Mapping of QTLs controlling epicotyl length in adzuki bean (Vigna angularis)

Abstract

Epicotyl length (ECL) of adzuki bean (Vigna angularis) affects the efficiency of mechanized weeding and harvest. The present study investigated the genetic factors controlling ECL. An F₂ population derived from a cross between the breeding line ‘Tokei1121’ (T1121, long epicotyls) and the cultivar ‘Erimo167’ (common epicotyls) was phenotyped for ECL and genotyped using simple sequence repeats (SSRs) and single-nucleotide polymorphism (SNP) markers. A molecular linkage map was generated and fifty-two segregating markers, including 27 SSRs and 25 SNPs, were located on seven linkage groups (LGs) at a LOD threshold value of 3.0. Four quantitative trait loci (QTLs) for ECL, with LOD scores of 4.0, 3.4, 4.8 and 6.4, were identified on LGs 2, 4, 7 and 10, respectively; together, these four QTLs accounted for 49.3% of the phenotypic variance. The segregation patterns observed in F₅ residual heterozygous lines at qECL10 revealed that a single recessive gene derived from T1121 contributed to the longer ECL phenotype. Using five insertion and deletion markers, this gene was fine mapped to a ~255 kb region near the end of LG10. These findings will facilitate marker-assisted selection for breeding in the adzuki bean and contribute to an understanding of the mechanisms associated with epicotyl elongation.

Introduction

The adzuki bean [Vigna angularis (Willd.) Ohwi & Ohashi] is a traditional legume grown for thousands of years in East Asia, especially in China, Japan and Korea, and is currently grown in more than 30 countries throughout the world (Verdcourt 1970, Zong et al. 2003). In Japan, the adzuki bean is the second most important legume crop after soybeans. More efficient adzuki bean cultivation requires mechanization, especially when weeding at the seedling stage and harvesting (Takenaka et al. 2009). The primary leaves of cultivars with insufficient epicotyl length (ECL) are buried by soil, resulting in loss of plants after mechanized weeding at the seedling stage (Shimada 2009). Moreover, cultivars with a short ECL have a low first pod height and this results in pods being cut, damaged, or not collected during mechanical harvesting (Shimada 2009). Therefore, developing cultivars with optimized ECL, 6 to 8 cm, has become an urgent requirement for adzuki bean breeders (Shimada 2009).

Mapping of genes and quantitative trait loci (QTLs) linked to important traits facilitates molecular breeding, including marker-assisted selection (MAS) for trait improvement. Implementation of molecular breeding for adzuki bean, however, lags behind that of other legume crops, including soybeans and common beans (Phaseolus vulgaris). Simple sequence repeats (SSRs), single nucleotide polymorphisms (SNPs) and insertion and deletions (InDels) are standard DNA markers currently utilized for gene mapping and MAS. Using these molecular markers, adzuki bean genetic linkage maps have been constructed from various F₂, BC₁F₁ and recombinant inbred lines (RILs), which act as mapping populations (Aoyama et al. 2011, Han et al. 2005, Horiuchi et al. 2015, Isemura et al. 2007, Kaga et al. 1996, 2000, 2008, Li et al. 2017, Liu et al. 2016, Yamamoto et al. 2016). Based on these linkage maps, genes and QTLs associated with several important traits have been identified, including traits related to flowering time, seeds, pods, stems and leaves (Aoyama et al. 2011, Horiuchi et al. 2015, Isemura et al. 2007, Kaga et al. 1996, 2000, 2008, Liu et al. 2016, Yamamoto et al. 2016), as well as brucid resistance (Somta et al. 2008).

ECL is a domestication-related trait in adzuki bean. Epicotyls of cultivars are longer than those of the wild parent, V. angularis var. nipponensis (Kaga et al. 2008). Using an F_2:3 population of a cross between V. angularis var. nipponensis and the cultivar ‘Kyoto-Dinagon’, two QTLs that control ECL were identified, Ecl3.1.1 on linkage group 1 (LG1) and Ecl3.9.1 on LG9 (Kaga et al. 2008). At these two loci, alleles from the cultivated-parent contributed to longer ECL. Using BC₁F_1:2, F₂ and F_2:3 populations of a cross between the wild species V. nepalensis and a landrace of adzuki bean cultivated in Japan, three QTLs that control ECL were identified, Ecl2.2.1 and Ecl2.1.2 on LG1 and Ecl1.2.1 on LG2 (Isemura et al. 2007). The alleles of the cultivated parent at these three QTLs were associated with increased ECL (Isemura et al. 2007). Several breeding lines with long epicotyls were developed recently in Japan (Sato unpublished data), but little is known about the genetic basis of the long epicotyl trait in modern adzuki bean cultivars and breeding lines.

The objectives of the present study were to define the genetic basis of ECL control in adzuki bean breeding lines, and cultivars and to develop tools that can be used in molecular breeding of adzuki bean. The linkage map, molecular markers and QTLs identified in this research will facilitate greater understanding of the mechanisms controlling ECL and molecular breeding for improvement of epicotyls in the adzuki bean.

