Characterization, identification and expression profiling of genome-wide R-genes in melon and their putative roles in bacterial fruit blotch resistance

Background Bacterial fruit blotch (BFB), a disease caused by Acidovorax citrulli, results in significant economic losses in melon. The causal QTLs and genes for resistance to this disease have yet to be identified. Resistance (R)-genes play vital roles in resistance to plant diseases. Since the complete genome sequence of melon is available and genome-wide identification of R-genes has been performed for this important crop, comprehensive expression profiling may lead to the identification of putative candidate genes that function in the response to BFB. Results We identified melon accessions that are resistant and susceptible to BFB through repeated bioassays and characterized all 70 R-genes in melon, including their gene structures, chromosomal locations, domain organizations, motif distributions, and syntenic relationships. Several disease resistance-related domains were identified, including NBS, TIR, LRR, CC, RLK, and DUF domains, and the genes were categorized based on the domains of their encoded proteins. In addition, we profiled the expression patterns of the genes in melon accessions with contrasting levels of BFB resistance at 12 h, 1 d, 3 d, and 6 d after inoculation with A. citrulli. Six R-genes exhibited consistent expression patterns (MELO3C023441, MELO3C016529, MELO3C022157, MELO3C022146, MELO3C025518, and MELO3C004303), with higher expression levels in the resistant vs. susceptible accession. Conclusion We identified six putative candidate R-genes against BFB in melon. Upon functional validation, these genes could be targeted for manipulation via breeding and biotechnological approaches to improve BFB resistance in melon in the future.

Melon is vulnerable to various biotic and abiotic stresses [7,8]. Bacterial fruit blotch (BFB) is a devastating disease of melon caused by Acidovorax citrulli, an aerobic, mesophilic, gram-negative, rod-shaped seedborne bacterium belonging to the beta subdivision of the Proteobacteria [9]. BFB has been reported in many countries and poses a serious threat to melon, as well as other cucurbit crops including prickly paddy melon, citron melon, cucumber, pumpkin, squash, several types of gourds, and watermelon [10][11][12][13][14][15][16]. BFB causes watersoaked lesions to form on cotyledons and leaves, leading to collapse and death. The lesions on fruits are small (1 cm diameter), irregular, and often sunken, progressing through the rind. The lesions then become necrotic, causing decay and cracks in the fruit. These lesions expose the plant to secondary infections and cause A. citrulli to colonize the pulp, eventually allowing the seed to become contaminated [17]. BFB causes 80-100% losses in production under favorable environmental situations, especially during the rainy season and in regions with highly fluctuating temperatures [18,19]. Although BFB is of great concern to farmers and seed companies, strategies for managing this disease are limited; chemical control measures are environmentally hazardous and only partially effective, and resistant commercial cultivars have not yet been developed [13,[20][21][22][23][24]. Host resistance represents the most cost effective and environmentally friendly approach for managing BFB [12]. However, no QTL or R-gene for this disease has thus far been identified in melon. Efforts to develop BFB-resistant melon genotypes would be greatly enhanced by the identification of functional R-genes.
An improved assembly and annotation of the melon (Cucumis melo L.) reference genome identified 70 Rgenes in melon [1,44,45]. In the current study, we investigated the expression patterns of R-genes throughout the melon genome in response to the BFB-causing bacterium A. citrulli in melon accessions contrasting in BFB resistance. The aim of this study was to identify putative candidate R-genes that confer resistance to BFB in melon.

Results
Genome-wide melon R-genes and their chromosomal distribution The latest version (v3.6.1) of the whole-genome sequence of the melon double haploid line DHL90 was constructed using an improved assembly and annotation. This sequence contains 70 R-genes [1,44]. We retrieved genomic information for these 70 R-genes, including their coding sequences and deduced amino acid sequences, from the cucurbit genome database (http:// cucurbitgenomics.org). Detailed genomic information about these R-genes, including their locations on chromosomes, is provided in Table 1. Chromosomal mapping of the 70 R-genes revealed that they are distributed across all 12 melon chromosomes, with 1 to 12 genes per chromosome ( Fig. 1; Table 1). Chromosome 9 (Chr09) contains the most R-genes (12), followed by Chr12 and Chr01 (10 and 9 genes, respectively). Chr11 contains the fewest R-genes (2), followed by Chr03 and Chr07 (3 genes each). The genes appear to be clustered, particularly in the telomere regions of chromosomes such as chr09, chr01, and chr04 ( Fig. 1).

