Functional divergence between soybean FLOWERING LOCUS T orthologues FT2a and FT5a in post-flowering stem growth.

Genes in the FLOWERING LOCUS T (FT) family integrate external and internal signals to control various aspects of plant development. In soybean (Glycine max), FT2a and FT5a play a major role in floral induction, but their roles in post-flowering reproductive development remain undetermined. Ectopic overexpression analyses revealed that FT2a and FT5a similarly induced flowering, but FT5a was markedly more effective than FT2a for the post-flowering termination of stem growth. The down-regulation of Dt1, a soybean orthologue of Arabidopsis TERMINAL FLOWER1, in shoot apices in early growing stages of FT5a-overexpressing plants was concomitant with highly up-regulated expression of APETALA1 orthologues. The Dt2 gene, a repressor of Dt1, was up-regulated similarly by the overexpression of FT2a and FT5a, suggesting that it was not involved in the control of stem termination by FT5a. In addition to the previously reported interaction with FDL19, a homologue of the Arabidopsis bZIP protein FD, both FT2a and FT5a interacted with FDL12, but only FT5a interacted with FDL06. Our results suggest that FT2a and FT5a have different functions in the control of post-flowering stem growth. A specific interaction of FT5a with FDL06 may play a key role in determining post-flowering stem growth in soybean.

Introduction

Groundbreaking studies of the orthologous genes FLOWERING LOCUS T (FT) in Arabidopsis and HEADING DATE 3a (Hd3a) in rice (Oryza sativa L.) have shown that they encode a florigen that is transported from leaves through the phloem to shoot apical meristems (SAMs) to induce the development of floral meristems (Corbesier ; Jaeger and Wigge, 2007; Mathieu ; Tamaki ; Notaguchi ). In SAMs, the FT/Hd3a protein interacts with the bZIP transcription factor FD/OsFD1 to activate transcription of floral integrator and meristem identity genes, such as the Arabidopsis genes SUPPRESSOR OF OVEREXPRESSION OF CONSTANS1 (SOC1) and APETALA1 (AP1) and their rice orthologues (Abe ; Wigge ; Corbesier ). There is growing evidence that this role and mechanism for induction of flowering may be widely conserved (Taoka ; Wickland and Hanzawa, 2015). Besides their common basic function as floral inducers, FT and Hd3a have broader roles in the control of other aspects of development. FT controls seed dormancy by integrating long-term temperature memories in fruit tissues of mother plants (Chen ), side shoot formation and elongation (Hiraoka ), stomatal opening (Kinoshita ), and maintenance of reproductive growth (Liu ). Hd3a controls vegetative growth, such as tillering (Tsuji ) and leaf development (Tsuji ).

Accumulating evidence also has revealed that the multifaceted function of FT/Hd3a may in part be attributed to their association with different transcription factors. In rice, Hd3a interacts with the 14-3-3 protein of the Gf14 family in the cytoplasm; the complex then translocates to the nucleus and associates with OsFD1 to form a tri-protein florigen activation complex, which promotes flowering (Taoka ) and tillering through lateral bud outgrowth (Tsuji ). The Hd3a–14-3-3 complex also binds OsFD2, a homologue of OsFD1; this complex is not involved in the control of flowering, but functions in leaf development (Tsuji ). Two additional OsFD1 homologues, Hd3a BINDING REPRESSOR FACTOR1 (HBF1) and HBF2, form repressor complexes with 14-3-3 and either Hd3a or RICE FLOWERING LOCUS T1 (RFT1); these complexes reduce the expression of Early heading date 1, an activator of Hd3a and RFT1 expression, and as a consequence reduce Hd3a and RFT1 expression to delay flowering and repress the MADS genes in SAMs (Brambilla ).

Another group of transcription factors that can interact with FT proteins and may be important for their functional divergence are the TEOSINTE BRANCHED 1, CYCLOIDEA AND PCF (TCP) family proteins (Ho and Weigel, 2014). Amino acid substitutions on the external loop of FT have been shown to convert its function from a floral activator to a repressor, similar to TERMINAL FLOWER1 (TFL1) (Hanzawa ; Ho and Weigel, 2014). These distinct functions may be partly determined by interactions with different TCP family proteins, which show sequence-dependent binding to the external loop of FT, without interfering with its interaction with 14-3-3 and FD proteins (Ho and Weigel, 2014).

