chromosome 1 of the reference sequence (Chinese Long, ver. 3), and the other major QTL Cy-2 (LOD score = 22.2) was detected at the physical position of 17,866,822 on chromosome 2. Cy-1 and Cy-2 explained 20.0% and 30.7% of the total phenotypic variation, respectively.

Fig. 3
figure 3

Linkage analysis of the logarithm of the odds (LOD) scores of ToLCNDV-ES resistance in F2 populations generated by crossing the begomovirus-susceptible cultivar ‘Sagami Hanjiro Fushinari’ (SHF; susceptible) and No.44 (resistant). The red line plotted indicates the LOD scores of the disease severity index (DSI) at 30 days post inoculation (dpi) and the blue line indicates the level of ToLCNDV-ES viral DNA accumulation at 30 dpi for A first round of the SHF × No.44 F2 population (n = 187) and B second round of SHF × No.44 F2 population (n = 143)

Full size image
Table 3 Quantitative trait loci (QTL) identified in the F2 populations of SHF × No.44
Full size table

In the second F2 population, 143 out of 149 individuals tested positive (96%) for ToLCNDV-ES infection at 30 dpi. Phenotypic segregations were as shown in Fig. S1B, and the segregation ratio indicated that ToLCNDV-ES resistance was conferred by a single recessive gene (Table S1). Linkage analysis of ToLCNDV-ES resistance was performed using 455 SNPs discovered by RAD-seq (n = 143). The number of linkage groups was the same as the chromosome number of cucumber, and the linkage map size was 623.3 cM (average marker distance = 2.0 cM). A CIM analysis of the SHF × No.44 F2 population detected three significant QTLs on chromosomes 1, 2, and 6 (Fig. 3B, Table 3). The highest peaks and narrowest intervals of QTLs were detected for the DSI score at 30 dpi. Among the three QTLs detected, two major QTLs had genomic regions that overlapped with Cy-1 and Cy-2 identified in the first round of QTL analysis. The Cy-1 QTL (LOD score = 13.7) was detected at position 24,211,834 on chromosome 1, and the Cy-2 QTL (LOD score = 21.6) was detected at position 17,695,694 on chromosome 2. Cy-1 and Cy-2 explained 18.0% and 36.3% of the total phenotypic variation, respectively. Additionally, the third minor QTL Cy-6 (LOD score = 8.1) located at the physical position 11,725,560 on chromosome 6 explained 8.5% of the total phenotypic variation.

The major QTLs Cy-1 and Cy-2 that were consistently detected in the two rounds of genetic mapping were further analyzed. Individuals with different allelic combinations of Cy-1 and Cy-2 were identified from the first-round F2 population on the basis of the genotypes of the closest markers at the LOD peaks. Plants homozygous for the No.44 allele at marker S2_17866822 on chromosome 2 were more resistant to ToLCNDV-ES and had lower virus titers than plants heterozygous or homozygous for the SHF allele. These results indicated that Cy-2 acts recessively to increase resistance to ToLCNDV-ES (Fig. S2A and B). In light of these results, this locus is hereinafter referred to as cy-2. In contrast, except for a single case, plants homozygous and heterozygous for the No.44 allele at marker S1_24837399 on chromosome 1 were more resistant to ToLCNDV-ES and had lower virus titers than those of plants homozygous for the SHF allele, indicating that Cy-1 acts dominantly (Fig. S2A and B). In addition, epistatic interaction between Cy-1 and cy-2 was detected (p = 0.017) for viral DNA accumulation (Fig. S2B). The effect of the No.44 allele at Cy-1 to increase resistance (i.e., decrease DSI score) was observed in two genotype classes, homozygous for the No.44 allele at the cy-2 and heterozygous, but not in the class homozygous for the SHF allele at the cy-2. In the second-round F2 population, a similar dominant nature of Cy-1 with the first-round F2 population was detected except for homozygous for the No.44 allele at the cy-2 (Fig. S2C). No statistically significant epistatic interaction between Cy-1 and cy-2 was detected in this case (p = 0.460). The overlapping regions of Cy-1 and cy-2 identified in the two rounds of linkage analyses included 468 and 573 candidate genes, respectively (Table S2).

