Genetic and molecular bases of photoperiod responses of flowering in soybean.

Flowering is one of the most important processes involved in crop adaptation and productivity. A number of major genes and quantitative trait loci (QTLs) for flowering have been reported in soybean (Glycine max). These genes and QTLs interact with one another and with the environment to greatly influence not only flowering and maturity but also plant morphology, final yield, and stress tolerance. The information available on the soybean genome sequence and on the molecular bases of flowering in Arabidopsis will undoubtedly facilitate the molecular dissection of flowering in soybean. Here, we review the present status of our understanding of the genetic and molecular mechanisms of flowering in soybean. We also discuss our identification of orthologs of Arabidopsis flowering genes from among the 46,367 genes annotated in the publicly available soybean genome database Phytozome Glyma 1.0. We emphasize the usefulness of a combined approach including QTL analysis, fine mapping, and use of candidate gene information from model plant species in genetic and molecular studies of soybean flowering.

Introduction

Soybean (Glycine max (L.) Merr.) is grown over a wide range of latitudes, from equatorial to at least 50 degrees north and 35 degrees south. However, the cultivation area of each cultivar is restricted to a very narrow range of latitudes. The wide adaptability of soybean has been created by natural variation in the major genes and quantitative trait loci (QTLs) controlling flowering. At present, nine major genes have been reported to control time to flowering and maturity in soybean: E1 and E2 (Bernard 1971), E3 (Buzzell 1971), E4 (Buzzell and Voldeng 1980), E5 (McBlain and Bernard 1987), E6 (Bonato and Vello 1999), E7 (Cober and Voldeng 2001a), E8 (Cober ), and J (Ray ). Linkage analyses have located these genes to molecular linkage groups (MLGs) C1 (Gm04) for E8 (Cober ), C2 (Gm06) for E1 and E7 (Cober and Voldeng 2001a, Molnar ), I (Gm20) for E4 (Abe , Molnar ), L (Gm19) for E3 (Molnar ) and O (Gm10) for E2 (Akkaya , Cregan ). At all of the loci except for E6 and J, dominant alleles delay time to flowering to different extents, interacting with the environment and with genotypes at other loci. The recessive alleles e6 and j were identified in crosses with late-flowering cultivars carrying a long-juvenile trait to condition later flowering (Bonato and Vello 1999, Ray ). In addition to these major genes, many QTLs controlling time to flowering have been reported (Chapman , Cheng , Funatsuki , Githiri , Keim , Khan , Komatsu , Lee , Liu , 2011, Liu and Abe 2010, Mansur , Orf , Poopronpan , Tasma , Wang , Watanabe , Yamanaka , Zhang ). Some of these QTLs most likely correspond to one of the known major genes, such as E1, E2, E3, E4, or E8 (Cheng , Funatsuki , Githiri , Khan , Liu and Abe 2010, Watanabe , Yamanaka ). Some of these QTLs are described in the Soybase database (http://soybase.org/). The major genes and QTLs for flowering often influence agronomic traits other than flowering and maturity, such as plant height and yield (Chapman , Cober and Morrison 2010, Lee , Mansur , Wang , Zhang ), degree of cleistogamy (Khan , Takahashi and Abe 1994), and seed coat pigmentation and cracking caused by chilling stress (Githiri , Takahashi and Abe 1999). Understanding of their molecular bases and their interactions with the environment may therefore be necessary to determine genotypic combinations that will lead to a higher or more stable yield in the cropping season of a particular region. In this review, we summarize the results obtained from previous studies of major genes and QTLs for flowering and those from recent molecular dissections of the maturity genes E2, E3 and E4 and of soybean orthologs of the Arabidopsis FLOWERING LOCUS T gene. We also describe soybean orthologs for Arabidopsis flowering genes deposited in the Williams 82 genome database, and we discuss the resolution of QTL mapping and the positioning of these orthologs on genetic and physical maps. The information on soybean orthologs of Arabidopsis flowering genes should be helpful in the search for candidate genes for targeted major loci and QTLs for flowering in soybean, and in the development of functional DNA markers for use in breeding programs.

Genetic bases for responses of flowering to artificially induced long daylength

Most soybean cultivars have a short-daylength (SD) requirement for floral induction. Flowering is usually suppressed under long-daylength (LD) conditions but induced when the daylength is shorter than a critical length. This sensitivity to photoperiod varies among cultivars: in particular, it is weak or absent in soybean cultivars adapted to high latitudes. Four major genes (E1, E3, E4 and E7) have been well characterized for their responses to LD artificially induced by fluorescent and incandescent lamps with different red to far-red quantum (R : FR) ratios (Buzzell 1971, Buzzell and Voldeng 1980, Cober , 1996b, 2001, Cober and Voldeng 2001a, 2001b, Kilen and Hartwig 1971, Saindon , 1989b). The E3 locus was first identified by extending natural daylength to 20 h with the use of cool-white fluorescent lamps with a high R : FR ratio; the e3e3 recessive homozygote alone can initiate flowering under fluorescence-induced LD (FLD) (Buzzell 1971, Kilen and Hartwig 1971). The E4 locus was identified by extending the natural daylength to 20 h with incandescent lamps with a low R : FR ratio (Buzzell and Voldeng 1980). A homozygous recessive e4e4 genotype is necessary for plants homozygous for the e3 allele to flower under incandescence-induced LD (ILD) without any marked delay in flowering (Buzzell and Voldeng 1980, Saindon ). However, the e4e4 genotype cannot on its own confer insensitivity to FLD. E3 and E4 most likely control flowering under LD conditions with a wide range of R : FR ratios in a non-additive manner.

