SLAF marker based QTL mapping of fruit-related traits reveals a major-effect candidate locus ff2.1 for flesh firmness in melon

CHEN Ke-xin, DAl Dong-yang, WANG Ling, YANG Li-min, Ll Dan-dan, WANG Chao, Jl Peng, SHENG Yun-yan

Department of Horticulture and Landscape Architecture, Heilongjiang Bayi Agricultural University, Daqing 163309, P.R.China

Abstract Flesh firmness (FF) is an important and complex trait for melon breeders and consumers.However, the genetic mechanism underlying FF is unclear.Here, a soft fruit melon (P5) and a hard fruit melon (P10) were crossed to generate F2, and the FF and fruit-related traits were recorded for two years.By performing quantitative trait locus (QTL) specificlocus amplified fragment (SLAF) (QTL-SLAF) sequencing and molecular marker-linkage analysis, 112 844 SLAF markers were identified, and 5 919 SNPs were used to construct a genetic linkage map with a total genetic distance of 1 356.49 cM.Ten FF- and fruit-related QTLs were identified.Consistent QTLs were detected for fruit length (FL) and fruit diameter (FD) in both years, and QTLs for single fruit weight (SFW) were detected on two separate chromosomes in both years.For FF, the consistent major locus (ff2.1) was located in a 0.17-Mb candidate region on chromosome 2.Using 429 F2 individuals derived from a cross between P5 and P10, we refined the ff2.1 locus to a 28.3-kb region harboring three functional genes.These results provide not only a new candidate QTL for melon FF breeding but also a theoretical foundation for research on the mechanism underlying melon gene function.

Keywords: QTL mapping, flesh firmness, fruit-related traits, melon, SLAF sequencing

1.lntroduction

Melon (CucumismeloL.), which belongs to the Cucurbitaceae family, is an important diploid species(2n=12) in terms of fruit production and consumption worldwide (Farcuhet al.2020).In total, 13.557 million tons of melon fruit were produced in China in 2019, and approximately 3 350 tons were exported in 2020, worth 45.798 million USD.Melon fruit has various features that affect marketability, especially texture, color and shape.Flesh firmness (FF) is one of the key traits for melon fruit quality, consumer acceptance and fruit transportability(Nimmakayalaetal2016).FF is a complex trait that involves pericarp structure and cell wall characteristics(Cha?bet al.2007; Uluisiket al.2016; Zaridet al.2021).Melon has been used as a model species for studying fruit ripening because both climacteric and nonclimacteric varieties are available (Ayubet al.1996).Research has been conducted on fruit ripening and softening,particularly with respect to the relationship between fruit ripening and cell wall enzyme activities (Goulao and Oliveira 2008; Toivonen 2008), and studies have indicated that the texture of melon fruit may be established during early fruit development (Cha?bet al.2007).

Quantitative trait loci (QTLs) and genes for FF have been widely studied in tomato (SolanumlycopersicumL.)and watermelon (Citrulluslanatus) (Guoet al.2015; Gaoet al.2020; Yanget al.2021).Regarding the regulation and molecular mechanism of climacteric, tomato fruit flesh is the most widely studied model.In tomato,several QTLs associated with FF have been identified(Tanksleyet al.1996; Fraryet al.2003; Chapmanet al.2012).Recently, several genes have been shown to play key roles in FF.These genes include the pectate lyase(PL) gene and the tomato AGAMOUS-Like 1 (TAGL1)gene, whose silencing may influence cuticle thickness and whose activity affects FF (Giménezet al.2015).More recently, theqFIRMSKIN1(qFIS1) gene, which encodes a GA2 oxidase, was identified, and its mutation led to increased bioactive gibberellin contents, enhanced cutin and wax biosynthesis, as well as increased FF and shelf life (Liet al.2020).In watermelon, rind hardness(RH) and related traits were determined using bulked segregant analysis sequencing (BSA-seq), and two RH QTLs were identified (Sunet al.2020; Yanget al.2021).Unlike tomato fruit flesh, the softening of melon fruit flesh is partially controlled by ethylene, and FF is associated with the fruit ripening progress (Morenoet al.2008;Galpazet al.2018; Zaridet al.2021).The most widely used genotype for studying FF was developed using the nonclimacteric accession PI161375 (ssp.agrestis) and the nonclimacteric Spanish cultivar PS as recurrent parents.Several QTLs associated with FF have been identified in melon (Morenoet al.2008; Galpazet al.2018; Pereiraet al.2020), and these QTLs are located across all 12 melon chromosomes.Research on the interactions between QTLs for ethylene biosynthesis during melon fruit ripening revealed a QTL for FF in the same region asETHQV6.3, which indicated that FF is probably due to the pleiotropic effect ofETHQV6.3(Vegaset al.2013).