Fine mapping of the ToLCNDV-ES resistance QTL Cy-1

We further finely mapped the ToLCNDV-ES resistance QTL Cy-1 (Fig. 4). Two KASP markers (Cuc24327273-KASP and Cuc25380586-KASP) were developed and five recombinants (No.8, No.12, No.270, No.420, and No.437) were screened from the newly prepared F2 population (n = 752) to narrow down the candidate region. The recombination points were verified using 13 additional markers, consisting of eight KASP and five high-resolution melting (HRM) markers. F3 populations were obtained by self-pollinating these F2 recombinants. The phenotypes of three F3 populations derived from the F2 individuals, No.270, No.437, and No.12, did not fit the genotypes of three markers, namely Cuc24895001-KASP, Cuc24978218-KASP, and Cuc25061147-HRM, respectively. While the phenotypes of two F3 populations derived from the F2 individuals, No.8 and No.420, did not fit the genotypes of two markers, namely Cuc25270268-KASP and Cuc25380586-KASP, respectively. From these results, the ToLCNDV-ES resistance QTL Cy-1 was mapped to a 209-kb region between the Cuc25061147-HRM marker and the Cuc25270268-KASP marker. In the Chinese Long genome (ver.3), 24 putative genes were identified in this target region (Table 4). Whole-genome resequencing of the two parents revealed that, compared with those of SHF and Chinese Long, only two genes, CsaV3_1G039730 and CsaV3_1G039870, had nonsynonymous substitutions in their open reading frames (ORFs) in No.44. Furthermore, the transcript level of CsaV3_1G039730 was significantly greater in ToLCNDV-ES-infected and mock-inoculated No.44 than in ToLCNDV-ES-infected and mock-inoculated SHF plants, as determined by RNA-seq analysis (Table 4). On the basis of these results, CsaV3_1G039730, which encodes an RNA-dependent RNA polymerase (RDR), was identified as a potential candidate gene conferring resistance to ToLCNDV-ES within QTL Cy-1 linked to ToLCNDV-ES resistance.

Fig. 4
figure 4

Fine mapping of candidate genes at Cy-1 on chromosome 1. Cy-1 was mapped between the Cuc25061147-HRM and Cuc25270268-KASP markers on chromosome 1. The red, blue, and gray bars indicate No.44 allele (homozygous dominant), SHF allele (homozygous recessive), and heterozygous genotype, respectively. F3 populations derived from the selfing of F2 recombinants (No.8, No.12, No.270, No.420, and No.437) were genotyped using one of three markers, namely, Cuc25061147-HRM, Cuc25193180-KASP, or Cuc25380586-KASP, as indicated by the yellow circle in the figure, to assess the genotype of the target region. On the right side of the figure, the genotype and ToLCNDV-ES resistance phenotype are indicated for F3 population: red, blue, and gray indicate No.44 allele (homozygous recessive), SHF allele (homozygous dominant), and heterozygous genotype, respectively. On the lower side of the figure, 24 predicted genes are indicated as black boxes. SHF indicates the begomovirus-susceptible cultivar ‘Sagami Hanjiro Fushinari’. Different letters indicate significant differences in the disease severity index (DSI) between the means of F3 individuals (Bonferroni–Dunn test; p < 0.05)

Full size image
Table 4 Predicted ORFs located in the candidate region
Full size table

Analysis of CsRDR3 encoding an RNA-dependent RNA polymerase

The full-length sequences of CsaV3_1G039730, CsRDR3, from No.44 and SHF were isolated by reverse-transcription PCR (RT‒PCR) and analyzed. These analyses revealed that CsRDR3 consisted of 19 exons (Fig. 5A). CsRDR3 showed high sequence similarly with SlRDR3 (Ty-1/Ty-3) of tomato and CaRDR3a (Pepy-2) of capsicum (Fig. 5B). Seven nonsynonymous substitutions were detected in the amino acid sequence of CsRDR3 in No.44 compared with its counterpart in SHF (Fig. 5B). In addition, compared with those in SHF, an approximately 1.9-kb deletion and a 62-bp insertion were identified in the CsRDR3 promoter region in No.44 (Fig. 5A). A BLAST analysis of this 1.9-kb fragment in the genomic DNA of SHF and Chinese Long revealed no similarity to any annotated sequences, including transposable elements. Multiplex PCR successfully amplified a 454-bp band from plants homozygous for the No.44-allele, an 805-bp band from plants homozygous for the SHF-allele, and 454-bp and 805-bp bands from heterozygous plants (Fig. 5A). This co-dominant DNA marker can be used for genotyping the resistant and susceptible alleles of CsRDR3. According to the results of the phylogenetic analysis, the RDRs from different plant species formed two clusters, namely α and γ -clades of RDR (Fig. 6). Phylogenetic analysis showed that the CsRDR3s of No.44 and SHF exhibited high amino acid sequence similarity with RDR3, RDR4, and RDR5 of Arabidopsis thaliana, which are γ-clade RDRs. The RDRs of tomato (SlRDR3), pepper (SlRDR3a), potato (StRDR3), and tobacco (NtRDR3) constituted an independent clade from the CsRDR3s within the γ-clade RDRs.

