Single-base deletion in GmCHR5 increases the genistein-to-daidzein ratio in soybean seed.

Novel mutant alleles related to isoflavone content are useful for breeding programs to improve the disease resistance and nutritional content of soybean. However, identification of mutant alleles from high-density mutant libraries is expensive and time-consuming because soybean has a large, complicated genome. Here, we identified the gene responsible for increased genistein-to-daidzein ratio in seed of the mutant line F333ES017D9. For this purpose, we used a time- and cost-effective approach based on selective genotyping of a small number of F2 plants showing the mutant phenotype with nearest-neighboring-nucleotide substitution-high-resolution melting analysis markers, followed by alignment of short reads obtained by next-generation sequencing analysis with the identified locus. In the mutant line, GmCHR5 harbored a single-base deletion that caused a change in the substrate flow in the isoflavone biosynthetic pathway towards genistein. Mutated GmCHR5 was expressed at a lower level during seed development than wild-type GmCHR5. Ectopic overexpression of GmCHR5 increased the production of daidzein derivatives in both the wild-type and mutant plants. The present strategy will be useful for accelerating identification of mutant alleles responsible for traits of interest in agronomically important crops.

Introduction

In leguminous plants such as soybean (Glycine max (L.) Merrill), isoflavones and their derivatives play important roles in microbe–plant interactions related to the establishment of symbiosis (Subramanian ) and the biosynthesis of phytoalexins for defense against pathogens (Welle ), and have health benefits as part of the human diet (Zaheer and Akhtar 2017). These compounds are synthesized through the phenylpropanoid pathway (Supplemental Fig. 1), which involves multiple biosynthetic enzymes (Hahlbrock and Scheel 1989). In the first step, chalcone synthase (CHS) catalyzes the conversion of p-coumaroyl-coenzyme A and malonyl-coenzyme A to naringenin chalcone (Ayabe ). At the same time, CHS and chalcone reductase (CHR) catalyze the synthesis of isoliquiritigenin chalcone from the same substrates (Bomati , Welle and Grisebach 1988). Next, naringenin chalcone and isoliquiritigenin chalcone are converted to the tricyclic flavanones naringenin and liquiritigenin, respectively, by the enzyme chalcone isomerase (CHI) (Ralston ). Finally, naringenin and liquiritigenin are converted to the major aglycone isoflavones genistein and daidzein by an isoflavone-specific enzyme, isoflavone synthase, which catalyzes the conversion of flavanones to their corresponding isoflavones via 2,3-aryl ring migration (Akashi , Jung , Steele ).

In soybean, duplication and inversion of CHS genes located in the I locus results in siRNA-mediated silencing of genes responsible for yellow seed coat (Kurauchi , Tuteja ). Single-base deletion in the gene encoding flavonoid 3ʹ-hydroxylase (T locus) changes pubescence color from tawny to gray (Toda ). Soybean flower color is regulated by the genes encoding flavonoid 3ʹ5ʹ-hydroxylase (W1 locus) (Zabala and Vodkin 2007), R2R3-type MYB transcription factor (W2 locus) (Takahashi ), dihydroflavonol 4-reductase 2 (W4 locus) (Yan ), flavonol synthase (Wm locus) (Takahashi ), and flavanone 3-hydroxylase (Wp locus) (Zabala and Vodkin 2005). Thus, many enzymes and transcription factors related to the flavonoid biosynthetic pathway are associated with the seed coat and flower color of soybean.

Traditionally, map-based cloning approaches use populations obtained from a cross between two parental lines with different genetic backgrounds. However, it is difficult to predict the genotype associated with a target trait in an individual because the number of segregated loci and the effect of each locus are often unknown. Therefore, additional materials such as backcross inbred lines and residual heterozygous lines where the target locus is segregated as a single Mendelian factor are often needed (Yamanaka ). In contrast, mutant lines share almost the same genetic background with the wild-type line, which allows the use of mutant libraries to examine the function of genes related to specific biological traits. Using mutant libraries, many genes have been isolated from model plants, such as Arabidopsis (Azpiroz-Leehan and Feldmann 1997), Lotus japonicus (Urbański ), and rice (Wang ). However, constructing a high-density mutant library is time-consuming and labor-intensive for crops such as soybean due to their long growing seasons and the large area needed to cultivate a large number of individuals. To date, several mutant libraries for soybean have been reported (Anai 2012, Cooper , Li , Tsuda ), but the use of such libraries has largely been restricted to identification of genes related to flowering (Cao , Watanabe , 2011, Xia ) and seed fatty acid biosynthesis (Anai , 2012, Hoshino , 2014).