Next-generation sequencing (NGS) technology has facilitated QTL detection and gene analysis because of the knowledge generated for hundreds of thousands of singlenucleotide polymorphisms (SNPs) (Bhatet al.2016).NGS techniques, including bulked segregant analysis(BSA) and genome-wide association studies (GWASs),have been widely used to detect major QTLs in tomato,cucumber, tetraploid rose, and watermelon (Chapmanet al.2012; Huet al.2018; Xuet al.2018; Yiet al.2020;Chaoet al.2021; Yanget al.2021).The melon genome sequence and annotated genes were made available for the first time in 2012 at http://melonomics.net (Garcia-Maset al.2012), and more than 90% of the scaffold assembly is anchored in the new version (Castaneraet al.2020).The availability of the melon genome facilitates mapbased cloning of the genes involved in fruit-related trait research (Sahuet al.2020).Specific length amplified fragment sequencing (SLAF) is a new and efficient method for constructing high-resolution genetic maps;and this method has been applied to sesame (Zhanget al.2013), soybean (Qiet al.2014; Liet al.2020),common carp (Sunet al.2013) and cucumber (Zhuet al.2016).More recently, QTLs of gummy stem blight (GSB)resistance genes were identified using SLAF sequencing(SLAF-seq) and BSA techniques (Gaoet al.2020), and flowering-related QTLs were identified in melon using SLAF-seq (Wanget al.2017).

Fruit maturity-related traits, including fruit ripening,FF, flesh thickness, fruit color, and fruit juice content,are important in melon (Kesh and Kaushik 2021).In the present research, genotype data for 119 F2melon plants were generated, and SNPs were discovered using SLAF-seq.A high-density genetic map of melon was also constructed using a large F2population.The FF QTLs detected across three years (generations) were also investigated.This work provides novel insights into the genetic control of fruit firmness that may be useful in the application of marker-assisted melon breeding to improve melon fruit quality.

2.Materials and methods

2.1.Materials

The melon inbred line P5 was obtained from Daqing Agricultural Institute, China.It is thin skinned and the fruit is soft (FF of ～3.4 kg cm–2) with a single-fruit weight(SFW) of ～0.45 kg, a 12-cm fruit length (FL) and an 8.3-cm fruit diameter (FD).P5 was crossed with melon line P10 from Zetian Seed Company of Heilongjiang Province,China, which has fruit that are moderately large (～0.74 kg SFW, 16 cm FL and 10 cm FD) and hard (FF of 8.5 kg cm–2) (Fig.1).We obtained permission from Zhengzhou Fruit Research Institute, Chinese Academy of Agricultural Sciences to collect the plants for this study.In total, 119 F2and F2:3families were grown in a greenhouse in autumn 2020 in Sanya and in spring 2021 in Daqing, respectively.A large group (429) of individual F2plants was planted in an open field in Daqing in the spring of 2021.

2.2.Phenotypic data collection

Fig.1 Variations of fruit related traits in P5, P10 and their derivatives.Mature fruits of the two parental lines and their F1 are shown in A.Note the significant variations in fruit length, fruit diameter, fruit weight and flesh firmness between the parental melon lines.B is an image from the 2021 field trial showing variations in fruit related traits in the F2 plants.C1 and C2 are representative fruit images among different F3 generations from the 2020 and 2021 tests for variations in fruit traits from 11 plants in each line.

Phenotypic data from the F2:3families in 2020 and 2021 were collected in the two environments for 2 years(autumn of 2020 in Sanya, Hainan Province, 108.56°latitude, 18.27° longitude; spring of 2021 in Daqing,Heilongjiang Province, 125.03° latitude, 46.58° longitude).Parental lines and F1plants were included in all the experiments with 10 plants.Members of the F3population were arranged in a randomized complete block design,and a large population comprising 429 individual F2plants(429 F2-2021) was also planted in the spring of 2021 in Daqing at the Heilongjiang Bayi Agricultural experimental station.The F3plants in 2020 and 2021 were grown in open fields, with 150 cm row spacing and 40 cm plant spacing.Each family comprised ～15 individual plants,and each plant produced 2–3 fruits.FL and FD were measured for a single fruit (cm), SFW was measured for mature fruit (g), and FF was measured using a hand penetrometer fitter equipped with a 5-mm cylindrical probe (GY-4 fruit pressure tester, China).All phenotypic data were recorded as the averages of three replicates.