Gene structures, domain organizations, and motif distribution of R-genes in melon
We analyzed the exon-intron structures of all 70 melon R-genes by comparing their coding sequences with the corresponding genomic sequences using the online tool GSDS2.0 (http://gsds.cbi.pku.edu.cn/). The highest number of exons (22) was in the gene MELO3C013803, followed by 18 in MELO3C007367 (Additional file 1: Fig.  S1). Among the 70 R-genes, 21 were mono-exonic, while 12 and 4 genes were bi-and tri-exonic, respectively.
We analyzed the conserved motifs of these 70 R-genes using the MEME Suite (http://meme-uite.org/tools/ meme). A total of 20 conserved motifs were detected in these 70 R-genes, each comprising more than 14 amino acids. The greatest number of motifs was identified in the LRR domain-encoding gene MELO3C002394, whereas the fewest were detected in MELO3C029505, MELO3C023580, and MELO3C006801, which are LRR-, CC-, and DUF-domain-encoding genes, respectively. The distribution of these conserved motifs, along with the motif sequences, is described in Fig. 2.

Microsynteny of melon R-genes with genes in the watermelon and cucumber genomes
We analyzed the microsyntenic relationships of the 70 R-genes from melon (Cucumis melo) with genes in the watermelon (Citrullus lanatus) and cucumber (Cucumis sativus) genomes using the Circos tool. Most R-genes from melon were homologous to R-genes from watermelon and cucumber. However, watermelon R-genes on chromosomes 11 and 12 lacked homologues in melon (Fig. 3). By contrast, all 70 R-genes in melon had homologues in all chromosomes of cucumber.
Expression profiles of melon R-genes in response to A. citrulli inoculation We investigated the expression patterns of the 70 melon R-genes in the leaves of resistant and susceptible melon seedlings at 12 h, 1 d, 3 d, and 6 d of inoculation with A. citrulli strain NIHHS15-280 via qRT-PCR. Several genes showed differential expression in the resistant vs.
susceptible accession at different time points. A general trend of low expression for these genes was observed in the susceptible accession (Fig. 4). On the contrary, most of the genes were significantly induced within 12 h of A. citrulli infection in the resistant accession and showed a general increase in expression in this accession. By contrast, in the susceptible accession, the expression of these genes fluctuated, with little or no expression at the 12 h time point. Heatmap analysis of the expression data identified a sub-cluster of six genes (MELO3C023441, MELO3C016529, MELO3C022157, MELO3C022146, MELO3C025518, and MELO3C004303) that showed contrasting trends of expression in the resistant vs. susceptible accession, with progressively increasing expression after inoculation with A. citrulli in the resistant but not the susceptible accession (Fig. 4). Extensive analysis of these six genes indicated that the expression of four genes (MELO3C023441, MELO3C004303, MELO3C022146, and MELO3C025518) increased in the resistant accession with increasing time after inoculation with A. citrulli (Fig. 5). In the susceptible accession, the expressions of these genes were very low in the initial hours after inoculation and did not show significant increase over time after inoculation. In the resistant accession, the expression of these four genes (MELO3C023441, MELO3C004303, MELO3C022146, and MELO3C025518) peaked at 6 d after inoculation, with levels approximately 8-, 8-, 10-, and 7-fold those of the control samples, respectively. In the susceptible accession, the expression of two of these genes did not increase in response to A. citrulli inoculation, whereas the expression of two genes (MELO3C022157 and MELO3C016529) generally increased in response to inoculation, but to a lesser extent than in the resistant accession. The expression of these two genes increased until 3 d after inoculation (5-fold in MELO3C016529 and 2.5fold in MELO3C022157), followed by a decrease to their lowest levels at 6 d post inoculation (Fig. 5).