Materials and Methods

Progenitor lines and Field experiments

An F₂ mapping population of 95 individuals was derived from a cross between the adzuki bean breeding line ‘Tokei1121’ (T1121), with longer epicotyls, and cultivar ‘Erimo167’, with common epicotyls (cross ID#1408). No cultivar or breeding line in the pedigree of T1121 (Supplemental Fig. 1) had long epicotyls, suggesting that ECL in T1121 had been improved by pyramiding multiple QTLs for ECL. ‘Erimo167’ is an isogenic line containing a brown stem rot resistance gene on the genetic background of cultivar ‘Erimo-shouzu’. Both parental lines were developed by the breeding programs at the Tokachi Agricultural Experimental Station (TAES), Hokkaido Research Organization, Japan. One F₁ seed was selfed to generate the F₂ generation and selfed again to produce F_2:3 families #1408-81-4.

The F₂ population and 10 plants of each parent were grown in the experimental field of the TAES from late May to early October 2016. In 2017, 10 plants of each parental line were grown in the experimental field of the TAES from late May to early October. Single seeds were sown with 60 cm row spacing and 10 cm plant spacing. ECL of each plant was measured 38 days after sowing, at which time epicotyls were fully extended.

Genotyping with SSR markers

Polymorphisms among parental strains were detected using 196 previously described SSR markers (Han et al. 2005). In addition, the SSR motifs in QTL regions on LG7 and LG10 were amplified using five primer pairs developed in house based on whole genome sequence data (Vigna Genome Server, VIGGS; https://viggs.dna.affrc.go.jp/download, Supplemental Table 1). The SSR motifs were identified using Gramene SSRIT (Temnykh et al. 2001, available at https://archive.gramene.org/db/markers/ssrtool). Primers were designed with Primer3Plus (https://primer3plus.com/cgi-bin/dev/primer3plus.cgi). Total DNA was extracted from fresh leaves of parental strains and each F₂ plant as described (Monna et al. 2002). Sequences of interest were PCR amplified in 10 μL reaction mixtures containing 2 μL (20 ng) total DNA, 5 μL GoTaq^® Master Mix (Promega, Madison, WI, USA), 0.5 μL of each forward and reverse primer (10 pmol), and 2 μL H₂O. The amplification conditions consisted of an initial denaturation at 95°C for 2 min; 35 cycles of denaturation at 95°C for 15 sec, annealing at the annealing temperature for 15 sec and extension at 72°C for 30 sec; and a final extension step at 72°C for 2 min. PCR products were separated on 4% agarose gels or on 6% acrylamide gels (acrylamide:N,Nʹ-methylenebisacrylamide = 29:1) and stained with ethidium bromide.

DNA extraction and restriction-site associated DNA sequencing

Genomic DNA was extracted from 95 individual F₂ plants and five plants of each parent using a CTAB method (Murray and Thompson 1980). DNA concentrations were quantified using the QubitTM dsDNA BR Assay Kit (Life Technologies, Carlsbad, CA, USA) with a Qubit fluorometer 2.0 (Invitrogen, Grand Island, NY, USA), according to the manufacturers’ directions and were adjusted to a final concentration of 20 ng μL^–1.

The double-digest restriction-site associated DNA (RAD) library (Peterson et al. 2012) was generated as described (Sakaguchi et al. 2015) with slight modifications. Genomic DNA (10 ng) was digested with EcoRI-HF and BglII (New England Biolabs, Ipswich, MA, USA), ligated to barcoded adapters and purified with AMPure^®XP (Beckman Coulter, Pasadena, CA, USA). Libraries were amplified using KAPA HiFi HS ReadyMix (Kapa Biosystems, Wilmington, MA, USA) and the PCR products were purified using AMPure^®XP. Fragments 200–1,000 bp in size were selected with E-Gel Size Select (Life Technologies, Carlsbad, CA, USA); the average size of the selected fragments was 337 bp (CV 21.0%). The library was constructed by Clockmics Inc. (Osaka, Japan) and sequenced with 50 bp single-end reads in one lane of an Illumina HiSeq2000 (Illumina, San Diego, CA, USA) by Macrogen (Seoul, South Korea).

Identification of SNPs from RAD-seq and genotyping

RAD-tag extraction and genotyping were performed using Stacks v.1.46 software (http://catchenlab.life.illinois.edu/stacks/) (Catchen et al. 2011). To discard low quality reads, the raw data were cleaned using the process_radtags program. The cleaned reads were mapped to the V. angularis v1.0 reference genome (https://viggs.dna.affrc.go.jp/download) using Bowtie 2 software (Langmead and Salzberg 2012) with default parameters. Only the alignments of unique targets were subjected to assembly using the pstack program of Stacks with the default condition (minimum depth of coverage, 3). The samples processed by pstack were used to create a set of consensus loci and merging alleles using the cstacks program with no options selected. The SNP genotype of each plant was determined using the sstacks program.