The other known flowering loci, E1, E7 and E8, are also involved in the control of sensitivity to ILD, particularly in the double-recessive e3e3e4e4 genetic background (Cober , 2001, 2010, Cober and Voldeng 2001a, 2001b). E1 has the largest effect on flowering (Bernard 1971, McBlain , Upadhyay ). However, a near-isogenic line (NIL) of cv. Harosoy homozygous for E1, e3 and e4 (OT93-28) initiated flowering at the same time as the NIL homozygous for e1, e3 and e4 (OT85-9) under FLD, suggesting that E1 does not influence time to flowering under LD with a high R : FR ratio (Cober ). In contrast, E1 exhibits a marked inhibitory effect on flowering under ILD with a R : FR ratio of less than 1.0 (Cober , Thakare ). A homozygous recessive e7e7 genotype further weakens the response of plants homozygous for e1, e3 and e4 to ILD (Cober and Voldeng 2001a, Cober ). However, the e7e7 genotype does not confer a complete loss of photoperiod sensitivity (Cober ). Another recessive allele, e8, is involved in the genetic difference in flowering time observed between breeding lines of genotype e1e1e3e3e4e4e7e7 (Cober ).

In addition to the double-recessive genotype at E3 and E4 (e3e3e4e4), another genetic mechanism is also involved in the control of ILD insensitivity. Abe found that a Japanese early-maturing cultivar, Sakamotowase, had a genetic system for ILD insensitivity different from that of the Japanese early-maturing cultivar Miharudaizu (e3e3e4e4). Mapping analysis indicated that both cultivars differed in their genotypes at the E1 and E4 loci. The genotype at E3 was assumed to be e3e3 in both cultivars because of their insensitivity to FLD; this was confirmed in a later study with the use of functional DNA markers (Liu and Abe 2010). Testcrosses with a Harosoy NIL for e3 (e1e1e3e3E4E4) further revealed that Sakamotowase has a novel gene for ILD insensitivity at, or tightly linked to, the E1 locus (Liu and Abe 2010). Therefore, at least two different systems are involved in the genetic control of ILD insensitivity in soybean.

Interaction between major genes and QTLs for flowering

Major genes and QTLs for flowering often interact with one another to determine time to flowering. For example, the effects of some major genes such as E2 (qFT2) and E3 (qFT3) are weakened or masked in early-flowering genetic backgrounds, such as those conditioned by a recessive allele at the E1 locus (Upadhyay , Watanabe , Yamanaka ). The two QTLs qFT2 and qFT3, detected in a cross between a Japanese cultivar, Misuzudaizu, and a Chinese forage soybean line, Moshido Gong 503, exhibited only a small allelic effect on flowering time under an early-maturing background conditioned by the recessive allele at qFT1 (e1e1), but the allelic effects became marked in a late-maturing background (E1E1) (Yamanaka ). Similarly, using cv. Clark NILs for the E1, E2, and E3 loci, Upadhyay found no effect of allelic substitutions at either E2 or E3 in an e1e1 background, whereas the effect of the E1 allele was marked and almost the same as that of the E2 and E3 alleles combined. Furthermore, the E2 and E3 alleles each interact positively with the E1 allele to enhance the photoperiod-sensitivity (Upadhyay ). A similar genetic interaction was observed in a combination of E4 with later-maturing genetic backgrounds (Saindon ). A marked allelic effect at E4 was observed in a segregating family with the E1E1 genotype (Abe ). Accordingly, the E1 gene appears to control time to flowering epistatically over the other E genes.

E3 and E4 encode phytochrome A proteins

The different responses of E genes to LD conditions with different R : FR ratios have suggested that some of these genes are involved in phytochrome A (phyA)-regulated floral induction in soybean (Cober , 2001). Liu analyzed the sequence variation in a phyA homolog (GmphyA2) between NILs that were photoperiod sensitive and insensitive for E4. They found that a Ty1/copia-like retrotransposon designated SORE-1 was inserted in the first exon of the GmphyA2 gene of photoperiod-insensitive lines carrying the recessive e4 allele (Kanazawa , Liu ). This insertion resulted in a premature stop codon causing a truncated and dysfunctional protein. Genetic mapping analysis confirmed that GmphyA2 cosegregated with E4 on MLG I (Gm20) (Abe , Liu ). Furthermore, the NIL for e4 showed an impaired deetiolation (greening) response under continuous FR-light conditions, as found in phyA null mutants of Arabidopsis (Neff and Chory 1998), rice (Takano , 2005), and pea (Weller , 2001). Taking these findings together, Liu concluded that the E4 gene encodes the GmphyA2 protein and that the recessive e4 allele is a loss-of function allele.

Soybean possesses a homoeologous copy of GmphyA2, namely GmphyA1, in MLG O (Gm10) (Choi , Liu ). The function of GmphyA1 remains undetermined, because no genetic variant is available yet at this locus. However, two findings may indicate that GmphyA1, like E4, is involved in both de-etiolation response and flowering under FR-enriched LD conditions. First, the phyA function of the e4 allele in the de-etiolation response was not completely lost, whereas the phyA null mutants of Arabidopsis, pea, and rice all showed a complete loss of the de-etiolation response under continuous FR light (Neff and Chory 1998, Takano , 2005, Weller , 2001), suggesting that another phyA copy has a redundant function to E4. Second, a Harosoy NIL for double-recessive alleles at E3 and E4 (e3e3e4e4) did not respond to LD with a relatively high R : FR ratio (1.0–5.0) but showed delayed flowering under LD with a low R : FR ratio (<1.0) (Cober , 2001). These redundant functions for de-etiolation and flowering suggest that GmphyA1 itself functions redundantly with E4 in both de-etiolation responses and photoperiod responses under FR-enriched light. However, sequence analyses of both homoeologs in wild (G. soja) and cultivated soybeans revealed that the nucleotide diversity at non-synonymous sites was lower in GmphyA1 than in GmphyA2, although the diversity at synonymous sites and non-coding regions was almost the same in the two loci, suggesting that GmphyA1 has been subject to more intense purifying selection than GmphyA2 (our unpublished data). Some degree of subfunctionalization may thus have occurred between the two phyA homoeologs. Further studies using dysfunctional mutants will be needed to determine the function of GmphyA1 in photoperiodic responses of flowering.