The genetic basis of the diversity of isoflavone content in soybean has been widely evaluated by quantitative trait locus (QTL) analysis using populations obtained from bi-parental lines displaying contrasting isoflavone content phenotypes or by genome-wide association study using germplasms (Akond , Cai , Gutierrez-Gonzalez , 2011, Han , Kassem , Li , Meksem , Pei , Primomo , Wang , Yoshikawa , Zeng ). A total of 293 QTLs related to isoflavones (61 for daidzein, 69 for genistein, 72 for glycitein, 91 for total isoflavone content) are currently listed in the SoyBase database (April 2019; https://www.soybase.org/). We have reported that isoflavone accumulation is tissue dependent; we observed weak correlations between the isoflavone content of cotyledon and other parts of the seed (plumule, epicotyl, hypocotyl, and radicle), indicating that several genetic factors control isoflavone accumulation in soybean seed (Watanabe ). The genes responsible for the reported QTLs are yet to be fully elucidated because a complicated range of environmental factors—including temperature during seed development, water availability in soil, and cultivation area—also affect the isoflavone content of soybean seed (Hoeck , Rasolohery , Tsukamoto ).

Here, we characterized a mutant soybean line with altered isoflavone content in seed (i.e., increased genistein derivative content and decreased daidzein derivative content) compared with the wild-type line. We then identified the gene responsible for this mutation by using a combination of a rapid mapping procedure and next-generation sequencing (NGS) analysis. Finally, we discuss the usefulness of this mutant allele for improving the isoflavone content of soybean seed.

Materials and Methods

Mutant screening

Mutant line F333ES017D9 was identified from an ethyl methanesulfonate–treated ‘Fukuyutaka’ mutant library of 3331 lines reported previously for the isolation of soybean flowering time mutants (Xia ). Among these lines, we choose 2720 lines that produced enough seeds (>20); five seeds from each of the 2720 lines were crushed and 100 mg of seed powder was used for extraction of isoflavones (see Measurement of isoflavone content). We selected F333ES017D9 as a candidate line with a high proportion of malonylgenistin in the total isoflavone content in its seeds; we propagated seeds of this line and confirmed the stability of the mutant phenotype.

Plant materials

Two Japanese soybean cultivars and one mutant line were used as parental lines: ‘Fukuyutaka’ (FUKU), ‘Toyoshirome’ (TOYO), and mutant line F333ES017D9 (MUT). FUKU and TOYO were bred at the National Agriculture Research Organization (Ibaraki, Japan). MUT was identified as described above in Mutant screening. To confirm the reliability of the mutant phenotype, FUKU and MUT (three plants per line) were grown in the summer of 2017 in a field owned by Saga University, Japan (33°14ʹN 130°17Eʹ).

For mapping experiments and to estimate the effects of the mutated gene, three populations were obtained by crossing TOYO and MUT (POP1 and POP2) and by crossing FUKU and MUT (POP3). POP1 comprised 118 F2 plants and the three parental lines (single plant per parental line was used for phenotyping) grown under natural day-length conditions in a field owned by Saga University. The seeds were sown on 19 July 2018 and the plants were harvested the following November. Seeds obtained from each F2 plant were used for genotyping and phenotyping. POP2 was made from F3 seeds from POP1, and a single seed from each of 47 F2 individuals of POP1 was used for evaluation of the effect of the mutated gene on tissue-specific isoflavone accumulation. POP3 comprised 62 F2 plants grown in a greenhouse. The seeds were sown on 1 February 2019, with approximately 30 plants in a single 128-well nursery tray, and were grown under extended-light conditions (14.5 h) until the end of March to avoid the induction of flowering under short-day conditions. Seeds were harvested from all 62 plants, and a single F3 seed from each plant was used for genotyping and phenotyping.

DNA extraction and DNA marker analysis

Genomic DNA was extracted from seed powder (approx. 50 mg) by using the CTAB method (Murray and Thompson 1980) and an automated DNA extraction machine (GENE PREP STAR PI-480; Kurabo Industries Ltd., Osaka, Japan) in accordance with the manufacturer’s instructions. All genotyping experiments were performed with DNA diluted 40 times in distilled water. To construct a linkage map of the mutated gene, we used nearest-neighboring-nucleotide substitution–high-resolution melting (NNNs-HRM) analysis and primers designed to detect single-nucleotide polymorphisms (SNPs), as reported (Yamagata ). To identify the mutated locus, a modified selective genotyping method (Darvasi 1997) was used together with 376 primer pairs designed using SNPs between FUKU and TOYO (Watanabe ). A LightCycler 96 system (Roche Diagnostics K.K., Tokyo, Japan) equipped with the manufacturer’s software (LightCycler 96 Software ver. 1.1.0.1320) was used to detect polymorphisms in polymerase chain reaction (PCR) fragments containing SNPs for NNNs-HRM markers, and to determine the genotypes of individuals; the PCR conditions for genotyping were as described previously (Yamagata ). We also designed 8 additional NNNS-HRM markers linked to the mutated locus from the SNP list between FUKU and TOYO for QTL analysis. The primers used for NNNs-HRM markers (Yamagata ) are listed in Supplemental Table 1.