Soybean (Glycine max), a facultative short-day (SD) plant, has 12 FT-like (FTL) genes (Kong ; Li ; Wu ). Among them, FT2a and FT5a have been extensively studied as floral inducers, because their expression patterns closely follow photoperiodic changes (Kong ), their overexpression promotes flowering even under non-inductive conditions (Sun ; Nan ; Guo ), and their down-regulation by RNAi (Guo ) and a loss-of-function mutation created by genome editing (Cai ) delay flowering. Nan analysed the interaction of FT2a and FT5a with soybean FD-like (FDL) proteins (FDL08, FDL15, and FDL19) containing the classical SAP motif (RXXS/TAP), a putative binding sequence for FT (Abe ; Taoka ). Of these proteins, only FDL19 physically interacted with FT2a and FT5a; it directly bound cis-elements in the promoter of an AP1 orthologue (AP1a) and induced the expression of AP1a, other AP1 orthologues, and other floral integrator and meristem identity genes such as SOC1 and LEAFY (LFY) orthologues (Nan ). The FT/FD–AP1 module is thus conserved in soybean, as in Arabidopsis and rice (Nan ). Photoperiod controls not only floral induction but also post-flowering reproductive growth such as stem termination and pod development in soybean (Han ; Xu ; Nico ). The stem determinacy in soybean is controlled by at least two genes, Dt1 and Dt2 (Bernard, 1972). Dt1 is an orthologue of Arabidopsis TFL1 and loss-of-function dt1 mutants have a determinate primary inflorescence (Liu ; Tian ). The transcript abundance of Dt1 is under the control of two phytochrome A genes, E3 and E4 (Liu ; Watanabe ); it is up-regulated and maintained at varying degrees under LD conditions depending on the E3 and E4 genotypes (Xu ). Dt2 is an orthologue of the FRUITFULL-like (FUL) MADS-domain gene AGAMOUS-like 79 and acts in a dominant manner to confer a semi-indeterminate habit on a Dt1 background (Ping ). It forms a complex with a SOC1 orthologue (SOC1a) at SAM to repress Dt1 expression by binding to its promoter (Liu ).

Although ectopic expression analyses and different expression profiles have suggested functional diversification among the soybean FT genes (Wang ), our understanding of their functions and diversity remains limited. In this study, we demonstrate that FT2a and FT5a differentially determine the fate of the SAM after floral induction, and identify a distinct role for FT5a in termination of post-flowering shoot growth. We present evidence that this role can occur independently of the shoot determinacy gene Dt2 and may involve distinct interactions with FD-like proteins.

Materials and methods

Plant materials and growth conditions

We used the wild-type (WT) photoperiod-sensitive indeterminate soybean (Glycine max (L.) Merr.) cultivar Williams 82 (e1-as/e2/E3/E4/E9/Dt1/dt2) and plants overexpressing 35S-driven FT2a (35S:FT2a) or FT5a (35S:FT5a) developed by Nan . Plants were grown under SD (12 h light/12 h dark) or long day (LD; 16 h light/8 h dark) conditions in a growth chamber (25 °C) or growth room (23 °C). For floral induction, WT plants were exposed to SD conditions at 25 °C for 12 d after emergence. Another photoperiod-sensitive indeterminate cultivar, Harosoy (e1-as/e2/E3/E4/E9/Dt1/dt2), and its near-isogenic line for the semi-determinate gene Dt2 (L62-364; Harosoy-Dt2) were used to determine the effect of FT5a overexpression driven by Apple latent spherical virus (ALSV) on stem termination and the expression of floral and shoot meristem identity genes. ALSV-infected plants were grown in 14 h day length (25 °C day, 20 °C night) for morphological observation and 16 h day length (23 °C) for expression analyses. Flowering time and number of nodes on the main stem were recorded for each plant.

RNA isolation and expression analysis

Shoot tips were sampled at Zeitgeber time 3 at 8 d after emergence (DAE) from FT-overexpressing plants (35S:FT2a, 35S:FT5a, and ALSV:FT5a-infected Harosoy and Harosoy-Dt2) and WT plants (Williams 82 and ALSV-infected Harosoy and Harosoy-Dt2) and at 30 DAE from 35S:FT2a and WT plants. All samples were immediately frozen in liquid nitrogen and stored at −80 °C. Total RNA was isolated from each shoot tip with Trizol Reagent (Invitrogen/Thermo Fisher Scientific, Waltham, MA, USA). cDNA was synthesized from total RNA (1 µg) using an oligo(dT)20 primer with M-MLV Reverse Transcriptase (Invitrogen). The abundance of shoot and floral meristem identity gene transcripts was determined by quantitative real-time PCR (qRT-PCR). Each qRT-PCR mixture (20 µl) contained 0.05 µl of the cDNA synthesis reaction mixture, 5 µl of 1.2 µM primer premix, and 10 µl of SYBR Premix ExTaq Perfect Real Time (TaKaRa, Kyoto, Japan). A CFX96 Real-Time System (Bio-Rad, Hercules, CA, USA) was used. The PCR cycling conditions were 95 °C for 3 min followed by 40 cycles of 95 °C for 10 s, 58 °C for 30 s, 72 °C for 20 s, and 78 °C for 2 s. To monitor the formation of primer dimers, fluorescence was quantified before and after the incubation at 78 °C. The β-tubulin gene was used as an internal control. A reaction mixture without reverse transcriptase was used to confirm the absence of genomic DNA contamination. For each transcript, amplification of a single DNA fragment was confirmed by melting curve analysis and gel electrophoresis of the PCR products. Averages and standard errors of relative expression levels were calculated for three independent plants. Primers used for expression analyses are listed in Supplementary Table S1 at JXB online.