The E3 gene was also identified as a phyA homolog by fine-mapping around a QTL for flowering time (qFT3) (Watanabe ). qFT3 is one of three major QTLs detected in a cross between Misuzudaizu and Moshido Gong 503, and on the basis of map position it has been suggested as a candidate for the maturity gene E3 in MLG L (Gm19) (Watanabe ). Fine-mapping by using a residual heterozygous line (RHL) derived from this cross delineated qFT3 within a 93-kb region of a single TAC clone in which a phyA homolog, GmphyA3, was located. Sequence analyses of GmphyA3 revealed one amino acid (AA) substitution between the parental lines: the early-flowering allele from Moshido Gong 503 possesses an AA substitution from glycine to arginine at an AA site of phyA that is conserved across diverse plant species. Furthermore, an NIL of Harosoy homozygous for e3 contained a truncated protein caused by deletion of a segment covering a genomic region 13 kb long, beginning in the fourth exon and extending downstream. The identity between E3 and GmphyA3 was confirmed by using an artificially induced mutant lacking a 40-bp segment in the first exon; this mutation was detected by TILLING (Targeting Induced Local Lesions In Genomes). The mutant allele encoded a truncated protein and flowered earlier than the parental variety Bay under FLD (Watanabe ). The control of photoperiodic response of flowering to FLD by e3 is therefore attributed to a dysfunctional GmphyA3 allele. Unlike the E4 locus, however, the E3 locus on its own is not involved in the control of de-etiolation response under continuous R or FR light (Liu ).

As in the case of the homoeologs GmphyA1 and GmphyA2, soybean possesses a homoeolog of GmphyA3, namely GmphyA4, in MLG N (Gm03) (Watanabe ). However, the GmphyA4 sequence of cv. Williams 82, a cultivar used for whole-genome sequencing, is most likely dysfunctional because of a deletion in the third exon (Watanabe ). Furthermore, neither a major gene nor a QTL controlling flowering time has so far been reported near the genomic position of GmphyA4.

The phyA protein is an effective FR sensor that is involved, directly and/or via interactions with other photo-receptors, in various developmental processes such as seed germination, de-etiolation, and phototropic responses in etiolated seedlings; it is also involved in early neighbor detection, shade perception, resetting of circadian rhythms, and flowering in light-grown plants (reviewed by Casal ). In addition, Franklin and Franklin and Whitelam (2007) revealed that phyA also functions as an R-light photoreceptor, particularly in R light with high photon irradiance. The different responses of E3 and E4 to LD conditions with different R : FR ratios suggest that the two genes participate in different aspects of the phyA functions controlled by the Arabidopsis phyA gene.

The possible roles of the other photoreceptors, such as phytochrome B (phyB) and cryptochrome (CRY), in the photoperiodic pathway of flowering have not been fully addressed in soybean. Zhang revealed that a soybean CRY1 ortholog, GmCRY1a, rescued the Arabidopsis late-flowering cry2 mutant in ectopic expression analysis with a CaMV35S::GFP-GmCRY1a construct, suggesting that the GmCRY1a protein promotes floral initiation. Furthermore, the GmCRY1a protein exhibited a pattern of photoperiod-dependent rhythmic expression that was correlated with the photoperiodic flowering and latitudinal distribution cline of soybean cultivars. However, the genetic variation affecting the circadian expression pattern remains unknown and is suggested to reside outside GmCRY1a itself (Zhang ). Recently, Cheng found a QTL near the region of MLG C1 (Gm04) in which E8 and GmCRY1a are located (Cober , Matsumura ). It is thus necessary to determine whether the natural variation in GmCRY1a expression is the cause of differences in flowering time.

E2 is a soybean ortholog of the Arabidopsis GIGANTEA gene

A candidate gene for E2 was identified through map-based cloning of qFT2, a QTL for flowering detected in a region of MLG O (Gm10) where E2 was previously mapped (Akkaya , Cregan ), in a cross between Misuzudaizu and Moshido Gong 503 (Watanabe ). By fine-mapping of the progeny of an RHL derived from this cross, Watanabe successfully mapped qFT2 within a 94-Kbp region in a single BAC clone containing a Williams 82 genomic region in which nine annotated genes were predicted. One of the genes, Glyma10g36600, showed a high degree of similarity to the Arabidopsis GIGANTEA (GI) gene. Sequence analyses revealed that the Glyma10g36600 sequences in Misuzudaizu, the donor parent for the early-flowering allele of qFT2, and cv. Harosoy, which carries a recessive e2 allele, contained a premature stop codon caused by a single nucleotide substitution in exon 10 and would therefore produce a truncated and dysfunctional GI-like protein; in contrast, the sequences in Moshido Gong 503, the donor for the late-flowering allele of qFT2, and a Harosoy NIL for E2 did not contain the premature stop codon. These results suggested that qFT2 (E2) encodes a soybean GI ortholog. This hypothesis was further supported by the analysis of a GI mutant detected by TILLING from an EMS-mutagenesis population of cv. Bay. The mutant line, which harbored a premature stop codon in exon 10, flowered earlier than Bay (E2/E2) (Watanabe ).

GI encodes a nuclear-localized membrane protein that functions upstream of CONSTANS (CO) and FLOWERING LOCUS T (FT) in Arabidopsis (Fowler , Huq , Koornneef , Mizoguchi ). GI coupled with a blue-light receptor protein (FLAVIN BINDING, KELCH REPEAT, F-BOX 1 [FKF1]) forms a blue-light-dependent complex that degrades a repressor protein (CYCLING DOF FACTOR 1 [CDF1]) that binds the promoter region of CO; degradation of the repressor protein thereby induces CO expression (Imaizumi , 2005, Nelson , Sawa ). Another function of GI is the regulation of a CO-independent pathway that cooperates with other transcriptional factors such as TARGET OF EGR1 PROTEIN 1 (TOE1) to control FT expression via microRNAs (Jung ). Furthermore, GI also directly controls the expression of FT in Arabidopsis (Sawa and Kay 2011). As in other plant species, it is reasonable to assume that GI-regulated pathways, which may be either CO-dependent or CO-independent, are involved in control of photoperiodic flowering in soybean as well.

