Identification of Early Tomato Fruit Ripening Loci by QTL-seq

QTL-seq has been successfully studied in identifying major QTLs, markers, and candidate genes associated with traits that are important for crop improvement. Tomato earliness is an economically important trait and is a major current research focus recently. This paper reports the identification of tomato early ripening fruit locus facilitated by QTL-seq using a novel next-generation sequencing technology. Two DNA pools of phenotypes of F2 offspring from crosses between the Bone ММ (early ripening fruit, P1) and 071-440 (late ripening fruit, P2) cultivars of (Solanum lycopersicum) were bulked for sequencing and alignment analysis. Sequencing results revealed 434 SNP markers on chromosome 11, a candidate QTL at position 52,048,208 bp (named er-fruit) and a candidate gene, Solyc11g071510.1.1. The “er-fruit” as confirmed by the traditional QTL method was related to the early fruit ripening trait in tomato. Additionally, BLAST analysis to known homologies for Solyc11g071510.1.1 gene encodes glycoside hydrolases (GHs). GHs are functionally associated with cell wall degradation, fruit softening and ripening. Thus, GHs may be important in fruit softening, stimulating early fruit ripening in tomato. Our results confirmed that QTL-seq is effective method to identify candidate QTL loci, candidate genes and candidate markers.


Introduction
Earliness in tomato is one of the factors that needs much concern in recent years due to climatic changes and increased world's population.The ability to bring their products earlier to the market in the season can produce better income for growers (Kevany et al., 2008).Earliness in tomatoes consists of three stages; (1) flowering time, (2) fruit setting time, and (3) fruit ripening time (Powers, 1941).The environmental factors such as temperature and light intensity play a significant role in the expression of any components for early maturity (Kerr, 1955;Adams et al., 2001).It has been reported "Early Cherry' alleles caused reductions in both ripening time and fruit weight by using RAPD marker analysis in F 2 population derived from a cross between Lycopersicon esculentum'E6203' (normal ripening) and Lycopersicon esculentum'Early Cherry' (early ripening) (Doganlar et al., 2000).
Early fruit ripening is commercially important and effective trait for tomato (Gur et al., 2010).A QTL (dw1) of the tomato that linked to phenotypic traits, increased yield (quantitative) and earliness (qualitative) have been identified although it caused a decline in fruit firmness (Inai et al., 2006).The tomato is classified as a climacteric fruit that needs phytohormone ethylene to ripen and it also coordinates expression of thousands of genes regulating fruit softening and increasing color development, sugars, acids, and aroma production (Klee & Giovannoni, 2011).The important fruit ripening phenotypes have been distinguished by rin, nor, Nr and Cnr mutants that have been provided novel insights into the control of ripening processes (Thompson et al., 1999).In addition, the cell wall modification for softening of the fruit tissues is affected by transcriptional factors nor, rin, and ethylene receptor Never-ripe (Nr) because the transcription level of cell wall degrading enzymes polygalacturonase and pectate-lyase were not observed in rin, nor, and Nr mutants during tomato fruit ripening (Osorio et al., 2011).Smith and Gross (2000) proposed that a member of glycoside hydrolase family 35, β-galactosidase II, may be involved in Gal metabolism during cell wall degradation for softening of tomato fruit, conversion of chloroplasts into chromoplasts, fruit growth, and senescence.
Next-generation sequencing (NGS) technology was proved as a quick accurate and successful method of genome analysis (Takagi et al., 2013a) which involves categorizing molecular markers associated to target genes or genotyping a pair of bulked DNA samples from two dissimilar extreme phenotypes and connecting the markers with QTLs related with chosen traits of research interest (Michelmore et al., 1991;Giovannoni et al., 1991;Mansur et al., 1993;Darvasi & Soller, 1994).The new approach has been proposed as a means of developing rapid QTL map through the MutMap (Abe et al., 2012), MutMap-Gap (Takagi et al., 2013b), Mutmap+ (Fekih et al., 2013), and QTL-seq (Takagi et al., 2013a) approaches.QTL-seq has developed (Fekih et al., 2013;Takagi et al., 2013a) to replace traditional QTL mapping which is labour-intensive, time-consuming and involves substantial costs associated with the development of DNA markers, genotyping and the generation of a large number of progenies during advanced segregating generations (Takagi et al., 2013a).Moreover, rapid identification of the QTL region (marker and candidate gene) associated with the traits of interest can be performed in the F2 population.The QTL-seq has been employed previously to identify QTLs underlying disease resistance traits in rice (Takagi et al. 2013a), the early flowering trait in cucumber (Lu et al., 2014), seed weight trait in the chickpea (Das et al., 2015), fruit weight, locule number (Illa-Berenguer et al., 2015) and early flowering traits in tomato (Ruangrak et al., 2018).In the present study, we used QTL-seq to identify the QTL for early ripening trait in tomato progenies of cross between naturally selected Bone ММ (earliness) and 071-440 (lateness) cultivars.

