Mutant Tof11 alleles are highly accumulated in early planting-adaptable Japanese summer type soybeans

Abstract

To avoid crop failure because of climate change, soybean (Glycine max (L.) Merrill) cultivars adaptable to early planting are required in western Japan. Because current Japanese cultivars may not be adaptable, genetic resources with high early-planting adaptability, and their genetic information must be developed. In the present study, summer type (ST) soybeans developed for early planting were used as plant materials. We examined their phenological characteristics and short reproductive period as an indicator of early planting adaptability and performed genetic studies. Biparental quantitative trait loci (QTL) analysis of a representative ST cultivar revealed a principal QTL for the reproductive period duration on chromosome 11. The results of resequencing analysis suggested that circadian clock-related Tof11 (soybean orthologue of PRR3) is a candidate QTL. Additionally, all 25 early planting-adaptable germplasms evaluated in this study possessed mutant alleles in Tof11, whereas 15 conventional cultivars only had wild-type alleles. These results suggest that mutant alleles in Tof11 are important genetic factors in the high adaptability to early planting of these soybeans, and thus, these alleles were acquired and accumulated in the ST soybean population.

Introduction

In East Asian countries, including Japan, the frequency of extreme rainfall events during the Meiyu-Baiu season (mid-June to mid-July), has increased because of climate change (Takahashi and Fujinami 2021). As the conventional sowing season for soybean (Glycine max (L.) Merrill) in western Japan is during this period, poor weather and soil conditions often lead to crop failure. To avoid cropping season-dependent instability, early planting should be performed. However, Japanese conventional cultivars do not necessarily achieve high yields under early planting conditions in this region (Matsuo et al. 2016). A key factor influencing adaptability is the properties of a cultivar’s reproductive stage. For high adaptability to early planting, the ability to produce and develop fertile pods under long photoperiod conditions may be required. Although it remains unclear how this ability is promoted, Xu et al. (2013) found that the reproductive period (RP) and pod developmental speed under long day-length conditions were influenced by genes related to day-length sensitivity (Hereafter, the R1–R8 stages are defined as the RP; the developmental stage is defined as described by Fehr and Caviness (1977)). Thus, reproductive development after flowering may be inhibited by long day-length conditions similar to flowering induction, although the response is genotype-dependent. Furthermore, a prolonged RP may indicate inhibition of reproductive development, whereas a short RP may be a sign of appropriate development. Indeed, early planting-adaptable cultivars in the US have a relatively short RP compared with those of low-adaptable Japanese cultivars (Matsuo et al. 2016). Therefore, the short RP under long day-length conditions should be evaluated to understand the genetic basis of adaptability to early planting.

Summer type (ST) soybean germplasms in Japan are important plant materials for developing early planting-adaptable cultivars. In this study, we regard soybean landraces and cultivars that have been used for early planting cultivation in mid- to south-area of Japan (from Kanto to Kyushu district) as ST soybeans. In some reports (for example Nagata 1949), early landraces and cultivars in high latitudinal Japan are also included in ST soybeans, but we have excluded them from the category considering the purpose of this study. ST lines were the predominant soybean crops in southwestern Japan (Kyushu district) from 1945 to 1965 (Gotoh 1983). Their proper planting dates in this region are mid- to late-April, and the harvest season is from early- to mid-August (Fukui and Arai 1951, Konno 1970), supporting their high adaptability to early planting. Additionally, ST soybeans have a relatively short RP as a unique phenological characteristic (Fukui and Arai 1951). However, the genetic basis of their RP has not been extensively investigated. Genetic studies of ST soybean are needed to verify the causal relationship between the short RP and high adaptability to early planting, and to effectively utilize early planting-adaptable genetic resources in new breeding programs.

In this study, the main genetic factors associated with the short RP of a representative ST cultivar were evaluated via biparental quantitative trait loci (QTL) analysis. Mutations in an identified candidate gene corresponding to the QTL were analyzed through resequencing analysis using next-generation sequencing (NGS). In addition, allele frequencies in the candidate gene (Tof11) influencing the short RP of Japanese ST soybean were examined to further understand the genetic basis of the RP duration.

Materials and Methods

QTL analysis of RP

QTL analysis was conducted to detect genetic factor(s) associated with ST soybean’s unique reproductive growth characteristics. A population derived from a cross of ‘TAMAHOMARE’ (registered as JP29184 in Genebank Project of National Agriculture and Food Research Organization (NARO)) and ‘KOGANEDAIZU’ (JP29504) was developed. ‘KOGANEDAIZU’ is a representative Japanese ST cultivar that exhibits a short RP under the long-day conditions in southwestern Japan (Fig. 1). ‘TAMAHOMARE’ is a cultivar with conventional phenological characteristics suitable for normal planting dates (late-June to mid-July) in western Japan (Fig. 1). A previous study (Tsubokura et al. 2014) and our original research (data no shown) revealed that ‘KOGANEDAIZU’ and ‘TAMAHOMARE’ had the same genotypes at the four major maturity genes (E1, E2, E3, and E4).

Fig. 1.

Difference in maturation progress among conventional cultivar ‘TAMAHOMARE’ and Japanese summer type (ST) soybeans. Pot No. 1: ‘TAMAHOMARE’, No. 2: ‘KOGANEDAIZU’, No. 3: ‘BONMINORI’, No. 4: ‘HIGOMUSUME’, and No. 5: ‘KIMUSUME’. Pot Nos. 2–5 contained Japanese ST soybean germplasms. Dates of sowing and photography were August 5 and October 25 (81 days after sowing), 2022, respectively. The plants flowered within three days (days 33–35 after sowing); however, the reproductive development of ST lines including ‘KOGANEDAIZU’ rapidly progressed compared to that of the conventional cultivar ‘TAMAHOMARE’. E1–E4 showed identical genotypes in these samples (see Table 1).