2.3.DNA extraction and SLAF-seq

Leaves from the parental lines and 119 individual F2-2020 plants were extracted using the cetyl-trimethylammonium bromide (CTAB) method (Luanet al.2010).The leaf DNA was diluted to 75–100 ng mL–1and then quantified with an ND-2000 Spectrophotometer (NanoDrop, Wilmington, DE,USA).A total of 119 F2-2020 plants and parental lines were used for genotyping with SLAF-seq (Daiet al.2022).The purified product was sequenced using the Illumina HiSeq 2500 System by Biomark Inc., China (http://www.biomarker.com.cn), in which the melon (Melon (DHL92)genome 3.6.1) (http://cucurbitgenomics.org/organism/18)genome sequencing data constituted a control library for evaluating the accuracy of the SLAF library construction.The SNP information of parent sequence data sets supporting the results of this article is available in the NCBI SRA database (https://www.ncbi.nlm.nih.gov/) under project accession number SUB11147277.

2.4.Sequencing data evaluation

The original sequencing data were filtered before analysis to ensure the quality of the information analysis.Reads containing linker sequences and reads whose N content was >10% of the total length were removed;reads with lengths ranging from 4–103 bp were selected for analysis, and the total length was 2×100 bp.Each sample sequencing dataset was analyzed for the number of reads, Q30 content and GC content.Based on the general genetic coding rules, the parental genotypes were aa and bb, and for the F2mapping population,SLAF markers with aa×bb patterns were used for map construction.

2.5.Genetic map construction and QTL mapping

First, the recombination rates and maximum likelihood(MLOD) values between markers were calculated.Then, the molecular markers were divided into different linkage groups (LGs) according to their MLOD value,and each LG represented a chromosome.Based on the chromosomes, a genetic map was constructed using the MLOD method, yielding an initial version of the marker sequences.Mapping was performed using the Kosambi function.R/QTL Software was used for the composite interval mapping (CIM) analysis, and the threshold of QTLs was determined by performing permutation tests(P<0.05).The LOD threshold was determined as the 95%confidence interval according to 1 000 permutations, and QTLs for which LOD=2.5 were considered putative QTLs related to traits in the genomic region.

2.6.Narrow mapping of the QTL of ff2.1 for the candidate region

Twelve cleaved amplified polymorphic sequence (CAPS)markers were designed by resequencing data from the parental lines using Primer 5 Software (Table 1), a genetic map was re-constructed using 429 F2-2021 individuals,and recombination F2individuals were used to narrow down thefft2.1QTL region.The PCR reactions were in a final volume of 10 μL, consisting of 1 μL of template DNA, 2 μL of 2×TaqPCR Mix, 1 μL each of forward and reverse primers, and double-distilled water for the remainder.The PCR amplification program was as follows: predenaturation at 94°C for 7 min; 30 cycles of denaturation at 94°C for 1 min, renaturation at 60°C for 30 s,with a 0.5°C reduction per cycle, and extension at 72°C for 90 s; and a final extension at 72°C for 10 min.

DNA sequence analysis for the candidate region used Softberry to predict candidate gene annotations, and BLAST was used to determine the functions of the genes in the Cucurbitaceae Database (http://cucurbitgenomics.org/).

Quantitative real-time PCR (qRT-PCR) was performed on a CFX96 Touch qRT-PCR machine (CFX96 Touch, Bio-Rad, USA) in conjunction with SYBR Green PCR Master Mix (TIANGEN Corp., China).RNA from the flesh of fruit at different developmental stages from the parental lines was extracted; and the stages corresponded to 10 days after pollination (DAP), 15 DAP, 20 DAP and 25 DAP.qRT-PCR was performed on a 10-μL mixture consisting of 1 μL of template cDNA (100 ng μL–1), 5 μL of 2× SYBR Green PCR Master Mix, 0.75 μL of forward and reverse primers (10 μmol L–1) and 2.5 μL of ddH2O.The qRT-PCR conditions were as follows: 95°C for 3 min; 40 cycles of 10 s denaturation at 95°C followed by 30 s of annealing and extension at 60°C; and a melting cycle of 0.5°C increases per 10 s from 65 to 95°C.All the samples were analyzed for three technical and three biological replicates,and samples without template were used as controls.The values from the triplicate reactions were averaged, and the relative expression levels were measured by means of the 2–??CTmethod (Livak and Schmittgen 2001).

