The dissection of R genes and locus Pc5.1 in Phytophthora capsici infection provides a novel view of disease resistance in peppers

Phytophthora capsici root rot (PRR) is a disastrous disease in peppers (Capsicum spp.) caused by soilborne oomycete with typical symptoms of necrosis and constriction at the basal stem and consequent plant wilting. Most studies on the QTL mapping of P. capsici resistance suggested a consensus broad-spectrum QTL on chromosome 5 named Pc.5.1 regardless of P. capsici isolates and resistant resources. In addition, all these reports proposed NBS-ARC domain genes as candidate genes controlling resistance. We screened out 10 PRR-resistant resources from 160 Capsicum germplasm and inspected the response of locus Pc.5.1 and NBS-ARC genes during P. capsici infection by comparing the root transcriptomes of resistant pepper 305R and susceptible pepper 372S. To dissect the structure of Pc.5.1, we anchored genetic markers onto pepper genomic sequence and made an extended Pc5.1 (Ext-Pc5.1) located at 8.35 Mb–38.13 Mb on chromosome 5 which covered all Pc5.1 reported in publications. A total of 571 NBS-ARC genes were mined from the genome of pepper CM334 and 34 genes were significantly affected by P. capsici infection in either 305R or 372S. Only 5 inducible NBS-ARC genes had LRR domains and none of them was positioned at Ext-Pc5.1. Ext-Pc5.1 did show strong response to P. capsici infection and there were a total of 44 differentially expressed genes (DEGs), but no candidate genes proposed by previous publications was included. Snakin-1 (SN1), a well-known antimicrobial peptide gene located at Pc5.1, was significantly decreased in 372S but not in 305R. Moreover, there was an impressive upregulation of sugar pathway genes in 305R, which was confirmed by metabolite analysis of roots. The biological processes of histone methylation, histone phosphorylation, DNA methylation, and nucleosome assembly were strongly activated in 305R but not in 372S, indicating an epigenetic-related defense mechanism. Those NBS-ARC genes that were suggested to contribute to Pc5.1 in previous publications did not show any significant response in P. capsici infection and there were no significant differences of these genes in transcription levels between 305R and 372S. Other pathogen defense-related genes like SN1 might account for Pc5.1. Our study also proposed the important role of sugar and epigenetic regulation in the defense against P. capsici.


Background
Oomycete Phytophthora capsici is a soilborn pathogen fungus that causes fruit rot, stem blight, foliar blight, and particularly root rot in peppers depending on the disease occurrence position [1]. P. capsici root rot (PRR) is a devastating pepper disease with typical symptoms of necrosis and constriction at the basal stem and consequent plant wilting. P. capsici basically spread via soil and splashing water in the form of microzoospores, and the disease PRR can break out very quickly in summer 1-2 days after rainfall due to field ponding. In protected cultivation, root rot occurs frequently 1-2 months after transplantation, especially in the case of continuous cropping. Resistance breeding is the first choice to prevent disease damage. A resistance genetic source PI201234 was first found in pepper [2]. The best known source is 'Criollo de Morelos-334' (CM334), which is also the sequencing material due to its perfect resistance [3]. It is still important to explore more genetic resources of P. capsici resistance.
Disease PRR was first reported in 1918 in the US, which was later found to be caused by P. capsica, a new fungus species [4]. Hence, phytopathologists and microbiologists have made great efforts to understand its pathogenic features [5]. P. capsici has even become a model pathogen in the study of plant-microbe interactions due to its wide range of hosts, including potato, tomato, cucurbits, beans, Arabidopsis and tobacco [6][7][8].
On the other side of host defense, plants develop PTI to detect nonspecific pathogen/microbe-associated molecular patterns (P/MAMPs), and effector-triggered immunity (ETI) which is resistance specific and accompanied by a hypersensitive response (HR) [14]. In the ETI system, NBS (nucleotide binding site)-ARC (apoptosis, R proteins, CED-4)-LRR (leucine rich repeat) proteins recognize pathogenic effectors and trigger downstream defense processes, including a rapid and strong oxidative burst, pathogenesis-related (PR) gene expression, and accumulation of antimicrobial compounds. NBS-ARC-LRR protein genes constitute the predominant majority of disease resistance genes (R genes). Several doses of R genes have been amplified in peppers by degenerate primers [15,16]. However, most R genes are still unknown because higher plants typically have hundreds of R genes. As demonstrated in potato [17] and many other higher plants, pepper R gene proteins should also optionally have conserved domains of toll/interleukin-1 receptor (TIR), coiled-coil (CC), and resistance to powdery mildew 8 (RPW8) in addition to NBS, ARC and LRR. Among these above domains, the NBS-ARC domain is the most conserved and is widely used to identify R genes.