Plant Materials and Phenotypic Evaluation
S. lycopersicum cv.Bone MM (Earliness (E); P1 from Russia) and 071-440 (Lateness (L); P2 from China) were used as parents (Figure A1).The genetic backgrounds of Bone MM and 071-440 are extremely different for first fruit ripening characteristics.For the phenotypic evaluation, the first fruit ripening time was visually scored by counting the days from the first flower opening (anthesis) to the first fruit ripening of each plant, developing 90% red color on fruit surface.The data were used for frequency distribution analysis.F2 progeny showing two extremes (early and late) of first fruit ripening times were isolated and pooled into two bulks (each bulk comprising 30 individuals).The experiment was performed in the tunnel type green house (at day/night average temperatures of 28 °C/15 °C) at the Institute of Vegetables and Flowers, Chinese Academy of Agricultural Sciences, Beijing, China (39.96°N, 116.33°E).

QTL-seq Analysis
Two DNA bulks of extreme early (41-45 days after anthesis) and late (55-59 days after anthesis) ripening times categories were extracted as equal volumes of DNA samples from the F2 progeny by following previously described DNA isolation methods (Abe et al., 2012;Takagi et al., 2013a).The genomic DNA extraction was performed from fresh tomato leaves using the Cetyl Trimethyl Ammonium Bromide (CTAB) method.The whole-genome sequencing was performed using an Illumina Genome IIx sequencer.Pair-end sequencing libraries (read length 100 bp) with 500 bp insert sizes were prepared for sequencing.The short reads were aligned to the S. lycopersicum reference genome sequence (//ftp.ensemblgenomes.org/pub/plants/release-22/fasta/solanum_lycopersicum/dna/Solanum_lycopersicum.SL2.40.22.dna.toplevel.fa)with BWA software (Li & Durbin, 2009).SNP-calling was performed using SAM tools software (Li et al., 2009) and then converted into SNP-index as reported (Abe et al., 2012;Fekih et al., 2013;Takagi et al., 2013a).A given result was based on short reads harbouring the SNP being different from the reference sequence (Fekih et al., 2013;Lu et al., 2014).A SNP-index of E-and L-Ripening bulks was subtracted to obtain a ∆(SNP-index).Fisher's exact test (Fisher, 1922) was used to evaluate the statistical significance of the ∆(SNP-index) values.The detection of functionally annotated putative SNPs and the annotation of the candidate polymorphic marker locus were performed using ANNOVAR software (Wang et al., 2010).

