Fine mapping and gene silencing pinpoint Capana10g002229 as a strong candidate gene regulating the deciduous character of ripe pepper fruit (Capsicum spp.)

The pepper S locus, which controls the deciduous character of ripe fruit, was first fine mapped into an interval with a physical length of ~ 38.03 kb on chromosome P10. Capana10g002229, encoding a polygalacturonase, was proposed as a strong candidate gene based on sequence comparison, expression pattern analysis and virus-induced gene silencing (VIGS). The deciduous character of ripe fruit, which is controlled by the dominant S locus, is a domesticated trait with potential value in the pepper processing industry (Capsicum spp.). However, the gene associated with the S locus has not been identified. Here, one major QTL designated S10.1 was detected by using the F2 population (n = 155) derived from BA3 (Capsicum annuum) × YNXML (Capsicum frutescens) and was further verified in an intraspecific backcross population (n = 254) derived from the cross between BB3 (C. annuum) and its wild relative Chiltepin (C. annuum var. glabriusculum) with BB3 as the recurrent parent. Then, a large BC1F2 population derived from the self-pollination of BB3 × (BB3 × Chiltepin) individuals and comprising 4217 individuals was used to screen the recombinants, and the S locus was ultimately delimited into a 38.03-kb region on chromosome P10 harbouring four annotated genes. Capana10g002229, encoding a polygalacturonase (PG), was proposed as the best candidate gene for S based on sequence comparison and expression pattern analyses. Downregulation of Capana10g002229 in fruits through VIGS significantly delayed fruit softening and abscission from the fruit-receptacle junction. Taken together, the results show that Capana10g002229 could be regarded as a strong candidate gene associated with the S locus in pepper. These findings not only lay a foundation for deciphering the molecular mechanisms underlying pepper domestication but also provide a strategy for genetic improvement of the deciduous character of ripe fruit using a marker-assisted selection approach.


Introduction
Plants always exhibit a deciduous character, which is a naturally prevalent phenomenon that occurs in various organs, such as leaves, flowers, pedicels, fruits and seeds (Sexton and Roberts 1982;Dong et al. 2021;Yan et al. 2021). The deciduous character can be induced by various mechanisms, including external environmental stress, internal programmed senescence and ripening (Taylor and Whitelaw 2001;Ma et al. 2021). The ripening-induced abscission of fruits or seeds can effectively facilitate propagation, allowing the establishment of new and independent individuals.
Pepper (Capsicum spp.) is among the most important vegetable crops worldwide. Furthermore, ripe pepper fruits are important raw materials for preparing a variety of condiments, such as chilli powder, chilli sauce and chilli oil, as well as for the extraction of highly valuable secondary metabolites, including capsaicin and capsorubin (Hernández-Ortega et al. 2012;Ding et al. 2016). During the growth and development of pepper plants, immature fruits occasionally undergo physiological shedding at the junction of pedicel and stem, and ripe fruit shedding occurs not only at this site (Setiamihardja 1991) but also at the fruit-receptacle junction. Ripe fruit shedding from the fruit-receptacle junction occurs mainly in wild pepper plants with small erect fruits (Smith 1951), and this wild trait can contribute to good adaptation for seed dispersal by birds (Tewksbury et al. 2008;Carlo and Tewksbury 2014).
The deciduous character of ripe fruits of wild pepper has been replaced by a nondeciduous character after domestication of most cultivated peppers; domestication is not only a prerequisite for and guarantee of high yield but also is beneficial for postharvest preservation and storage of the fruit (Smith 1951). With the rapid development of the pepper processing industry in recent years, there is a growing demand for special varieties suitable for postharvest processing, where the removal of peduncles is the first step. Until now, this step has relied mainly on time-consuming and labour-intensive manual clipping or mechanical excision with a high attrition rate. Furthermore, mechanical excision often results in woody peduncle tissue residues in the ground powder, reducing the quality and colour intensity of the product. Therefore, breeding a pepper variety with a natural easy fruit shedding character may be an effective way to solve this problem from the perspective of genetic improvement.
Previous studies have reported that the inheritance of the deciduous character of ripe pepper fruit is controlled by a single dominant locus, named S (Smith 1951). The S locus was preliminarily mapped to the bottom part of chromosome 10 in the region corresponding to that in which a polygalacturonase (PG) gene was previously mapped in tomato. A partial sequence of the pepper PG gene (246 bp) was cloned by using two primers from the tomato fruit PG gene. Then, a strong correlation between the pepper PG gene and S locus was established based on genetic linkage, mRNA and protein expression analyses (Rao and Paran 2003). Similarly, with the homology-based gene cloning strategy, through sequence comparison of PG between a soft flesh mutant and wild-type pepper, a SNP in the intron VIII 3′ splice junction was found to be genetically cosegregated with the fruit softening character, and the differential levels of water-soluble pectin suggested that the mutated PG gene may play a critical role in the nonsoftening fruit of wild-type pepper (Kim et al. 2014). Nevertheless, forward genetic evidence showing PG to be the causal gene determining the deciduous character of ripe pepper fruit is still lacking. Furthermore, the biological function of PG in pepper has not yet been investigated.
In this study, the pepper S locus was initially genetically mapped using an interspecific F 2 population derived from the crossing of the cytoplasmic male sterility (CMS) line BA3 and the Capsicum frutescens restorer line YNXML based on a high-density SNP map constructed previously (Cheng et al. 2016). To overcome the limitations of interspecific reproduction isolation and CMS, another intraspecific backcross segregation population (BC 1 F 1 ) stemmed from a cross between BB3 and Chiltepin with the former as the recurrent parent was constructed to verify the initial genetic mapping result. Then, a large BC 1 F 2 population was used to screen recombinants for fine mapping of the S locus. Finally, sequence comparison, expression analysis and virus-induced gene silencing (VIGS) experiments were performed to verify the functions of important candidates.