3.Results

3.1.Phenotypic traits

Averages of all the phenotypic traits of fruit for ～10 fruit from the parental lines were recorded for the two years.The average fruit weight (FW) of P5 (average of 403.2 g)was significantly lower than that of P10 (average of 851.45 g).For FF, the mature fruit of P5 was soft, with a flesh firmness of 3.6 kg cm–2, and the mature fruit of P10 was hard, with a flesh firmness of 8.5 kg cm–2(Fig.1).The average values of FL in the two years were 12.1 and 16.9 cm for P5 and P10, respectively, and the average values of FD in the two years were 8.3 and 9.9 cm for the parental lines, but the difference was not significant at the 0.05 level.For the F2population, 119 and 429 individuals were investigated in 2020 and 2021, respectively.The results indicated that the average FF and SFW in 2021 were more than those in 2020, but the FL and FD were not different.The average FF was 5.4 kg cm–2in 2020 and 5.7 kg cm–2in 2021.These differences were observed because the plant number was significantly different between the two years.The SFW was 798.9 g in 2020 and 814.4 g in 2021.For the F3families, the average FF ranged from 1.4 to 10.4 kg cm–2in 2020 and from 2.3 to 7.3 kg cm–2in 2021.The average FL was 15.8 cm in 2020 and 17.2 cm in 2021, and the FD was 8.5 cm in 2020 and 8.8 cm in 2021 (Table 2 and Fig.1).The SFW ranged from 257.4 to 1 002.1 g in 2020 and from 272.3 to 1 120.1 g in 2021.All traits showed continuous variation in the F3population, and the average values of SFW, FF, FL and FD in the F1population were 855.9 g, 5.4 kg cm–2, 16.45 cm and 8.5 cm, respectively.Similarly, the average FL ranged from 4.4 to 16.3 cm and from 5.3 to 17.1 cm, with FDs ranging from 4.3 to 10.3 cm and from 4.9 to 13.7 cm for the F3population during the two years (Fig.2).

Table 1 Molecular primer sequences for fine mapping

The frequency distributions of FF, SFW, FL and FD in the 429 F2-2021 were largely normal, indicating the quantitative nature of these traits (Fig.3).The FF and FD of the F1plants were close to the midparental value, but the SFW of F1was closer to that of P10.The FL of the F1population was much greater than that of P10, whose fruits were long (Table 2).The SFW of the F1plants seemed to show some heterotic effects.

Correlations between the different traits in the two years were examined.Spearman’s rank correlation coefficients (rs) between FF/SFW, FL, and FD are presented in Table 3.Significant correlations between SFW and FL and FD in the two years were detected(Table 3), and a correlation between FL and FD was also detected in both years (Table 3).However, FF and the other traits were not correlated in either year.

3.2.SLAF-seq analysis and genetic map construction

In total, 627 311 039 clean reads were obtained by analyzing both parents and 119 F2-2020 offspring.For P5,30 578 577 clean reads with a sequencing Q30 of 93.56%were obtained, and for P10, 30 080 137 clean reads with a sequencing Q30 of 93.70% were obtained.TheRsaI andHaeIII enzymes were used to digest the genomic DNA of the parents, and the F2plants presented an enzyme digestion efficiency of 94.61%.A total of 186 893 SLAF markers were obtained, and 112 844 SLAF markers were obtained after standard filtering.Finally, 5 919 SNP markers were used to construct genetic linkage maps.

After quality assessments, 5 845 SNP markers were ultimately used to construct a high-density genetic map of the 12 melon chromosomes.The total genetic distance of the map was 1 356.49 cM, and the average genetic distance between two adjacent markers was 0.537 cM(Table 4 and Fig.4).Significant differences were observed in areas with 45 to 938 markers.Chromosome(Chr.) 4 had only 45 markers, whereas Chr.11 had 938 markers; the genetic lengths ranged from 85 to 130 Mb,and the average distances between two markers ranged from 0.2 to 0.6 cM.All the SLAF markers were consistent with the melon genomic data, and a high-collinearity consensus map with an average Spearman coefficient of 99.49% was obtained (Fig.4).

3.3.QTL mapping of fruit-related traits

Ten QTLs for FF and fruit-related traits were identified during the two years.Two QTLs for SFW (2020-sfw5.1and2021-sfw4.1) were detected on Chr.5 and Chr.4in 2020 and 2021, respectively.The LOD values and phenotypic variances were 3.2 and 13.16%, respectively,in 2020 and 3.0 and 16.53%, respectively, in 2021, but each QTL was detected in only a single year (Table 5).All alleles from the P10 melon QTL contributed to increased SFW (i.e., positive additive effects).QTL2020-sfw5.1was located between 2 020 187 and 2 377 674 in the melon genome, which spans a 0.36 Mb region that contains 42 candidate genes.QTL2021-sfw4.1was located between 2 888 303 and 4 092 402 in the melon genome, which spans a 1.20 Mb region that contains 142 candidate genes.QTL2020-sfw5.1had negative additive effects that might reduce FW, but2021-sfw4.1exerted positive additive effects that contributed to increased fruit elongation.