Discussion
Here, we identified R-genes with putative roles in resistance to BFB disease in melon by profiling the genomewide expression patterns of R-genes from melon in response to inoculation with A. citrulli. Disease resistance in plants involves the interaction between specific disease resistance (R)-genes in plants and avirulence (avr) genes of the pathogen which is known as gene-for-gene model [55,56]. Most plant R-genes belong to a superfamily of genes encoding proteins with an NBS or LRR domain, an N-terminal TIR or CC domain, or an RLK/ RLP domain [29,57]. A meta-analysis of the 314 cloned plant R-genes revealed that 191 (61%) such genes are NBS-LRR genes and 60 (19%) genes are RLKs/RLPs [58]. NBS domains bind to and hydrolyze adenosine triphosphate (ATP) or guanosine triphosphate (GTP) and are involved in signaling; LRR domains are highly adaptable structural domains that are responsible for protein-protein interactions and play an important role in plantpathogen recognition [59]; TIR domains provide pathogen specificity and plant defense responses, while CC domains are involved in pathogen recognition and signaling; and RLK domains play roles in signaling and plant defense responses. In melon, four resistance gene homologue sequences were previously reported that contained 14 TIR-NBS-LRR genes [60,61]. A study of the first complete genome sequence of melon identified 411 putative R-genes, including 161 RLKs, 110 RLP (receptor-like proteins) genes, 19 RLK-GNK2 (kinases containing an additional antifungal protein ginkbilobin-2 domain) genes, and 81 genes containing canonical resistance domains, such as NBS, LRR, and TIR domains [1]. Among these genes, 25 were homologous to Pto genes from tomato and 15 were homologous to Mlo genes from barley [62,63]. After further improvements in the assembly and annotation of the melon (Cucumis melo L.) reference genome, 70 Rgenes were ultimately identified in melon [44]. Our comprehensive in-silico analysis of the 70 melon R-genes revealed that they encode proteins with several disease resistance-related domains, including LRR, NBS, TIR, NB-ARC, CC, RLK, and DUF domains ( Table 2). These genes are distributed across all melon chromosomes, and some are clustered in the telomeric regions of a few chromosomes (Fig. 1). The clustering of R-genes is an evolutionarily conserved defense mechanism in plants wherein recombination in closely located genes creates new motif combinations, which generates novel resistance specificities and broadens plant resistance to different diseases [42,64]. R-gene clusters that provide resistance to multiple diseases have been reported for angular leaf spot, downy mildew, and anthracnose diseases in cucumber [65] and for blackleg, sclerotinia stem rot, and clubroot diseases in B. napus [66][67][68] and B. rapa [66]. In melon, a 1 Mb region on chromosome five contains the highest density of R-genes [69]. In addition, a cluster of 13 TNL genes is located in the same region as the melon Vat resistance gene [70], and another cluster of 7 TNL genes is located in the region harboring the Fom-1 resistance gene [71]. The Vat locus encodes a CC-NBS-LRR protein that confers resistance to aphid and aphid-mediated viruses in melon. The loss of two highly conserved LRRs is linked with susceptibility to these viruses [72]. In addition, the Fusarium wilt resistance locus Fom-2 is a TIR-NBS-LRR gene [73]. Expression patterns of the genome-wide R-genes are thus studied to identify any potential candidate genes against A. citrulli.
Six melon genes were highly expressed in the BFBresistant accession. Of these genes, three (MELO3C016529, MELO3C022157, and MELO3C022146) are TNL genes, two (MELO3C023441 and MELO3C025518) are LRR genes, and one (MELO3C005452) is an NBS-LRR gene ( Table 2). These genes were highly expressed at 6 d after inoculation (Fig. 5), which is consistent with our observation that BFB symptoms first appeared at 6-7 d in a susceptible accession [74]. Expression analysis upon infection with A. citrulli indicated a general trend of low expression for most R-genes in susceptible accession. By contrast, a set of genes including MELO3C023441, MELO3C004303, MELO3C022146, and MELO3C025518 were expressed at much higher levels, and MELO3C022157 and MELO3C016529 were expressed at relatively higher levels, (Fig. 5) in the resistant accession. Such higher expression in response to A. citrulli in the resistant accession indicates the potential involvement of these R-genes in BFB resistance in melon.
Several comparative transcriptomic studies have been reported in melon [75][76][77], but few studies have focused exclusively on expression profiling of R-genes against phytopathogenic agents in melon. For example, RNA-seq assessment of the changes in transcript levels at different time points in Phytophthora capsici-inoculated tissues of resistant and susceptible melon genotypes provided a basis for identifying candidate resistant genes [78]. Comparative transcriptome analysis identified ten genes that were differentially expressed in resistant and susceptible cultivars of melon in response to powdery mildew [79]. In addition, a study of the MLO (mildew resistance locus o) gene family in melon revealed candidate genes that might play roles in susceptibility to powdery mildew [80]. In watermelon, six NBS-encoding R-genes were identified as candidates for gummy stem blight (GSB) resistance [81,82]. Finally, markers have been developed for detecting both GSB and BFB resistance in melon based on the sequence polymorphism in the TIR-NBS-LRR gene MELO3C022157 [81,83]. Notably, all six candidate Rgenes identified in the current study have corresponding homologues in watermelon and cucumber (Fig. 3). The  roles of these genes in BFB resistance in these two crops remain to be investigated.

Conclusions
We identified six putative candidate genes that might play roles in resistance to BFB in melon. This is the first report of candidate genes for BFB resistance in melon. Our findings provide a basis for further functional studies to validate the exact roles of these genes. In addition, causal sequence polymorphisms could be identified in these genes, leading to the development of markers for BFB resistance. Our findings will thus be useful for improving the BFB resistance trait in melon.

A. citrulli: collection, culture, and inoculum preparation
A. citrulli strain NIHHS15-280 was obtained from the National Institute of Horticultural and Herbal Science (NIHHS), South Korea. The bacterium was cultured on Petri plates containing 20 ml King's B (KB) medium supplemented with 100 μg ml − 1 ampicillin for 36-48 h at 28°C [84] until bacterial colonies formed. For all inoculations, a bacterial suspension was prepared by covering the culture plates with 5 ml of sterile, double distilled (DD) water and gently scraping the surface of the KB medium using an L-shaped rubber spreader to an optical density (OD) of 1.0 at 600 nm, as measured using a NanoDrop ND-1000 Spectrophotometer. The bacterial suspension was diluted to a final concentration of~1 × 10 6 colony forming units (cfu) mL − 1 .