Fig. 5
figure 5

Molecular genetic identification of CsRDR3. A Gene structure of CsRDR3 in ‘Sagami Hanjiro Fushinari’ (SHF; susceptible) and No.44 (resistant). Closed black boxes indicate exons identified from comparisons of the transcript and genome sequences of CsRDR3. In No.44, the promoter region of CsRDR3 had a 1.9-kb deletion and 62-bp insertion. Agarose gel showing the PCR amplicons of CsRDR3 dereived from No.44, SHF, and F1 plants using an Indel marker targeting the promoter region of CsRDR3. L indicates the 1-kb DNA ladder. Uncropped full length gel image is provided in the Supplementary Information as Supplementary Fig. S3. B Multiple sequence alignment of CsRDR3 in No.44 and SHF and its homologs SlRDR (Solanum lycopersicum) and CaRDR3a (Capsicum annuum) using Clustal Omega. The red underline represents the DFDGD motif, and the red colored letters indicate amino acid substitutions in No.44 compared with those in SHF

Full size image
Fig. 6
figure 6

Neighbor-joining tree of the protein sequences of CsRDR3 and RDRs of other plant species. Bootstrap values were calculated based on 1000 replicates and are indicated at nodes

Full size image

Virus-induced gene silencing of CsRDR3

All the analyses so far support CsRDR3 as a strong candidate for the gene conferring ToLCNDV-ES resistance at the QTL Cy-1. To further verify this hypothesis, we conducted virus induced gene silencing (VIGS) of CsRDR3 to confirm its involvement in ToLCNDV-ES resistance. In an initial experiment in which CsRDR3 was silenced in No.44 by VIGS, changes in resistance were not detected because of the effect of another major ToLCNDV-ES resistance QTL cy-2 on chromosome 2. To address this problem, we used two chromosome segment substitution lines, No.148 and No.489, which were homozygous for the No.44 genotype at the target region on chromosome 1 and homozygous for the SHF genotype at the target region on chromosome 2. Moreover, we used CsRDR3-specific gene fragments as inserts for the apple latent spherical virus (ALSV) vector to avoid the off-target effects of VIGS.

For VIGS of CsRDR3 in No.148, the ALSV construct pBICAL2::CsRDR3-189 was used. The characteristic photobleaching effect appeared in the topmost leaves after 28 dpi and was subsequently enhanced in the phytoene desaturase (PDS) gene-silenced No.148 plants (Fig. 7A). Plants with mixed ToLCNDV-ES and ALSV infections, as confirmed by PCR, were selected for further analysis. The control No.148 plants with mixed infection of ToLCNDV-ES and wild-type ALSV were resistant to ToLCNDV-ES. In contrast, the CsRDR3-silenced No.148 plants exhibited mild leaf-yellowing symptoms and accumulated nine times more ToLCNDV-ES viral DNA, which was significantly greater than that of the control plants. The transcript levels of CsRDR3 tended to be lower in CsRDR3-silenced No.148 plants than in control plants at 40 dpi, but with no statistical differences. Although the first experiment partially supported our hypothesis that CsRDR3 is involved in ToLCNDV-ES resistance, further experiments are needed.