Plant materials
An interspecific F 2 population comprising 297 individuals was previously developed through artificial hybridization of the CMS line BA3 (Capsicum annuum) and the C. frutescens restorer line YNXML and then by self-crossing of the hybrid. The red ripe fruits of YNXML were easily detached from the fruit-receptacle junction, while those of BA3 obtained from artificial pollination were firmly attached to the fruit-receptacle junction. This interspecific F 2 population was previously genotyped with the pepper SNP array, namely CapSNP15K, and a high-density interspecific SNP map which was named BY-SNP was constructed (Cheng et al. 2016). Of these, 142 and 155 individuals, together with the BB3, YNXML and BA3 × YNXML (F 1 ), were grown successively in the autumn of 2011 and 2012 at the Zengcheng Experimental Station, South China Agricultural University (SCAU), Guangzhou, China (23° N, 113° E). Among a subgroup (n = 155) of the interspecific F 2 population planted in 2012, we successfully identified the deciduous character of ripe fruit for only 79 individuals, as the remaining could not flower or bear fruit normally due to the limitations of interspecific reproduction isolation and CMS.
To overcome the limitations of reproduction isolation and sterility, BB3 (C. annuum), which is a near-isogenic maintainer line of BA3, and the wild pepper relative Chiltepin (C. annuum var. glabriusculum), the red ripe fruits of which were deciduous, were used to construct a backcrossing population (BC 1 F 1 ) with BB3 as the recurrent parent, namely BB3 × (BB3 × Chiltepin). Targeted genetic mapping based on the BC 1 F 1 segregation population was performed to verify the initial mapping of the S locus based on the BY-SNP map, and the S locus was initially located between the markers FD-177 and FD-113. The BB3, Chiltepin, BB3 × Chiltepin (F 1 ) and BB3 × (BB3 × Chiltepin) (BC 1 F 1 , n = 254) populations were grown at the SCAU Main Campus Teaching & Research Base, Guangzhou, China (23° N, 113° E), in 2017. To further fine map the S locus, a BC 1 F 2 population comprising 4217 individuals was developed by selfpollination from the BC 1 F 1 individuals, which were simultaneously heterozygous at markers FD-177 and FD-113.