Traditional QTL Analysis
To verify the results of the QTL-seq, conventional QTL analysis using InDel (insertion or deletion) markers was used.Two hundred and three InDel markers were identified from chromosome 11 (

Phenot
The early f fruit) of 12 The freque anthesis to whereas F Figure 1.early and  These 21 markers were applied to the segregating population for QTL analysis.MQM mapping analysis identified a major QTL for early fruit ripening time delimited by two InDel markers Er-InDel178 and Er-InDel191, which were physically located in the region range of 51,965,954 to 52,146,442 Mb on chromosome 11 (Figure 3(b)).A LOD threshold value of 3.8 was used for declaration of a QTL.The LOD values in this region ranged from 1.06 to 7.94 with the highest peak at marker locus Fl-InDel178 (7.94).This interval was corresponded to the genomic region identified by the QTL-seq method (Figures 2(a)-2(c)).In addition, the candidate gene was aligned to the Solyc11g071350.1.1,encodes glycoside hydrolases (GHs), which was analyzed by BLAST through the Sol Genomics Network (SGN; http://solgenomics.net/)website (Fernandez-Pozo et al., 2015).

Discussion
QTL-seq is a powerful tool for identifying candidate QTL loci and candidate genes using NGS technology as previously reported (Takagi et al., 2013a;Lu et al., 2014;Das et al., 2015;Illa-Berenguer et al., 2015).This study is aimed to rapidly identify the candidate QTL locus and gene related to the early fruit ripening of the tomato using QTL-seq.Our aim was achieved successfully using naturally selected varieties from Russia (earliness) and China (lateness).The results of phenotyping and the distribution of early fruit ripening time demonstrated that multiple genes control fruit ripening time because the frequency distribution is close to a normal (Gaussian) distribution (Takagi et al., 2013a) (Figure 1).Thus, results suggest that the F2 population can further benefit from the use of QTL-seq analysis.The QTL mapping results confirmed the QTL-seq analysis, supporting the proposition of the QTL located on 52,048,208 bp was a major QTL associated with the early ripening fruit phenotype.
The normal distribution of the F2 population clearly allowed the performance of QTL-seq, which is based on the crossing of two parents that have extreme phenotypic differences followed by selfing of F1 individuals to generate F2 progeny.Takagi et al. (2013a) suggested that an F2 population is much easier to generate than RILs of complex generations.DNA samples of F2 individuals showing extreme phenotypes, i.e. those exhibiting the earliest and latest extreme values of fruit ripening phenotype were bulked in an equal ratio and subjected to whole genome sequencing.In this study, the high base accuracy of Q30 varied from 90.08% (for L-Ripening bulk) to 92.29% (P1), with an average of 91.30% (Table B5) suggested that the sequencing data of all the samples corresponded to low error probabilities and sufficiently high quality.Alignment analysis of the sequencing data showed a candidate QTL located on 52,048,208 bp on the Solyc11g071510.1.1 gene on chromosome 11 and this was confirmed by the result of the traditional QTL method which was consistent with the QTL-seq analysis.
Furthermore, the result of BLAST protein function analysis suggested that this candidate gene encodes glycoside hydrolases (GHs).GHs function as common degradation enzymes with a bond between a carbohydrate, a protein, lipid or another moiety, and are found in many kinds of organisms such as archaea, bacteria, animals and plants (Tyler et al., 2010).Consequently, genes encoding GHs are comparatively abundant in plants where they are involved in processes of starch metabolism, defense, and cell-wall remodeling (Tyler et al., 2010).GH genes play important roles in synthesizing carbohydrate-active enzymes in photosynthesis and in constructing carbohydrate-rich cell walls (Coutinho et al., 2003).Other functions of GHs in plants include pathogen defense, the degradation of starch, and hormone signalling (Minic, 2008).GH genes express to regulate functions in plant cell wall synthesis, renovation, and degradation (Minic & Jouanin, 2006;Lopez-Casado et al., 2008).In this context, GHs which participate in the degradation of cell wall polysaccharides are also implicated in the governance of plant cell wall loosening, the regulation of growth and development, germination, abscission, cell adhesion and fruit ripening (Fischer & Bennett, 1991;Minic, 2008).GH genes also play an important role during fruit ripening, with multiple enzymes promoting the disassembly of cell wall polysaccharides or polysaccharide domains and contribute to modifications in cell wall construction.The most characterized and studied cell wall degrading proteins in fruits were reviewed by Owino, Ambuko, and Mathooko (2005).These include GH enzymes such as polygalacturonases (PGs), ß-D-galactosidases, endo-ß-1,4-D-glucanases, and to a lesser extent endo-ß-mannanases, ß-D-xylosidases, α-D-galactosidase, and XET (Minic, 2008).ß-galactosidase II plays an important role in degrading galactan and the rise in its activity through tomato ripening suggests a possible role for this enzyme in tomato softening (Smith & Gross, 2000).During fruit ripening, pectin and some hemicellulosic polysaccharides gradually develop solubility and depolymerize by the release of neutral sugar residues from side chains of matrix polysaccharides (Huber & O'Donoghue, 1993;Brummell & Labavitch, 1997).