Table 2 Means of the fruit traits in the two years1)

Fig.2 Fruit trait performance of the two parents and their F1, and frequency distribution of among F3 populations in the two years.Data for flesh firmness (FF), single fruit weight (SFW), fruit length (FL) and fruit diameter (FD) were collected in the F3 populations from the P5×P10 cross across two years (2020 and 2021).

For FL, two QTLs during the two years (2020-fl2.1and2021-fl2.1, with LOD values of 2.9 and 6.1, respectively)were detected on Chr.2 and explained 5.09 and 23.81%,respectively, of the phenotypic variance.These two QTLs were located at approximately the same position on Chr.2.2020-fl2.1was located between 941 136 and 967 546 in the melon genome, which spans 0.3 Mb that contains fivecandidate genes, while2021-fl2.1was located between 822 796 and 854 529 in the melon genome, which spans a 0.3-Mb region that contains five candidate genes.These two loci were close to each other and may not be separate, and the same QTL for a larger population was used for further mapping.

Table 3 Correlation coefficients of single fruit weight (SFW),fruit firmness (FF), fruit length (FL), and fruit diameter (FD) in the two years

Table 4 Genetic map construction by SLAF sequencing

Fig.4 Linkage map and chromosomal locations of QTL for flesh firmness (FF).QTL analyses for FF, single fruit weight (SFW),fruit length (FL) and fruit diameter (FD) based on data from the 2020 and 2021 F3 experiments.Previous studies of FF are also listed to compare the loci.Numbers to the left of each chromosome (Chr.) show the map length in centi Morgans (cM).Vertical bars represent the 2.5-LOD support interval of each QTL.The original QTL names from each publication are used.QTL with asterisks are major- or moderate-effect QTL (R2>10%) in the respective studies.Red-colored QTL for the present research include FF, SFW, FL, and FD.Blue-colored FF QTL from Pereira et al.(2020) and Vegas et al.(2013); dark purple-colored FF QTL from Amanullah et al.(2018); yellow-colored FF QTL from Moreno et al.(2008), and green-colored FF QTL from Dai et al.(2022) in our research group.

Furthermore, two QTLs for FD in the two years were detected at the same position (2020-fd5.1and2021-fd5.1)on the same chromosome, indicating that this QTL was a stable and major QTL.The LOD values for FD were 4.1 and 2.5, and the percentages of phenotypic variance were 17.14 and 10.98%, respectively (Table 5 and Fig.4).For both FL and FD, the alleles2020-fl2.1and2021-fl2.1contributed to reductions in FL and FD (negative additive effects).Notably,2020-fd5.1and 2021-fd5.1were located between 1 565 061 and 1 788 118 in the melon genome and contained 19 candidate genes.

Four QTLs for FF were detected on Chr.2, Chr.5 and Chr.10.The LOD values of these QTLs ranged from 4.8 to 10.8, and the percentages of phenotypic variance explained (PVE) by these QTLs ranged from 8.32 to 30.11% (Table 5).The QTLs on Chr.2 (2020-ff2.1and2021-ff2.1) were detected at the same position and these QTLs explained more than 15% of the phenotypic variance, indicating that these QTLs were stable and major (Fig.5-A).Importantly,2020-ff2.1,2020-ff5.1and2021-ff2.1contributed to a reduction in FF (negative additive effects), but2020-ff10.1contributed to an increase in FF (positive additive effects).The2020-ff2.1and2021-ff2.1regions spanned 0.17 Mb and contained 16 candidate genes.For the F2:3populations, some test entries presented SFW and FF values that were larger than those of P10 and the F1, which might be explained by the opposite effects of the QTLs detected.

3.4.Validation and refinement of the ff2.1 QTL in a large population

SLAF-QTL was performed on 119 F2-2020 genotypes produced from a P5×P10 cross and the offspring of these 119 plants in the F2:3population across two years to confirm the effects on FF.The QTL for FF,ff2.1, was reanalyzed for a larger F2population comprising 429 individuals in 2021.These individuals were obtained from the 2020 F2experiment, with the 119 sequencedplants excluded.Twelve CAPS markers were designed according to the resequencing data of the parental lines and used to confirm the position offf2.1between SNP426 and SNP753 in the melon genome(the genome position was 23 531 942 to 23 699 258) (Table 1).One stable and major QTL was detected with a relatively high LOD score,and the position of the peak was consistent with those detected using the F2:3population.The LOD profiles forff2.1-F2are illustrated in Fig.5-B and C.Furthermore, the flanking region of this locus was analyzed in 36 recombinant plants.