The F₂ and F_2:3 populations from the cross described above were grown in an experimental field at the Western Region Agricultural Research Center (Kinki, Chugoku and Shikoku Regions), NARO, at Zentsuji City (34°13ʹN, 133°46ʹE) in 2017 and 2018, respectively. The parents were grown in the same field in both years. The materials were sown in mid- to late-June (June 23, 2017; June 18, 2018), and their flowering and maturation dates were determined based on the soybean growth stage definition reported by Fehr and Caviness (1977). The sowing dates were fixed mid- to late-June to prevent green stem disorder; sowing before mid-June can lead to defects in conventional cultivars and some of their progenies, making it difficult to evaluate the intrinsic maturation date of the materials (Hill et al. 2006).

To construct a linkage map for QTL analysis, total DNA was extracted from the leaves of each F₂ individual as described by Edwards et al. (1991). The genotypes of a set of simple sequence repeat (SSR) markers (Sayama et al. 2011) were examined. Additionally, five SSR loci (BARCSOYSSR_11_0520, 0708, 0867, 1109, and 1492; Song et al. 2010) were used for map construction of chromosome (Chr-) 11. In total, 62 marker loci were polymorphic between the parents and were used for QTL analysis. Polymerase chain reaction (PCR) was performed as described previously (Sayama et al. 2011, Song et al. 2010), and the PCR products were electrophoresed on a LabChip GX Touch (PerkinElmer, Waltham, MA, USA). Based on the marker genotype data, a linkage map was constructed using MAPMAKER/EXP 3.0b software (Lincoln et al. 1993). Based on the phenotype and linkage data, composite interval mapping was performed with QTL Cartographor 2.5_011 (Wang et al. 2012). The logarithm of odds (LOD) threshold for QTL detection equivalent to a 5% type I error rate was determined through 1000 permutation tests (Churchill and Doerge 1994).

Resequencing analysis of ‘KOGANEDAIZU’ and ‘TAMAHOMARE’

The results of QTL analysis suggested that a genetic factor located between GMES1909 and BARCSOYSSR_11_1078 strongly affect the RP. Identifying candidate genes of the QTL is important for understanding early planting adaptability and short RP. Thus, we evaluated the physical positions of some genes on Chr-11 that may influence soybean maturity (Watanabe et al. 2012) using the Phytozome v13 website (https://phytozome-next.jgi.doe.gov/). To narrow down the candidate genes, we performed genome resequencing analysis of the two parental lines, ‘KOGANEDAIZU’ and ‘TAMAHOMARE’. For resequencing, total DNA was extracted from the leaves of these plants using a DNeasy Plant Mini Kit (Qiagen, Hilden, Germany). NGS sequencing was outsourced to Eurofins Genomics Japan (Tokyo, Japan). Read data over 30× coverage of the whole genome were obtained using an Illumina HiSeq X Ten system (Illumina, San Diego, CA, USA) for both lines. The data were deposited in the DNA Data Bank of Japan Sequence Read Archive under accession number DRA015319. To detect sequence variation between the two lines, the read data were processed using CLC Genomics Workbench v22 (Qiagen Arhus A/S). Primer sequences were trimmed from the FASTQ data. The trimmed sequences were aligned to the reference genome (Gmax_580_Wm82.a4.v1 assembly) using the ‘Map Reads to Reference’ and ‘Local Realignment’ functions. The reference genome data were obtained from the Phytozome v13 website. After alignment, the ‘Basic Variant Detection’ function was used to call the variation. The sequences, including the reference sequence, were compared using a ‘truck list’ comprised of the variant called result files. The default settings of the software were used as processing parameters for read data alignment and variant calling.

Estimation of Tof11 effects on RP

Our QTL and resequencing analyses revealed Glyma.11G148362 (Tof11) as the most probable candidate gene responsible for the short RP of ‘KOGANEDAIZU’. To estimate this effect, the relationship between the Tof11 genotype and RP duration was investigated using the segregating population used for QTL analyses. The population was classified into three subgroups (TAMAHOMARE homozygous, heterozygous, and KOGANEDAIZU homozygous) according to the Tof11 genotype in the F₂ generation, and the mean RP durations of the subgroups in F₂ and F_2:3 were compared. Genotyping of Tof11 was carried out using the high-resolution melting (HRM) method (Vossen et al. 2009) on a LightCycler 96 System (Roche Diagnostics, Basel, Switzerland). Primers for genotyping were constructed based on sequence variation in ‘TAMAHOMARE’ and ‘KOGANEDAIZU’ (one-base deletion at nucleotide 20,364 of Tof11 gene region containing the 5ʹ-untranslated region). Primer sequences and sizes of the expected amplicons are shown in Supplemental Table 1. AmpliTaq Gold 360 Master Mix (Thermo Fisher Scientific, Waltham, MA, USA) containing DNA polymerase was used for the PCR reactions, according to the manufacturer’s instructions. SYTO 9 Green Fluorescent Nucleic Acid Stain (Thermo Fisher Scientific) was used as the intercalator at a final concentration of 1.5 μM. The PCR thermocycling conditions were the same as those described for the BARCSOYSSR primers (Song et al. 2010). Statistical analysis was performed using the Rcmdr-package (Fox 2005) in R (Ihaka and Gentleman 1996).