Measurement of isoflavone content

Isoflavones were extracted from whole seed and from isolated seed tissues. For screening of the mutant library, 5 freeze-dried whole seeds were crushed into a fine powder with a powder mill with a single replication. For analysis of FUKU and MUT, 10 freeze-dried whole seeds were crushed into a fine powder with three biological replications. For the analysis of POP1, 10 seeds from each F2 individual, FUKU, TOYO, and MUT were mixed together. For the analysis of POP2 and POP3, a single F3 seed from each plant was collected, the seed coat was removed, and the cotyledon and “hypocotyl” (plumule, epicotyl, hypocotyl, and radicle) were separated with a knife. Each freeze-dried cotyledon was crushed into fine powder with a powder mill. Samples of powdered cotyledons were used for DNA extraction (50 mg) and for isoflavone extraction (50 mg). Isoflavones were extracted from the hypocotyls of POP2 and POP3 seeds. Freeze-dried hypocotyls were weighed for calculation of isoflavone content by weight, placed in a 3-mL plastic tube provided by Yasuikikai (Osaka, Japan) with 5-mm metal beads (Yasuikikai), and then crushed into a fine powder with a Multibeads shocker (Yasuikikai). Then, powder obtained from whole seed (100 mg) or cotyledon (50 mg), or all of the powder obtained from hypocotyl, was suspended in 1 mL of extraction buffer containing 70% ethanol (v/v) and 0.1% acetic acid (v/v). Extraction and separation by high-pressure liquid chromatography (Jasco Corp., Tokyo, Japan) were conducted as described previously (Watanabe ). Extraction was performed three times, and the combined extract (approx. 3 mL) was filtered through a 0.45-μm nitrocellulose membrane (GE Healthcare UK Ltd., Buckinghamshire, England). Filtrate samples (3 μL) were analyzed by high-pressure liquid chromatography on a Hydrosphere C18 column (YMC Co., Ltd., Kyoto, Japan). External standard curves for daidzein, daidzin, malonyldaidzin, glycitein, glycitin, malonylglycitin, genistein, genistin, and malonylgenistin (all purchased from Wako, Osaka, Japan) were constructed, and peak areas for the extracted isoflavones were calculated according to their retention times and the areas of the standards. To evaluate mutant phenotype, G/D score was used, which was defined as the ratio of genistein-type isoflavones (genistin [Ge] and malonygenistin [Mge]) to daidzein-type isoflavones (daidzin [Da] and malonyldaidzin [Mda]). If the amount of an isoflavone was under the detectable level, we excluded that component from the statistical analysis.

NGS analysis

DNA used for whole-genome sequencing analysis was extracted from young leaves before full expansion, for parental lines FUKU and TOYO or from combined powdered whole seed from the 16 POP1 individuals with the highest G/D scores (mixed-MUT) . NGS analysis was performed on our behalf by Novogene (Beijing, China). Approximately 17-Gbp sequences were obtained for each sample. A soybean reference sequence (data version, Glyma 1.0) was obtained from a public database (https://phytozome.jgi.doe.gov). FASTQ files containing sequence data were used to align short reads to a candidate region (58.4–61.4 Mbp region of Gm18) by using the Bowtie 2 program (ver. 2.3.5.1, Langmead and Salzberg 2012). We used the SAMtools suite (ver. 0.1.18, Li ) to detect polymorphisms among the three extracted DNAs. Default parameters were used for short read alignment and to extract polymorphisms. We used IGV viewer (ver. 2.4.10, Robinson ) to manually confirm each polymorphism between the sequences of FUKU, TOYO, and mixed-MUT.

Expression analysis using developing seed from parental lines

Parental lines FUKU and MUT were grown in a greenhouse (February to June 2018) under a long-day photoperiod (14.5 h/9.5 h) until the end of March. Developing seed was collected randomly from several plants at 40, 55, and 70 days after flowering (DAF). Total RNA was isolated from developing seed (50–100 mg) with a Total RNA Extraction Kit Mini Plant (SciTrove, Tokyo, Japan) with additional rDNase I (Takara Bio Inc., Shiga, Japan) treatment. cDNA was synthesized from 1 μg total RNA with ReverTra Ace (Toyobo, Osaka, Japan). A 5-μL aliquot (approx. 25 ng) of cDNA was used as a template. Quantitative real-time PCR was conducted as follows in a LightCycler 96 system: 95°C for 5 min for activation of enzyme, 95°C for 15 s, 60°C for 15 s, 72°C for 20 s, for a total of 40 amplification cycles. EvaGreen Dye (20× in water; Biotium, Inc., Fremont, CA, USA) was used as the fluorescent dye, and dNTPs (Takara), homemade recombinant Taq polymerase, the PCR buffer mentioned in a previous study (Yamagata ), and 1 pmol of each gene-specific primer were used to evaluate transcript levels of target genes. Expression relative to GmACT2/7 (Glyma.19G147900), the internal reference gene, was calculated by using the 2-ΔΔCt method (Schmittgen and Livak 2008). Data from four technical replicates were analyzed. The primers used are listed in Supplemental Table 1.

Construct preparation for ectopic expression of GmCHR5

The full-length coding sequence of GmCHR5 (Glyma.18G285800) was amplified from FUKU seed cDNA by PCR using KOD FX Neo polymerase (Toyobo). The PCR product was cloned into the entry vector pENTERTM/D-TOPO via the TOPO Cloning reaction (Thermo Fisher Scientific, Waltham, MA, USA) and sequenced. The cloned fragment was transferred into the destination vector pUB-GW-GFP (Maekawa ) by the Gateway LR recombination reaction (Thermo Fisher Scientific) and designated GmCHR5-OE. GmCHR5-OE and the pUB-GW-GFP vector were introduced into Agrobacterium rhizogenes strain LBA1334 by a freeze–thaw method (Holsters ). The primers for cloning of the GmCHR5 gene are listed in Supplemental Table 1.