Fig. 7
figure 7

Functional analysis of CsRDR3 by virus-induced gene silencing (VIGS). Effect of VIGS on ToLCNDV-ES caused symptoms, relative ToLCNDV-ES viral DNA accumulation levels, and CsRDR3 transcript levels in A No.148 and B No.489 plants at 40 days post inoculation (dpi). Lines No.148 and No.489 were homozygous for the No.44 (resistant) genotype in the target region on chromosome 1 containing Cy-1 and for the ‘Sagami Hanjiro Fushinari’ (SHF; susceptible) genotype in the target region on chromosome 2 containing cy-2. Young upper leaves of ToLCNDV-ES-infected plants were collected at 40 dpi for analysis of viral DNA accumulation by qPCR. Normalized relative viral DNA values were calculated using the 25S rRNA gene. The numbers of biological replicates are indicated on each bar. The data are presented as the mean ± standard error (SE). Significant differences between means (Student’s t test; p < 0.05) are indicated with asterisks and N.S. indicates not significant

Full size image

Since the efficiency of VIGS by ALSV is reportedly affected by the genotype of the host plant and the size of the fragment inserted into the virus vector [49, 50], we used another chromosome segment substitution line, No.489, and ALSV construct pBICAL2::CsRDR3-102, which has a shorter fragment from a different part of the CsRDR3 gene (Fig. 7B). The PDS gene-silenced No.489 plants started to show a photobleached phenotype in the developing leaves at 28 dpi. Since the growth of No.489 was more vigorous than that of No.148 under the same experimental conditions, we cut back the plants at 28 dpi and used the newly elongating lateral branches for analyses. At 40 dpi, the PDS-silenced No.489 plants exhibited a highly uniform photobleached phenotype. The control No.489 plants with mixed infection of ToLCNDV-ES and wild-type ALSV were resistant to ToLCNDV-ES. In contrast, the CsRDR3-silenced No.489 plants exhibited severe leaf curling and yellowing symptoms. In addition, there were 51-fold more viral DNA of ToLCNDV-ES accumulated in the CsRDR3-silenced No.489 plants than in the control plants. Moreover, the CsRDR3 transcript level was significantly lower in the CsRDR3-silenced No.489 plants than in the control No.489 plants. We also performed VIGS for other candidate genes that had either nonsynonymous substitution in the ORF or difference in gene expression in ToLCNDV-ES-infected plants (Table 4), such as CsaV3_1G039720, CsaV3_1G039750, CsaV3_1G039790, CsaV3_1G039810, CsaV3_1G039870, and CsaV3_1G039910, but no loss-of-resistance phenotypes were observed in the gene-silenced plants (supplementary Fig. S4). In conclusion, that CsRDR3 is most likely the gene underlying ToLCNDV-ES resistance at QTL Cy-1 in No.44.

Discussion

ToLCNDV was first isolated in India from diseased tomato plants [14]. Since then, its distribution has been reported from the Indian subcontinent, Southeast Asia, the Middle East, and the Mediterranean Basin [8]. The ToLCNDV isolates from Spain, Italy, Morocco, Algeria, and Tunisia in the Mediterranean Basin are monophyletic and form a nested clade within a much more diverse group of virus isolates from the Indian subcontinent. This strongly indicates that the ToLCNDV population in the Mediterranean Basin was founded by a single virus that originated from the Indian subcontinent [7, 8, 16]. A similar phylogenetic pattern was displayed for Southeast Asian ToLCNDV isolates from Thailand, Laos, Indonesia, and Taiwan. In the present study, we demonstrated that No.44 is resistant to the ToLCNDV-ES isolate from Spain and to the highly pathogenic Indonesian isolate ToLCNDV-IDN, which are the representative ToLCNDV isolates from the Mediterranean Basin and Southeast Asia, respectively.