Investigation of Tof11 allele frequencies in Japanese ST germplasms

Our QTL analysis and resequencing analysis revealed Glyma.11G148362 (Tof11) as the most probable candidate gene responsible for the short RP of ‘KOGANEDAIZU’. Based on these results, the relationship between Tof11 and early planting adaptability was investigated. The genetic features Tof11 in ST soybean germplasms were analyzed. Some alleles in Tof11 in the natural soybean population have been reported (Lu et al. 2020), whereas in the current study, a novel allele was detected. We performed genotyping analysis of Tof11 and other loci related to flowering and maturation (E1–E4). As plant materials, we used 21 cultivars or landraces considered ST, according to prior reports. Representative STs are shown in Fig. 1. We chose the materials based on four previous reports; three of these reports, except for that by Furutani et al. (1962), are in Japanese and thus were excluded from references in this article. In addition to the ST germplasms, 15 Japanese conventional cultivars from the western region and four cultivars from the US used in a previous study of early planting (Matsuo et al. 2016) were examined for comparison. Seeds from the ST germplasms and eight Japanese conventional cultivars were obtained from the Genebank Project of NARO. The other conventional Japanese cultivars were directly obtained from breeders, as they have not been registered in Genebank. The seeds of US cultivars were obtained from the US Department of Agriculture. The names of the materials and their registration numbers (JP number of NARO Genebank or PI number of US Department of Agriculture) are shown in Table 1, together with the genotyping results.

Table 1. Genotypes of five genes related to phenological characteristics

Cultivar name	Cultivar attribution	Registration number	Tof11 genotype	E1 genotype	E2 genotype	E3 genotype	E4 genotype
HIGO DAIZU	Summer	JP27550	tof11-a3	E1	e2-ns	E3	E4
KIMUSUME	Summer	JP27552	tof11-a3	E1	e2-ns	e3-tr	E4
SHIN 3	Summer	JP28239	tof11-a2	E1	e2-ns	E3	E4
SHIROME	Summer	JP28379	tof11-a3	E1	e2-ns	e3-tr	E4
BONMINORI	Summer	JP28493	tof11-a2	E1	e2-ns	e3-tr	E4
AOCHI	Summer	JP29369	tof11-a3	E1	e2-ns	e3-tr	E4
MATSUURA1)	Summer	JP29470	tof11-a3	E1	e2-ns	E3	E4
HIGOMUSUME	Summer	JP29475	tof11-a3	E1	e2-ns	e3-tr	E4
KISAYA	Summer	JP29485	tof11-a3	E1	e2-ns	E3	E4
FUUFU DAIZU	Summer	JP29487	tof11-a3	E1	e2-ns	e3-tr	E4
KAIRYOU SHIROME	Summer	JP29499	tof11-a3	E1	e2-ns	e3-tr	E4
KOGANEDAIZU1)	Summer	JP29504	tof11-a2	E1	e2-ns	e3-tr	E4
SAYOHIME	Summer	JP29505	tof11-a3	E1	e2-ns	E3	E4
FUJIMUSUME	Summer	JP29506	tof11-a2	E1	e2-ns	e3-tr	E4
WASE NATSU	Summer	JP29508	tof11-a3	E1	e2-ns	e3-tr	E4
KASUGA ZAIRAI	Summer	JP29509	tof11-a3	E1	e2-ns	e3-tr	E4
KAIRYOU GIONBOU	Summer	JP29510	tof11-a3	E1	e2-ns	e3-tr	E4
ICHIGOU WASE	Summer	JP29517	tof11-a3	E1	e2-ns	e3-tr	E4
SANGOU WASE	Summer	JP29519	tof11-a3	E1	e2-ns	e3-tr	E4
ORIHIME	Summer	JP29623	tof11-a3	E1	e2-ns	e3-tr	E4
SHIROSAYA 1 (NOUKEN)	Summer	JP250733	tof11-a3	E1	e2-ns	e3-tr	E4
ENREI1)	Conventional	JP28862	Tof11	E1	e2-ns	e3-tr	E4
TAMAHOMARE	Conventional	JP29184	Tof11	E1	e2-ns	e3-tr	E4
AKIYOSHI	Conventional	JP29578	Tof11	E1	E2	E3	E4
FUKUYUTAKA1)	Conventional	JP29668	Tof11	E1	E2	E3	E4
AKISHIROME1)	Conventional	JP29669	Tof11	E1	e2-ns	E3	E4
TOYOSHIROME	Conventional	JP29681	Tof11	E1	e2-ns	E3	E4
TACHINAGAHA1)	Conventional	JP67666	Tof11	E1	e2-ns	e3-tr	E4
NISHIMUSUME	Conventional	JP76377	Tof11	E1	e2-ns	E3	E4
SACHIYUTAKA1)	Conventional	—	Tof11	E1	E2	e3-tr	E4
AKIMARO	Conventional	—	Tof11	E1	e2-ns	E3	E4
TATSUMARO	Conventional	—	Tof11	E1	E2	e3-tr	E4
KOGANESAYAKA	Conventional	—	Tof11	E1	E2	e3-tr	E4
HATSUSAYAKA	Conventional	—	Tof11	E1	e2-ns	E3	E4
KOTOYUTAKA	Conventional	—	Tof11	E1	e2-ns	E3	E4
SATONOHOHOEMI	Conventional	—	Tof11	E1	e2-ns	E3	E4
Spry	US cultivar	PI 553051	tof11-a1	E1	E2	E3	E4
5002T	US cultivar	PI 634193	tof11-a1	E1	E2	E3	E4
UA 4805	US cultivar	PI 639187	tof11-a1	E1	E2	E3	E4
LD00-3309	US cultivar	PI 639740	tof11-a1	e1-as	E2	E3	E4
Williams 821)	For comparison	PI 518671	tof11-a1	e1-as	E2	E3	E4
PEKING1)	For comparison	JP28432	tof11-a1	E1	E2	E3	E4

1) The HRM results of E1 to E4 genotyping were the same as the gel-based results reported by Tsubokura et al. (2014).

According to Lu et al. (2020), three alleles in Tof11, wild-type (Tof11), tof11-1 and tof11-8 are predominant in Japanese cultivars and landraces. tof11-1 is characterized by a single base deletion in exon 9. In the Gmax_580_Wm82.a4.v1 assembly, the deletion site is at nucleotide 21,248 of the gene region containing the 5ʹ untranslated region. tof11-8 has a five-base deletion in exon 7, corresponding to nucleotides 20,015–20,019. We also detected a new mutant allele in ‘KOGANEDAIZU’ via resequencing analysis. The ‘KOGANEDAIZU type allele is characterized by a one-base deletion of nucleotide 20,364. Based on these results, the objective mutation sites analyzed via genotyping were limited to nucleotides 20,015–20,019, 20,364, and 21,248.