Soybean hairy root transformation

Soybean transgenic hairy roots were generated as reported previously (Chen ) with some modifications. Briefly, FUKU and MUT seeds were sterilized with chlorine gas for 7 to 8 h, transferred to MS medium without sucrose (Murashige and Skoog 1962), and germinated for 5 to 7 days under a 16 h/8 h photoperiod at 22 to 25°C. LBA1334 harboring GmCHR5-OE or pUB-GW-GFP vector were re-suspended in sucrose-free MS liquid medium containing freshly prepared 100 μM acetosyringone (Sigma-Aldrich, St. Louis, MO, USA) to a final OD600 of 0.6 and used for inoculation. Dark-green, healthy cotyledons were separated from the hypocotyl to make two cotyledons. The point where the cotyledon and hypocotyl join was gently wounded with a sterile blade and soaked in the medium containing LBA1334 for several minutes. Co-cultivation was conducted in the dark at 22 to 25°C for 3 days. Cotyledons were then incubated on the induction culture medium in the dark at 25°C for 3 to 4 weeks. Transgenic hairy roots with strong green fluorescent protein (GFP) signal were selected, weighed, placed in a 3-mL plastic tube with 5-mm metal beads (Yasuikikai), frozen in liquid nitrogen, crushed with a Multibeads shocker, and used for isoflavone extraction as described above.

QTL analysis and statistical analysis

A linkage map of Gm18 was constructed in AntMap software (ver. 1.2, Iwata and Ninomiya 2006) using the default parameters for 96 plants from POP1 (selected in the order of their sample ID numbers). QTLs were identified by single-interval mapping in R/qtl software (ver. 1.44-9, Broman ). Linear regression analysis was used to confirm the effects of the QTLs by using the genotype of the closest DNA marker in 96 individuals of POP1 and their isoflavone content. Linear regression analysis was also used to estimate the effect of the mutated locus by using the genotype of the DNA marker targeting the polymorphism detected in the candidate gene (GmCHR5_1bp_del; see results section) in 96 individuals in POP1, 47 individuals in POP2, and 62 individuals in POP3, and their isoflavone content. Statistical analysis for the ectopic expression of GmCHR5 was conducted with single-factor ANOVA and Tukey’s post-hoc test, with P = 0.05 considered the cut-off for statistical significance. All statistical analyses were performed in R software (ver. 3.6.1, R Development Core Team 2008).

Results

Identification of a mutant line with altered seed isoflavone content

We screened 2720 out of 3331 lines in an ethyl methanesulfonate–treated FUKU mutant library (Xia ) and obtained several lines with a unique isoflavone phenotype. One of those lines, F333ES017D9 (MUT), showed a much higher malonylgenistin (Mge) content (72.5% of total isoflavone context) in whole seed compared with the average value of the mutant library (50.7 ± 6.92%, Supplemental Fig. 2). MUT showed a very high G/D score (4.31) compared with the average value of the mutant library (1.4 ± 0.44, Supplemental Fig. 2). We confirmed the mutant phenotype by propagating seeds obtained from plants grown in the 2017 summer season and found that MUT also showed a significantly decreased daidzin (Da), malonyldaidzin (Mda), glycitin, and malonylglycitin (Mgly) content in whole seed (Table 1). The G/D score of MUT (9.1) was significantly higher than that of FUKU (2.4; Table 1). However, there was no significant difference in total isoflavone content between MUT and FUK (Table 1).

Table 1.

Phenotypes of FUKU and MUT for whole-seed isoflavone content (μg/g)

Isoflavone	FUKU (n = 3)			MUT (n = 3)		P-value Student’s t-test
	Average	SDa		Average	SD
Daidzin	10.0	0.10		3.4	1.95	P < 0.01
Genistin	33.4	6.43		44.5	10.05	nsc
Glycitin	14.1	1.77		6.7	1.36	P < 0.01
Malonyldaidzin	394.2	34.71		148.4	76.70	P < 0.05
Malonylgenistin	917.7	114.10		1132.1	210.11	ns
Malonylglycitin	52.4	4.73		17.0	2.59	P < 0.01
G/D scoreb	2.4	0.30		9.1	2.53	P < 0.05
Total isoflavone	1421.8	133.84		1352.1	295.27	ns

SD, standard deviation.

G/D score, ratio of genistein-type isoflavones (genistin + malonylgenistin) to daidzein-type isoflavones (daidzin + malonyldaidzin).

ns, not significant.