Development and genotyping of InDel markers

To perform fine mapping of a QTL on LG10, InDel markers were developed based on resequencing data from the adzuki bean cultivar ‘Chihaya-hime’, with common epicotyls, and the breeding line ‘Tokei161’, with longer epicotyls (Mori unpublished data). InDels between ‘Chihaya-hime’ and ‘Tokei161’ greater than 10 bp in size were selected to develop PCR-based markers. Primers 20–27 nucleotides in length (target length, 22 nucleotides), with melting temperatures (Tm) of 58–61°C (target Tm, 60°C) and yielding products 100–300 bps in length, were designed using Primer3Plus software (https://primer3plus.com/cgi-bin/dev/primer3plus.cgi). Five primer pairs were used (Supplemental Table 2), and the protocol for marker analysis was identical to that for SSR analysis.

Construction of a linkage map and identification of QTLs

A linkage map was constructed from marker genotypes of the F₂ population in JoinMap^® 4 (van Ooijen 2006), with a LOD threshold score set to 3.0. Genetic distances were estimated using the Kosambi map function (Kosambi 1943). QTLs analyses were performed using MapQTL^® 6 (van Ooijen 2002). First, putative QTLs (LOD >3.1) were detected by the simple interval mapping (SIM). In the next step, a cofactor marker which had the highest LOD score at each putative QTL was selected, and multiple QTL model (MQM) mapping was performed. The LOD threshold for a significant QTL (P < 0.05) was based on the results of 1,000 permutations (Churchill and Doerge 1994). Genetic parameters, including the percentage of phenotypic variance explained, the additive effect, and dominance effect of each QTL, were estimated using MapQTL. Chi-square tests were performed to examine segregation ratios at marker loci for deviation from the expected ratio of 1:2:1, and skewing was determined.

Detection of epistatic interactions

To analyze digenic interactions among QTLs, the appropriate F₂ plants were classified into nine genotypic classes representing all combinations of the three genotypes (homozygous A, heterozygous, homozygous B) at each of two molecular markers, each of which was tightly linked to two target QTLs. Mean ECLs were compared among the nine genotype classes by two-way analysis of variance (ANOVA). A P-value of <0.05 for the interaction between genotypes at the two linkage groups was take as evidence for a digenic interaction. Genotypic classes were determined by using the molecular markers CEDC009 (LG2), CEDG103 (LG4), AZ07_28.0M3 (LG7) and SNP_19605 (LG10), which were tightly linked to QTLs identified by QTL mapping.

Fine mapping

One SSR marker, AZ10_26.6M2; two InDel markers, AZ10_27.5M_InDel and AZ10_28.5M_InDel, flanking the QTL on LG10; and one SSR marker, AZ07_26.5M5, flanking the QTL on LG7 in the progeny of F_2:3 families #1408-81-4 were genotyped. This process led to the identification of several plants that had recombinations at the LG10 region and were homozygous for the T1121 allele at AZ07_26.5M5 (qECL7). The detailed locus of qECL10 was determined by the comparison between their phenotype and graphical genotypes among recombinants and nonrecombinant control plants harboring homozygous allele for three markers, AZ10_26.6M2, AZ10_27.5M_InDel and AZ10_28.5M_InDel. These residual heterozygous lines (RHLs) at qECL10, as well as the parental lines, were grown in a growth chamber with a 16 h/8 h light/dark cycle maintained at 22°C during the day and 15°C during the night and with photosynthetic photon flux density during the day of 360 μmol m^–2 s^–1. Nine seeds of each parental line and 34 to 38 F₅ seeds derived from two self-pollinated F₄ RHLs were sown at 5 cm × 5 cm × 5.5 cm depth with soil compost.

Four InDel markers, AZ10_28.1M_InDel, AZ10_28.4M_InDel, AZ10_28.5M_InDel, and AZ10_28.6M_InDel, flanking the QTL on LG10 were genotype in 131 F₅ individuals derived from the self-pollinated F₄ RHLs. These RHLs at the QTL on LG10 were grown in a growth chamber using the conditions described above. Ten seeds of each parental line and the F₅ seeds were sown at 5 cm × 5 cm × 5.5 cm depth with soil compost.

Results

Field data analysis

In the field, T1121 had the longer ECLs and higher first pod height than ‘Erimo167’ (Fig. 1). The ECLs of T1121 were 1.1 cm longer than those of ‘Erimo167’ in 2016 and 1.6 cm longer in 2017 (P ≤ 0.001; Table 1). The ECLs of F₂ individuals showed a continuous segregation and ranged from 1.8 cm to 7.0 cm with transgressive segregation, suggesting that ECL is controlled by multiple genes (Fig. 2, Table 1).

Fig. 1.

Plant architecture of the long epicotyl adzuki bean breeding line, ‘Tokei1121’ (T1121; left), and the cultivar ‘Erimo167’ (right), grown in the experimental field of Tokachi Agricultural Experimental Station (A) 38 days after sowing and (B) at pod maturity in 2018.

Table 1. Epicotyl length of the adzuki bean breeding line ‘Tokei1121’ and the cultivar ‘Erimo167’ measured after epicotyl elongation had stopped in 2016 and 2017. Data are mean ± SD and asterisks (***) indicate a significant difference between the means in each year at a significance level of P < 0.001.