Soybean FLOWERING LOCUS T orthologs

One of the striking findings obtained from the extensive molecular dissections of flowering in Arabidopsis and rice is that the product of FT, FT protein, is a florigen that moves through the phloem to the shoot apex (Corbesier , Jaeger and Wigge 2007, Mathieu , Notaguchi , Tamaki ), and its function is highly conserved across unrelated species (Böhlenius , Hayama , Hsu , Izawa , Kojima , Lifschitz , Yan ). Kong found that soybean possesses at least ten FT homologs, which consist of five sets of tandemly linked gene pairs. These pairs are separated into three clades, each corresponding to one of three clades of pea (Pisum sativum) FT genes, PsFTa, PsFTb and PsFTc (Hecht ).

Expression analyses of cv. Harosoy and its NILs grown in SD and LD conditions have indicated that two of the ten FT homologs, GmFT2a (Glyma16g26660) and GmFT5a (Glyma16g04830), showed highly upregulated expression under SD (inductive conditions for flowering), but highly suppressed expression under LD (non-inductive conditions) (Kong , Thakare ). Ectopic expression analyses of GmFT2a and GmFT5a driven by the CaMV35S promoter showed that these genes can promote floral initiation in Arabidopsis ecotype Colombia (Col-0) and complement the function of FT mutants ft-1 and ft-3, providing additional evidence that the GmFT2a and GmFT5a gene products function as florigens in Arabidopsis (Kong , Thakare ). Similarly, Arabidopsis FT ectopically expressed in soybean by using the Apple latent spherical virus vector can promote flowering in both indeterminate and determinate soybean cultivars under non-inductive conditions (Yamagishi and Yoshikawa 2010). These results indicate that FT is a key player in floral initiation in soybean as well.

Expression of GmFT2a and GmFT5a is under the control of phyA homologs E3 and E4 (Kong ). An NIL of cv. Harosoy homozygous for phyA mutants e3 and e4 showed a high level of expression of both GmFT2a and GmFT5a under LD, whereas expression of both genes was highly suppressed in the photoperiod-sensitive cv. Harosoy (E3E3E4E4). In Arabidopsis, the combination of phyA and CRY2 promotes flowering through stabilization of the CO protein (Valverde ). This promotive function of phytochrome A in flowering is also observed in pea (an LD plant) and rice (an SD plant); phyA mutants in both species delayed flowering under inductive light conditions (Takano , Weller ). This is in contrast to the soybean e3 and e4 mutant alleles, which cause no flowering delay under inductive (SD) conditions (Cober , Cober and Voldeng 2001b). Furthermore, night-break experiments in rice demonstrate that transcription of Hd3a (a rice FT ortholog) is determined mainly by light-signal trans-duction dependent on PHYB, not PHYA (Ishikawa , 2009). The relative roles of photoreceptors in photo-periodic flowering may therefore vary among plant species. The genetic variation in photoperiodic expression of the soybean FT homologs is most likely attributable to allelic variation of each of the two phyA homologs.

An SD-to-LD transfer experiment further demonstrated the difference in response to photoperiod between GmFT2a and GmFT5a. Expression of GmFT2a was strictly regulated by photoperiodic changes from SD to LD, whereas the response of GmFT5a to photoperiodic changes was gradual, and its expression was retained at low levels even after the plants were transferred to LD (Kong ). These findings suggest that, in addition to the phyA-mediated photo-period response, a second regulatory mechanism may also be involved in the differences in expression pattern between GmFT2a and GmFT5a. Under the phyA-mediated photo-periodic regulation system, GmFT2a and GmFT5a may redundantly and strongly induce flowering under shorter daylengths, but under longer daylengths GmFT5a alone may promote flowering in a photoperiod-independent manner. These two FT homologs may therefore coordinately control flowering in soybean.

In addition to E3 and E4, E2 influences the mRNA abundance of FT homologs. Watanabe found a clear association between flowering time and the GmFT2a expression in two sets of NILs for the E2 locus; dysfunctional e2 alleles promoted GmFT2a expression and conditioned earlier flowering. However, they could not observe significant differences in the GmFT5a expression between the NILs. These results suggest that the E2 gene (GmGIa) mainly controls flowering time through the regulation of GmFT2a (Watanabe ). The different responses to photoperiodic changes observed between GmFT2a and GmFT5a (Kong ) may thus be caused by involvement of the GI (E2)-regulated pathway in GmFT2a expression, but not in GmFT5a expression. More detailed studies are needed to test this hypothesis. On the other hand, Thakare found no difference in the expression of several orthologs of Arabidopsis flowering-time genes, including FT, CO, GI, and TIMING OF CAB EXPRESSION 1 (TOC1), between genotypes E1E1 and e1e1 (both in an e3e3e4e4 genetic background) in young seedlings 8 days after planting under ILD. However, differences in the expression of GmFT2a and GmFT5a between the E1E1 and e1e1 genotypes became marked 10 days after planting under these conditions: the E1 allele inhibited the expression of both FT homologs compared with the e1 allele (Thakare ).