Overexpression of FT5a by ALSV

ALSV is transmitted to the next generation at a rate of 20–30%, which is particularly suitable for overexpressing or knocking down target gene expression from early seedling stages (Yamagishi and Yoshikawa, 2009). The coding sequence (CDS) of FT5a (516 bp) was amplified by PCR with the primer pairs FT5a-XhoI+ (5′-CCGCTCGAGATGGCACGGGAGAACCCTC-3′; the XhoI site is italicized) and FT5a-BamHI− (5′-CGCGGATCCATATCTCCTTCCACCGCAAC-3′; the BamHI site is italicized) using cDNA from Harosoy as a template. The PCR products were digested with XhoI and BamHI and ligated into the ALSV binary vector pEALSR2mL5mR5 digested with the same enzymes, as described by Li . The constructs pEALSR2mL5mR5FT5a and pEALSR1 were mechanically inoculated into Chenopodium quinoa (Li ), and the resulting virus was designated ALSV:FT5a. ALSV without the FT5a sequence was used as a control. Seeds of Harosoy and Harosoy-Dt2 were inoculated with ALSV:FT5a or ALSV using a Helios Gene Gun System (Bio-Rad) as described by Yamagishi and Yoshikawa (2009) and germinated in the dark (25 °C). The seedlings were transplanted into PVC pots filled with nursery soil and grown in a growth room (25 °C/20 °C, 14 h light/10 h dark). The flowering time and number of nodes on the main stem were recorded for each plant, and the seeds were harvested.

Yeast two-hybrid assays

Yeast two-hybrid (Y2H) assays were performed as described by Nan . The yeast cloning vectors pGBKT7 and pGADT7, the control vectors pGADT7-T and pGBKT7-53, and the yeast strain Y2H were obtained from Clontech (TaKaRa). The pGBKT7 vectors containing the coding sequences (CDSs) of FT2a or FT5a and the pGADT7 vectors containing the CDSs of FDL08, FDL15, FDL19, or Arabidopsis FD were developed by Nan . The CDSs of Dt1 and four FDLs (FDL06, FDL12, FDL13, and FDL20) were amplified from Harosoy cDNA. The CDS of Arabidopsis FT was amplified from Col-0. The CDSs of Dt1 and FT were inserted into the pGBKT7 vector to generate GAL4 DNA-binding domain fusions to be used as bait. The CDSs of the four FDL genes were cloned into pGADT7 to generate GAL4 DNA activation domain fusions to be used as prey. The bait and prey plasmids were co-transformed into Y2H using the lithium acetate method (Gietz and Woods, 2002) and selected on synthetic dropout (SD) medium lacking Leu and Trp (SD/−L−T). After 4 d of incubation at 30 °C, the cells were re-plated on selection plates with SD medium lacking Ade, His, Leu, and Trp, but containing the X-α-Gal substrate (SD/−A−H−L−T+X-α-Gal) for the interaction test. The primers and restriction sites used to generate the bait and prey constructs are presented in Supplementary Table S2.

Bimolecular fluorescence complementation

The CDSs of Dt1 and FT were inserted into the pUC_SPYNE vector, which encodes the N-terminal half of yellow fluorescent protein (YFP). The cDNAs of the four FDL genes (FDL06, FDL12, FDL13, and FDL20) were inserted into the pUC_SPYCE vector, which encodes the C-terminal half of YFP. The pUC_SPYNE vectors containing the CDSs of FT2a or FT5a and the pUC_SPYCE vectors containing the CDSs of FDL08, FDL15, FDL19, or FD were developed by Nan . The recombined pUC_SPYNE and pUC_SPYCE plasmids were co-transfected into Arabidopsis protoplasts using polyethylene glycol as described by Nan . YFP fluorescence was detected 24 h after transfection under a confocal laser-scanning microscope (LSM 510 Meta; Carl Zeiss, Jena, Germany). The primer pairs are listed in Supplementary Table S2.

Results

Overexpression of FT2a and FT5a affects stem growth differently

We first examined flowering time and expression levels of FT2a or FT5a in the indeterminate cultivar Williams 82 (WT) and the 35S:FT2a and 35S:FT5a transgenic plants. As reported by Nan , the 35S:FT2a and 35S:FT5a plants flowered earlier than WT plants under LD conditions (Supplementary Fig. S1A). Transcript levels of FT2a were markedly higher in all 35S:FT2a plants tested than in WT plants (Supplementary Fig. S1B), whereas those of FT5a were more variable; two of the six 35S:FT5a plants tested (no. 4 and no. 6) had relatively high FT5a expression (Supplementary Fig. S1C). We used the progeny of these two 35S:FT5a plants and a randomly selected 35S:FT2a plant in subsequent analyses. As expected, the progeny of 35S:FT2a and 35S:FT5a plants flowered earlier than WT plants under LD conditions (Fig. 1A, B), consistent with the previous report of Nan .