Year	T1121	Erimo167	Differences (T1121-Erimo167)	F2
2016	4.5 ± 0.5	3.4 ± 0.3	1.1***	3.8 ± 0.9
2017	4.5 ± 0.5	3.0 ± 0.1	1.6***

Fig. 2.

Frequency distribution of epicotyl length in F₂ individuals of a cross between the adzuki bean breeding line ‘Tokei1121’ (T1121) and the cultivar ‘Erimo167’. Fully-elongated epicotyl length was measured 38 days after sowing. Mean ECL of the parents is indicated by unfilled (‘Erimo167’) and filled (T1121) triangles and standard deviation of the mean by horizontal lines.

Microsatellite marker analysis

Of the 196 adzuki bean SSR primer pairs examined in the F₂ population, only 26 pairs (13.2%) amplified single polymorphic bands that segregated as co-dominant (1:2:1) markers. In addition, five novel SSR primer pairs on LG7 and LG10 amplified co-dominant single polymorphic bands that segregated 1:2:1.

SNP discovery via RAD-seq

We obtained a total of 120.1 million raw reads of 50 bp, with a yield greater than 6.1 Gb. Using the sequence data of the two parental samples, a catalog containing 27,294 RAD-tags was built for SNP discovery. Comparison of the parental samples yielded 215 putative SNP markers. After correcting for sequencing errors, 119 SNP loci were assigned to LGs in the Adzuki bean reference genome. SNP density was dependent on LG, varying between 22.8 Mb and 2.2 Mb per SNP. LG9 showed the highest density, 2.2 Mb per SNP, whereas, LGs 2, 3 and 5 each had fewer than five SNPs (Supplemental Fig. 2). Twenty-five SNPs detected in more than 70% of F₂ individuals (≥67 individual plants) were identified and used for subsequent linkage mapping.

Construction of a linkage map

A total of 52 segregating markers, including 27 SSRs and 25 SNP markers, which were used for linkage mapping and generated 7 LGs (LGs 1, 2, 4, 7, 8, 9 and 10) at a LOD threshold value of 3.0 (Fig. 3). LG8 and LG10 each contained a gap that separated these two linkage groups into two parts. The total length of the 7 LGs in the map was 373.0 cM. These linkage maps covered approximately 33% of the whole adzuki genome.

Fig. 3.

Linkage map of adzuki bean based on an F₂ population derived from a cross between the adzuki bean breeding line ‘Tokei1121’ and the cultivar ‘Erimo167’ using the indicated SSR and SNP markers. Each linkage group (LG) is labeled with a number at the top. Arrowheads and white boxes to the left of individual linkage groups show LOD peaks of QTL and 2-LOD support intervals, respectively.

Detection of putative QTLs

At a LOD threshold >3.1, four QTLs for ECL were detected (Table 2, Fig. 3). The QTL with the highest LOD score, 6.4, was detected at the 44.0 cM position (nearest marker SNP_19605) on LG10. This QTL, which was responsible for an estimated 17.8% phenotypic variance explained (PVE), was named qECL10 (QTL for ECL located on LG10). The T1121 allele at qECL10 increased ECL, being responsible for an additive effect of 0.6 cm, calculated as half the difference between the homozygous phenotype of each parental line. A second QTL, qECL7, was detected at the 3.9 cM position (nearest marker AZ07_28.0M3) on LG7 (LOD, 4.8; PVE, 12.7%), with the T1121 allele at qECL7 also increasing ECL (additive effect, 0.5 cm). A third QTL, qECL2, was detected at the 29.4 cM position (nearest marker CEDC009) on LG2 (LOD, 4.0; PVE, 10.3%), with the T1121 allele at qECL2 increasing ECL (additive effect, 0.2 cm). A fourth QTL, qECL4, was detected at the 64.9 cM position (nearest marker CEDG103) on LG4 (LOD, 3.4; PVE, 8.5%); the ‘Erimo167’ allele at qECL4 increased ECL (additive effect, 0.3 cm). Together, these four QTLs accounted for 49.3% of the phenotypic variance between T1121 and ‘Erimo167’.

Table 2. Putative QTLs for epicotyl length detected in an F₂ population from a cross between the long epicotyl breeding line ‘Tokei1121’ and cultivar ‘Erimo167’

QTL	Linkage group	LOD peak positiona (cM)	Nearest marker	Marker interval	LODb	AEc	DEd	PVEe
qECL2	2	29.4	CEDC009	CEDG050–CEDC009	4.0	0.2	–0.5	10.3
qECL4	4	64.9	CEDG103	CEDG232–CEDG005	3.4	–0.3	–0.4	8.5
qECL7	7	3.9	AZ07_28.0M3	AZ07_26.5M5–SNP_14295	4.8	0.5	–0.1	12.7
qECL10	10	44.0	SNP19605	SNP_19602–CEDG134	6.4	0.6	–0.1	17.8

^a LOD peak position shows the position of the arrowhead in Fig. 3.

^b LOD: logarithm of odds.

^c AE: additive effect; positive values indicate that positive allele is derived from T1121, while negative value indicates that positive allele is derived from ‘Erimo167’.