Soybean orthologs of Arabidopsis flowering genes

Extensive molecular dissections of flowering by using artificially induced mutants in Arabidopsis have revealed that at least 100 genes are involved (Ehrenreich , Hetch , Quecini ). Natural variation in flowering time in major crops such as rice, wheat, and pea has been often reported to result from the variation in orthologs of Arabidopsis flowering genes. Examples include Hd1 (CO) and Hd3a (FT) in rice (Kojima , Yano ), Vrn1 (APETALA1) and Vrn3 (FT) in wheat (Yan , 2006), and LATE FLOWERING (TERMINAL FLOWER 1; TFL1), LATE BLOOMER1 (GI) and GIGAS (FT) in pea (Foucher , Hecht , 2011). Genomic information on soybean orthologs of Arabidopsis flowering genes, such as the number of orthologs and their genomic positions, may therefore provide useful clues for dissecting the molecular bases of flowering in soybean. Several studies have already identified and characterized the soybean orthologs of Arabidopsis photoreceptors, clock-associated genes, and flower-identity genes as flowering genes (Kong , Liu , 2008, 2010, Matsumura , Tasma and Shoemaker 2003, Thakare , Thakare , Tian , Watanabe , 2011, Xue , Zhang ).

We extracted the orthologs of 109 non-overlapping Arabidopsis flowering genes (cited by Ehrenreich , Hetch , Quecini ) from the Williams 82 genome database (Phytozome Glyma 1.0; http://www.phytozome.net/). We detected a total of 333 orthologs of 92 Arabidopsis genes from among a total of 46,367 annotated genes (Table 1, Supplemental Table 1 and Supplemental Fig. 1). This survey indicated that soybean possesses orthologs for most of the Arabidopsis flowering genes. It also highlights a striking but expected feature resulting from the paleopolyploidy of the soybean genome (Cannon and Shoemaker 2012, Schmutz ): soybean clearly has multiple copies of most of the Arabidopsis flowering genes. Furthermore, relatively large syntenic blocks exist in the homoeologous regions of three pairs of chromosomes, Gm04 (MLG C1) and Gm 06 (MLG C2), which contain two sets of blocks of 6 and 8 orthologs each Gm03 (MLG N) and Gm19 (MLG L), which contain blocks of 12 orthologs each and Gm10 (MLG O) and Gm20 (MLG I), which contain blocks of 5 orthologs each (Fig. 1). The functions of these multiple orthologs in the control of soybean flowering should be clarified in further studies. As suggested by functional analyses of the multiple homologs of phyA (Liu , Watanabe ), FT (Kong ) and TFL1 (Liu , Tian ), it is reasonable to speculate that, within each set of duplicated genes, each gene has a function either redundant to, or differentiated from, the others, thus generating more diverse and more complex flowering behaviors in soybean.