The sites were genotyped using the high-resolution melting method (Vossen et al. 2009) on a LightCycler 96 System (Roche Diagnostics, Basel, Switzerland). DNA was extracted from the seeds using Template Prepper for DNA reagent (NIPPON GENE, Tokyo, Japan) according to the manufacturer’s instructions. The PCR was performed as described above for the Tof11 effect. Primer sequences and sizes of expected amplicons are shown in Supplemental Table 1.

E1–E4 were genotyped using high-resolution melting. Primer sequences for the four genes are shown in Supplemental Table 1. All primer sets were designated based on each allele sequence reported by Tsubokura et al. (2014). Allele names were also as described in the previous study. The primer set for the E1 locus can detect the e1-as and e1-nl alleles. However, the e1-af allele cannot be distinguished from E1 using these primers, as the frequency of e1-af is very low (Tsubokura et al. 2014, Xu et al. 2013). Therefore, we considered all alleles that differed from e1-as and e1-nl as E1. The primer set for E2 locus can distinguish e2-ns from the others (E2-in and E2-dl). Two alleles that differed from e2-ns, E2-in and E2-dl, were expected but were considered as equally functional and named as ‘E2-’ (Tsubokura et al. 2014); therefore, we treated them as the same ‘E2’ allele. The primer set for the E3 locus can distinguish the e3-tr allele from the others (E3-Ha, E3-Mi, and e3-Mo). As for the E3 locus, we collectively considered the other allele that differed from e3-tr to be ‘E3’ because E3-Ha and E3-Mi were equally functional, and the frequency of e3-Mo was very low (Tsubokura et al. 2014). The primer set for E4 locus can detect the e4-SORE-1 allele. Lines that did not contain e4-SORE-1 were considered E4. The experimental conditions for PCR were the same as those described above for Tof11 genotyping.

Results

QTL analysis of RP

To detect genetic factor(s) associated with the short RP in ST soybeans, QTL analysis was performed using the F₂ and F_2:3 populations. A linkage map composed of 13 linkage groups, spanning 1258 cM was constructed. The average intermarker distance was 25.7 cM. Using the composite interval mapping method, a significant LOD peak was detected around the GMES0027 locus on Chr-11 in both F₂ and F_2:3. The LOD curve for Chr-11 and estimated effects of the locus are shown in Fig. 2. The estimated additive effects in the QTL analysis suggest that the allele from the paternal parent ‘KOGANEDAIZU’ lead to the short RP. The marker interval corresponding to the QTL regions GMES1909 and BARCSOYSSR_11_0708 spanned 26.4 cM. The physical distance between the two marker loci was 4.64 Mbps in the Gmax_580_Wm82.a4.v1 assembly. In this interval, 546 genes (loci) were annotated (data not shown).

Fig. 2.

Quantitative trait loci (QTL) for duration of reproductive period (RP) on chromosome 11 and physical position of marker and candidate gene loci. The additive effects shown in the small table were equivalent to the effect of the ‘TAMAHOMARE’ allele. The r² values indicate the variation explained by the QTL among the total phenotype variation. Horizontally long blank box shown in the bottom of the figure is a schematic of soybean chromosome 11. The ‘BSS’ in the marker locus name indicates ‘BARCSOYSSR’. Tof11 shown under the chromosome schema is the candidate gene for the QTL.

To estimate the effects of the detected QTL, the relationship between the frequency distribution of the RP and GMES0027 genotype was examined (Fig. 3). The mode of the duration was 96–99 days for the F₂ population. The long-duration parent ‘TAMAHOMARE’ exhibited a slightly longer duration (101 days) than the mode value. The tail of the short duration side was extended and contained many individuals fixed for the ‘KOGANEDAIZU’ allele at the GMES0027 locus. In contrast, most individuals fixed for the ‘TAMAHOMARE’ allele were near the mode (96–99 days) or on the longer duration side. Thus, the duration of the RP may be strongly influenced by genetic factors flanking the SSR locus GMES0027, and an allele in ‘KOGANEDAIZU’ leads to a short RP. The F_2:3 population showed a random distribution with three peaks. However, the ‘KOGANEDAIZU’ and ‘TAMAHOMARE’ type lines were concentrated on the short and long duration sides, respectively, supporting the results of QTL analysis.

Fig. 3.

Frequency distribution of reproductive period (RP) duration and genotype of GMES0027. A: F₂ population of ‘TAMAHOMARE’ × ‘KOGANEDAIZU’. The down arrows in the top of the figure indicate RP duration of the parental lines. B: F_2:3 population, which is the progeny of the F₂ population shown in A. The genotype of a parental F₂ individual represents that of its progeny line.

In addition to the QTL of RP duration, some QTLs of duration from sowing to flowering (R1 stage of Fehr and Caviness (1977)) and that from sowing to maturing (R8 stage of Fehr and Caviness (1977)) were detected (Supplemental Table 2). Hereafter, the periods from sowing to R1 and from sowing to R8 are referred to as the vegetative period (VP) and total growth period (TGP), respectively. A stable QTL of VP that was significant in both F₂ and F_2:3 was detected on Chr-08. For TGP, a stable QTL on Chr-11 was detected. The position of the TGP QTL was almost the same as that of RP described above. The correlation coefficients between RP and VP in F₂ and F_2:3 were –0.12 and 0.45, respectively. Only F_2:3 exhibited a statistically significant correlation at the 0.1% level. The coefficients of determination were relatively low (0.01 and 0.20, respectively), however, the correlation coefficient was significant. These results suggested that the relationship between RP and VP was relatively weak and that the genetic factor(s) controlling RP and VP may be different, in particular in F₂.