^d DE: dominance effect.

^e PVE: percentage of the total phenotypic variation explained by each QTL.

Epistatic interactions between QTLs

Digenic interactions between pairs of QTLs responsible for the ECL of nine genotype classes were tested by two-way ANOVA (Fig. 4). A clear epistatic interaction was observed between qECL10 and qECL7 (P = 0.002). The effect of the T1121 allele at qECL10 (increasing ECL) was observed in the class homozygous for the T1121 allele at qECL7, but not in the heterozygous and homozygous classes for the ‘Erimo167’ allele at qECL7. In addition, an epistatic interaction was observed between qECL2 and qECL4 (P = 0.022). The effect of the ‘Erimo167’ allele at qECL4 (increasing ECL) was observed in the homozygous classes for the T1121 allele, whereas the ‘Erimo167’ allele at qECL4 had a relatively small effect on the homozygous class of the ‘Erimo167’ allele at qECL2. In contrast, the ‘Erimo167’ allele at qECL4 had no effect on the genotype class heterozygous at qECL2. Genetic effects of qECL2 were additive to those of qECL7 (P = 0.101) and qECL10 (P = 0.160). In addition, the T1121 allele at qECL4 was responsible for the genetic effects of all genotype classes of qECL7 (P = 0.579) and qECL10 (P = 0.170), indicating that qECL4 did not interact epistatically with qECL7 and/or qECL10.

Fig. 4.

Mean epicotyl length for nine genotype classes in an F₂ population derived from a cross between the breeding line ‘Tokei1121’ (T1121) and the cultivar ‘Erimo167’ (E167). Genotypes were determined using the closely linked markers, CEDC009 (qECL2), CEDG103 (qECL4), AZ07_28.0M3 (qECL7) and SNP_19605 (qECL10). Black, shaded and white bars indicate plants homozygous for the ‘Tokei1121’ allele, heterozygous plants and plants homozygous for the ‘Erimo167’ allele, respectively. Differences in mean ECL between the nine genotypic classes were assessed by 2-way analysis of variance and P-values for the interaction term in each comparison are indicated in the figure. The data are presented as means and error bars indicate standard deviation. The number on each bar plot shows that the number of plants classified into each genotype group.

Fine mapping of qECL10

We targeted the largest effect QTL, qECL10, for fine mapping. Based on the digenic epistatic interaction between qECL10 and qECL7, we first screened a single F₃ plant, F₃#1408-81-4, which was heterozygous at AZ10_26.6M2 and CEDG134 for qECL10 and homozygous for the T1211 allele at AZ07_26.5M5 and AZ07_30.0M5 (qECL7). We identified two F₄ recombinants between AZ10_26.6M2 and CEDG134, F₄#81-4-433 and -459, which were derived from self-pollinated F₃#1408-81-4. Finally, we genotyped three markers, AZ10_26.6M2, AZ10_27.5M_InDel and AZ10_28.5M_InDel, and screened F₅ plants homozygous at the qECL10 region with recombinant and nonrecombinant types. As a result of phenotypic analysis in #81-4-459, the plants classified T1121-homozygous (9.9 ± 0.7 cm) at AZ10_27.5M_InDel and AZ10_28.5M_InDel showed significantly longer ECLs than that of ‘Erimo167’-homozygous (4.7 ± 0.5 cm). In #81-4-433, T1121-homozygous plants at AZ10_28.5M_InDel showed long ECLs, 10.3 ± 1.3 cm, whereas ‘Erimo167’-homozygous plants had common ECLs (5.1 ± 0.6 cm), similar to ‘Erimo167’. Based on the marker genotypes and ECL, the qECL10 region was found to be located between AZ10_27.5M_InDel at the 27,485,959 bp position and the end of LG10 (Fig. 5).

Fig. 5.

Genotypes of three DNA markers at the qECL10 region on LG10 and epicotyl length in the F₅ RHLs. Plants were homozygous for the ‘Tokei1121’ (black boxes) and ‘Erimo167’ (white boxes) alleles at the three markers. P-values indicate level of significance for difference in ECL between pairs of genotypes.

Next, we screened 131 F₅ individuals derived from self-pollination of RHL F₄#81-4-472. These plants were heterozygous at AZ10_28.5M_InDel (qECL10) and homozygous for the T1121 allele at AZ07_26.5M5 and AZ07_30.0M5 (qECL7). The ECLs of T1121 (9.8 ± 1.8 cm) was significantly longer than that of ‘Erimo167’ (5.5 ± 0.9 cm) (Supplemental Fig. 3). The ECLs of these 131 F₅ individuals exhibited a bimodal distribution, ranging from 2.3 cm to 12.9 cm (means = 5.6 ± 2.3 cm). These 131 plants could be separated into two groups, consisting of 95 plants with shorter (<5.9 cm, means = 4.3 ± 0.6 cm) epicotyls and 36 plants with longer (>7.3 cm, means = 9.1 ± 0.9 cm) epicotyls (Supplemental Fig. 3). This segregation fit a monogenic 3:1 ratio (χ² = 0.43, P = 0.51), indicating that longer epicotyls were controlled by a single recessive gene from T1121. Finally, qECL10 was mapped to 1.5 cM from AZ10_28.6M_InDel, at the distal end of LG10 (Fig. 6), a genomic region containing 37 putative genes (Vigna Genome Server, VIGGS: https://viggs.dna.affrc.go.jp/download, Supplemental Table 3).