Table 1

The list of soybean orthologs for arabidopsis flowering genes

Gene	Abbreviation	Gene function	Soybean homologous genes1)	Characterized genes in soybean
AT1G18090		5′-3′ exonuclease family protein	Glyma07g11320
AT2G25920		3′-5′ exonuclease domain-containing protein/K homology domain-containing protein/KH domain-containing protein (TAIR : AT2G25910.2)	Glyma02g01760, Glyma10g01830
AT5G62640		proline-rich family protein	Glyma05g01510, Glyma17g10370
AT5G62040		PEBP (phosphatidylethanolamine-binding protein) family protein
AT1G68050	ADO3, FKF1	flavin-binding, kelch repeat, f box 1	Glyma05g34530, Glyma08g05130
AT4G18960	AG	K-box region and MADS-box transcription factor family protein	Glyma05g29590, Glyma08g12730, Glyma13g29510, Glyma15g09500
AT3G61120	AGL13	AGAMOUS-like 13
AT4G11880	AGL14	AGAMOUS-like 14	Glyma05g03660, Glyma07g08820, Glyma17g14190
AT3G57230	AGL16	AGAMOUS-like 16	Glyma02g38120, Glyma08g06990, Glyma14g36240
AT2G22630	AGL17	AGAMOUS-like 17	Glyma01g02520, Glyma15g06490
AT4G22950	AGL19, GL19	AGAMOUS-like 19
AT2G45660	AGL20, SOC1, ATSOC1	AGAMOUS-like 20	Glyma03g02200, Glyma07g08830, Glyma09g40230, Glyma18g45780
AT4G37940	AGL21	AGAMOUS-like 21	Glyma01g02530, Glyma02g38090
AT4G24540	AGL24	AGAMOUS-like 24
AT2G45650	AGL6	AGAMOUS-like 6	Glyma03g02210, Glyma05g07350, Glyma07g08890, Glyma09g27450, Glyma16g32540
AT5G60910	AGL8, FUL	AGAMOUS-like 8	Glyma04g31800, Glyma04g31810, Glyma05g07380, Glyma06g22650, Glyma08g27680, Glyma17g08890, Glyma18g50910
AT2G14210	ANR1, AGL44	AGAMOUS-like 44	Glyma09g33450, Glyma13g32810
AT1G69120	AP1, AGL7	K-box region and MADS-box transcription factor family protein	Glyma01g08150, Glyma02g13420, Glyma08g36380, Glyma16g13070
AT4G36920	AP2, FLO2, FL1	Integrase-type DNA-binding superfamily protein	Glyma01g39520, Glyma03g33470, Glyma05g18170, Glyma10g22390, Glyma11g05720, Glyma17g18640, Glyma19g36200
AT3G54340	AP3, ATAP3	K-box region and MADS-box transcription factor family protein	Glyma01g37470, Glyma04g02980, Glyma06g02990, Glyma11g07820, Glyma12g13560, Glyma15g23610, Glyma16g17450, Glyma18g33910
AT5G24470	APRR5, PRR5	pseudo-response regulator 5	Glyma03g42220, Glyma04g40640, Glyma06g14150, Glyma07g05530, Glyma16g02050, Glyma19g44970
AT2G46790	APRR9, PRR9, TL1	pseudo-response regulator 9
AT2G27550	ATC	centroradialis	Glyma10g08340, Glyma12g30940, Glyma13g22030, Glyma13g39360
AT5G24930	ATCOL4, COL4	CONSTANS-like 4	Glyma04g06240, Glyma06g06300, Glyma07g08920, Glyma08g24550, Glyma12g10320, Glyma14g21260, Glyma18g11180, Glyma18g11400
AT5G57660	ATCOL5, COL5	CONSTANS-like 5	Glyma13g01290, Glyma17g07420
AT3G05120	ATGID1A, GID1A	alpha/beta-Hydrolases superfamily protein
AT3G63010	ATGID1B, GID1B	alpha/beta-Hydrolases superfamily protein	Glyma02g17010, Glyma03g30460, Glyma10g02790
AT5G27320	ATGID1C, GID1C	alpha/beta-Hydrolases superfamily protein	Glyma10g29910, Glyma13g25900, Glyma20g37430
AT2G17770	BZIP27	basic region/leucine zipper motif 27	Glyma01g36810
AT2G46830	CCA1	circadian clock associated 1	Glyma07g05410
AT5G62430	CDF1	cycling DOF factor 1
AT5G15840	CO, FG	B-box type zinc finger protein with CCT domain
AT5G15850	COL1, ATCOL1	CONSTANS-like 1
AT3G02380	COL2, ATCOL2	CONSTANS-like 2	Glyma08g28370, Glyma13g07030, Glyma18g51320, Glyma19g05170
AT2G24790	COL3, ATCOL3	CONSTANS-like 3
AT4G08920	CRY1, BLU1, HY4, OOP2, ATCRY1	cryptochrome 1	Glyma04g11010, Glyma06g10830, Glyma13g01810, Glyma14g35020
AT1G04400	CRY2, FHA, AT-PHH1, PHH1, ATCRY2	cryptochrome 2	Glyma02g00830, Glyma10g32390, Glyma18g07770, Glyma20g35220
AT1G18100	E12A11, MFT	PEBP (phosphatidylethanolamine-binding protein) family protein	Glyma05g34030, Glyma08g05650
AT4G22140	EBS	PHD finger family protein/bromo-adjacent homology (BAH) domain-containing protein	Glyma06g34850, Glyma12g20180, Glyma12g35680, Glyma13g34740, Glyma19g23530
AT2G25930	ELF3, PYK20	hydroxyproline-rich glycoprotein family protein	Glyma04g05280, Glyma07g01600, Glyma08g21110, Glyma14g10530, Glyma17g34980
AT2G40080	ELF4	Protein of unknown function (DUF1313)	Glyma11g35270, Glyma14g06480, Glyma18g03130
AT5G04240	ELF6	Zinc finger (C2H2 type) family protein/transcription factor jumonji (jmj) family protein	Glyma10g35350, Glyma20g32160
AT1G79730	ELF7	hydroxyproline-rich glycoprotein family protein	Glyma07g01830, Glyma08g21490
AT2G06210	ELF8, VIP6	binding	Glyma05g24180, Glyma09g07980, Glyma15g19450
AT5G11530	EMF1	embryonic flower 1 (EMF1)	Glyma04g08680, Glyma06g08790
AT5G51230	EMF2, VEF2, CYR1, AtEMF2	VEFS-Box of polycomb protein	Glyma10g23370, Glyma10g23420, Glyma11g03950, Glyma20g16880
AT4G15880	ESD4, ATESD4	Cysteine proteinases superfamily protein	Glyma06g37220, Glyma07g37640, Glyma09g04970, Glyma15g15890, Glyma17g03010
AT4G16280	FCA	RNA binding;abscisic acid binding	Glyma17g03960
AT4G35900	FD, FD-1, atbzip14	Basic-leucine zipper (bZIP) transcription factor family protein
AT2G33835	FES1	Zinc finger C-x8-C-x5-C-x3-H type family protein	Glyma13g31050, Glyma15g08320
AT5G10140	FLC, FLF, AGL25	K-box region and MADS-box transcription factor family protein	Glyma05g28130
AT3G10390	FLD	Flavin containing amine oxidoreductase family protein	Glyma02g18610
AT3G04610	FLK	RNA-binding KH domain-containing protein	Glyma02g15850, Glyma03g31670, Glyma03g40840, Glyma10g03910, Glyma19g34470, Glyma19g43540
AT2G43410	FPA	RNA binding	Glyma11g13490, Glyma12g05490, Glyma13g42060, Glyma15g03330
AT5G24860	FPF1, ATFPF1	flowering promoting factor 1	Glyma04g07900, Glyma09g05080, Glyma09g05090, Glyma14g17260, Glyma15g15730, Glyma17g29720
AT4G00650	FRI, FLA	FRIGIDA-like protein	Glyma04g38060, Glyma06g17010
AT5G16320	FRL1	FRIGIDA like 1	Glyma02g46680, Glyma08g43760, Glyma18g09060
AT1G31814	FRL2	FRIGIDA like 2