High-throughput sequencing technology has enabled us to obtain DNA polymorphisms easily, even for non-model plant species. Meanwhile, successful genetic mapping of the target gene largely depends on accurate phenotyping of target traits which requires deliberate and laborious work but also needs improved operational efficiency. To take the balance between accuracy and efficiency, ToLCNDV-ES was inoculated to F2 individuals by agroinfiltration rather than grafting in our linkage analyses. Graft-inoculation of begomoviruses is a promising method to deliver virus to evaluating plant material as shown in Table 2 and our previous studies [32, 33, 58]. However, grafting requires laborious work and takes a relatively long time to obtain phenotype data compared to agroinfiltration. In our agroinfiltration experiments shown in Table 1, ToLCNDV-ES infected No.44 plants in 26–27%. However, ToLCNDV-ES successfully infected F2 populations in 92–96%. This may be partially due to the trained skill of the researcher through the experience of agroinfiltration, however, it was still inevitable for inoculation escapes, 6 out of 203 F2 individuals and 6 out of 149 F2 individuals in the first and second rounds of linkage analyses, which we excluded these individuals from our study through PCR-based diagnosis. These experimental procedures, visual-based 0–4 arbitrary symptom severity rating method for DSI score, and fluctuation of gene frequency in the seed bulk may have affected the segregation ratio either singly or in combination. It should be noted that two rounds of phenotyping with two independent F2 populations consistently mapped two major QTLs for ToLCNDV-ES resistance on chromosomes 1 and 2 in CIM analyses. Moreover, further fine-mapping using F3 populations narrowed the candidate region of chromosome 1 to a genomic fragment containing 24 genes. We cannot dismiss the possibility that the other 23 candidate genes, in addition to RDR in the target genomic region, also contributed to ToLCNDV-ES resistance in No.44 according to our forward genetic analysis. However, the results of whole-genome sequencing, RNA-seq analyses, and previous reports regarding begomovirus resistance in tomato and capsicum conferred by Ty-1/Ty-3 and Pepy-2 encoding RDRs [33, 35], strongly indicated that RDR is most likely the gene responsible for ToLCNDV resistance.

To strengthen the scientific basis of our hypothesis, we conducted reverse genetic analyses by using ALSV-based VIGS in cucumber. The efficiency of VIGS was greater in the second experiment in which chromosome segment substitution line No.489 and the ALSV construct pBICAL2::CsRDR3-102 were used than the first experiment, in which chromosome segment substitution line No.148 and the ALSV construct pBICAL2::CsRDR3-189 were used. There are several possible explanations for these results. Because it is reported that the effectiveness of ALSV-mediated VIGS is genotype dependent in soybean (Glycine max) [49], the difference in genotype between chromosome segment substitution cucumber line No.489 and No.148 may have affected the VIGS efficiency of CsRDR3. The length and/or position of inserted sequences are also reported to affect the efficiency of VIGS [50]. Since we used CsRDR3 at different positions and fragment lengths for pBICAL2::CsRDR3-102 and pBICAL2::CsRDR3-189, this may have resulted in differences in VIGS efficiency between the two experiments. Notably, there was good correspondence between the VIGS efficiency of CsRDR3 and the symptomology observed or the accumulating ToLCNDV-ES viral DNA. Moreover, because the VIGS of other candidate genes, such as CsaV3_1G039720, CsaV3_1G039750, CsaV3_1G039790, CsaV3_1G039810, CsaV3_1G039870, and CsaV3_1G039910, induced no loss of resistance, as did the CsRDR3 (CsaV3_1G039730)-silenced plants, we concluded that CsRDR3 is the gene responsible for ToLCNDV-ES resistance at the major QTL Cy-1.

Several begomovirus resistance genes have been cloned from Solanaceae plants using elaborate forward and reverse genetic methods. In tomato, the TYLCV resistance genes Ty-1/Ty-3 encode RDRs, Ty-2 encodes a nucleotide-binding leucine-rich repeat (NB-LRR) protein, and ty-5 encodes Pelota [34,35,36]. More recently, we identified pepy-1 encoding Pelota and Pepy-2 encoding an RDR that confer resistance to bipartite begomoviruses in capsicum [32, 33]. The resistance gene Ty-1, which is widely used in the breeding of F1 hybrid cultivars of tomato plants, also confers resistance to the curtovirus, the leafhopper-transmitted beet curly top virus, suggesting that RDR-based resistance represents broad resistance to geminiviruses [51]. Previously conducted genetic mapping of ToLCNDV-ES resistance in cucumbers revealed QTLs in the overlapping genomic region of chromosome 1, as detected in our study [44,45,46]. These previous findings indicate that Cy-1, which harbors CsRDR3, confers resistance to ToLCNDV-ES in multiple cucumber resources, not only in No.44.