Fig. 6.

Location of qECL10 on LG10. A Linkage map and the location of qECL10 using the F₂ population derived from ‘Tokei1121’ and ‘Erimo167’. B Physical map of qECL10 using the F₅ self-pollinated plants of RHLs F₄#81-4-433 and -459. C Fine mapping of the qECL10 gene using the F₅ self-pollinated plants (n = 131) of RHL F₄#81-4-472. Genetic distance is shown on the left (A, C). DNA markers are shown on the right (A, B, C). The numbers in parentheses below the DNA marker indicate their physical map position (bp) of Shumari (A, B, C).

Discussion

New loci controlling ECL in adzuki bean

Gene mapping and QTL detection are very useful for gene cloning, MAS breeding and trait improvement. The present study described the molecular dissection of important QTLs, qECL10, qECL7, qECL2 and qECL4, that contribute to ECL in adzuki bean. ECL was increased by the T1121 alleles at qECL10, qECL7 and qECL2 and by the ‘Erimo167’ allele at qECL4. To our knowledge, the present study is the first report to clarify the genetic basis of epicotyl length of adzuki bean cultivars and/or breeding lines. The QTLs for ECL on LG10, LG7 and LG4 have not been described previously. QTL controlling ECL were identified in prior research that examined genetic differences between the adzuki bean, V. angularis var. angularis, and its presumed ancestor, V. angularis var. nipponensis, and a wild relative, V. nepalensis (Isemura et al. 2007, Kaga et al. 2008). Using BC₁F_1:2, F₂ and F_2:3 plants resulting from crosses between the wild species V. nepalensis and cultivated adzuki bean, three QTLs responsible for ECL, Ecl2.2.1 (LG1), Ecl2.1.2 (LG1) and Ecl1.2.1 (LG2), were identified (Isemura et al. 2007). In each case, alleles from the cultivated parent increased ECL (Isemura et al. 2007). Additional QTL, Ecl3.1.1 on LG1 and Ecl3.9.1 on LG9, were identified in an F_2:3 population of a cross between V. angularis var. angularis and V. angularis var. nipponensis (Kaga et al. 2008). Alleles from the cultivated parent at both loci increased ECL. Ecl1.2.1 and qECL2 are close to the common SSR marker, CEDG009, on LG2 (Isemura et al. 2007, Fig. 3). Future genetic studies or gene cloning are required to determine if the ECL-related gene linked to qECL2 is the same as that linked to Ecl1.2.1.

A molecular linkage map was constructed using an F₂ population of a cross between the breeding line T1121 and the cultivar ‘Erimo167’. Despite genome-wide screening for polymorphic SSRs and SNPs via the RAD-seq procedure, only 27 SSR and 25 SNP markers were detected. Therefore, linkage maps constructed in this study covered only approximately 33% of the whole adzuki genome. In contrast, a total of 1,571 SNPs was identified via RAD-seq from the Chinese cultivar ‘Ass001’ and a wild adzuki bean (accession no. CWA108) (Li et al. 2017). SNP genotyping, in this case, provided detailed information about the genomic composition of cultivars and distinguished cultivars from wild adzuki beans (Li et al. 2017). In order to increase the coverage rate of linkage maps, it will be necessary to develop new DNA markers, such as CAPS and dCAPS, using the SNP information obtained by Rad-seq analysis. In the present study, however, the two parent strains have ‘Erimo-shozu’ as a common progenitor and were developed by the same breeding program. It seems plausible that substantial portions of the genomic regions containing few or no polymorphisms, including LGs 3, 4, 5 and 11, are genetically identical. These results suggest that whole-genome re-sequencing data may provide more information when using genetically close breeding lines to identify useful genes.

The present study narrowed down the qECL10 region to a 255 kb interval at the end of LG10 (Fig. 5). This physical position of qECL10 encompassed a total of 37 putative genes (Supplemental Table 3), including Vigan.10G248300.01 (similar to BRASSINOSTEROID INSENSITIVE 1-associated receptor kinase 1-like [XP_003538632.1, Glycine max]). BRASSINOSTEROID INSENSITIVE 1-associated receptor kinase 1 is involved in the brassinosteroid (BR) signaling pathway and positively regulates the BR-dependent plant growth pathway in Arabidopsis (reviewed by Planas-Riverola et al. 2019). BR hormones have been shown to be essential for cell elongation, including hypocotyl elongation, in Arabidopsis (Clouse and Sasse 1998). Thus, this gene is an attractive candidate for the gene at qECL10 that contributes to the long ECL phenotype in T1121. Fine mapping of the other QTL and further molecular research are needed before the genes controlling ECL are identified definitively.