Phenotyping
The deciduous character of ripe fruit was evaluated and recorded for at least five fruits after ripening (55 days after pollination) by a handhold-based method. Fruits adhering tightly to the fruit-receptacle junction after ripening were identified as the nondeciduous type, while fruits that were soft and juicy, and easily detached from the fruit-receptacle junction after ripening were judged to be the deciduous type.

Genetic mapping
Initial genetic mapping of the S locus was performed by using the interval mapping (IM) method and the nonparametric Kruskal-Wallis (KW) test, both of which were integrated into MapQTL6.0 software (Van Ooijen 2009), based on the deciduous character of ripe fruit of 79 individuals and the BY-SNP framework map constructed in our previous study (Cheng et al. 2016). The threshold of the LOD score (= 5.4) was estimated using a permutation test with 1000 iterations at a genome-wide significance level (P value) of 0.01, and the threshold for the K* statistic score (= 18.64) was set at a stringent significance level (P value) of 0.0001. A high confidence candidate interval for the S locus was determined based on the results of both methods with the criteria K* ≥ 45 and LOD ≥ 15.65 (5-LOD-drop). Linkage mapping was performed by using JoinMap4.0 software (Van Ooijen 2006).

Marker development and genotyping
Insertion and deletion (InDel) variations were identified by aligning the resequencing reads of BB3 to the Chiltepin reference genome (Qin et al. 2014) with SOAPindel (http:// soap. genom ics. org. cn/). Primers for anchoring the InDel sites were designed with Primer3web (http:// prime r3. ut. ee/). Genomic DNA was extracted from young leaves by using the modified CTAB method (Murray and Thompson 1980). Primer pairs showing specific amplification and polymorphism between BB3 and Chiltepin were used to genotype the segregation population. PCRs and procedures were performed according to the method described previously . Electrophoretic separation of PCR products was performed with a 6% polyacrylamide gel.

Gene cloning
RNA samples were subjected to reverse transcription by using a cDNA synthesis kit as described in our previous study . The specific primers for amplifying the candidate genes were designed with Primer3web (http:// prime r3. ut. ee/), and the sequences are listed in Supplementary Table S1. The method of gene cloning was consistent with a previous study ).

Expression analysis
The expression levels of candidate genes in fruits of parents at three different developmental stages, including the mature green stage (33 days after pollination), breaker stage (44 days after pollination) and red ripe stage (55 days after pollination) ( Fig. 1), were examined by quantitative real-time PCR (qRT-PCR). In addition, roots, stems, leaves and flowers were collected for tissue-specific expression analysis of Capana10g002229. The qRT-PCR method was Fig. 1 Morphology of fruits at three different developmental stages. Mature green stage, ~ 33 days after pollination; breaker stage, ~ 44 days after pollination; red ripe stage, ~ 55 days after pollination consistent with a previous study . All reactions were performed with three biological replicates and three technical replicates, and relative expression levels were analysed using the 2 −∆∆CT method (Livak and Schmittgen 2001). The specific primers used for qRT-PCR analysis are listed in Supplementary Table S2. The RT-PCR analysis of Capana10g002229 was performed with the Capa-na10g002229-cDNA primer pairs (Supplementary Table S1) at the green, breaker and red ripe stage in both BB3 and Chiltepin. The target bands were detected using 1% agarose gel electrophoresis.

VIGS vector construction
The tobacco rattle virus (TRV)-mediated VIGS system was used to downregulate gene expression in this study. The target region was selected by using the SGN VIGS online tool (https:// vigs. solge nomics. net/). A 300-bp fragment of PDS (phytoene desaturase gene, as a positive control, LOC107861625) and a 245-bp fragment of Capa-na10g002229 were amplified using specific primers that carried an upstream BamHI restriction site and a downstream EcoRI restriction site (Supplementary Table S3). The plasmid pTRV2 was double digested with BamHI and EcoRI, and then, pTRV2 restriction fragments and target gene fragments were ligated together by T4 DNA polymerase to generate the silencing vectors pTRV2:PDS and pTRV2:Capana10g002229.