Fig. 1.

Flowering time and stem growth of WT and the progeny of 35S:FT2a and 35S:FT5a transgenic plants under LD conditions (16 h light/8 h dark). (A) Plant morphology and terminal-flower formation at shoot apices, when flowering was initiated in FT-overexpressing plants. (B) Numbers of days from emergence to flowering. (C) Numbers of nodes in a main stem. The number of nodes in WT was recorded when 35S:FT2a plants produced terminal flowers. Error bar, standard deviation (n=6). Different letters indicate statistical significance at 5% level by Tukey–Kramer test.

We also observed a strong effect of 35S:FT5a on shoot determinacy under LD conditions, with plants terminating stem growth at the third or fourth node (average, 3.2; Fig. 1A, C), compared with WT, which produced terminal flowers at node 17 or 18 (Supplementary Fig. S2). In contrast, despite a similar flowering time, 35S:FT2a plants had a much weaker effect on determinacy; terminal flowers were produced at the 12th to 14th node (average, 13.4). When grown in SD conditions, WT and 35S:FT2a plants flowered on average at 26.2 and 23.3 DAE, respectively, and terminated stem growth on average at 8.0 and 5.3 nodes, respectively (Supplementary Fig. S3). 35S:FT5a plants flowered at almost the same time (22.0 DAE) as 35S:FT2a plants, but terminated stem growth at the third node in SD conditions (Supplementary Fig. S4). Therefore, FT5a had a noticeable effect on stem growths irrespective of day length conditions, whereas FT2a itself had an ability to terminate stem growth, but its effect was small, compared with that of FT5a.

Overexpression of FT5a affects stem growth independent of Dt2 genotype

The indeterminate cultivar Williams 82 possesses a genotype of Dt1/dt2 at the two stem growth habit loci. Since the dt2 allele is considered to be a hypomorphic allele whose expression is not up-regulated to a level sufficient to suppress Dt1 expression (Ping ), we compared the effects of FT5a overexpression on post-flowering stem growth between Harosoy (Dt1/dt2) and Harosoy-Dt2 (Dt1/Dt2) plants overexpressing FT5a by the ALSV vector. The ALSV:FT5a-infected Harosoy and Harosoy-Dt2 plants terminated stem growth earlier than those infected with the empty ALSV vector; they produced three or more pods at the stem tip, whereas those infected with the empty ALSV produced no pods at the stem tip (Fig. 2A). To assess the effects of FT5a overexpression more precisely, we used the progeny of the infected plants, which were systemically infected with ALSV:FT5a or ALSV from the seedling stage. The progeny of ALSV:FT5a-infected plants also flowered earlier and terminated stem growth at a lower node than the progeny of ALSV-infected controls in both dt2 and Dt2 genetic backgrounds (Fig. 2B, C). The plants carrying ALSV:FT5a terminated stem growth at five to seven nodes (average, 7.0 in Harosoy, 5.5 in Harosoy-Dt2), significantly fewer than in the ALSV-infected progeny of both Harosoy (12.5) and Harosoy-Dt2 (10.6). As expected, Harosoy-Dt2 flowered and terminated stem growth slightly but significantly earlier than Harosoy in both ALSV:FT5a-infected and ALSV-infected progeny, but the effect of FT5a overexpression was almost the same in the dt2 and Dt2 backgrounds. Therefore, it appears that FT5a can act independently of the Dt2 genotype to terminate stem growth.

Fig. 2.

Stem termination in ALSV- and ALSV:FT5a-infected Harosoy (HA) and its NIL for Dt2 (H-Dt2) plants. (A) Plant stature and pod formation at shoot apex in an ALSV:FT5a- and ALSV-infected Harosoy plants in a day length of 14 h. Photos were taken at 58 d after transplanting infected seeds. The ALSV:FT5a plant terminated stem growth at a lower node position (yellow arrow) and produced seven pods in a terminal inflorescence (top panel). (B, C) Average numbers of days from emergence to the opening of the first flower (B) and average number of nodes in a main stem (C) in the progeny of infected plants. Open bar, plants infected with the empty ALSV; closed bar, plants infected with ALSV:FT5a. Different letters indicate statistical significance at 5% level by Tukey–Kramer test. Error bar, standard deviation.