QTL mapping of pepper resistance to P. capsici was first reported in 1996 [18]. In this study, 13 QTLs were identified using the F2 mapping population of Perennial and YOLO Wonder, and one QTL linked to molecular marker TG483 on chromosome 5 had a major effect on resistance, which explained 41-55% of the total variance. Since then, many studies have confirmed these QTLs on chromosome 5 using different genetic resources (mostly CM334), mapping populations, and mapping strategies [19][20][21][22][23]. Based on the above studies, Mallard et al. (2013) constructed three consensus QTLs on chromosome 5 by using anchor markers and meta-analysis [24]. Meta-Pc5.1 and Meta-Pc5.3 were positioned close to teach other on the short arm of the chromosome and Meta-Pc5.2 was on the long arm. Recent QTL mapping work again confirmed the major QTL on the short arm of chromosome 5 [25][26][27]. Now, it is very clear that the major QTL Pc5.1 is a broad-spectrum QTL that controls resistance to all P. capsici. All the reports proposed R genes at Pc5.1 as candidate genes. However, the detailed genetic mechanism remains unknown, and the function of these R genes needs to be characterized. The pepper genome sequences of CM334 and Zunla were independently released in 2014 [28,29], which enabled a thorough dissection of QTL structure.
In this study, we identified NBS-ARC candidate genes by mining the genomic sequence and profiled the responses of these genes in P. capsici infection. We also constructed an extended Pc5.1 (Ext-Pc5.1) to cover all reported QTLs from different QTL mapping works and profiled the responses of the genes on this locus in P. capsici infection. The comparison of root metabolites and root transcriptome between resistant and sensitive peppers in P. capsici infection renewed our understanding about the roles of R genes and QTL Pc5.1and provide new insights in P. capsica-resistance.

Resistance assessment of Capsicum germplasm
Pepper seedlings with 6 leaves were inoculated with P. capsici by injecting zoospores into the soil around the basal stem (Fig. 1). A total of 160 germplasm materials were subjected to the resistance assessment. As a result, we identified 10 materials of high resistance (HR), 7 materials of resistance (R), 31 materials of moderate resistance (MR), and 112 materials of nonresistance (NR) (Additional file 1: Table S1). The HR pepper germplasm showed comparable resistance to CM334. The ten HR germplasms included 2 bell peppers, 5 cayenne peppers, 1 cluster pepper, 1 linear pepper (var. annuum L. dactylus M), and 1 upward pepper (var. conoide (Mill.) Isish). We selected four accession (304R, linear pepper; 305R, upward pepper; 370S, cone pepper; 372S, cayenne pepper) to be used in the following experiments that represented different pepper types.

Primary metabolites in infected roots
Ethanol extract from inoculated roots was subjected to GC-MS, which revealed dynamic changes in primary metabolites responding to P. capsici infection (Fig. 2). All inner standards were salinized, which indicates total and successful derivatization. Resistant accessions 304R and 305R show different alteration profiles to 370S and 372S in respect of sugar contents. There was a sharp increase of sucrose at 3 days post inoculation (3 dpi) in resistant peppers but decrease in susceptible peppers. Similarly, tagatose, fructose and mannose were strongly increased in 304R and 305R but decrease in susceptible peppers particularly 370S. In addition, propanetricarboxylic acid and butanedioic acid were reduced quickly after P. capsici inoculation in 370S and 372S but not in resistant materials. No additional consensus differences between resistant peppers and susceptible peppers were observed for the remaining compounds. The robust response of sugar contents may enhance the resistance against P. capsici.