Resequencing analysis of ‘KOGANEDAIZU’ and ‘TAMAHOMARE’

QTL analysis showed that genetic factors that strongly impact the RP duration were present in the central part of Chr-11. Although more than 500 genes were annotated in the QTL region, three candidate genes (Glyma.11G136600, Glyma.11G148362 (Tof11), and Glyma.11G149825 in the Gmax_580_Wm82.a4.v1 assembly) associated with phonological characteristics were selected based on a previous study (Watanabe et al. 2012) and their sequences were obtained via resequencing with NGS.

Glyma.11G149825 did not show sequence variation between ‘KOGANEDAIZU’ and ‘TAMAHOMARE’ and thus was excluded from further analysis. In contrast, sequence variations predicted to affect gene function were detected in Glyma.11G136600 and Tof11 in the two soybean lines. Two missense single-base substitutions in the coding sequence were detected in Glyma.11G136600 of ‘KOGANEDAIZU’ (Supplemental Fig. 1). Tof11 in ‘KOGANEDAIZU’ has a one-base deletion at the junction of exon 7 and intron 7, whereas ‘TAMAHOMARE’ showed the same sequence as the reference at this site (Fig. 4). According to Brown (1999), the nucleotide sequence ‘GU’ in pre-mRNA is an important 5ʹ splice site of the intron. Therefore, this deletion may seriously affect mRNA splicing.

Fig. 4.

Variation of Tof11 structure among ‘TAMAHOMARE’, ‘KOGANEDAIZU’, and reference cultivar ‘Williams 82’. Black boxes, shaded boxes, and horizontal lines indicate exons, untranslated regions (UTRs), and introns, respectively. The numbers around the schema are the nucleotide order counted from the tip of 5ʹ side of the gene region including the UTR. The asterisk indicates a five-base deletion site used for genotyping. The sequence at the site did not vary among the three lines shown in this figure.

Estimation of Tof11 effect on RP and VP

The population derived from ‘TAMAHOMARE’ × ‘KOGANEDAIZU’ used for QTL analysis was classified into three subgroups according to the Tof11 genotype in the F₂ generation. ‘TAMAHOMARE’ homozygous, heterozygous, and ‘KOGANEDAIZU’ homozygous subgroups comprised 32, 71, and 30 individuals (F₂) and lines (F_2:3), respectively. The genotypes of Tof11 and GMES0027 that were used in QTL analysis and located close to Tof11 (approximately 260 Kbps apart) were very similar (only eight individuals in 133 had recombination between the two loci). This indicated that the results of our HRM genotyping method were reliable.

The RP and VP durations of the subgroups are shown in Fig. 5. RP duration was significantly different among the three subgroups in the F₂ and F_2:3 populations. It is possible that a large proportion of the difference between parental lines is controlled by Tof11. In contrast, the relationship between the Tof11 genotype and VP duration was ambiguous. No significant differences among the subgroups were detected in F₂. Although the difference in F_2:3 was significant, the numerical value of this difference was small.

Fig. 5.

Relationship between growth periods and Tof11 genotype in the ‘TAMAHOMARE’ × ‘KOGANEDAIZU’ population. The F₂ and F_2:3 progenies used in the QTL analysis were classified into three subgroups according to the genotype of Tof11 in the F₂ generation, and the mean duration of the growth period was compared. ‘D’, ‘SD’, and ‘S’ shown in boxes indicate mean of duration, standard deviation of duration, and statistically significance among the three subgroups, respectively. Different alphabet for the ‘S’ indicates significant difference at 0.1% level by Tukey’s multiple comparison test. Values of parental lines are also shown for comparison.

Tof11 allele frequencies in Japanese ST germplasms

To investigate the relationship between the phenological characteristics of ST soybean and Tof11, Tof11 and four E series genes (E1–E4) were genotyped (Table 1). HRM was used for genotyping. Nine germplasms of our study and a previous study using gel-based genotyping (Tsubokura et al. 2014) exhibited the same genotype (Table 1). This indicated that our HRM genotyping method for E1 to E4 was reliable.

Four genotypes were detected for Tof11. We named an allele exhibiting a one-base deletion at nucleotide 21,248 as tof11-a1 and the allele with a one-base deletion at nucleotide 20,364 (KOGANEDAIZU type) as tof11-a2. The allele with a five-base deletion of nucleotides 20,015–20,019 was named as tof11-a3. tof11-a1 and tof11-a3 may correspond to tof11-1 and tof11-8 described by Lu et al. (2020), respectively, but we named them independently because our genotyping was not based on resequencing, and thus unknown sequence differences may exist other sites.

Among the ST germplasms, no lines contained wild-type Tof11. All lines possessed an allele containing a mutation that may be related to decreased gene function (tof11-a2 and tof11-a3). Approximately 80% of the ST lines (17/21) contained tof11-a3, whereas the others possessed tof11-a2. In contrast with the ST lines, all conventional Japanese cultivars showed a wild-type Tof11 allele. All four US cultivars showed the same tof11-a1 sequence as in the reference genome (‘Williams 82’). tof11-a1 was detected only in US cultivars, although some Japanese cultivars have been found to contain this deletion (Lu et al. 2020). According to the Tof11 genotypes, the plant materials were classified into three groups: ST germplasms (tof11-a2 or tof11-a3), conventional Japanese cultivars (Tof11), and US cultivars (tof11-a1).

The ST lines exhibited differences in the genotype of E2 compared with the other plant materials. All ST lines contained the e2 allele, whereas conventional Japanese cultivars showed variable genotypes, and all US cultivars had the wild-type E2 allele. For E1, E3, and E4, no clear difference in allele frequency was detected between the ST germplasms and other plant materials.