Agrobacterium-mediated inoculation in fruit
pTRV1, pTRV2 and pTRV2 derivatives were introduced into Agrobacterium tumefaciens strain GV3101 by using the freeze-thaw method. Agrobacterium tumefaciens cells of pTRV2 vectors carrying target genes were mixed with pTRV1 in a 1:1 ratio (OD 600 = 0.8). The young fruits of BB3 × Chiltepin (F 1 ) at 14 to 18 days after flowering were used for Agrobacterium infiltration with a 1 mL needleless syringe. Uninfected and pTRV2:PDS-infected fruits were used as the blank and positive control, respectively. All individuals were kept in a dark chamber at 20 °C for 2 days. Then, they were moved into a growth room at 22 °C (day)/18 °C (night) with a 16 h light/8 h dark photoperiod cycle with 60% relative humidity.

Inheritance of the deciduous character of ripe pepper fruit
The fruits of Chiltepin at both the mature green and breaker stages were nondeciduous, similar to those of BB3, while at the red ripe stage, the fruits of Chiltepin showed a deciduous character. In contrast, all fruits from BA3 and BB3 individuals displayed a nondeciduous character at any of the three fruit developmental stages. The F 1 (BB3 × Chiltepin) individuals presented a deciduous character of ripe fruit, similar to that of Chiltepin (Fig. 1).
Out of the 79 investigated F 2 individuals derived from the interspecific cross between BA3 and YNXML, 40 individuals exhibited a deciduous character of ripe fruit. Chi-square tests showed that the observed deciduous/nondeciduous segregation (40/39) significantly deviated from the expected monogenic inheritance model (χ 2 = 0, P < 0.01), which could be ascribed to the small population size, interspecific reproduction isolation and CMS. For the BB3 × (BB3 × Chiltepin) population, all 254 individuals flowered and fruited normally. Of these, 124 individuals had deciduous fruits, similar to those of Chiltepin, while 130 individuals had nondeciduous fruits, similar to those of BB3. The observed deciduous/nondeciduous segregation fit the expected 1:1 ratio (χ 2 = 0.14, P > 0.05), indicating that the deciduous character of ripe pepper fruit was controlled by a single dominant gene, which was designated S.

Fine mapping of the pepper S locus
Although the number of individuals with definite phenotypes was small, we performed linkage mapping of the pepper S locus based on the deciduous character of ripe fruit of 79 individuals and the BY-SNP framework map. The results of both the IM and KW methods showed that a unique region at the end of chromosome P10 was closely associated with the deciduous character of ripe pepper fruit (Fig. 2). The peak marker located in this association region was scaf-fold9868.65254, with K* and LOD values of 54.588 and 20.65, respectively (Supplementary Table S4). By combining the results of the above two methods, the S locus could be preliminarily delimited to a high confidence interval between scaffold1125.777179 and scaffold20383.65648, with a physical length of ~ 5.06 Mb (Supplementary  Table S4).
To verify the initial mapping result, a total of eight polymorphic InDel markers were uniformly developed within the 5.06 Mb candidate interval (Supplementary Table S5), and then, they were used to genotype the 254 individuals of the BB3 × (BB3 × Chiltepin) population. Linkage analysis indicated that all eight markers had a tight linkage with the S locus, with genetic distances ranging from 0 to 2.8 cM (Fig. 3a). Among them, four markers, namely FD-244, FD-33, FD-280 and FD-188, were found to cosegregate with the S locus. Consequently, we could localize the S locus into a ~ 224.90-kb region between markers FD-177 and FD-113 (Fig. 3a). Preliminary mapping and fine mapping of the S locus controlling the deciduous character of ripe fruit. a The QTL was preliminarily mapped to a region between InDel markers FD-177 and FD-113 using the BC 1 F 1 population (n = 254). b Candidate region of the S locus based on the genotyping results in a large BC 1 F 2 population (n = 4217). Numbers below the line represent the number of recombinants. The red markers cosegregated with S. c Genotyping results of 13 recombinants. r1-r13 represent the 13 recombinants. B, F 1 and C represent BB3, (BB3 × Chiltepin) and Chiltepin, respectively. White, grid and black rectangles denote the homozygous BB3 genotype (ND, nondeciduous), heterozygous F 1 genotype (D, deciduous) and homozygous Chiltepin genotype (D, deciduous), respectively (colour figure online) To further fine map the pepper S locus, a large BB3 × (BB3 × Chiltepin) BC 1 F 2 population consisting of 4217 individuals was developed and used for screening recombination occurring within the candidate region by using the markers FD-177 and FD-113 as flanking markers (Fig. 3b). As a result, a total of 13 recombinant individuals were identified and then genotyped with the abovementioned four cosegregating markers and nine newly developed InDel markers (Supplementary Table S6). The results showed that FD-33 and FD-640 cosegregated with S ( Fig. 3b; Supplementary Figure S1), indicating that the S gene could be delimited to a 38.03-kb region between markers FD-564 and FD-634 (Fig. 3b, c).