Transcriptome of infected roots
We performed RNA-seq using roots of 305R and 372S to profile the dynamic response of the major QTL and NBS-LRR genes that may contribute to resistance against P. capsici (Additional file 2: Table S2). The transcriptome at 3 dpi was compared with that at 0 dpi to identify differentially expressed genes (DEGs) caused by P. capsici. As a result, a total of 3073 and 1743 DEGs were identified in 305R and 372S, respectively ( Fig. 3a; Additional file 3: Table S3; Additional file 4: Table S4). For both 372S and 305R, there were more upregulated DEGs than downregulated DEGs. There were many more DEGs in 305R than in 372S, indicating a strong defense response in 305R. This finding is interesting when considering that a visible symptom was noted for 372S, but no change in appearance was noted for 305R.
In KEGG enrichment analysis, the largest differences were noted in the pathways of valine, leucine and isoleucine degradation (ko00280, downregulated) and starch and sucrose metabolism (ko00500, upregulated) in 305R as well as carotenoid biosynthesis (ko00906, upregulated) and plant hormone signal transduction (04075) in 372S (Table 1; Additional file 8: Figure S1). Pathogen infection repressed the expression of ethylene signal transduction genes in 372S and disturbed other phytohormone signal pathways, including auxin, cytokinine, gibberellin, abscisic acid, brassinosteroid, jasmonic acid, and salicylic acid. Significant enrichments of both phenylpropanoid biosynthesis and glutathione metabolism were found in 305R and 372S.
In the GO enrichment analysis, only 3 significant enrichments were shared by 305R and 372S, indicating very different responses of the transcriptome to P. capsici (Table 2; Additional file 9: Figure S2). Notably, 20 DEGs in 305R were enriched under the GO term "response to endoplasmic reticulum (ER) stress", whereas that number was 7 in 372S, indicating a differential response in ER stress-mediated plant immunity. As a successful defense, 305R also shows an impressive response inside nuclear processes including nucleosome assembly and DNA replication initiation, epigenetic processes including, chromatin silencing by small RNA, methylationdependent chromatin silencing, histone methylation and phosphorylation, and DNA methylation. A total of 117 genes were assigned with epigenetic-related biological processes among which 42 genes were significantly affected by P. capsici in 305R while that number was 4 in 372S (Additional file 5: Table S5). We found many interesting DEGs responding to P. capsici in 305R, e.g., Histone, ATP-dependent DNA helicase, Chromatin structure-remodeling complex protein, NBS-LRR and Pentatricopeptide repeat-containing protein that may generate phasiRNAs in dicots [30]. Based on the results above of the enrichment analysis, we further compared the expression of DEGs under several interesting KEGG pathways or GO terms ( Fig. 3be). DEGs involved in "endoplasmic reticulum stress" and epigenetic modification were notably upregulated in 305R. Interestingly, under the GO term "fungus response", 12 DEGs out of 19 DEGs were downregulated  Figure S4). This corresponds well with increased sugar compounds in metabolite analysis.

Identification of NBS-ARC genes and their responses to P. capsici
A total of 823 candidate NBS-ARC domain proteins were identified by searching the HMM file (PF00931) against the whole-genome peptide sequences. The number increased to 1226 with an E value< 0.01 when using the pepper-specific HMM file. Finally, we identified 571 NBS-ARC domain proteins after removing short amino acid sequences. Complete assessment using the CD-Search Tool indicated 390 proteins with a complete NBS-ARC domain. As expected, all the NB-ARC domain genes were clustered among the pepper genomes, particularly at chromosome arms (Additional file 12: Figure  S5). These proteins were grouped according to the repetition and position of NBS-ARC, TIR, CC, LRR, RPW8 as well as coiled coil domain of the potato virus