Identification of the locus of the mutated gene

We used selective genotyping and QTL analysis to identify the locus of the mutated gene in an F2 population obtained from the cross between MUT and TOYO (POP1). We expected to find segregation of genetic loci related to isoflavone content, besides the mutated gene, within the population because of the difference in genetic background between FUKU and TOYO. The G/D scores were FUKU (1.5), TOYO (1.6), and MUT (6.2) for plants grown in the 2018 season, and we concluded that G/D score was a suitable trait for evaluation of the effect of the mutant gene. The G/D score in POP1 ranged from 1.1 to 23.0 (Fig. 1), indicating that the MUT allele responsible for a high G/D score was possibly segregated in POP1, and that G/D score was affected by other segregating loci between FUKU and TOYO or by environmental factors because of the continuous distribution of G/D score. We considered that QTL analysis would be useful for identifying the mutated locus in POP1 and applied a selective genotyping approach to find the chromosomal region associated with the mutated locus in MUT. We selected the 10 plants in POP1 with the highest G/D score for selective genotyping, and screened the SNP loci linked to the mutant locus based on the genotype of the selected plants. A homozygous FUKU genotype determined with NNNs-HRM markers would indicate a strong link between the analyzed DNA marker and the mutated loci. In contrast, DNA markers not linked to the mutated locus would show a random genotype (homozygous for FUKU or TOYO, or heterozygous). In total, 348 of 376 (92.6%) NNNs-HRM markers covering the whole soybean genome showed polymorphisms between FUKU and TOYO. Among them, we identified only two markers, Gm18_60238623 and Gm18_60832874 (marker names are based on chromosome number and physical position), showing complete coincidence with the FUKU genotype in all selected plants.

Fig. 1.

Distribution of G/D scores in POP1. Each POP1 plant is ordered according to its G/D score. G/D scores of the parental lines FUKU, TOYO, and MUT are shown in black. Asterisks indicate the 10 plants with the highest G/D score, which are discussed in the main text.

We then constructed a linkage map for soybean chromosome 18 (Gm18) with additional NNNs-HRM markers. Linkage between the mutated gene and these markers was confirmed by QTL analysis using 96 plants from POP1. An LOD (logarithm of the odds) peak was detected at 29 cM (closest marker, Gm18_61024346 at 35.2 cM; Fig. 2A). We predicted the genetic effects of this mutated gene by linear regression analysis with the closest marker (Gm18_61024346) and found that this locus could explain 54.9% of the phenotypic variance of G/D score in POP1 (Fig. 2B). The mutated gene was located within an interval between Gm18_58420729 and Gm18_61355250.

Fig. 2.

Identification of mutant locus by QTL analysis of POP1. (a) LOD score curve for the linkage map of part of Gm18. Marker names are presented as the combination of chromosome name and physical position. (B) Boxplot showing phenotypic classification with the DNA marker closest to the mutated locus. HET, heterozygous.

Identification of the responsible gene by NGS analysis

Next, we identified the gene responsible for the target mutation by using the MutMap technique (Abe ) for a targeted chromosomal region and NGS whole-genome sequence data for FUKU, TOYO, and mixed DNA of POP1 plants showing the MUT phenotype (mixed-MUT). A total of 72.1 K short reads (150-bp paired-end sequences) for FUKU, 71.3 K for TOYO, and 63.8 K for mixed-MUT were obtained. The estimated sequence volume for each sample was 17.4 (mixed-MUT), 19.4 (TOYO), and 19.7 times (FUKU) that of the reference soybean genome (1.1 Gbp; Schmutz ). From the result of QTL analysis of Gm18, we extracted short reads mapped to a 58.4- to 61.4-Mbp region covering the mutated gene locus. Three nucleotide substitutions and two deletions were polymorphic between FUKU and MUT (Supplemental Table 2). Among these mutated sites, a single-base deletion was located in the third exon of Glyma18g52250 (gene ID in Wm82.a2.v1 database, Glyma.18G285800). In the MUT allele, the guanine nucleotide at position 620 from the start codon of the coding sequence was deleted, resulting in a protein of 212 amino acids (aa) instead of 394 aa for the wild-type protein (Fig. 3). This gene was annotated in the database (https://phytozome.jgi.doe.gov/pz/portal.html) as 6ʹ-deoxychalcone synthase; the product of this gene (GmCHR5) shows chalcone reductase activity and is responsible for control the relative concentrations of daidzein and genistein derivatives in soybean seed (Mameda ). We hypothesized that the mutation of GmCHR5 resulted in reduced substrate flow for daidzein synthesis and increased substrate flow for genistein synthesis.

Fig. 3.

Gene structure and location of the single-base deletion in MUT GmCHR5 (Glyma.18G285800). (A) Exons and untranslated regions of GmCHR5 are shown as black and white rectangles, respectively. The interval between two vertical lines is 250 bp. (B) Nucleotide sequence around the 1-bp deletion (white box) and the amino acid sequence (single capital letters) from residue 201 from the start codon of GmCHR5 mRNA.

Genetic effects of the 1-bp deletion in GmCHR5 on isoflavone content

We next confirmed the reliability of the NGS data and the effect of the mutant allele by using several mapping populations. We developed additional HRM markers targeting the 1-bp deletion in GmCHR5 (hereafter GmCHR5_1bp_del). The melting temperature of the PCR product with this deletion is lower than that of the wild type (Supplemental Fig. 3). GmCHR5_1bp_del was mapped to the locus between Gm18_ Gm18_58420729 and Gm18_61024346. Only two recombinations were observed between GmCHR5_1bp_del and Gm18_61024346.