Candidate gene sequence variation between BB3 and Chiltepin
Based on the Zunla-1 reference genome, four protein-coding genes were annotated within the fine mapped region, namely Capana10g002229, Capana10g002230, Capa-na10g002231 and Capana10g002232 (Table 1). Of them, Capana10g002229 was orthologous to the Solanum lycopersicum polygalacturonase-2 (PG2) gene (Supplementary Figure S2), which is involved in cell wall metabolism, specifically in polyuronide degradation (Watson et al. 1994). Capana10g002230, Capana10g002231 and Capa-na10g002232 encoded proteins that were homologous to the Arabidopsis thaliana zinc finger AN1 and C2H2 domaincontaining stress-associated protein 11 (SAP11), cycloartenol-C-24-methyltransferase (SMT1) and homologue of U3 small nucleolar RNA-associated protein 18 (At5g14050), respectively. Of these, SMT1 has been implicated as being involved in the sterol biosynthesis pathway (Diener et al. 2000), while the functions of the other two are unknown.
Sequence comparison analysis between the parent lines BB3 and Chiltepin showed that there was no variation in Capana10g002230 and Capana10g002232 (Supplementary Figures S3 and S4). Two SNPs were found in the cDNA sequence of Capana10g002231, but there was no change in the deduced protein sequences between them (Supplementary Figures S5 and S6). For Capana10g002229, one nonsynonymous SNP and one 7-bp InDel were found between the cDNA sequences of BB3 and Chiltepin (Supplementary Figure S7). We further compared the full-length DNA sequences of Capana10g002229 between BB3 and Chiltepin. Consequently, a total of 34 genetic variations, including 23 SNPs and 11 InDels, were identified in the Capana10g002229 region (Supplementary Table S7). Of these, one SNP in the intron VIII 3′ splice acceptor site, corresponding to a G in Chiltepin and an A in BB3, was found. This SNP leads to a premature stop codon in BB3 (Supplementary Figure S8).

Characterization of candidate gene expression patterns
The cDNA sequences of Capana10g002229 could be amplified by RT-PCR with primer pairs covering the fulllength cDNA of PG (Supplementary Figure S7) at the breaker and red ripe stages in both BB3 and Chiltepin, no product was obtained at the mature green stage (33 days after pollination) in both BB3 and Chiltepin (Supplementary Figure S9). qRT-PCR expression pattern analysis during fruit development showed that Cap-ana10g002229 was strongly expressed in the Chiltepin fruits at the red ripe stage compared to the mature green and breaker stages. For the fruits of BB3, although minor expression of Capana10g002229 was found at both the breaker and red ripe stages, there was basically unchanged between the two stages. Notably, the expression level of Capana10g002229 at the red ripe stage was significantly higher in the Chiltepin fruits than in the BB3 fruits (Fig. 4a). At the mature green and red ripe stages, the expression of Capana10g002230 was higher in the Chiltepin fruits than in the BB3 fruits, but there was no significant change from the mature green to red ripe stage for either Chiltepin or BB3 (Fig. 4b). Capana10g002231 was expressed at very low levels during BB3 fruit development, and relatively high expression was observed in Chiltepin at all three developmental stages; however, similar to that of Capana10g002230, the expression of Capa-na10g002231 in Chiltepin showed no significant change from the mature green to the red ripe stages (Fig. 4c). With regard to Capana10g002232, the expression level was essentially the same between BB3 and Chiltepin at both the breaker and red ripe stage (Fig. 4d). In summary, among the four candidates, Capa-na10g002229 was the only gene that exhibited an expression pattern that was consistent with the phenotypic changes during fruit development. Thus, considering the results of sequence variation analysis, Capana10g002229 could be proposed as a strong candidate gene for S. In addition, expression analysis in various tissues showed that Capana10g002229 was expressed at very low levels in roots, stems, leaves and flowers (Supplementary Figure  S10), indicating that it may be expressed specifically in fruits. Thus, Capana10g002229 was selected for further functional verification through VIGS analysis.