X resistance (RX-CC, abbreviated as Cx in this study) ( Table 3; Additional file 6: Table S6; Additional file 7: Table S7). The conserved domains and motifs as well as the gene structure of all the NBS-ARC genes were analyzed (Additional file 13: Figure S6). In the pepper genome, there are only 3 TIR-NBS genes, which is notably fewer than in other higher plants. In  The polygenetic tree indicates that NBS-ARC domain genes in the same cluster on chromosomes have high identity, e.g., genes on chromosomes 6 and 11 (Fig. 5a). Interestingly, genes with long branches, e.g., CA04g19370, CA04g09960, CA00g93130, and CA02g25810, might experience the acquisition of CC, Cx, LRR, or RPW8 domains (Fig. 5b). In total, 32 NBS-ARC genes exhibit a significant response to P. capsici infection, which were mainly clustered on chromosomes 3, 5, and 7 (Fig. 6a). Among them, only 2 had a CC domain, 1 had an RPCW8 domain, and 5 had an LRR domain (Fig. 6b). The 5 NB-LRR genes are probably P. capsici isolate-specific.  Response of QTL Pc5.1 to P. capsici QTL Pc5.1 is known a major and broad spectrum QTL [24]. We converted the genetic positions of molecular markers into physical positions in the CM334 genome by BLAST primer sequences or marker sequences against genome sequences (Fig. 7a). We coordinated the Meta-Pc5.1 locus (between markers C2_ At1g33970 and C2_At3g51010) and the adjacent Meta-Pc5.3 locus (between markers TG483 and TG437). This chromosome segment positioned 8.35 Mb -38.13 Mb on chromosome 5 (between markers T1261 and C2_At2g01770) is taken as an extended Pc5.1 (Ext-Pc5.1) in this study. A total of 44 DEGs were identified at Ext-Pc5.1 among which 34 DEGs were identified in 305R and 18 in 372S, indicating a stronger response in 305R than in 372S. In a detail, there were 11 DEGs at Meta-Pc5.1, 10 DEGs at Meta-Pc5.3 and 23 DEGs at the surrounding and conjunction regions. At Ext-Pc5.1, there were a total of 14 NBS-ARC genes but only one (CA05g04300) of them responded to P. capsici, which was induced in both 350R and 372S. Moreover, this R gene positioned at Meta- All the biological processes had a ks value < 0.001 Pc5.3 and it is N-type. All the candidate genes including R genes and other genes at Pc5.1 proposed in literatures did not show any significant response to P. capsici infection (see discussion) [25][26][27]. And there was no significant difference of these candidate genes in transcription level between 305R and 372S (data not shown). At Meta-Pc5.1, we found 3 DEGs showed different dynamic profile in the P. capsica-infection including snakin-1 (SN1), RING finger protein, and a transcription cofactor. Basically, their expression level was increased in 305R and decreased in 372S after inoculation (Fig. 7b). Snakin-1 (SN1) is a well-known pathogen defense-related gene, which is characterized by its in vitro activity and targeted pathogens in wide-spectrum [31]. No SNP at coding regions was detected by comparing SN1 transcripts of 372S and 305R (data not shown).

Discussion
Transcriptome comparison between resistant material and susceptible material enables us to discriminate the gene response, which should account for the difference in resistance. A recent study of the CM334 root transcriptome proposed that the upregulation of the phenylpropanoid biosynthesis pathway may play a role in resistance (Li, 2020). We also observed the difference in the response of phenylpropanoid biosynthesis pathway between 305R and 372S (Additional file 9: Fig. S3). This differential response is not P. capsica-specific and it looks like a common phenomenon in plant resistance against diseases [32,33]. Indeed, many other clear differences exist between 305R and 372S in addition to the phenylpropanoid pathway. Valine, leucine and isoleucine are branched-chain amino acids (BCAAs) and play roles in plant growth, development, defense, and flavor [34]. The degradation pathway of BCAAs mainly occurs in mitochondria, and its downregulation may help maintain increased BCAA levels and enhance defense.