Expression of shoot and floral meristem identity genes in FT-overexpressing plants

Since the overexpression of FT2a and FT5a terminated stem growth to different extents, we next tested whether the difference in phenotype might be reflected in distinct effects on downstream genes. We examined the expression of shoot and floral meristem identity genes in shoot apices of 8-day-old (primary leaves expanded) WT, 35S:FT2a, and 35S:FT5a plants. The genes analysed were Dt1 and orthologues of FUL (FUL1a, FUL1b, FUL2a, FUL2b, FUL3a and FUL3b (Dt2)), AP1 (AP1a, AP1b and AP1c), LFY (LFY2), and SOC1 (SOC1a and SOC1b). The expression of Dt1 was strongly repressed in 35S:FT5a plants, but essentially unaffected by FT2a overexpression (Fig. 3). In contrast, the Dt2, FUL orthologues, and AP1c were all significantly up-regulated in both 35S:FT2a and 35S:FT5a plants, but AP1a, AP1b, and LFY were significantly up-regulated only in 35S:FT5a plants. In particular, up-regulation of the three AP1 orthologues was much stronger in 35S:FT5a than in 35S:FT2a plants. The expression levels of SOC1a and SOC1b were not affected in either 35S:FT2a or 35S:FT5a plants. Interestingly, both 35S:FT2a and 35S:FT5a plants significantly up-regulated the expression level of Dt2, but the expression level of Dt1 was only down-regulated in 35S:FT5a. Those results suggest that the inhibition of Dt1 expression in 35S:FT5a plants was not due to the up-regulation of Dt2, but rather likely to highly-up-regulated expression of AP1 orthologues.

Fig. 3.

Expression of floral and shoot meristem identity genes in shoot apices of WT, 35S:FT2a and 35S:FT5a transgenic plants at 8 d after emergence. Values are given relative to β-tubulin transcript levels. Error bar, standard error of the mean of three biological replicates (independent plants). *P<0.05 (Student’s t test).

We also analysed the expression profiles of these genes in shoot apices of 30-day-old WT and 35S:FT2a plants (five to six trifoliate leaves expanded), because the 35S:FT2a plants retained vegetative growth at shoot apices in the late growing stage (Fig. 1; Supplementary Fig. S2). The expression levels of floral and shoot identity genes except for Dt1 mostly increased in 30-day-old plants (Fig. 4), compared with those of 8-day-old plants (Fig. 2). The overexpression of FT2a in 30-day-old plants significantly up-regulated the FUL1a, FUL1b, AP1b, and AP1c expression, whereas there was no significant difference in the expression levels for Dt2, three FUL and two SOC1 orthologues, and LFY2. In particular, the expression of three AP1 orthologues was up-regulated to levels similar to those of the 8-day-old 35S:FT5a plants. The expression level of Dt1 at 30 DAE was slightly but not significantly down-regulated in 35S:FT2a plants compared with WT plants.

Fig. 4.

Expression profiles of floral and shoot meristem identity genes in shoot apices of WT and 35S:FT2a transgenic plants in a later growing stage (30 DAE). Values are given relative to β-tubulin transcription levels. Error bar, standard error of the mean of three biological replicates (independent plants). *P<0.05 (Student’s t test).

The effects of FT5a overexpression on the expression profiles of shoot and floral meristem identity genes were further evaluated in ALSV:FT5a-infected progeny of Harosoy and Harosoy-Dt2. As observed in 35S:FT5a plants (Fig. 3), the overexpression of FT5a significantly down-regulated Dt1, but up-regulated Dt2 and three AP1 orthologues to an equivalent extent in the shoot apices of 8-day-old Harosoy and Harosoy-Dt2 plants (Fig. 5). Dt2 and dt2 alleles thus responded similarly to FT5a overexpression.

Fig. 5.

Expression profiles of floral and shoot meristem identity genes in shoot apices of ALSV- and ALSV:FT5a-infected Harosoy (HA) and Harosoy-Dt2 (H-Dt2) plants at 8 DAE under the LD condition. Values are given relative to β-tubulin transcription levels. Relative accumulation of ALSV carrying FT5a is presented for FT5a. Error bar, standard error of the mean of three biological replicates (independent plants). *P<0.05 (Student’s t test).

Differences in the interactions of FT2a, FT5a, and TFL1b with FDLs

To determine whether the differential activity of FT2a and FT5a for stem growth termination and expression of shoot and floral meristem identity genes might reflect differential patterns of interaction with other partners, we tested the interaction of FT2a and FT5a with four FDL proteins: FDL06, FDL12, FDL13, and FDL20 (Nan ), using yeast two-hybrid (Y2H) and bimolecular fluorescence complementation (BiFC) assays. FT2a and FT5a both interacted with FDL12, but only FT5a interacted with FDL06; they did not interact with FDL13 and FDL20 (Fig. 6A). To validate these results, we conducted in vivo BiFC analyses. The results confirmed that FDL12 interacted with FT2a and FT5a, but FDL06 interacted only with FT5a in the nuclei of Arabidopsis protoplasts (Fig. 6B). Because Arabidopsis TFL1 interacts with FD (Abe , Wigge , Hanano and Goto, 2011, Huang ; Ho and Weigel, 2014, Ryu ), we also tested the interaction of Dt1 (TFL1b) with FD and seven FDLs, including the three analysed by Nan . The Y2H and BiFC assays revealed that Dt1 interacted with FD and FDL12 but not with any other FDL proteins in vitro or in vivo (Fig. 7). The Y2H assay further revealed that Arabidopsis FT interacted with FDL06, FDL12, and FDL19 (Supplementary Fig. S5).