In plants, the long double-stranded RNA molecules are cleaved to produce siRNAs by RNA-silencing pathways. The RDR protein participates in these pathways as a part of the antiviral defense mechanism. RDRs share a special conserved RDR catalytic domain, and eukaryotic RDRs can be grouped into RDRα, RDRβ, and RDRγ [52]. Members of the RDRβ clade are not found in plants but are conserved mainly among fungi [53]. In the genome of A. thaliana, there are six RDRs which are grouped into the RDRα and RDRγ clades [54]. Arabidopsis α-type RDRs have the DLDGD motif of eukaryotic RDRs but are not essential for the production of viral siRNAs in geminivirus-infected Arabidopsis [54, 55]. The functions of the three RDRγs in Arabidopsis with a DFDGD amino acid motif in the catalytic domain have not yet been described. In contrast, the function of γ-type RDRs related to geminivirus resistance in Solanaceae is better understood. According to our phylogenetic analysis, the cucumber CsRDR3 was clustered with γ-type RDRs, specifically SlRDR3 (Ty-1/Ty-3) of tomato and CaRDR3a (Pepy-2) of capsicum. Furthermore, it was demonstrated that the replication of begomoviral DNA is restricted by RDR via a transcriptional gene silencing pathway in tomato plants harboring Ty-1 [37]. We speculate that CsRDR3 confers ToLCNDV resistance via a similar mechanism in cucumber. Analyses of the transcript levels and sequence differences in CsRDR3 between ToLCNDV-resistant No.44 and susceptible SHF revealed higher gene transcript levels in No.44, as well as multiple amino acid substitutions in its CsRDR3a protein. Similar observations have been reported for Ty-1 and its encoded RDR [35, 56]. However, further research is needed to determine whether one or both of these differences in CsRDR3 is required for conferring resistance. Since γ-type RDR was demonstrated to confer resistance to begomovirus not only in Solanaceae but also in cucumber, other vegetable crops in cucurbit may also harbor a resistance allele of γ-type RDR. From a practical viewpoint, the multiplex PCR-based system developed in this study to distinguish between resistant and susceptible alleles of CsRDR3 will be practically beneficial for DNA marker-assisted breeding of ToLCNDV resistance in cucumber using No.44 as a resistant source.

In this study, our genetic mapping analyses revealed three QTLs for ToLCNDV-ES resistance on chromosomes 1, 2, and 6 in the cucumber genome. However, the minor QTL on chromosome 6 was detected only in the second-round mapping population, and its effect on resistance was rather small. Moreover, cy-2 on chromosome 2 was identified as a major QTL conferring resistance to ToLCNDV-ES. The ty-5 and pepy-1 genes, both of which encode Pelota, are well-known recessive genes involved in resistance to begomoviruses [32, 34]. No Pelota-encoding gene was present in the target region of cy-2, suggesting that this region contains a novel recessive resistance gene that is potentially useful as a plant disease susceptibility (S) gene. This gene could be the target of genome editing or artificial mutant screening for the application of resistance mechanisms to other crop species. Fine mapping and reverse genetic analysis of cy-2 are now in progress. These analyses will aid in DNA marker-assisted breeding for durable and wide-spectrum begomovirus resistance when combined with Cy-1 and will provide a theoretical basis for the practical application of this resistance mechanism in various crops.

Conclusions

We conducted fine-mapping of a major QTL Cy-1 on chromosome 1 for ToLCNDV-ES resistance in cucumber accession No.44. Among 24 genes located on the candidate region, CsRDR3, an ortholog of Ty-1/Ty-3 of tomato and Pepy-2 of capsicum, was found to be a strong candidate conferring ToLCNDV-ES resistance. Reverse genetic analysis of CsRDR3 by VIGS also supported that this gene is responsible for ToLCNDV-ES resistance. DNA marker developed in this study to distinguish the resistant and susceptible alleles of CsRDR3 will be practically beneficial for marker-assisted breeding of ToLCNDV resistance in cucumber using No.44 as a resistant source.

Methods

Plant material

The cucumber (C. sativus) accession No.44 originating from South Asia, which is resistant to ToLCNDV, and the susceptible cultivars ‘Natsu Suzumi’ (NS) (Takii seeds, Kyoto, Kyoto, Japan) and ‘Sagami Hanjiro Fushinari’ (SHF) were used in this study. The F2 and F3 populations generated by crossing SHF and No.44 were used for genetic mapping. Plants were cultivated in a growth room with loose temperature control (25–30 °C day, 23–25 °C night) and a photoperiod of 13-h light/11-h dark.