Discussion

Via our QTL analysis, a locus strongly affecting the RP duration was detected in Chr-11. A QTL that also strongly impacts the RP has been detected around this position in different plant materials (Komatsu et al. 2012). Kong et al. (2018) also detected a QTL involved in the duration of RP in the central part of Chr-11, although its effect may depend on the environmental conditions. Thus, soybean may contain genetic factor(s) in Chr-11 that regulate the duration of the RP. We evaluated the genetic factor detected in the current study to enable further studies on the adaptability to early planting and short RP of soybean.

Resequencing analysis of ‘KOGANEDAIZU’ and ‘TAMAHOMARE’, as the parents in our QTL analysis, was performed using NGS. Candidate gene were selected based on the results of QTL analysis and a report on genes related to the flowering time in soybean. We identified three candidate genes related to the short RP (Glyma.11G136600, Glyma.11G148362 (Tof11), and Glyma.11G149825 in the Gmax_580_Wm82.a4.v1 assembly), which were further narrowed down based on the physical position, sequence variation, and estimated influence on gene function.

Glyma.11G149825 was excluded from the candidate because sequence variation from the parental lines was not detected. In contrast, some sequence variations were detected in Glyma.11G136600. The two missense mutations detected in the coding sequence may lead to differences in gene function. However, the ‘KOGANEDAIZU’ haplotype was widely detected in late maturing Japanese cultivars or landraces that were strongly expected to have a comparably long RP, according to the ‘TASUKE’ (Kumagai et al. 2013, 2019) website (https://daizutasuke275-core.daizu.dna.affrc.go.jp/). In addition, the physical position of the gene is very close to the marker locus GMES1909 (approximately 250 Kbps), and thus was separated from the LOD peak in our QTL analysis (see Fig. 2). Particularly, for F_2:3, a gap of the LOD peak and GMES1909 locus was around 30 cM. Based on these results, it is unlikely that Glyma.11G136600 is responsible for the differences in the RP duration.

In contrast to the two genes described above, Tof11 likely affects RP in soybeans. A single-base deletion (GT to G-) at the 5ʹ-splice site of the 7th intron was detected in ‘KOGANEDAIZU’. In Arabidopsis, almost all introns had GU (GT in the genome) at the 5ʹ-splice site and AG at the 3ʹ-splice site (Nickolai et al. 2006). According to a previous study, the frequency of non-canonical (not GU-AG) splice sites is 0.7% and 1.4% in Arabidopsis and rice, respectively. In rice, a one-base substitution at the GT site that promotes different alleles at the waxy locus has been reported (Isshiki et al. 1998). From these facts, it is suggested that the GT in the plant intron 5ʹ-splice site plays a very important role with regard to mRNA splicing. Therefore, it is likely that the single-base deletion at the GT of ‘KOGANEDAIZU’ leads to the loss or repression of gene function of Tof11.

Tof11 is close (approximately 260 Kbps) to the marker locus GMES0027 which is nearest the two LOD peaks detected in QTL analysis (see Fig. 2). Tof11 is considered an orthologue of PRR3 of Arabidopsis thaliana and its function has been determined to some degree (reviewed by Li and Lam 2020). Tof11 acts as repressor of soybean CCA1 and LHY which repress E1. As E1 is a repressor of GmFT2a (a florigen gene in soybean), malfunction of Tof11 accelerates flowering. Tof11 has not been reported to participate in controlling the RP duration but is likely involved in this process, as the E1-related day-length sensing system also prolongs the RP duration under long day-length conditions (Xu et al. 2013). CCA1 and LHY are key components of the circadian clock (reviewed by Barak et al. 2000) in addition to acting as repressors of E1; therefore, Tof11 may influence the RP duration independently of E1.

The effects of Tof11 on RP and VP were estimated using the same population for QTL analysis. The results indicated that Tof11 may have a considerable impact on the duration of RP, whereas it may have a limited influence on VP duration. This estimation coincided with the results of the QTL analysis for VP and the correlation analysis. Because the estimation was obtained from progeny of the biparental cross, we cannot exclude the possibility that some genetic factors linked to Tof11 affect the duration. Therefore, this result is not suitable for the identification of genes corresponding to the RP QTL on Chr-11. However, the results of our study, including QTL and resequencing analyses, support the hypothesis that Tof11 is the responsible gene. Although we cannot conclude that the detected QTL for the short RP is identical to Tof11, we focused on Tof11 as the most probable candidate gene.

To investigate the relationship between Tof11 and early planting adaptability, the genotypes of materials suitable for early planting were genotyped, including Japanese ST germplasms and some US cultivars (Matsuo et al. 2016). All ST germplasms and US cultivars possessed the mutant type allele that leads to Tof11 protein to loss or represses the protein function. In contrast, Japanese conventional cultivars possess only the wild-type allele. These results suggest that the mutated Tof11 allele enables suitable reproductive growth of early planting-adaptable soybeans under long day-length conditions. Moreover, mutant alleles may be distinguishing factors in ST soybean. Through isozyme and seed protein analysis, Hirata et al. (1996) estimated that the ST germplasms in the Kyushu district were formed within Japan from various origins, which remains unclear. ST germplasms were likely formed through extensive crossing and selection processes, during which mutant alleles accumulated in the population to improve the adaptability to early planting.

We observed differences in the Tof11 allele between ST lines and US cultivars. The ST lines contained tof11-a2 or tof11-a3, whereas the US lines possessed only tof11-a1, although we only evaluated four US lines. Interestingly, the frequency of deletion at nucleotide 21,248 (corresponding to tof11-a1) is higher than that of the five-base deletion at nucleotides 20,015–20,019 (corresponding to tof11-a3) in the Japanese cultivar (Lu et al. 2020). Although the reason for this result remains unclear, differences in the effects of alleles on adaptability and in cultivating conditions between the plant material’s origin may explain the genotype variation. Early planting conditions may not identical between the US and Japan. According to Vann et al. (2021), in North Carolina, which has a similar latitude to western Japan, most early planting is performed in May rather than in April. In contrast, Japanese ST soybeans are typically planted in April. The time lag can lead to differences in environmental conditions such as day length during the RP. tof11-a1 may be suitable for mild early planting (planting in May) and in-time planting (in June), whereas tof11-a2 and tof11-a3 can be used only for extremely early planting (in April).