Fig. 6.

Interaction of four FDLs with FT2a and FT5a. (A) Yeast two-hybrid assays. After co-transformation of the baits and preys, an equal amount of yeast clones were plated on SD−Leu−Trp (SD/−L−T) and SD−Ade−His−Leu−Trp+X-α-Gal (SD/−A−H−L−T+X-α-Gal) selective plates, and the plates were incubated at 30 °C until the emergence of the yeast clones. (B) Bimolecular fluorescence complementation. Arabidopsis protoplasts co-transformed with constructs of FTs or FDs fused to the N-terminal (YN) and C-terminal (YC) halves of YFP, respectively (as indicated), were imaged using a confocal microscope after incubation at room temperature (20–25 °C) over 18 h. Images are shown as YFP, merged YFP, and bright field. Scale bar, 20 µm.

Fig. 7.

Interaction of FD and 7 FDLs with Dt1 (TFL1b). (A) Yeast two-hybrid assays. After co-transformation of the baits and preys, an equal amount of yeast clones were plated on SD−Leu−Trp (SD/−L−T) and SD−Ade−His−Leu−Trp+X-α-Gal (SD/−A−H−L−T+X-α-Gal) selective plates, and the plates were incubated at 30 °C until the emergence of the yeast clones. (B) Bimolecular fluorescence complementation assays. Arabidopsis protoplasts co-transformed with constructs of Dt1 or FD/FDLs fused to the N-terminal (YN) and C-terminal (YC) halves of YFP, respectively (as indicated), were imaged using a confocal microscope after incubation at room temperature (20–25 °C) over 18 h. Images are shown as YFP, merged YFP, and bright field. Scale bar, 20 µm.

To determine whether the effects of FT2a and FT5a overexpression might in part reflect their differential transcriptional regulation of FDL genes, we evaluated the expression profiles of FDL06, FDL12, and FDL19 in 8-day-old plants in Williams 82, 35S:FT2a and 35S:FT5a plants, and ALSV:FT5a-infected progeny of Harosoy and Harosoy-Dt2. The expression of the three FDL genes was not affected by the overexpression of FT2a or FT5a (Fig. 8), suggesting that their distinct roles in stem termination are likely to result from differences in their inherent activities and interactions, rather than differential effects on expression of their FDL interaction partners.

Fig. 8.

Expression profiles of FDL genes at shoot apices in wild type, and FT2a and FT5a overexpressing plants. Expression analyses were carried out (A) at 30 DAE for 35S:FT2a and 8 DAE for 35S:FT5a transgenic plants, and (B) for ALSV:FT5a-infected Harosoy (HA) and Harosoy-Dt2 (H-Dt2) plants. Values are given relative to β-tubulin transcript levels. Error bar, standard error of the mean of three biological replicates (independent plants). *P<0.05 (Student’s t test).

Discussion

FT2a and FT5a differentially control post-flowering stem growth

FT and FTL genes have multifaceted functions in various aspects of plant development and growth. Soybean FT orthologues FT2a and FT5a are major floral inducers (Kong ; Nan ), but their functional roles in traits other than flowering remain unexplored. Our ectopic overexpression analyses revealed that both FT2a and FT5a promote post-flowering stem termination, but their effects differ in magnitude (Fig. 1). The overexpression of FT5a strongly terminated stem growth concomitant with floral induction. This phenomenon was similar to that reported by Yamagishi and Yoshikawa (2011), who found that the overexpression of Arabidopsis FT by the ALSV vector in the indeterminate soybean cultivar Dewamusume leads to production of terminal flowers at the third-node stage. In sharp contrast, the 35S:FT2a plants retained vegetative growth of the SAM while flowering at lateral axils until terminal flowers developed at the 12th to 14th node, although stem growth terminated slightly earlier than in the WT plants under both LD and SD conditions (Fig. 1; Supplementary Figs S2, S3). These findings suggest a functional divergence between FT2a and FT5a in terms of their effect on post-flowering stem growth: although both genes terminate stem growth, the effect of FT5a is much stronger than that of FT2a.