Downregulation of Capana10g002229 delayed fruit softening and shedding
The universal primer pair TRV2-AF406991 (Supplementary  Table S3) was used to determine whether the TRV vectors had been successfully transferred into fruits (Fig. 5a). Of the 30 fruits injected with TRV2:PDS-containing A. tumefaciens solution, 22 fruits were discarded due to decay after 2 weeks of injection, and the remaining 8 fruits developed normally and expanded. All the fruits showed symptoms of PDS gene silencing, that is, the fruit colour changed from dark green to yellow-green (Supplementary Figure S11). With regard to the TRV2:Capana10g002229-infiltrated fruits, 48 out of 150 developed and expanded normally. Of these, a total of 9 fruits displayed different softening and shedding processes compared with the blank control; 8 fruits in which the Fig. 4 Comparative expression analysis of four candidate genes in fruits at three developmental stages between BB3 and Chiltepin. Mature green stage, ~ 33 days after pollination; breaker stage, ~ 44 days after pollination; red ripe stage, ~ 55 days after pollination expression levels of Capana10g002229 were significantly downregulated (Fig. 5b) showed delayed softening and shedding ~ 10 days later than the blank control (Fig. 5c), although 6 fruits turned fully red (Fig. 5d), and 2 fruits were half red and half green near the injection site (Fig. 5e). More intriguingly, the remaining fruit showed harder flesh and a nondeciduous character for more than 72 days after pollination, until it dried up (Fig. 5f).