The allele composition of E series genes may be related to the RP duration or early planting adaptability. All ST germplasms exhibit E1E1 e2e2 E4E4. Particularly, e2 fixation is characteristic in ST lines, as the locus is fixed for E2 or segregates in other material groups (Japanese conventional and US cultivars). E2 has been predicted to influence the RP duration and time to flowering (Komatsu et al. 2012). The e2 allele may be required for both flower induction and normal reproductive development under long day-length conditions. The e2 allele should be further evaluated to determine its role in early planting adaptability. E2 is a soybean ortholog of GIGANTEA in Arabidopsis and considered to be related to the circadian clock (Watanabe et al. 2011). As Tof11 is also a circadian clock-related regulator, E2 and Tof11 may interact to control the RP duration. Epistasis between E2 and GmPRR3s (Li et al. 2019) and between E2 and another genetic factor related to the circadian clock (Su et al. 2021) has been suggested to affect the soybean flowering time or whole growth period, supporting that E2 and Tof11 interact to regulate the RP.

The relationship among the allele composition of Tof11 and E2, RP duration, and early planting adaptability should be verified using ideal plant materials such as near-isogenic lines of Tof11 and E2. As a DNA marker system is available for allele recognition, plant materials can be easily developed in a short period. The materials and marker system are also useful for breeding programs. Studies of the relationship between the allele composition and its effects on RP may improve the understanding of the role of the circadian clock in soybean reproductive development rather than only in flower induction.

Author Contribution Statement

K.K.: Quantitative trait loci analysis, resequencing analysis, genotyping of Tof11 and E series genes. T.S.: Quantitative trait loci analysis, genotyping of Tof11 and E series genes. K.Y.: Quantitative trait loci analysis. Y.T.: Evaluation of flowering and maturation date of segregating population, quantitative trait loci analysis.

Acknowledgments

We thank Dr. Akito Kaga of the Institute of Crop Science, National Agriculture and Food Research Organization for detailed and helpful instruction on NGS data processing using CLC Genomics Workbench software. We appreciate the support of the members of our institute. This study was financially supported by NARO original funding.