We next used linear regression analysis to further examine the genetic effects of GmCHR5_1bp_del on isoflavone phenotype in POP1, POP2, and POP3 (Tables 2–4). GmCHR5_1bp_del explained 59.7% (G/D scores), 46.7% (Mda), 37.2% (Da), 33.7% (Mge), 17.0% (Ge), and 13.7% (Mgly) of phenotypic variances for isoflavones in POP1 (Table 2). In dissected seed tissue from POP2, GmCHR5_1bp_del increased the G/D score of cotyledon and hypocotyl tissues, and 42.7% of the cotyledon G/D score variance and 59.9% of that of the hypocotyl could be explained by this mutation (Table 3). Increased production of genistein derivatives (Ge and Mge) and decreased production of daidzein derivatives (Da and Mda) was clear in both cotyledon and hypocotyl tissues in the MUT compared with the TOYO genotype (Table 3). We also confirmed the genetic effects of GmCHR5_1bp_del in the backcross population POP3, which was grown in a greenhouse in winter. We detected a significant association between G/D score in cotyledon (24.0%) and in hypocotyl (28.3%) (Table 4).

Table 2.

Effect of GmCHR5_1bp_del on isoflavone content (μg/g) in whole seed of POP1

Number of plants	Genotype
	MUT homozygous 19			Heterozygous 53			TOYO homozygous 24
Isoflavone	Average	SDa		Average	SD		Average	SD		F-value	PVEb	P-value
Daidzin	9.4	5.2		26.8	9.7		39.1	19.8		27.0	37.2%	P < 0.001
Genistin	149.6	52.9		131.3	32.8		100.3	32.7		9.5	17.0%	P < 0.001
Malonyldaidzin	395.7	205.5		970.4	291.1		1307.9	451.6		40.7	46.7%	P < 0.001
Malonylgenistin	3035.8	555.3		2622.9	456.3		2091.3	309.8		23.7	33.7%	P < 0.001
Malonylglycitin	50.0	27.5		69.9	26.5		85.1	35.7		7.4	13.7%	P < 0.01
G/D scorec	10.3	5.6		3.0	0.8		1.8	0.4		68.9	59.7%	P < 0.001

SD, standard deviation.

PVE, proportion of phenotypic variance explained.

G/D score, ratio of genistein-type isoflavones (genistin + malonylgenistin) to daidzein-type isoflavones (daidzin + malonyldaidzin).

Table 3.

Effect of GmCHR5_1bp_del on isoflavone content (μg/g) in dissected seed tissues from POP2

Number of plants	Genotype
	MUT homozygous 13			Heterozygous 14			TOYO homozygous 20
Isoflavone	Average	SDa		Average	SD		Average	SD		F-value	PVEb	P-value
In cotyledon
Daidzin	7.7	5.8		34.8	19.6		48.5	31.3		12.1	35.4%	P < 0.001
Genistin	164.4	83.3		153.4	58.0		123.7	52.6		1.8	nsc	ns
Malonyldaidzin	254.9	151.5		1012.6	439.2		1322.9	653.5		19.7	46.7%	P < 0.001
Malonylgenistin	3043.9	821.1		2649.1	720.4		2283.7	625.3		4.7	17.3%	P < 0.05
Malonylglycitin	10.6	3.0		10.0	1.4		9.5	3.8		0.5	ns	ns
G/D scored	16.5	14.1		3.1	1.1		2.3	1.4		16.4	42.7%	P < 0.001
In hypocotyl
Daidzin	134.7	78.2		306.0	114.1		310.0	82.3		17.8	44.2%	P < 0.001
Genistin	350.3	143.3		203.1	107.4		129.3	65.7		18.3	44.8%	P < 0.001
Glycitin	584.1	433.9		1306.5	720.0		1476.2	699.7		8.4	27.3%	P < 0.001
Malonyldaidzin	5542.3	2344.0		10548.7	3469.7		10965.1	2970.9		15.6	41.0%	P < 0.001
Malonylgenistin	10087.1	3203.8		4877.9	1787.3		3565.9	1520.0		37.9	62.8%	P < 0.001
Malonylglycitin	2294.6	1103.3		4402.1	1960.5		4854.9	1867.8		9.8	30.4%	P < 0.001
G/D score	2.3	1.2		0.6	0.3		0.4	0.2		33.6	59.9%	P < 0.001

SD, standard deviation.

PVE, proportion of phenotypic variance explained.

ns, not significant.

G/D score, ratio of genistein-type isoflavones (genistin + malonylgenistin) to daidzein-type isoflavones (daidzin + malonyldaidzin).

Table 4.