Concludsion
In summary of this study, as confirmed by traditional QTL and BLAST protein function analysis, QTL-seq detection found that a GH gene is related to the early fruit ripening trait in the tomato as GH genes are functionally associated with cell wall degradation, fruit softening and ripening fruit.Thus, GHs may be important in fruit softening that stimulate early fruit ripening of tomato.These results established that QTL-seq is rapid and effective method to identify candidate QTL loci, candidate genes and candidate markers.In addition, our results are important for plant breeding and crop improvement because early ripening is not only one of the major earliness traits in tomato but also one of the important agronomical traits in crop plants.Note. 1 The number of SNPs in an exon region, including stop gain, stop loss, non-synonymous and synonymous; 2 Introduction of a stop codon; 3 Loss of a stop codon; 4 Missense non-synonymous regions; 5 The number of SNPs presumed to be silent; 6 The number of SNPs in introns; 7 Splicing regions are located in the splice site (near the exon/intron boundaries of the 2 bp intron); 8 1 Kb downstream region; 9 1 Kb upstream region; 10 1 Kb upstream/downstream gene; 11 The number of SNPs in regions between genes; 12 Transitions (ts) are interchanges between a purine base and another purine (A↔G) or replacement of a pyrimidine with another pyrimidine (C↔T); 13 Transversions (tv) are interchanges between the purine and pyrimidine bases (T↔A, T↔G, C↔A, C↔G); 14 ts/tv; Transition/transversion ratio.Note. 1 The number of SNPs in an exon region, including stop gain, stop loss, non-synonymous and synonymous;2 Introduction of a stop codon;3 Loss of a stop codon;4 Missense non-synonymous regions;5 The number of SNPs presumed to be silent; 6 The number of SNPs in introns;7 Splicing regions are located in the splice site (near the exon/intron boundaries of the 2 bp intron); 8 1 Kb downstream region; 9 1Kb upstream region; 10 1 Kb upstream/downstream gene; 11 The number of SNPs in regions between genes; 12 Transitions (ts) are interchanges between a purine base and another purine (A↔G) or replacement of a pyrimidine with another pyrimidine (C↔T); 13 Transversions (tv) are interchanges between the purine and pyrimidine bases (T↔A, T↔G, C↔A, C↔G); 14 ts/tv; Transition/transversion ratio.

Takagi
Fig

Note. 1
Effective sequencing data; 2 Comparison to the reference genome of the read numbers (including single end alignment and double end alignment); 3 Reference genome reads divided by the effective sequencing data; 4 The average sequencing depth; 5 A reference genome has at least 1 base covered per site accounting the genome; 6 A reference genome has at least 4 bases covered per site accounting the genome.

Table 1 .
in a traditional QTL analysis.Among 203 InDel markers from chromosome 11, 21 InDel markers were polymorphic between the E-and L-Ripening bulks as shown in Table1.The information of InDel markers used in the traditional QTL analysis

Table B1 .
InDel markers were identified from chromosome 11

Table B2 .
Sequencing depth and coverage

Table B3 .
SNP detection and annotation

Table B4 .
The annotations of the candidate polymorphic marker loci

Table B5 .
Sequencing data quality