Transition from vegetative to reproductive growth in the SAM depends on the abundance of the Dt1 transcript (Liu ; Xu ). The expression of Dt1 in indeterminate plants is inhibited under SD conditions, but is up-regulated to a varying extent under LD conditions depending on the E3 and E4 genotypes and in accordance with the timing of stem growth termination (Xu ). In this study, Dt1 expression was strongly inhibited by the overexpression of FT5a but not FT2a (Figs 3, 4). In Arabidopsis, TFL1 expression is repressed by binding of AP1 to the CArG box and of LFY to the LFY-binding site, both in the 3′ intergenic region (Serrano-Mislata ). The overexpression of FT2a and FT5a up-regulated the expression of AP1 orthologues (AP1a–c) and LFY2 (Figs 3, 4), as previously reported (Nan ). However, the expression levels of AP1 orthologues were much higher in 35S:FT5a than in 35S:FT2a plants at the primary leaf stage (Fig. 3). Later, expression of AP1s in 35S:FT2a plants was up-regulated to levels similar to those in 35S:FT5a plants in the primary leaf stage (Fig. 4), and stem growth was terminated earlier in the 35S:FT2a plants than in WT plants (Supplementary Fig. S2). FT2a and FT5a had similar effects on floral induction at axillary meristems, but only FT5a terminated vegetative growths at SAM shortly after floral induction. It may be plausible to assume different thresholds of expression levels of AP1s in the floral induction at axillary meristems and the termination of stem growth at SAM (Fig. 9). The different regulation of expression of AP1s may therefore be responsible for the different effects of FT2a and FT5a on stem growth, although a further study is needed to confirm the function of AP1 orthologues in soybean.

Fig. 9.

Different functions of FT2a and FT5a through interactions with FDLs in the control of flowering and stem termination in soybean. (A) FT2a binds to FDL19, which induces expression of AP1s by directly binding the promoter (Nan ). The expression level of AP1s induced by the FT2a–FDL19 complex may be sufficient to promote floral induction at axillary meristems. (B) FT5a binds to both FDL19 and FDL06. These complexes induce expression of AP1s to higher levels than the FT2a–FDL19 complex. The additional induction of expression of AP1s by these complexes may be essential to terminate stem growth by suppressing the Dt1 expression at shoot apical meristem. FL, flowering; ST, stem termination.

Dt2 is another regulator of Dt1 expression; it forms a complex with SOC1a at the SAM to repress Dt1 expression by binding to its promoter (Liu ). In this study, Dt2 expression was up-regulated in 35S:FT2a and 35S:FT5a plants to a similar extent (Fig. 3). Williams 82 has a dt2 allele, whose expression is not up-regulated to a level sufficient to suppress Dt1 expression (Ping ). We thus compared the effects of FT5a overexpression on post-flowering stem growth and expression level of Dt2 at the primary leaf stage between ALSV:FT5a-infected Harosoy and Harosoy-Dt2 plants (Figs 2, 5). As observed in 35S:FT5a plants, the overexpression of FT5a by ALSV up-regulated Dt2 expression and accelerated termination of stem growth in both Harosoy and Harosoy-Dt2. Harosoy-Dt2 flowered and terminated stem growth slightly earlier than Harosoy, but no marked difference was detected in Dt2 expression levels between them in both ALSV:FT5a and ALSV infected plants (Figs 2, 5). Taken together, our results suggest that Dt2 is not involved in post-flowering stem growth controlled by FT5a. FT5a may rather determine the cell fate in the SAM mainly through up-regulation of AP1 orthologues. Furthermore, the function of Dt2 on stem determinacy might be partly under the age-dependent control.

Different interactions of FT2a, FT5a, and Dt1 with soybean FDLs

FT/Hd3a exported from leaves forms a tri-protein complex with 14-3-3 and FD/OsFD in the SAM; in this complex, different FD–FDL combinations mediate diverse and multifaceted functions of FT/Hd3a in various aspects of plant development and morphogenesis (Taoka ; Tsuji , 2015; Ho and Weigel, 2014; Brambilla ). In our study and Nan , FT2a and FT5a both interacted with FDL12 and FDL19, whereas only FT5a interacted with FDL06 (Fig. 6). Because FDL06 exclusively interacts with FT5a, it is tempting to assume that the difference in the interactions with FDL06 explains different effects of FT2a and FT5a on post-flowering stem growth (Fig. 9). The findings obtained from the expression analyses suggest that the expression level of AP1s induced by the FT2a–FDL19 complex was sufficient to promote floral induction at axillary meristems (Fig. 3). The FT5a–FDL19 and FT5a–FDL06 complexes may induce the expression of AP1s to higher levels than the FT2a–FDL19 complex. The additional induction of expression of AP1s by these complexes may be essential to terminate stem growth by suppressing Dt1 expression at the SAM. Overexpression of FT2a and FT5a had no notable effect on the expression of genes for the three FDL proteins interacting with FT2a and FT5a; in particular, FDL06 and FDL19 were expressed at relatively high and similar levels in early-growing stages in WT and FT2a/FT5a-overexpressing plants under LD conditions (Fig. 8). In Arabidopsis, FD is already expressed in the shoot apex before floral induction; once delivered to the SAM, FT forms a complex with FD, which in turn activates floral identity genes (Wigge ). In our study, overexpression of FT5a strongly terminated stem growth of indeterminate plants, concomitant with floral induction, despite similar abundance of the FDL06 transcript in FT5a-overexpressing and WT plants. Therefore, even in earlier growing stages, FDL06 may have already accumulated in the SAM to a level sufficient for interaction with FT5a, and an increase in the FT5a–FDL06 complex caused by the highly up-regulated FT5a expression may repress Dt1 expression via the up-regulation of AP1 orthologues to terminate stem growth. Alternatively, transcription factors other than FDLs, such as TCP family proteins, may be involved in stem termination mediated by FT5a, as reported in Arabidopsis (Ho and Weigel, 2014). Further study is needed to determine the function of FDL06 and its downstream genes such as AP1 orthologues in the control of post-flowering stem growth.