Epistasis between qECL10 and qECL7

In addition to understanding the biological functions of a single QTL, it is important to clarify the genetic interactions among QTLs. The present study found that epistatic interactions are important genetic determinants of ECL in adzuki bean. Among the digenic gene combinations of the four QTLs, two epistatic interactions, between qECL10 and qECL7, and between qECL2 and qECL4, were detected. These epistatic interactions suggested that these four QTLs control epicotyl elongation via two distinct pathways, with one pathway involving qECL10 and qECL7 and the other involving qECL2 and qECL4. More comprehensive characterization of these QTLs in near-isogenic lines (NILs) is needed to determine the manner of their genetic interactions. We are currently developing NILs, with relatively small chromosomal segments containing each QTL.

Transgressive segregation was observed in ECLs of F₂ individuals of a cross between T1121 and ‘Erimo167’ (Fig. 2). QTL mapping identified positive alleles originating from both parents, including the T1121 alleles at qECL10, qECL7 and qECL2 and the ‘Erimo167’ allele at qECL4, indicating that these QTLs have complementary effects. Transgressive segregation may therefore be due, at least in part, to the activity of loci with complementary additive effects differentially present in parental lines.

Marker-assisted selection to improve ECL

Adzuki bean cultivars that have 6–8 cm ECL optimized for mechanization, especially weeding at the seedling stage and harvesting, are needed urgently (Shimada 2009). Genetic interactions between qECL10 and qECL7, and qECL2 and qECL4 suggested that QTL pyramiding may be effective for generating novel cultivars with desirable ECL (Fig. 4). The present study described the development of a codominant InDel marker, AZ10_28.6M_InDel, located close to qECL10, and a codominant SSR marker, AZ07_26.5M5, located close to qECL7 (Figs. 3, 5). In addition, we identified adzuki SSR markers, CEDC009 and CEDG103, linked to qECL2 and qECL4, respectively. In the future study, we need to clarify the four QTLs appeared to be stable across genetic backgrounds and across environments.

Author Contribution Statement

MM, HS and KK designated and managed the project. MM and KK wrote the manuscript. KM, TK, DK, YK, TY and HN participated in trait investigations, data analysis, and field experiments. AJN performed RADseq analysis. PB contributed helpful suggestions. KK is a corresponding author to conduct genetic analysis. MM is a corresponding author to conduct genomic-based analysis related to handling NGS data. All authors approved the final version of the manuscript.

Acknowledgments

The authors thank Drs. H. Shimada, H. Miura, K. Onishi and J. Kasuga for their valuable comments throughout the present study. This study was funded by the Japan Beans Fund Association (KK and HS) and JSPS KAKENHI Grant Number JP17K07618 (MM).