AT1G65480	FT	PEBP (phosphatidylethanolamine-binding protein) family protein	Glyma02g07650, Glyma08g28470, Glyma08g47810, Glyma08g47820, Glyma16g04830, Glyma16g04840, Glyma16g26660, Glyma16g26690, Glyma18g53670, Glyma19g28390, Glyma19g28400	FT5a, FT2a
AT2G19520	FVE, ACG1, MSI4, NFC4, NFC04, ATMSI4	Transducin family protein/WD-40 repeat family protein	Glyma09g07120, Glyma13g42660, Glyma15g02770, Glyma15g18450
AT5G13480	FY	Transducin/WD40 repeat-like superfamily protein	Glyma13g26820, Glyma15g37830
AT4G02780	GA1, ABC33, ATCPS1, CPS, CPS1	Terpenoid cyclases/Protein prenyltransferases superfamily protein	Glyma03g31080, Glyma03g31110, Glyma19g33950
AT1G14920	GAI, RGA2	GRAS family transcription factor family protein	Glyma02g08240, Glyma05g03020, Glyma05g27190, Glyma08g10140, Glyma20g34260
AT1G22770	GI, FB	gigantea protein (GI)	Glyma09g07240, Glyma10g36600, Glyma20g30980	E2
AT2G39810	HOS1	ubiquitin-protein ligases	Glyma01g41590, Glyma11g03840
AT5G23150	HUA2	Tudor/PWWP/MBT domain-containing protein	Glyma11g10670, Glyma12g02980
AT4G02560	LD	Homeodomain-like superfamily protein	Glyma03g36970, Glyma19g39620
AT5G61850	LFY, LFY3	floral meristem identity control protein LEAFY (LFY)	Glyma04g37900, Glyma06g17170, Glyma20g19600
AT1G01060	LHY, LHY1	Homeodomain-like superfamily protein	Glyma03g42260, Glyma16g01980, Glyma19g45030
AT2G18915	LKP2, ADO2	LOV KELCH protein 2
AT5G06100	MYB33, ATMYB33	myb domain protein 33	Glyma04g15150, Glyma13g04030, Glyma20g11040
AT3G46640	PCL1	Homeodomain-like superfamily protein	Glyma01g36730, Glyma11g14490, Glyma12g06410
AT1G25540	PFT1	phytochrome and flowering time regulatory protein (PFT1)	Glyma01g21710, Glyma02g10880
AT1G09570	PHYA, FHY2, FRE1, HY8	phytochrome A	Glyma03g38620, Glyma10g28170, Glyma19g41210, Glyma20g22160	E3, E4
AT2G18790	PHYB, HY3, OOP1	phytochrome B	Glyma09g03990, Glyma15g14980
AT4G16250	PHYD	phytochrome D
AT4G18130	PHYE	phytochrome E	Glyma09g11600, Glyma15g23400
AT5G20240	PI	K-box region and MADS-box transcription factor family protein	Glyma04g42420, Glyma06g12380, Glyma13g09660, Glyma14g24590
AT3G12810	PIE1, SRCAP, chr13	SNF2 domain-containing protein/helicase domain-containing protein	Glyma02g29380, Glyma09g17220
AT1G09530	PIF3, POC1, PAP3	phytochrome interacting factor 3	Glyma02g00980, Glyma03g38390, Glyma03g38670, Glyma10g28290, Glyma19g40980, Glyma19g41260, Glyma20g22280
AT3G59060	PIL6, PIF5	phytochrome interacting factor 3-like 6
AT5G60100	PRR3	pseudo-response regulator 3	Glyma11g15560, Glyma11g15580
AT5G02810	PRR7, APRR7	pseudo-response regulator 7	Glyma10g05520, Glyma12g07860, Glyma13g19870
AT2G28550	RAP2.7	related to AP2.7	Glyma02g09600, Glyma11g15650, Glyma12g07800, Glyma13g40470, Glyma15g04930
AT3G48430	REF6	relative of early flowering 6	Glyma04g36620, Glyma04g36630, Glyma06g18290, Glyma06g18300
AT2G01570	RGA1, RGA	GRAS family transcription factor family protein	Glyma02g01530, Glyma04g21340, Glyma06g23940, Glyma10g33380, Glyma11g33720
AT1G66350	RGL1, RGL	RGA-like 1	Glyma08g25800, Glyma17g13680, Glyma18g43580, Glyma19g40440
AT3G03450	RGL2	RGA-like 2	Glyma09g04110, Glyma10g01570, Glyma15g15110, Glyma18g04500
AT5G15800	SEP1, AGL2	K-box region and MADS-box transcription factor family protein	Glyma01g08130, Glyma02g13400, Glyma08g27670, Glyma13g06730, Glyma18g50900, Glyma19g04320
AT3G02310	SEP2, AGL4	K-box region and MADS-box transcription factor family protein
AT1G24260	SEP3, AGL9	K-box region and MADS-box transcription factor family protein	Glyma05g28140, Glyma08g11120, Glyma10g38580, Glyma11g36890, Glyma18g00800, Glyma20g29250
AT2G03710	SEP4, AGL3	K-box region and MADS-box transcription factor family protein	Glyma08g11110
AT3G58780	SHP1	K-box region and MADS-box transcription factor family protein	Glyma08g42300, Glyma18g12590
AT2G42830	SHP2, AGL5	K-box region and MADS-box transcription factor family protein	Glyma02g45730, Glyma14g03100
AT4G24210	SLY1	F-box family protein	Glyma04g04540, Glyma06g04640, Glyma14g09560, Glyma17g08050, Glyma17g27170, Glyma17g35610
AT3G11540	SPY	Tetratricopeptide repeat (TPR)-like superfamily protein	Glyma02g36210, Glyma03g35610, Glyma10g08710, Glyma19g38230
AT4G09960	STK, AGL11	K-box region and MADS-box transcription factor family protein	Glyma04g43640, Glyma06g48270, Glyma08g06980
AT2G22540	SVP, AGL22	K-box region and MADS-box transcription factor family protein	Glyma01g02880, Glyma02g04710, Glyma06g10020, Glyma07g30040, Glyma08g07260, Glyma13g33040, Glyma15g06300, Glyma15g06310
AT5G03840	TFL1, TFL-1	PEBP (phosphatidylethanolamine-binding protein) family protein	Glyma03g35250, Glyma09g26550, Glyma16g32080, Glyma19g37890	Dt1
AT5G17690	TFL2, LHP1	like heterochromatin protein (LHP1)	Glyma03g23260, Glyma04g00340, Glyma06g00400, Glyma16g08860
AT5G61380	TOC1, APRR1, PRR1, AtTOC1	CCT motif-containing response regulator protein	Glyma04g33110, Glyma05g00880, Glyma06g21120, Glyma17g11040
AT4G20370	TSF	PEBP (phosphatidylethanolamine-binding protein) family protein	Glyma18g53680, Glyma18g53690
AT1G30950	UFO	F-box family protein	Glyma05g26460, Glyma08g09380
AT5G57380	VIN3	Fibronectin type III domain-containing protein
AT4G29830	VIP3	Transducin/WD40 repeat-like superfamily protein	Glyma13g16700, Glyma17g05990
AT5G61150	VIP4	leo1-like family protein	Glyma04g32540, Glyma06g21900
AT3G18990	VRN1, REM39	AP2/B3-like transcriptional factor family protein	Glyma01g11670, Glyma04g43620, Glyma07g21160, Glyma08g44640, Glyma08g44650, Glyma09g18790, Glyma09g20280, Glyma11g13210, Glyma11g13220, Glyma12g05250, Glyma16g05110, Glyma19g27950, Glyma20g01130, Glyma20g24220
AT3G24440	VRN5, VIL1	Fibronectin type III domain-containing protein	Glyma05g35280, Glyma07g09800, Glyma08g04440, Glyma09g32010
AT5G57360	ZTL, LKP1, ADO1, FKL2	Galactose oxidase/kelch repeat superfamily protein	Glyma09g06220, Glyma13g00860, Glyma15g17480, Glyma17g06950