Table 5 Information on the QTLs detected for fruit-related traits in F2 populations derived from the cross between P5 and P10 in the two years1)

For recombinant No.5, the FF was 10.4, and the phenotype of F2-5 was H.Alleles for SNP482 to SNP7426 of F2-5, showing that the left segment had originated from P10, while the right segment from SNP426 to SNP087 originated from chromosome P5.The phenotype and genotype of recombinant F2-5 indicated a left margin of the target region at SNP426.For recombinant No.120, the FF was 2.4, and the phenotype was A.Alleles for SNP0870 to SNP753 of F2-120, showing that the left segment had originated from P5,while the right segment for SNP753 to SNP482 was heterozygous.The phenotype and genotype of recombinant F2-120 dictated a right margin of the target region at SNP753.Taken together, these results suggested that SNP426 to SNP753 was the candidate region for FF.These results were consistent with those of the gene mapping study using the larger population (429) of F2plants (2021).This candidate region spanned 28.3 kb, after alignment to the melon reference genome (Melon (DHL92) v3.6.1 Genome, https://www.icugi.org/pub/cucurbit/genome/melon/v3.6.1/), which contained three functional genes: ABC transporter family protein (MELO3C017401),phospholipase A2-alpha (MELO3C017400), and defensin-like protein 3 (MELO3C029736) (Fig.5-C).

In this study, three candidate gene were aligned, so to understand the gene expression pattern, qRT-PCR analyses were performed on the parental lines at different fruit development stages, including 10 DAP, 15 DAP, 20 DAP and 25 DAP.The results indicated no significant differences in the expression ofMELO3C017401andMELO3C029736between the different developmental stages.However, the expression ofMELO3C017400at 15 DAP was significantly different between parental lines P5 and P10.At 10 DAP, gene expression was similar in P5 and P10, but at 15 DAP, the gene expression in P10 was approximately three times higher than in P5.As the fruit matured,MELO3C017400gene expression was substantially higher in P10, and was five times higher than in P5.At 20 DAP, gene expression peaked in P5 and P10, suggesting that this stage represented the key time point for fruit ripening and softening.For P5, the gene expression levels were similar, and no differences were observed at 20 and 25 DAP (data not shown).For P10, no differences were observed between these two stages, whereas significant differences were detected in P5 and P10 at 20 and 25 DAP(Fig.6).Gene expression peaked at 20 DAP in P5 and P10, after which the expression declined in P10 but remained at a similar level in P5.P5 and P10 fruit matured at 20 DAP and ripened and softened beginning at 30 DAP, after which the green fruit flesh turned whiteyellow beginning at 25 DAP (data not shown).Considering the qRTPCR gene expression data and phenotypes, 20–25 DAP was a key time point for the thin-skinned melon.

Fig.5 LOD curves of QTL for the flesh firmness (FF) traits detected in a large F2 population.A, QTL for FF by SLAF-seq technology in two years in chromosome 2 were detected with 429 F2 plants of a P5×P10 cross from the 2021 field trail.B, horizontal dashed lines are LOD thresholds for declaring significant QTL.Horizontal red bars define 3.0 LOD intervals of each QTL.Fine mapping based on 429 F2 plants with the phenotype of F2:3 in 2021 defined as the candidate locus to a 28.3 kb region between markers SNP426 and SNP753.C, the dark color codes are homozygous ‘P5’, the white color codes are homozygous ‘P10’, and the striped color codes are heterozygous genotypes.Three coding genes predicted in the 28.3 kb region, and the boxes represent genes.

4.Discussion

Many economically important traits of melon fruit, such as FW, FL, FD and FF, are controlled by QTLs (Wanget al.2016; Amanullahet al.2018; Pereiraet al.2018;Liuet al.2019).Based on high-throughput sequencing technology, SNPs and insertions-deletions (InDels) have been identified and used for QTL mapping inCucumis sativusL.(Zhuet al.2016),Vitis vinifera(Wanget al.2016),BrassicanapusL.(Wenet al.2018),Pisum sativum(Zhenget al.2018), andBenincasahispida(Jianget al.2015).In melon, various mapping populations,including F2plants, F2:3families, recombinant inbred lines(RILs), and near-isogenic lines (NILs), have been used for QTL mapping (Zalapaet al.2007; Cuevaset al.2008,2009; Wanget al.2016; Amanullahet al.2018; Pereiraet al.2018).In the present study, a high-density genetic map was constructed, and several QTLs were detected in the same region or in close regions of the same chromosome in the two study years, indicating that this high-density genetic map provides accurate and reliable measurements.