Discussion
The deciduous character of ripe pepper fruit is considered an ancient trait and is prevalent in the wild ancestors of C. annuum, such as Chiltepin (C. annuum var. glabriusculum) and some cultivars of C. frutescens (Rao and Paran 2003). In the field, we found that the Chiltepin fruits started to become deciduous at the late-breaker stage, and the deciduous character was always observed together with the soft flesh character, which is consistent with a previous report (Rao and Paran 2003). The accomplishment of pepper genome sequencing accelerated the discovery of target genes through a forward genetics approach (Kim et al. 2014;Qin et al. 2014;Hulse-Kemp et al. 2018). Herein, a quantitative trait locus (QTL) responsible for the deciduous character of ripe fruit, namely S10.1, was identified on pepper chromosome P10 by using an F 2 population derived from a cross of BA3 (C. annuum) and YNXML (C. frutescens). In addition, S10.1 was also responsible for the deciduous character of ripe fruit in the BB3 × (BB3 × Chiltepin) population (Fig. 3). This result implied that the deciduous character of ripe pepper fruit originated before species differentiation between the C. . c1-c4 represent the blank control group (no infection). p1-p4 represent the TRV2:PDS group. v1-v8 represent the TRV2:Capana10g002229 group. The first row gel map represents the RT-PCR amplification results for the Blank control and the infiltrated fruits with TRV2:PDS (Expected amplification product fragment size is 681 bp) and TRV2:Capana10g002229 (Expected amplification product fragment size is 626 bp) by using the universal primers TRV2-AF406991 (Supplementary Table S3). The second row gel map represents the RT-PCR amplification band (154 bp) of the internal reference gene Ca-Actin (GenBank: GQ339766) (Supplementary Table S2). The target bands were detected using 1% agarose gel electrophoresis. b Relative expression level of Capana10g002229 in the fruits of the control and TRV2:Capana10g002229 groups. *represents significance at the 0.05 level (Student's t test). c Comparison of the days of fruit softening and shedding between the control and TRV2:Capana10g002229 groups. d-f Red ripe fruit infected with TRV:Capana10g002229. White arrows indicate the infiltrated sites frutescens and C. annuum and pepper domestication from C. annuum var. glabriusculum to C. annuum var. annuum.
To our knowledge, it is the first time the S locus was delimited into a relatively precise candidate interval (38.03 kb) based on a solid forward genetic mapping strategy. During this process, two cosegregating InDel markers (FD-33 and FD-640) were developed ( Fig. 3 and Supplementary Figure S1), and they can be applied to identify the deciduous character of red ripe fruits at the seedling stage during the process of variety improvement. As a result, individuals with nondeciduous character of ripe fruit can be pulled out at seedling stage, saving a lot of labour and reducing the cost of managing a large number of seedlings compared to the traditional breeding methods.
Furthermore, four protein-coding genes were annotated within the fine mapped region (Table 1). Of these, Capa-na10g002229 was proposed as a strong candidate gene for S based on the results of sequence variation analysis and expression pattern analysis in fruits at different developmental stages ( Fig. 4 and Supplementary Figure S7). The cDNA sequence was amplified by RT-PCR with primer pairs covering the full-length cDNA of Capana10g002229 at the breaker and red ripe stages in both BB3 and Chiltepin, and only no product was obtained at the mature green stage in both BB3 and Chiltepin (Supplementary Figure S9). Furthermore, the qRT-PCR-based expression analysis also showed that the Capana10g002229 gene was expressed at the breaker and the red ripe stage in both BB3 and Chiltepin (Fig. 4a) although significant differences in expression are present. These results were inconsistent with the results of the previous study by Rao and Paran (2003), in which PG was neither expressed in any ripening stage including the mature green, breaker or red fruits of the cultivar Maor, nor in the mature green fruits of accession BG 2816, but its transcription was detected in the breaker and red ripe stages of BG2816. We attribute these discrepancies to the use of different materials with different genetic background.
In this study, one nonsynonymous SNP and one 7-bp InDel were found between the cDNA sequences of BB3 and Chiltepin, and both were found to be located in the conserved regions of Capana10g002229, encoding a polygalacturonase-2, belonging to the glycosyl hydrolase 28 family (Table 1, Supplementary Figure S7). Of these, the 7-bp deletion in the cDNA sequence of BB3 was derived from the SNP in the intron VIII 3′ splice acceptor site, changing a G in Chiltepin to A in BB3, leading to a premature stop codon in BB3 (Supplementary Figure S8). Interestingly, the same variation was found to be genetically cosegregated with the fruit softening character of the soft flesh mutant (Kim et al. 2014). Therefore, the SNP in the intron VIII 3′ splice acceptor site was considered an important candidate causal variation for the change in the deciduous character of ripe fruit; at present, however, the possibility of the nonsynonymous SNP in exon 3 being the causal variation cannot be ruled out. More evidence is still needed to explore which specific locus of variation is responsible for the phenotypic difference in the deciduous character of ripe fruit. Anyhow, our study provided more allele variation information for the Capana10g002229 (PG) in pepper.
Capana10g002229 was orthologous to the PG2 gene in tomato (Supplementary Figure S2), which is involved in cell wall metabolism, specifically in polyuronide degradation in fruit ripening and softening processes (Watson et al. 1994). To date, several PG genes related to fruit ripening, softening and cell separation have been isolated from tomato (Bird et al. 1988), peach (Pressey and Avants 1973), pear (Pressey and Avants 1976), papaya (Chan and Tam 1982) and Arabidopsis (Ogawa et al. 2009). In this study, we successfully downregulated the expression of Capana10g002229 in pepper fruits by direct agroinfiltration of young fruits for the first time. Downregulation of Capana10g002229 delayed softening and ripening for most infected fruits; however, they were still accompanied by softening and the deciduous character of ripe fruit, which may be related to the fact that VIGS alone could not completely silence the gene (Burch-smith et al. 2004). Similar phenomena were also observed in transgenic tomato with the antisense PG gene; in these plants, the expression level of PG was substantially reduced, but the fruits showed normal ripening and softening (Smith et al. 1990). The mechanism by which Capana10g002229 causes the deciduous character of ripe pepper fruit remains to be further studied.