Fig. 1

Syntenic blocks containing soybean orthologs of Arabidopsis flowering genes in homoeologous regions of different chromosomes. The orthologs, represented by Arabidopsis gene symbols, are shown in their positions on the soybean physical maps. The orthologs within each set of syntenic blocks are arranged in the same order, but the blocks are sometimes inverted relative to one another. st and en indicate start and end of chromosome, respectively.

Information on the physical position of orthologs to known Arabidopsis genes (Supplemental Table 1 and Supplemental Fig. 1) may help to identify candidate genes for targeted major loci and QTLs. Fig. 2 is an example showing the usefulness of physical map information in identifying candidate genes responsible for three flowering QTLs detected in a cross between Misuzudaizu and Moshido Gong 503 (Watanabe , 2009, 2011, Yamanaka , 2005). The positions of DNA markers tagging the three QTLs, which were originally detected in an RIL population derived from these two parents, delineated their genomic positions in specific regions of Gm06 (MLG C2), Gm10 (MLG O) and Gm19 (MLG L). Fine-mapping and QTL analysis detected two DNA markers separated by a genetic distance of 2 cM, Satt365 and Satt489, for qFT1 (Yamanaka ). The region flanked by the two markers corresponds to a physical distance of approximately 3 Mbp in a pericentromeric region where repetitive sequences are very rich and recombination is severely inhibited (Cannon and Shoemaker 2012, Schmutz ). The genome sequence information predicted only one ortholog of an Arabidopsis flowering gene, REPRESSOR OF GA1-3 (Glyma06g23940) in the region (Fig. 2). Considering the involvement of qFT1 (E1) in photoperiod sensitivity, however, this ortholog is not likely to be the responsible gene, although further studies are needed to confirm its identity. Similarly, QTL mapping placed qFT2 (E2) and qFT3 (E3) in regions on Gm10 (MLG O) and Gm19 (MLG L), respectively; each QTL is flanked by two SSR markers, which are separated by ca. 11 cM and 15 cM, respectively (Watanabe ). These regions contain one and three orthologs for qFT2 and qFT3, respectively, although the physical distances between the markers are still over 1.0 Mbp (Fig. 1). Fine-mapping studies further narrowed these regions into regions of less than 100 Kbp within single genomic DNA clones, and finally a single ortholog could be evaluated and identified as a candidate gene for each qFT (Watanabe , 2011). Hence, fine-mapping subsequent to QTL analysis, together with a candidate gene approach based on the physical positions of Arabidopsis orthologs, may facilitate identification and characterization of molecular major genes and QTLs for flowering in soybean, although novel genes that have no corresponding Arabidopsis flowering gene should not be excluded from consideration as candidate genes.

Fig. 2

Genetic and physical maps of SSR markers tagging three flowering-time QTLs detected in the cross between Misuzudaizu and Moshido Gong 503, and soybean orthologs of Arabidopsis flowering genes predicted in the delineated regions. Distance along the vertical bars indicates the physical distance reported in the Phytozome database. The genetic distance (cM) shown to the left of each SSR marker is cited from an integrated soybean map (Song ) and represents the distance from the end of linkage group to the marker. Dotted lines indicate the possible positions of QTLs inferred from the closest flanking DNA markers (Watanabe , Yamanaka ). Small white squares indicate genomic regions delineated by fine mapping, each containing a single soybean ortholog of an Arabidopsis flowering gene. These regions were subjected to further analyses to identify the genes underlying qFT2 and qFT3 (Watanabe , 2011). Orthologs of flowering genes with underlines indicate genes corresponding to the QTL (shown to the left).

Concluding remarks

It will probably not be so easy to identify the molecular bases of the major genes and QTLs underlying the natural variation in flowering time of soybean, because most of those genes and QTLs exist in multiple copies in the genome, interacting more or less with one another and with the environments in which the genes are evaluated. An understanding of the molecular bases for three major genes, E2, E3 and E4, enables us to develop functional DNA markers to estimate easily and accurately the genotypes at each of these loci. By identifying the genotypes of cultivars with functional DNA markers, we can compare the direct or indirect effects of specific maturity genes on flowering and plant yield by removing any influences from the segregation of other maturity genes or QTLs; thus, we can uncover the residual variation in flowering time left unexplained by these genes, and we can select appropriate parental lines for crossing to produce progeny segregating for one or a few genes. We can also evaluate the relative role of each maturity gene in the adaptation of soybean over a wide range of latitudes. A combined approach—QTL analysis followed by fine mapping and the use of candidate genes predicted from information on model plant species—should be applied to the diverse genetic resources in the soybean germplasm. An increased understanding of the genetic and molecular bases of flowering is expected to contribute to breeding for higher and more stable yields in soybean.