In our study, QTLs for FL (in 2020 and 2021) were located on Chr.2, and these QTLs spanned 0.03 Mb and contained four genes in both years.The QTLfl2.1was identified as a stable and accurate QTL for FL in the present study, and its PVE was greater than 15%,indicating that this locus is a major-effect QTL.The detected FL QTLs were consistent across the two years,which was also consistent with results from earlier publications (Harel-Bejaet al.2010; Díazet al.2014,2017).Compared with previous research,fft2.1was near the location found previously, in which a 38.4-kb region of fruit firmness was located in Chr.2 using F2plants from PS and PI124112 (Harel-Bejaet al.2010).By combining the results of previous research and this study for melon FL QTL mapping, Chr.2 is a candidate target region for melon FL.

Fig.6 Relative expression levels of three candidate genes in soft flesh firmness melon P5 and hard flesh firmness P10.A,expression levels of MELO3C017401.B, expression levels of MELO3C017400.C, expression levels of MELO3C029736 were quantified using the 2-ΔΔCT method.For the two parents,the expression levels of the respective genes are shown at 10, 15, 20 and 25 days after pollination (DAP).Each sampling was repeated three times, and 15 plants were mixed in equal amounts to form one replicate in the two parents.Data are mean±SE (n=3).＊＊, indicates an extremely significant difference,P<0.01.

FD QTLs were also consistently detected in the same region in the present study.Notably,2020-fd5.1and2021-fd5.1were located on Chr.5 between 1 630 691(1 565 061) and 178 818 in the melon genome, which spanned approximately 0.22 Mb.The use of the F2generation produced from an AR5×Earl favourite cross revealed FD QTLs on LG5 and LG10 (Fukinoet al.2008).FD QTLs in a RIL population resulting from a Vedrantais×Piel de Sapo cross were shown to be located on Chr.5 and Chr.8 (Pereiraet al.2018), and introgression lines (ILs) resulting from a Ginsenmakuwa×Vedrantais cross were used to determine that FD QTLs were located on Chr.1, Chr.5, Chr.6 and Chr.11 (Perpi?áet al.2016).Some QTLs for FD were located on Chr.5 when different populations (melon types) were used, indicating that Chr.5 may harbor an important region for FD.In previous melon research, fruit size (shape) and FW were strongly correlated with FD, and several detected QTLs for FD were collocated in the same region as QTLs for FW (Pereiraet al.2018; Lianet al.2021).In the present study, QTLs for SFW and FD were not consistently found in the same regions, but unlike in previously published research,fd5.1and especiallyfd5(Perpi?áet al.2016)were located near each other.

QTLs for SFW identified during 2020 and 2021 were located on Chr.4 and Chr.5, respectively; however, thesfw-2020andsfw-2021QTLs were not consistent in the two years.For the qualitative traits, the environment might affect performance, which may have led to the twoyear SFW difference.Previous research has investigated the QTLs for FW on Chr.2, Chr.5 and Chr.10 (Amanullahet al.2018), on Chr.3, Chr.4, Chr.5 and Chr.12 (Monforteet al.2004), Chr.2 and Chr.5 (Wanget al.2016), and on Chr.1, Chr.2, Chr.6 and Chr.11 (Pereiraet al.2018).FW always shows strong correlations with FL and FD(Amanullahet al.2018).Recently, theFWQU5.1QTL for FW was identified on Chr.5 in a RIL population, and this QTL resulted in an increased FW when the male parental allele was present.More recently, 57 FW QTLs identified in melon in previous studies were summarized,and these QTLs were distributed across all 12 melon chromosomes, with the QTLs on Chr.6 and Chr.8 subsequently researched in depth (Panet al.2020).All these results indicate that the genetic inheritance of FW is more complex.