Effect of GmCHR5_1bp_del on isoflavone content (μg/g) in a backcross population grown in a greenhouse (POP3)

Number of plants	Genotype
	MUT homozygous 23			Heterozygous 21			FUKU homozygous 18
Isoflavone	Average	SDa		Average	SD		Average	SD		F-value	PVEb	P-value
In cotyledon
Daidzin	1.7	2.8		3.6	2.9		6.8	4.1		10.4	29.0%	P < 0.001
Genistin	18.1	4.1		17.6	3.8		22.0	8.1		3.8	11.4%	P < 0.05
Glycitin	3.7	2.1		5.4	2.9		3.8	1.6		3.6	10.9%	P < 0.05
Malonyldaidzin	47.5	65.7		106.6	83.1		179.3	105.5		12.3	29.3%	P < 0.001
Malonylgenistin	946.2	1155.2		1078.7	413.8		358.9	316.9		1.6	nsc	ns
G/D scored	48.1	38.2		23.7	31.8		8.2	6.7		9.3	24.0%	P < 0.001
In hypocotyl
Daidzin	34.6	30.5		53.0	34.2		95.3	45.1		14.4	32.7%	P < 0.001
Genistin	452.7	176.5		409.7	124.5		308.6	124.7		5.0	14.6%	P < 0.01
Glycitin	107.1	65.7		53.5	69.5		51.0	56.7		3.0	ns	ns
Malonyldaidzin	1370.6	1051.2		1925.7	1053.4		3318.2	1243.5		16.0	35.2%	P < 0.001
Malonylgenistin	10838.2	3249.8		9346.5	2103.9		7855.9	2457.6		6.3	17.6%	P < 0.01
Malonylglycitin	252.1	202.6		372.0	243.7		643.4	304.7		12.7	30.1%	P < 0.001
G/D score	10.7	6.1		7.1	6.2		2.7	1.3		11.7	28.3%	P < 0.001

SD, standard deviation.

PVE, proportion of phenotypic variance explained.

ns, not significant.

G/D score, ratio of genistein-type isoflavones (genistin + malonylgenistin) to daidzein-type isoflavones (daidzin + malonyldaidzin).

These mapping experiments indicated that the 1-bp deletion in GmCHR5 was responsible for the altered production of daidzein and genistein derivatives in soybean seed tissues and that the effect of this mutation on G/D score was affected by the differences in growth conditions between the field (2018 summer season) and greenhouse (2019 winter season).

Expression of GmCHR5 during seed development in FUKU and MUT

A search of the RNA atlas in SoyBase (https://www.soybase.org) revealed that GmCHR5 is expressed during seed development at 25, 28, 35, and 42 DAF. We analyzed the expression profile of GmCHR5 by extracting RNA from developing seeds of FUKU and MUT at 40, 55, and 70 DAF. In both lines, constitutive GmCHR5 expression was detected at 40 and 55 DAF, but expression was reduced at 70 DAF (Fig. 4). GmCHR5 expression was much higher in FUKU (about 7.0 times at 40 and 55 DAF; 3.7 times at 70 DAF) than in MUT, suggesting that the mutation in GmCHR5 reduced its transcription or mRNA stability.

Fig. 4.

Expression of GmCHR5 in developing seed from FUKU and MUT. Expression levels are shown relative to the expression of the internal control gene GmACT2/7. P-value < 0.001 (two-way ANOVA) for the difference in relative expression between FUKU (solid line) and MUT (dashed line) at all three of the development stages indicated. DAF, days after flowering.

Ectopic expression of GmCHR5 in transgenic hairy roots

Control hairy roots of FUKU and MUT transformed with the pUB-GW-GFP vector showed a low G/D score (average G/D score for all transformants, 0.29 ± 0.13 [SD]) and Mda was the major isoflavone component of these hairy roots. FUKU (average G/D score, 0.27) showed a low G/D score compared with MUT (0.30, Fig. 5). We found a significant decrease in G/D score in 10 independent transgenic FUKU (0.08) and MUT (0.16) hairy roots showing a strong GFP signal and therefore overexpressing GmCHR5 (Fig. 5). These results indicated that ectopic expression of GmCHR5 increased the production of daidzein derivatives.

Fig. 5.

Ectopic expression of GmCHR5 in FUKU and MUT transgenic hairy roots. Values are means ± SD (10 independent hairy root lines). Different letters show significant differences by single-factor ANOVA and Tukey’s post-hoc test, P < 0.05.

Discussion

Here, we identified a gene, GmCHR5, mutated in an isoflavone soybean mutant with increased genistein derivative content and reduced daidzein derivative content in whole seed. Single-base deletion in GmCHR5 resulted in a truncated protein and caused the substrate flow to move away from the production of daidzein derivatives and toward the production of genistein derivatives in the isoflavone biosynthetic pathway. To identify the gene responsible for the mutant phenotype, we applied NNNs-HRM genotyping analysis to a small number (10) of F2 plants with the mutant phenotype; this analysis was followed by alignment of short reads obtained by NGS analysis to the identified QTL region. This procedure is time- and cost-effective in comparison with traditional linkage analysis using a linkage map covering the whole genome. The rapid mapping procedure demonstrated in the present study will be useful for accelerating the identification of genes in agronomically important crops that have large or complicated genomes.

In soybean, 14 CHR genes have been identified based on sequence homology, 11 of which are expected to be functional and have tissue-specific expression patterns (Sepiol ). The expression of Glyma.18G285800 (GmCHR5), Glyma.14G005700 (GmCHR1, named by Mameda ), and Glyma.02G307300 (GmCHR6) are induced by inoculation with the soybean pathogen Phytophthora sojae (Sepiol ). Physical interaction with isoflavone synthase was detected for GmCHR5 but not for GmCHR1 or 6 (Mameda , Nakayama ). The increased expression of GmCHR5 during seed development correlated well with the isoflavone accumulation pattern in developing seed (Mameda ). Together, these findings suggest that GmCHR5 plays a major role in the synthesis of daidzein derivatives in soybean seed.