Literature Cited

•  Aoyama,  S.,  K.  Onishi and  K.  Kato (2011) The genetically unstable dwarf locus in azuki bean (Vigna angularis (Willd.) Ohwi & Ohashi). J. Hered. 102: 604–609.
•  Catchen,  J.M.,  A.  Amores,  P.  Hohenlohe,  W.  Cresko and  J.H.  Postlethwait (2011) Stacks: building and genotyping loci de novo from short-read sequences. G3 (Bethesda) 1: 171–182.
•  Churchill,  G.A. and  R.W.  Doerge (1994) Empirical threshold values for quantitative trait mapping. Genetics 138: 963–971.
•  Clouse,  S.D. and  J.M.  Sasse (1998) Brassinosteroids: essential regulators of plant growth and development. Annu. Rev. Plant Physiol. Plant Mol. Biol. 49: 427–451.
•  Han,  O.K.,  A.  Kaga,  T.  Isemura,  X.W.  Wang,  N.  Tomooka and  D.A.  Vaughan (2005) A genetic linkage map for azuki bean [Vigna angularis (Willd.) Ohwi & Ohashi]. Theor. Appl. Genet. 111: 1278–1287.
•  Horiuchi,  Y.,  H.  Yamamoto,  R.  Ogura,  N.  Shimoda,  H.  Sato and  K.  Kato (2015) Genetic analysis and molecular mapping of genes controlling seed coat colour in adzuki bean (Vigna angularis). Euphytica 206: 609–617.
•  Isemura,  T.,  A.  Kaga,  S.  Konishi,  T.  Ando,  N.  Tomooka,  O.K.  Han and  D.A.  Vaughan (2007) Genome dissection of traits related to domestication in azuki bean (Vigna angularis) and comparison with other warm-season legumes. Ann. Bot. 100: 1053–1071.
•  Kaga,  A.,  M.  Ohnishi,  T.  Ishii and  O.  Kamijima (1996) A genetic linkage map of azuki bean constructed with molecular and morphological markers using an interspecific population (Vigna angularis × V. nakashimae). Theor. Appl. Genet. 93: 658–663.
•  Kaga,  A.,  T.  Ishii,  K.  Tsukimoto,  E.  Tokoro and  O.  Kamijima (2000) Comparative molecular mapping in Ceratotropis species using an interspecific cross between azuki bean (Vigna angularis) and rice bean (V. umbellata). Theor. Appl. Genet. 100: 207–213.
•  Kaga,  A.,  T.  Isemura,  N.  Tomooka and  D.A.  Vaughan (2008) The genetics of domestication of the azuki bean (Vigna angularis). Genetics 178: 1013–1036.
•  Kosambi,  D.D. (1943) The estimation of map distances from the recombination values. Ann. Eugen. 12: 172–175.
•  Langmead,  B. and  S.L.  Salzberg (2012) Fast gapped-read alignment with Bowtie 2. Nat. Methods 9: 357–359.
•  Li,  Y.,  K.  Yang,  W.  Yang,  L.  Chu,  C.  Chen,  B.  Zhao,  Y.  Li,  J.  Jian,  Z.  Yin,  T.  Wang et al. (2017) Identification of QTL and qualitative trait loci for agronomic traits using SNP markers in the adzuki bean. Front. Plant Sci. 8: 840.
•  Liu,  C.,  B.  Fan,  Z.  Cao,  Q.  Su,  Y.  Wang,  Z.  Zhang and  J.  Tian (2016) Development of a high-density genetic linkage map and identification of flowering time QTLs in adzuki bean (Vigna angularis). Sci. Rep. 6: 39523.
•  Monna,  L.,  X.  Lin,  S.  Kojima,  T.  Sasaki and  M.  Yano (2002) Genetic dissection of a genomic region for a quantitative trait locus, Hd3, into two loci, Hd3a and Hd3b, controlling heading date in rice. Theor. Appl. Genet. 104: 772–778.
•  Murray,  M. and  W.  Thompson (1980) Rapid isolation of high molecular weight plant DNA. Nucleic Acids Res. 8: 4321–4325.
•  Peterson,  B.K.,  J.N.  Weber,  E.H.  Kay,  H.S.  Fisher and  H.E.  Hoekstra (2012) Double digest RADseq: an inexpensive method for de novo SNP discovery and genotyping in model and non-model species. PLoS ONE 7: e37135.
•  Planas-Riverola,  A.,  A.  Gupta,  I.  Betegón-Putze,  N.  Bosch,  M.  Ibañes and  A.I.  Caño-Delgado (2019) Brassinosteroid signaling in plant development and adaptation to stress. Development 146: dev151894.
•  Sakaguchi,  S.,  T.  Sugino,  Y.  Tsumura,  M.  Ito,  M.D.  Crisp,  D.M.J.S.  Bowman,  A.J.  Nagano,  M.N.  Honjo,  M.  Yasugi,  H.  Kudoh et al. (2015) High-throughput linkage mapping of Australian white cypress pine (Callitris glaucophylla) and map transferability to related species. Tree Genet. Genomes 11: 121.
•  Shimada,  H. (2009) Evaluation of adaptability of long hypocotyl or long flower stalk lines of adzuki bean to mechanical weeding and combine harvesting. Bull. Hokkaido Pref. Agric. Exp. Stn. 94: 55–62 (in Japanese with English summary).
•  Somta,  P.,  A.  Kaga,  N.  Tomooka,  T.  Isemura,  D.A.  Vaughan and  P.  Srinives (2008) Mapping of quantitative trait loci for a new source of resistance to bruchids in the wild species Vigna nepalensis Tateishi & Maxted (Vigna subgenus Ceratotropis). Theor. Appl. Genet. 117: 621–628.
•  Takenaka,  H.,  J.  Kato,  H.  Sato,  K.  Sekiguchi,  H.  Momono,  K.  Murata,  H.  Shimada,  S.  Aoyama,  K.  Tomita and  M.  Minami (2009) Mechanization of adzuki bean harvesting and the quality of the products. Bull. Hokkaido Pref. Agric. Exp. Stn. 94: 65–79 (in Japanese with English summary).
•  Temnykh,  S.,  G.  DeClerck,  A.  Lukashova,  L.  Lipovich,  S.  Cartinhour and  S.  McCouch (2001) Computational and experimental analysis of microsatellites in rice (Oryza sativa L.): frequency, length variation, transposon associations, and genetic marker potential. Genome Res. 11: 1441–1452.
• van Ooijen, J.W. (2002) MapQTL^® 6, Software for the mapping of quantitative trait loci in experimental populations of diploid species. Kyazma B.V., Wageningen, The Netherlands.
• van Ooijen, J.W. (2006) JoinMap^® 4, Software for the calculation of genetic linkage maps in experimental populations. Kyazma B.V., Wageningen, The Netherlands.
•  Verdcourt,  B. (1970) Studies in the Leguminosae-Papilionoïdeae for the ‘flora of tropical East Africa’: IV. Kew Bull. 24: 507–569.
•  Yamamoto,  H.,  Y.  Horiuchi,  R.  Ogura,  H.  Sakai,  H.  Sato and  K.  Kato (2016) Identification and molecular mapping of Flowering Date1 (FD1), a major photoperiod insensitivity gene in the adzuki bean (Vigna angularis). Plant Breed. 135: 714–720.
•  Zong,  X.X.,  A.  Kaga,  N.  Tomooka,  X.W.  Wang,  O.K.  Han and  D.  Vaughan (2003) The genetic diversity of the Vigna angularis complex in Asia. Genome 46: 647–658.