Literature Cited

•  Barak,  S.,  E.M.  Tobin,  R.M.  Green,  C.  Andronis and  S.  Sugano (2000) All in good time: The Arabidopsis circadian clock. Trends Plant Sci 5: 517–522.
• Brown, T.A. (1999) Intron splicing. In: Brown, T.A. (ed.) Genomes, BIOS Scientific Publishers, Oxford, pp. 212–219.
•  Churchill,  G.A. and  R.W.  Doerge (1994) Empirical threshold values for quantitative trait mapping. Genetics 138: 963–971.
•  Edwards,  K.,  C.  Johnstone and  C.  Thompson (1991) A simple and rapid method for the preparation of plant genomic DNA for PCR analysis. Nucleic Acids Res 19: 1349.
• Fehr, W.R. and C.E. Caviness (1977) Stages of soybean development. Special report 80, Agriculture and Home Economics Experiment Station, Iowa State University, Ames, pp. 1–12.
•  Fox,  J. (2005) The R Commander: A basic-statistics graphical user interface to R. J Stat Softw 14: 1–42.
•  Fukui,  J. and  M.  Arai (1951) Ecological studies on Japanese soy-bean varieties. I. Classification of soy-bean varieties on the basis of the days from germination to blooming and from blooming to ripening with special reference to their geographical differentiation. Japan J Breed 1: 27–39 (in Japanese with English summary).
•  Furutani,  Y.,  K.  Yasuda,  Y.  Ide and  K.  Sakata (1962) Studies on the characteristics of the early soybean varieties 1. Varietical difference in their growing and ripening process. Bulletin of the Kyushu Agricultural Experiment Station 7: 339–352 (in Japanese with English summary).
• Gotoh, K. (1983) Historical review of soybean cultivation in Japan. Tropical agriculture research series. In: Proceedings of a symposium on tropical agriculture researches. 17, pp. 135–142.
•  Hill,  C.B.,  G.L.  Hartman,  R.  Esgar and  H.A.  Hobbs (2006) Field evaluation of green stem disorder in soybean cultivars. Crop Sci 46: 879–885.
•  Hirata,  T.,  M.  Kaneko,  J.  Abe and  Y.  Shimamoto (1996) Genetic differentiation between summer and autumn maturing cultivars of soybean (Glycine max (L.) Merrill) in Kyushu district of Japan. Euphytica 88: 47–53.
•  Ihaka,  R. and  R.  Gentleman (1996) R: a language for data analysis and graphics. J Comput Graph Stat 5: 299–314.
•  Isshiki,  M.,  K.  Morino,  M.  Nakajima,  R.J.  Okagaki,  S.R.  Wessler,  T.  Izawa and  K.  Shimamoto (1998) A naturally occurring functional allele of the rice waxy locus has a GT to TT mutation at the 5ʹ splice site of the first intron. Plant J 15: 133–138.
•  Komatsu,  K.,  T.Y.  Hwang,  M.  Takahashi,  T.  Sayama,  H.  Funatsuki,  N.  Oki and  M.  Ishimoto (2012) Identification of QTL controlling post-flowering period in soybean. Breed Sci 61: 646–652.
•  Kong,  L.,  S.  Lu,  Y.  Wang,  C.  Fang,  F.  Wang,  H.  Nan,  T.  Su,  S.  Li,  F.  Zhang,  X.  Li et al. (2018) Quantitative trait locus mapping of flowering time and maturity in soybean using next-generation sequencing-based analysis. Front Plant Sci 9: 995.
•  Konno,  S. (1970) Soybean cultivation in Japan. Jpn Agric Res Q 5: 21–27.
•  Kumagai,  M.,  J.  Kim,  R.  Itoh and  T.  Itoh (2013) TASUKE: A web-based visualization program for large-scale resequencing data. Bioinformatics 29: 1806–1808.
•  Kumagai,  M.,  D.  Nishikawa,  Y.  Kawahara,  H.  Wakimoto,  R.  Itoh,  N.  Tabei,  T.  Tanaka and  T.  Itoh (2019) TASUKE+: A web-based platform for exploring GWAS results and large-scale resequencing data. DNA Res 26: 445–452.
•  Li,  M.W.,  W.  Liu,  H.M.  Lam and  J.M.  Gendron (2019) Characterization of two growth period QTLs reveals modification of PRR3 genes during soybean domestication. Plant Cell Physiol 60: 407–420.
•  Li,  M.W. and  H.M.  Lam (2020) The modification of circadian clock components in soybean during domestication and improvement. Front Genet 11: 571188.
• Lincoln, S., M. Daly and E. Lander (1993) Constructing genetics maps with MAPMAKER/EXP 3.0. Whitehead Institute Technical Report, Whitehead Institute, Cambridge.
•  Lu,  S.,  L.  Dong,  C.  Fang,  S.  Liu,  L.  Kong,  Q.  Cheng,  L.  Chen,  T.  Su,  H.  Nan,  D.  Zhang et al. (2020) Stepwise selection on homeologous PRR genes controlling flowering and maturity during soybean domestication. Nat Genet 52: 428–436.
•  Matsuo,  N.,  K.  Fukami and  S.  Tsuchiya (2016) Effects of early planting and cultivars on the yield and agronomic traits of soybeans grown in southwestern Japan. Plant Prod Sci 19: 370–380.
•  Nagata,  T. (1949) Studies on the growth habit of so-called summer vers. autumn soy beans. (1) Classification of varieties in respect to the said habit. Jpn J Crop Sci 18: 131–134 (in Japanese with English summary).
•  Nickolai,  N.A.,  M.E.  Troukhan,  V.V.  Brover,  T.  Tatarinova,  R.B.  Flavell and  K.A.  Feldmann (2006) Features of Arabidopsis genes and genome discovered using full-length cDNAs. Plant Mol Biol 60: 69–85.
•  Sayama,  T.,  T.Y.  Hwang,  K.  Komatsu,  Y.  Takada,  M.  Takahashi,  S.  Kato,  H.  Sasama,  A.  Higashi,  Y.  Nakamoto,  H.  Funatsuki et al. (2011) Development and application of a whole-genome simple sequence repeat panel for high-throughput genotyping in soybean. DNA Res 18: 107–115.
•  Song,  Q.,  G.  Jia,  Y.  Zhu,  D.  Grant,  R.T.  Nelson,  E.-Y.  Hwang,  D.L.  Hyten and  P.B.  Cregan (2010) Abundance of SSR motifs and development of candidate polymorphic SSR markers (BARCSOYSSR_1.0) in soybean. Crop Sci 50: 1950–1960.
•  Su,  T.,  Y.  Wang,  S.  Li,  L.  Wang,  K.  Kou,  L.  Kong,  Q.  Cheng,  L.  Dong,  B.  Liu,  F.  Kong et al. (2021) A flowering time locus dependent on E2 in soybean. Mol Breed 41: 35.
•  Takahashi,  H.G. and  H.  Fujinami (2021) Recent decadal enhancement of Meiyu-Baiu heavy rainfall over East Asia. Sci Rep 11: 13665.
•  Tsubokura,  Y.,  S.  Watanabe,  Z.  Xia,  H.  Kanamori,  H.  Yamagata,  A.  Kaga,  Y.  Katayose,  J.  Abe,  M.  Ishimoto and  K.  Harada (2014) Natural variation in the genes responsible for maturity loci E1, E2, E3 and E4 in soybean. Ann Bot 113: 429–441.
•  Vann,  R.A.,  K.  Drake-Stowe,  G.S.  Buol and  E.J.  Dunphy (2021) Production practices that maximize soybean yield: What we have learned from the North Carolina soybean yield contest. Agron J 113: 3399–3417.
•  Vossen,  R.H.A.M.,  E.  Aten,  A.  Roos and  J.T.  den Dunnen (2009) High-resolution melting analysis (HRMA): More than just sequence variant screening. Hum Mutat 30: 860–866.
• Wang, S., C.J. Basten and Z.-B. Zeng (2012) Windows QTL Cartographer 2.5. Department of Statistics, North Carolina State University, Raleigh, NC https://brcwebportal.cos.ncsu.edu/qtlcart/WQTLCart.htm.
•  Watanabe,  S.,  Z.  Xia,  R.  Hideshima,  Y.  Tsubokura,  S.  Sato,  N.  Yamanaka,  R.  Takahashi,  T.  Anai,  S.  Tabata,  K.  Kitamura et al. (2011) A map-based cloning strategy employing a residual heterozygous line reveals that the GIGANTEA gene is involved in soybean maturity and flowering. Genetics 188: 395–407.
•  Watanabe,  S.,  K.  Harada and  J.  Abe (2012) Genetic and molecular bases of photoperiod responses of flowering in soybean. Breed Sci 61: 531–543.
•  Xu,  M.,  Z.  Xu,  B.  Liu,  F.  Kong,  Y.  Tsubokura,  S.  Watanabe,  Z.  Xia,  K.  Harada,  A.  Kanazawa,  T.  Yamada et al. (2013) Genetic variation in four maturity genes affects photoperiod insensitivity and PHYA-regulated post-flowering responses of soybean. BMC Plant Biol 13: 91.