Compared with 372S, which shows a major response in the process of metabolism and oxidation, the dynamic change in the 305R transcriptome is enriched in nucleic processes, such as cell proliferation, DNA replication, nucleosome assembly and transcription regulation, which provides an interesting view about pathogen resistance. We also observed a strong response of epigenetic modification about DNA, histone, and chromatin in 305R but not in 372S. Many reports support the participation of epigenetic mechanisms in plant tolerance against biotic stress [35,36]. NBS-LRR and pentatricopeptide repeat-containing protein were enriched in processes of DNA methylation and histone methylation, which may generate phased secondary small interfering RNAs (phasiRNAs). The interaction of R gene and miRNA-phasiRNA cascade contributes to the delicate control of R gene-mediated disease resistance [37]. It is very necessary to profile the dynamic epigenetic change of DNA methylation, histone modification, chromatin remodeling, and small RNA in P. capsica-infection in the future. Notably, sugars including sucrose, tagatose, fructose and mannose were upregulated by pathogen infection in 305R but not in 372S, which was due to the differential response of sugar pathway-related genes in transcriptome. Sugars play roles as energy, carbon sources, and signaling molecules in plant-pathogen interactions [38,39]. Some sugar pathway genes function as PR proteins, such as cell wall invertases (CWIs), which convert sucrose into glucose and fructose [40]. Many reports show the positive correlation of sugar accumulation with plant resistance against pathogenic microorganisms. For example, pretreatment with sucrose enhances resistance, and accordingly, rice plants overexpressing exotic PR genes accumulate more sucrose in leaves [41]. Apple adopts sorbitol to modulate resistance against fungal disease via NLR genes, and silencing synthesis results in a decrease in resistance [42]. The differential response of sugar accumulation and pathway genes was also observed in soybean infections when comparing resistant and susceptible materials [33]. Here, it is the first-time that sugars were reported to associate with P. capsica-resistance. Rapid upregulation of sugar after infection may be involved in lignification of the root cell wall or activate the downstream defense reaction as a part of plant immunity.
We identified a total of 571 NBS-ARC domain genes, 390 of which had a complete NBS-ARC domain. Before our study, only dozens of NBS-ARC genes were identified by degenerate PCR amplification [15,16]. This is due to either the limitation of degenerate PCR or the absence of genomic sequences at that time. Although the NBS-ARC domain is conserved, sequence variance due to rapid evolution of the gene family is noted; therefore, a large portion of genes cannot be detected by degenerate PCR method. Pepper has an impressively large number of NBS-ARC genes when compared with other reports using similar mining approaches [43]. LRR is the most variable domain of NBS-ARC protein and is critical for recognition specificity in ETI defense reaction [44]. There was a total of 204 NBS-ARC genes possess LRR domains (NB-LRR), however, only 5 of them show significant response. These five NB-LRR genes might be specific to P. capsici isolate used in this study. Moreover, there are many more P. capsici inducible NBS-ARC genes in 305R than in 372S, which might mean an effective ETI defense reaction in 305R.
Despite the effort in genetic analysis, very little is known about the detailed genetic mechanism. In the resistant resource PI201234, two genes, CA05g06820   proposed the late blight resistance genes CA05g05390, CA05g05440, and CA05g05720 [26]. All of these studies, together with the QTL on chromosome 10, led to a consensus about the role of the R genes in P. capsici resistance. Nevertheless, we did not detect any differential expression of all the candidate genes proposed at Pc5.1. Three other R genes (CA05g02520, CA05g02540, CA05g02840) at positions 5.47-6.74 Mb and a cluster of LRR receptor-like genes (CA05G02270, CA05G02310, CA05G02340, CA05G02350) located at 4.5-4.7 Mb showed significant response. However, they are too far from the consensus Meta-Pc5.1 and other published Pc5.1. Instead, they are likely located at a rare locus CC_ Pc5.4 [23]. These differential expressions are probably P. capsici-specific because their expression was not affected by inoculation with Phytophthora infestans, pepper mottle virus, or tobacco mosaic virus [46]. Reverse genetic studies stress the function of many other factors in P. capsici resistance, e.g., CaRGA2 [47], EDS1 [11], CaChiIV1 [48], CaAP2/ERF064 [49], and squamosal promoter binding protein (SBP) [50]. However, no significant response of these genes was identified at the locus Pc5.1. Snakin-1 (SN1) gene (CA05g05250) encoding a cysteine-rich antimicrobial peptide (AMP) might contribute to the broad-spectrum QTL. Snakin protein which was also called gibberellin stimulated peptide is firstly and thoroughly studied in potato since 1990s [51]. Many studies have proved that snakin proteins show strong antimicrobial activity in vivo and in vitro. Over  [52], alfalfa [53], and lettuce [54]. In pepper, CaSnakin protein was induced by rootknot nematode infection and play a role in resistance to nematode [55]. More genetic evidences need to be provided in the future to clarify it function in P. capsiciresistance.