CHR of Medicago sativa (312 aa, AAB41555 in a public database; https://www.ncbi.nlm.nih.gov/) is monomeric and has a (α/β)8-barrel structure (Bomati ). A BLAST comparison showed that aa 12–314 of GmCHR5 had a similar sequence to aa 11–312 of M. sativa CHR (262/303; 86% similarity). The mutation in the third exon of GmCHR5 resulted in production of a truncated protein (212 aa) lacking the C-terminal region containing aa residues related to NADP(H) binding (e.g., Pro-301 and Ile-298) (Bomati ). Further studies to evaluate the enzymatic activity of the MUT allele using expression vectors are necessary to confirm that the allele is null for CHR activity.

The formation of a metabolon comprising GmCHR5, isoflavone, and CHS on the surface of the endoplasmic reticulum is essential for rapid substrate transfer from CHS to CHR (Mameda , Waki ). Therefore, we can expect that GmCHR5 deficiency would increase naringenin production and decrease liquiritigenin production, which is supported by the high proportion of phenotypic variance explained (PVE) for G/D score we obtained for the mutant line. The loss-of-function of GmCHR5, however, did not lead to complete daidzein-derivative deficiency, likely because other CHR-encoding genes (e.g., GmCHR6, Glyma.16G219600, and Glyma.12G006800) are also expressed during early seed maturation (Sepiol ). The PVE for G/D score for the mutant allele varied among the experimental populations (Tables 2–4) and some POP1 plants showed G/D scores higher than that of MUT (Fig. 1). Together, these results indicate that other genetic and environmental factors affect the expression and/or enzymatic activity of GmCHR5.

In transgenic FUKU and MUT hairy roots, we observed a reduction in G/D score by GmCHR5-OE, indicating that exogenous GmCHR5 can contribute to the production of daidzein derivatives. In the mock (control vector only)-infected hairy roots, G/D scores were similar between FUKU and MUT, and low (0.27–0.3; Fig. 5) compared with seed (2.4–9.1, Table 1 and 1.5–6.2, Fig. 1). Endogenous daidzein was already synthesized in hairy roots, which may have decreased the difference between the G/D scores of FUKU and MUT. Soybean roots grown under a controlled environment showed a low G/D score and a significant difference between FUKU and MUT (0.037 for FUKU and 0.15 for MUT). These results indicate that the effect of the MUT allele in transgenic hairy roots was unstable and that multiple copies of soybean GmCHR would probably affect the biosynthesis of daidzein in a spatial or temporal manner. The silencing of GmCHR by RNAi in hairy roots decreases the production of daidzein derivatives and suppresses race-specific resistance against P. sojae (Graham ) because of reduced synthesis of the daidzein derivatives phytoalexin and glyceollin I. In the present study, we did not observe any negative effects of the GmCHR5 mutation on plant growth; however, whether this mutation decreases resistance to soybean diseases needs further examination.

Two QTLs have been reported around GmCHR5. The first, qSifc-a-18-3 (for total isoflavone content), detected by genome-wide association analysis using 366 Chinese soybean landraces, is located close to the SNP Gm18_57350313 and explains 2% of the PVE (Meng ). The second, qCDZ|proG_1, is a minor QTL that controls daidzein content and was detected between SSR (simple sequence repeat) markers Satt288 and Satt191 (Gm18, 55.4–58.7 Mbp in Glyma 1.0) by bi-parental QTL analysis of two Chinese soybean varieties (Han ). It is possible that the cause of these QTLs is a mutation of GmCHR5. Comparison of the sequences of GmCHR5 in soybean germplasms of wide geographical origin will be necessary to evaluate the genetic effects of GmCHR5 on the phenotypic diversity of isoflavone content within soybean. Further study using the mutant line identified in the present study will also be useful to uncover the biological activities of GmCHR5, because this mutant line has a uniform genetic background with the wild-type line.

The allele carrying a mutation of GmCHR5 found in the present study may be useful as a source allele to produce soybean lines that produce seeds with a high genistein-derivative content. Such lines would have high breeding value because of the potentially beneficial effects of genistein on inflammation, angiogenesis, and the development of metabolic diseases in humans (Mukund ).

In conclusion, we identified a novel mutant allele, GmCHR5, responsible for increased production of major aglycones and their derivatives in a soybean mutant line. This mutant allele could be a useful genetic material for future breeding programs. Our approach, selective genotyping with NNNs-HRM markers followed by NGS analysis of a restricted chromosomal region, is a time- and cost-effective means of identifying the genes responsible for a target phenotypic trait from a dense mutant library, especially for crops with large or complicated genomes.

Author Contribution Statement

SW, AS, FH, and TA designed the study. SW, WO, and RS collected the data. SW, RS, and TA analyzed the data.

Supplemental Figures

Supplemental Tables