Plant materials, growth conditions, and bioassays
The BFB-resistant (PI 353814) and -susceptible (PI 614596) melon accessions [74,85] used in this study were obtained from the U.S. National Plant Germplasm System (https://npgsweb.ars-grin.gov/gringlobal/search. aspx), USDA, USA. The seeds were sown in a commercial nursery soil mixture in 32-cell trays and grown in a controlled plant growth chamber at 25 ± 2°C, 16 h day length, relative humidity of 60%, and a light intensity of 440 μmoles/m 2 /s at bench level. After 3 weeks, the plants were transferred to a greenhouse.
Two weeks after germination, the plants were transferred to plastic pots and grown in a greenhouse at 24 ± 2°C with a relative humidity of 90% where the plants were inoculated with A. citrulli. The resistance status of the accessions was reconfirmed via bioassay (Fig. 6) as previously reported with minor modifications [86]. Plants at the 3-5 true-leaf stage (4-5 weeks old) were sprayed with bacterial suspensions until runoff in a greenhouse at 22 ± 2°C with a relative humidity of 96%. Plants were re-inoculated 3 d after the first inoculation to ensure that no plants had avoided inoculation and to eliminate false positives. Leaf samples from three biological replicates were collected at different time points (0 h, 12 h, 1 d, 3 d, and 6 d), immediately immersed in liquid nitrogen, and stored at − 80°C for RNA extraction and cDNA synthesis.

Total RNA isolation and cDNA synthesis
The melon leaves were ground to a powder in liquid nitrogen, and 100 mg of each sample with three biological replicates was subjected to total RNA extraction using the RNeasy Mini kit (Qiagen, Valencia, CA) following the manufacturer's instructions. Firststrand cDNA was synthesized from total RNA with a SuperScript III First-Strand Synthesis System kit (Invitrogen, Gaithersburg, MD).

Identification and in silico analysis of melon R-genes
Genomic information for all 70 R-genes, as reported in the improved assembly and annotated genome of melon [44], was retrieved from the cucurbit genomic database (http://cucurbitgenomics.org) (Additional file 1: Table  S1). The genes were subjected to a series of in silico analyses such as exon-intron structure, motif distribution, domain organization, chromosomal mapping, and microsynteny analyses (for specific analytical tools, see the Results section).

Primer design and quantitative RT-PCR analysis
Gene-specific primers for quantitative RT-PCR (qRT-PCR) were designed using Primer3Plus (https://primer3 plus.com/cgibin/dev/primer3plus.cgi) ( Table 1). The Fig. 6 Disease symptoms in the leaves of susceptible (PI 614596) and resistant (PI 353814) melon accessions at 12 d after inoculation with A. citrulli. All leaves were detached from the plants immediately before they were photographed expression patterns of the R-genes were analyzed by qRT-PCR in a LightCycler® instrument (Roche, Mannheim, Germany) following the manufacturer's instructions. The reactions were performed in a 10 μL volume consisting of 5 μL of 2x qPCRBIO SyGreen Mix Lo-ROX (PCR Biosystems, London, UK), 5 pmol of primers, and cDNA template diluted to the appropriate concentrations. The PCR conditions were as follows: 5 min at 95°C, followed by 3-step amplifications at 95°C for 15 s, 56°C for 15 s and 72°C for 20 s for 45 cycles. The mean expression levels of relevant genes were calculated by the 2 -ΔΔ Ct method [87] using the average value of three reference genes [2,8,88] as internal control.

Statistical analysis
Analysis of variance (ANOVA) and significance tests were carried out using the normalized gene expression values with MINITAB17 software (Minitab Inc., State College, PA, USA). Tukey's pairwise comparison test was employed to determine the mean separation of expression values. p values indicate statistically significant variations of expression.
Additional file 1: Table S1. Details of the primers designed for expression profiling of melon R-genes. Figure S1. Exon-intron structures of R-genes in melon genome-wide. Light red rectangles and black lines indicate exons and introns, respectively. Figure S2. Domain structures of the 70 R-genes in melon. The conserved domains were identified using the NCBI Conserved Domain Database (CDD) (https://www.ncbi.nlm.nih. gov/Structure/bwrpsb/bwrpsb.cgi). Detailed descriptions of these domains are provided in Table 2. Specific domains in each protein are shown in the diagram. Figure S3. Gene expression profiles of resistant and susceptible melon accessions at different time points normalized to melon Actin expression (CmACT7, 149 bp), as determined by qRT-PCR analysis.