Conclusions
We screened out 10 pepper materials showing high resistance to P. capsica. Primary metabolites of infected root were detected by GC-MS and sugars were greatly upregulated by P. capsica infection in 305R but not in 372S.We dissect the response of Ext-Pc5.1 and R genes in P. capsici infection in transcriptome analysis. A total of 570 NBS-ARC genes were identified in pepper CM334 genome, among which 34 genes were affected by the infection, but only 5 genes belonged to NB-LRR type. These NB-LRR DEGs were not positioned at Ext-Pc5.1 and all the candidate genes proposed in previous publications did not show any response in P. capsici infection. These results provide new insights about the role of R genes at the broad-spectrum Pc5.1 QTL. At least, we should not ascribe the P. capsici-resistance controlled by Pc5.1 to the transcriptional response of R genes. We propose SN1 at Meta-Pc5.1 as the candidate gene controlling the wide-spectrum resistance. In addition, we found interesting and very clear response of sugar metabolism and processes of epigenetic modification systems, which provide a new insight into the resistance mechanism.

Plant materials
Phytophthora-resistant peppers were screened from 160 germplasm materials. All pepper seedlings were cultured in plug trays with 50 cells in a culture room. Ten seedlings for each accession were surveyed in the resistance investigation at the stage of 6 leaves. The known pepper CM334 with high resistance was used as a positive control. Two resistant accessions, 304R and 305R, and two susceptible accessions, 370S and 372S, were identified and used in metabolite analysis and RNA-seq. Entire roots from each seedling were collected at 0 dpi, 3 dpi, and 5 dpi according to the development of the dynamic symptom. There was no symptom at 0 dpi; symptom appeared at 3 dpi; the whole plant became wilting at 5 dpi. Roots from 5 seedlings were mixed and homogenized as one sample. Triplicate samples were obtained at each sampling time point. All the samples were used to analyze the dynamic changes in root metabolites. The same samples at 0 dpi and 3 dpi were also used in RNA-seq.

P. capsici infection and resistance identification
The P. capsici isolate named LJYM1 was provided by Sichuan Academy of Agricultural Sciences, which were collected from Chengdu, China and was identified as physiological race 2 [56]. The inoculation method of disease PPR followed that described by Kimble [2]. Subcultures were performed on potato dextrose agar (PDA) medium at 27°C in the dark. Proliferation culture was performed on V8 medium in the dark until the plates were uniformly covered with mycelia. Zoospores were then induced in light and collected to make zoospore solutions with a concentration of 10 5 spores/ml. Five milliliters of zoospore solution was injected into the soil around the basal stem of pepper seedlings at the 6-leaf stage. Injury of disease PPR was assessed from 0 to 10 dpi following the standard method described by Li et al. (2006) [57]. Resistance assessment was performed according to the index of disease injury at 5 dpi. High resistance (HR) was determined by very low disease index (0-10); Resistance (R) was determined by low disease index (10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25); moderate resistance (MR) was determined by moderate disease index ; nonresistance (NR) was determined by high disease index (50-100).

Metabolite extraction and GC-MS analysis
The method of metabolite extraction and detection follows Desbrosses et al. (2005) with modification [58]. In brief, each root sample was dried by a vacuum freeze dryer at − 40°C, and 100 mg tissues were pulverized in a 2-ml tube using a slender glass pestle. Ground samples were suspended in 540 μl precooled methanol plus 60 μl ribitol in methanol (2 mg/ml), which was used as an internal standard. The tube was vortexed and then put in a water bath at 4°C for 30 min with ultrasonic treatment. After adding 600 μl of water and 300 μl of chloroform, the samples were vortexed and then centrifuged at 14,000 rpm for 10 min at room temperature. We then transferred 400 μl aliquots of the aqueous phase to 2 ml tubes and dried them at room temperature by vacuum centrifugation. The precipitate metabolites were suspended and derivatized by methoxyamination in 40 μl methoxyamine hydrochloride in pyridine (20 mg/ml) for 2 h at 37°C. Subsequent trimethylsilylation was performed at 37°C for 30 min after adding 70 μl MSTFA solution. The derivatized solution was filtered through a 45-μm membrane after adding 900 μl hexane. All chemicals and reagents were of analytical grade.