ToLCNDV inoculation and detection

Two ToLCNDV isolates were used in this study: the ToLCNDV-ES isolate ES-Alm-Cuc-16 (GenBank accession numbers for DNA-A: LC596380, DNA-B: LC596383) from Spain and the ToLCNDV-IDN isolate BACu-20 (DNA-A: LC511775, DNA-B: LC511780) from Indonesia [15, 57]. Full details about the infectious clones and the inoculation method have been reported elsewhere [15, 58]. In brief, agrobacteria bearing the plasmids pGreenII-p35S-ToLCNDV-[ES-Alm-Cuc-16]-DNA-A and pGreenII-p35S-ToLCNDV-[ES-Alm-Cuc-16]-DNA-B or pGreenII-p35S-ToLCNDV-[BACu-20]-DNA-A and pGreenII-p35S-ToLCNDV-[BACu-20]-DNA-B were cultured to an optical density of 0.3 and subsequently used to agroinfiltrate cotyledons of No.44, SHF, NS, and SHF × No.44 F1, F2, and F3 plants before the first true leaf developed. The disease symptoms of each plant were scored at 15 and/or 30 dpi as the disease severity index (DSI). The DSI ranged from 0 to 4 and was scored as follows: 0, no symptoms; 1, a few yellow spots on the leaf; 2, yellow spots across the whole leaf; 3, yellowing of the leaf with mild curling; and 4, intense yellowing and curling of the leaf. For DNA and RNA extraction, the young upper leaves were collected and stored at − 80 °C. Statistical analyses for DSI scores were conducted using the non-parametric Bonferroni–Dunn test or Mann–Whitney U test with Excel Toukei ver. 7.0, and a p value less than 0.05 was considered statistically significant.

An inoculation experiment was conducted to evaluate the graft transmission of ToLCNDV-ES to No.44 and SHF plants. At approximately 30 dpi, ToLCNDV-ES-infected symptomatic NS plants were used as a scion and grafted onto uninoculated No.44 and SHF plants (as rootstocks), and the newly elongated lateral branches from the rootstock were evaluated for ToLCNDV-ES resistance. Plants were grown in a greenhouse. Disease symptoms were surveyed at 30 and 60 days after grafting. The fifth and 10th true leaves were collected at 30 days after grafting, and the ninth and 20th true leaves were collected at 60 days after grafting for DNA extraction.

A Nucleon PhytoPure Kit (GE Healthcare, Little Chalfont, Buckinghamshire, UK) was used to extract DNA from cucumber leaves. Conventional PCR and qPCR were conducted to detect and quantify the viral DNA of ToLCNDV following the methods of [59]. The primer sequences and PCR conditions used in these analyses are listed in Supplementary Tables S3 and S4. Statistical analysis for accumulating viral DNA amount were performed by Student’s t test in Excel Toukei ver. 7.0 and a p value less than 0.05 was considered statistically significant.

QTL mapping of ToLCNDV resistance

In this study, two rounds of phenotyping were conducted for two independent F2 populations consisting of 187 and 143 individuals infected with ToLCNDV-ES, and the extracted DNA was subjected to the first and the second rounds of RAD-seq analysis. The RAD-seq libraries were created for F2 individuals and their parents and sequenced with the NovaSeq 6000 platform (Illumina, Hercules, CA, USA) following the methods of [60]. The obtained reads were trimmed and mapped to the genome sequence of Chinese Long, ver.3 (C. sativus), one of the latest cucumber genome sequences [61]. The precise methodology used for the data analysis is described elsewhere [32]. The genetic linkage maps were constructed from the single nucleotide polymorphism (SNP) RAD tags, and R/qtl was used to conduct QTL analyses by composite interval mapping (CIM) [62].

Fine mapping

The NovaSeq 6000 platform (Illumina) was used for whole-genome resequencing of No.44 and SHF, and the data analysis of the obtained reads was conducted following the methods reported in [32]. Additional SNP markers were developed for the KBiosciences KASPar assay (LGC Genomics GmbH, Berlin, Germany) to further narrow the region of the QTL, and recombinants were screened from F2 individuals. Five F2 recombinants, namely, No.8, No.12, No.270, No.420, and No.437, were screened, and the additional KASP markers and high-resolution melting (HRM) markers were used to identify the recombination points in each F2 individual. HRM analysis was conducted following the methods reported in [32]. The primer sequences and PCR conditions used for fine mapping is listed in Supplementary Table S5 and Table S4. The recombinants were self-pollinated to obtain F3 populations.