FF is an economically important trait related to fruit storage, particularly in recent years, and breeding efforts to enhance fruit melon shelf life are ongoing (Kesh and Kaushik 2021).To date, more than 20 QTLs for flesh/fruit firmness have been identified, which are present across approximately all melon chromosomes.Five QTLs for FF were discovered in melon NILs resulting from a cross between the nonclimacteric parental lines PI161375 and Piel de Sapo (Morenoet al.2008).Approximately five QTLs for fruit firmness have been discovered on Chr.2,Chr.3, Chr.5 and Chr.8, while two QTLs for FF (individually)were identified on Chr.3 and Chr.6 (Amanullahet al.2018).Recently,FIRQV2.1andff2.2were found to be located close to each other at the top of Chr.2 (Pereiraet al.2020), and these loci were near another locus(Morenoet al.2008).A major QTL of FF (FIRQV2.2) was also mapped to the end of Chr.2 (Pereiraet al.2020).More recently, Pereiraet al.(2021) identified a QTL(FIRQP2.1) located at the beginning of Chr.2 (0–16.42 Mb) that could be the main factor governing fruit flesh softening in nonclimacteric melon germplasm.Compared with our previous results concerning the QTLs for fruit firmness,FF2.1was found by SLAF-BSA technology to be in the same region asff2.1in an F2:3population resulting from a cross between the wild melon genotype 1244 and MS5, which has thick skinned fruit (Daiet al.2022) (Fig.4).The results indicated that chromosome 2 was still a target region for FF (Fig.5).All the results indicated that chromosome 2 was important for melon fruit firmness andff2.1was a consistent candidate region for FF in melon (Fig.4).

The candidate annotation gene phospholipase A2 (PLA2) is a precursor of eicosanoids, including prostaglandins (PGs) and leukotrienes (LTs).The secretory PLA2 (sPLA2) family has been shown to be involved in a number of biological processes (Murakami and Kudo 2002).Previous research has revealed interactions among hormones, phospholipases and several genes (Romeroet al.2013), and phospholipase D(PLD) activity increases the response to abscisic acid(ABA) in bean and to ethylene inArabidopsis(Romeroet al.2013).In the present study,CMPLA2(Cucumis melo) was suggested to be a candidate gene in QTLff2.1, and the gene expression data indicated a significant difference in expression between the parental lines.We aligned the gene sequences of P5 and P10, and compared the data with those in the DHL92 reference genome sequence database.This comparison identified one nonsynonymous mutation that resulted in the conversion of “ATA” to “GTA”, which induced a change in the protein sequence from “R” to “K”, and the sequence of P10 was the same as that of the reference genome database while the sequence of P5 was not (Appendix A).Nonetheless, we cannot conclude whether this mutation results in a unique function of the protein.Resequencing data for seven melon germplasms were collected to analyze the protein sequence variation and answer this question (Appendix A).Based on the sequencing results, three germplasms that produced hard fruit had an A (adenine) (with the resulting amino acid in the protein being R (arginine)), which was consistent with the results for P5 (fruit firmness of 8.5), while two germplasms whose fruit was soft had a G (guanine) (with the resulting amino acid in the protein being K), which was consistent with the results for P10 (Appendix A).We also checked the expression of theMELO3C017400gene in publicly accessible transcriptome data at https://cucurbitgenomics.org/feature/gene, a project PRJNA383830RPKM for tissues from melon Charentais,which indicated thatMELO3C017400is differentially expressed among plant organs (the results are available at https://cucurbitgenomics.org/feature/gene/MELO3C017400.2).In fruit, this gene is expressed somewhat lower than in leaf and flower, which indicated that theMELO3C017400gene may show specific expression in fruit.However, because limited melon transcriptome data were available in the database, we still need more solid results forMELO3C017400as a FF controlled gene, and the next step would be to conduct an in-depth comparison of more data to identify potential allelic variations.

5.Conclusion

In this study, QTL-SLAF sequencing and molecular marker-linkage analysis to identify a total of 112 844 SLAF markers and 5 919 SNPs.These markers were further employed to construct a genetic linkage map with a total genetic distance of 1 356.49 cM.Consistent QTLs were observed for fruit length and fruit diameter across multiple years.Additionally, QTLs associated with single fruit weight were found on separate chromosomes in both years.By analyzing 429 F2progeny from the cross between P5 and P10, researchers pinpointed theff2.1locus to a specific 28.3 kb region that encompasses three functional genes.This investigation offers insights into the genetic basis of melon fruit firmness and establishes a foundation for identifying melon fruit firmness genes.

Acknowledgements

We thank the Zhengzhou Fruit Research Institute,Chinese Academy of Agricultural Sciences for providing melon plant samples.This work was supported by the grants from the National Natural Science Foundation of China (31772330 and 32002043), the Natural Science Foundation of the Heilongjiang Province, China(LH2022C065) and the Heilongjiang Bayi Agricultural University Support Program for SanHengSanZong, China(TDJH202004).

Declaration of competing interest

The authors declare that they have no conflict of interest.

Appendixassociated with this paper is available on https://doi.org/10.1016/j.jia.2023.02.014