Qualitative detection of primary metabolites in roots was performed using an Agilent GC-MS (7890A-5975C) equipped with an HP-5MS column (30.0 m × 0.25 mm × 0.25 μm). One μl sample was injected in split-less mode. The oven temperature was initially maintained at 100°C for 1 min, increased to 184°C at a rate of 3°C/min, increased to 190°C at a rate of 0.5°C/min, and finally increased to 280°C at a rate of 15°C/min. Mass spectra were collected at a scanning range of 40-600 m/z. Each metabolite was identified by retention indices and comparison of mass spectra with a reference mass spectral library (NIST2011, Gaithersburg, MD, USA).
RNA-seq and differential gene expression caused by P. capsici infection Root samples were ground into power in liquid nitrogen, and 100 mg of each sample was used to extract total RNA using the TRIzol method as previously described [59]. One microgram of total RNA was purified via DNase I digestion. Poly(A) mRNAs were enriched using oligo (dT) magnetic particles. The obtained mRNA was fragmented via metal hydrolysis in 1× fragmentation buffer (Life Technologies) and then ligated to an RNA oligonucleotide adaptor. The ligation products were used to generate first-strand cDNA via reverse transcription (RT)-PCR. Then, short PCR was performed to amplify the cDNA to obtain sufficient quantities of DNA products. A single 'A' nucleotide was added to the 3′ ends of cDNAs to facilitate ligation with adapters. Sequencing was performed using an Illumina Genome Analyzer IIx according to the manufacturer's instructions.
Reads with adapters, reads with greater than 10% unknown bases, and low-quality reads were filtered to obtain clean reads. Clean reads were mapped to the CM334 genome (ftp://ftp.solgenomics.net/genomes/ Capsicum_annuum/) using HISAT2 [60] and assembled the transcripts with mapped reads using StringTie [61]. For the analysis of protein-coding genes, only uniquely mapped reads were used, and the transcript levels were calculated as fragments per kilobase of transcript per million fragments (FPKM) using the Cufflinks software package [62]. DEGs were identified using DEseq [63], and the criteria were FDR < 0.01 and fold-change≥2.

Identification of NBS-ARC domain genes
The genomic assembly and annotation of CM334 were downloaded from the Sol Genomics Network (ftp://ftp. solgenomics.net/genomes/Capsicum_annuum/). NBS-ARC candidate proteins were screened using HMMER v3 as described by Lozano et al. (2015) with minor alterations [43]. A hidden Markov model (HMM) of the NBS-ARC family (PF00931) from the Pfam database (http://pfam.sanger.ac.uk/) was used to search NBS-ARC domain proteins from genomic amino acid sequences using the function "hmmsearch" of HMMER v3. A pepper-specific NBS-ARC HMM was created from the obtained high-quality protein set (E-value< 1✕10 − 55 ) using the function "hmmbuild" of HMMER v3. The obtained specific HMM file was used to search NBS-ARC domain proteins again from genomic amino acid sequences using the function "hmmsearch". Candidate proteins with an E-value< 0.01 and lengths shorter than 200 bp were discarded. The HMMs of NBC-ARC (PF00931), TIR (PF01582), RPW8 (PF05659), and LRR (PF00560, PF07723, PF07725, and PF12799) from the Pfam database were merged into one HMM by the function "hmmpress" and used to mine the corresponding domains by the "hmmscan" HMMER v3 suite. Paircoliled 2 (http://cb.csail.mit.edu/cb/paircoil2/paircoil2. html) was used to identify the CC domain with a Pscore cutoff of 0.03 and minimum search window length of 21. Conserved domains were confirmed by the Batch Web CD-Search Tool (https://www.ncbi.nlm.nih.gov/ Structure/bwrpsb/bwrpsb.cgi) and Multiple Expectation for Motif Elicitation (MEME). All the identified NBS-ARC domain proteins were grouped according to the number and position of NBS-ARC, CC, TIR, RPW8, LRR, and RX-CC_like. Tree topology of NBS-ARC genes was constructed by MEGA7 [64]. The visualization of gene structure and domain pattern was done by using TBTools [65].