Analyses of the candidate gene

Sepasol-RNA I Super G extraction buffer (Nacalai Tesque, Kyoto, Japan) and High-Salt Solution for Precipitation (Plant) (Takara Bio, Shiga, Japan) were used to extract and purify the RNA from leaves collected at 30 dpi. Transcriptome profiling of three biological replicates for each treatment was conducted via RNA-seq using the NovaSeq 6000 platform (Illumina) for mock-inoculated and ToLCNDV-ES-infected No.44 and SHF plants according to the previously reported methods [33]. De novo assembly of reads obtained by RNA-seq was conducted by Trinity (v.2.8.3.) [63]. The ORF sequences of the candidate gene, CsRDR3, were amplified by RT‒PCR to confirm the result of De novo assembly. RT‒PCR was conducted following the previously reported methods using the cDNA template, KOD-plus Neo (Toyobo, Osaka, Japan), and the CsRDR3-specific primer pairs to amplify the ORF of the candidate gene [32]. The amplified PCR products were cloned into the TOPO vector (Thermo Fisher Scientific, MA, USA) and used for subsequent sequencing. Supplementary Table S3 contains the list of the sequences of the primers used for PCR. Multiple sequence comparisons of the predicted RDR amino acid sequences of A. thaliana, Oryza sativa (rice), Nicotiana tabacum, S. tuberosum, S. lycopersicum, and C. annuum were conducted using the MUSCLE program [64]. MEGA 7.0 was used to construct the phylogenetic tree via the neighbor‒joining method, with 1,000 bootstrap replicates [65].

Reverse genetic analysis by virus-induced gene silencing

Apple latent spherical virus (ALSV) vectors were used for VIGS [50, 66]. In brief, partial coding sequences of CsPDS (102 bp), CsRDR3 (CsaV3_1G039730) (189 bp) or CsRDR3 (CsaV3_1G039730) (102 bp) were amplified from No.44 with the CsPDS-102-Xho/Bam, CsRDR3-189-Xho/Bam or CsRDR3-102-Xho/Bam primers, respectively (Supplementary Table S3). The obtained amplicons digested with XhoI and BamHI were ligated with XhoI- and BamHI-digested pBICAL2 to construct pBICAL2::CsPDS, pBICAL2::CsRDR3-189, and pBICAL2::CsRDR3-102. To avoid the effects of off-target genes, a BLASTn analysis was conducted using PCR fragment sequences as queries to design the gene-specific primers for each gene.

For VIGS, we used two chromosome segment substitution lines, No.148 and No.489, which were homozygous for the No.44 genotype at the target region on chromosome 1 and homozygous for the SHF genotype at the target region on chromosome 2. Cultures of Agrobacterium containing pBICAL1 and pBICAL2, pBICAL2::CsPDS, pBICAL2::CsRDR3-189, or pBICAL2::CsRDR3-102 were mixed at a ratio of 1:1 for agroinfiltration. Before the first true leaf developed, cotyledons were agroinfiltrated with the bacterial suspension at an optical density of 0.3. One day after agroinfiltration of ALSV, ToLCNDV-ES was agroinoculated following the methods reported previously [15, 67]. For agroinoculation, a colony inoculation procedure was performed on the hypocotyl just below the apex of the seedlings using Agrobacterium transformed with pGreenII-p35S-ToLCNDV-[ES-Alm-Cuc-16]-DNA-A + B. No.489 plants were cut back at 28 dpi to allow for elongation of lateral branches, which were subsequently used for analysis. Symptom survey and collection of young upper leaves were conducted at 40 dpi.

The viral DNA-A of ToLCNDV-ES was detected by conventional PCR and quantified by qPCR. ALSV infection was confirmed by conventional PCR using the ALSV-F/R primer pair (Table S3). The transcript levels of CsRDR3 were analyzed via qRT‒PCR following the previously reported methods [32]. The primer sequences and PCR conditions used for real-time qRT‒PCR analyses are listed in Supplementary Tables S3 and S4. The CsActin reference gene was used to normalize the transcript level of the candidate gene, and 2−ΔΔCt method was applied to calculate relative gene transcript levels. Statistical analyses for gene expression and viral DNA accumulation were conducted using Student’s t test, and a p value less than 0.05 was considered statistically significant.