By analysing the soybean genome (G. max Wm82.a1 v. 1), Nan identified the top 18 FDL gene sequences producing high-scoring segment pairs with Arabidopsis FD, and classified them into the FD, wheat TaFDL2, and AREB/ABI5 clades. The soybean FDL proteins whose interactions with FT2a and FT5a were analysed here and by Nan belong to the wheat TaFDL2 (FDL08, FDL13, FDL15, FDL19, and FDL20) or AREB/ABI5 clade (FDL06 and FDL12). Because members of the AREB/ABFs clade are generally stress-related transcription factors (Kang ; Kim ; Fujita ; Furihata ; Yoshida ; Yoshida ), FDL06 may be a stress-related transcription factor in soybean. Functional study is needed to determine the roles of FDL06 not only in the control of stem growth habit, but also in tolerance to environmental stresses. Soybean genes encoding the FD-clade FDLs are FDL02, FDL04, and FDL0602 (Tsuji ; Nan ; Sussmilch ). However, we did not assay the interactions of FT2a and FT5a with these FDLs, because their expression was previously undetectable in leaves and shoot apices (Nan ).

Interestingly, FDL06 and FDL19 did not interact with Dt1 (TFL1b) (Fig. 7). In Arabidopsis, TFL1 competes with FT for interaction with FD (Abe , Wigge , Hanano and Goto, 2011, Huang ; Ho and Weigel, 2014, Ryu ). A member of the phosphatidylethanolamine‐binding protein family, BROTHER OF FT AND TFL1, delays flowering transition under high salinity by competing with FT for the same binding site on FD (Ryu ). Similarly, rice TFL1-like proteins RICE CENTRORADIALIS (RCN) 1 to 4 compete with Hd3a for 14-3-3 binding, and the tri-protein complex of RCN, 14-3-3, and OsFD1 represses reproductive growth antagonistically to the Hd3a-containing florigen activation complex (Kaneko-Suzuki ). Our findings may indicate that FT2a and FT5a control floral induction and stem termination without interference by Dt1 with their interactions with FDL proteins. In particular, the interaction of FDL06 with FT5a but not FT2a may play a key role in determining post-flowering stem growth in soybean. Fine-tuning of FT5a expression may therefore be critical for regulation of node development in the main stem and branches, and thus be directly related to seed yield. In contrast, FDL12 may be a common interactor for FT2a, FT5a, and Dt1. According to an atlas of soybean RNA sequences (Severin ), FDL12 is expressed mainly during late seed maturation, unlike FDL06 and FDL19, which are expressed at relatively low levels across various tissues and stages of seed maturation. FDL12 may thus play a distinct role in seed maturation, and not be involved in floral induction or stem termination at apical or lateral meristems.

Functional divergence of FT homologues among legume species

Garden pea has five FTL genes, which are classified into three groups, FTa (PsFTa1 and PsFTa2), FTb (PsFTb1 and PsFTb2), and FTc (PsFTc); all these genes except PsFTa2 promote flowering of the Arabidopsis ft mutant when ectopically expressed under the control of the CaMV 35S promoter (Hecht ). Similarly, Medicago truncatula has five FTL genes corresponding to each of the three groups (MtFTa1, MtFTa2, MtFTb1, MtFTb2, and MtFTc); MtFTa1, MtFTb1, and MtFTc promote flowering of the Arabidopsis ft mutant when ectopically expressed under the control of the CaMV 35S promoter (Laurie ). Intriguingly, both PsFTc and MtFTc induce terminal flowers in the Arabidopsis ft mutant more rapidly than the other homologues (Hecht ; Laurie ), effects similar to that of 35S:AtFT transformation (Teper-Bamnolker and Samach, 2005). FT2a and FT5a are classified into the FTa and FTc groups, respectively (Laurie ). However, ectopic expression analyses with the 35S promoter, and FT or TFL1 native promoters in Col-0, and ft and tfl1 mutants show that both FT2a and FT5a promote flowering and produce terminal flowers in Arabidopsis (Kong ; Thakare ; Fan ; Wang ). Therefore, the function of FT5a may be similar to those of the other FTc members (PsFTc in pea and MtFTc in M. truncatula), whereas the function of FT2a in stem growth may differ from that of its homologues (PsFTa1 and MtFTa1). However, the expression profiles of the FTc group in various tissues differ between soybean and the other two species: both PsFTc and MtFTc are expressed in shoot apices but not in leaves (Hecht ; Laurie ), whereas FT5a is expressed in both (Kong ). Therefore, after the divergence of soybean (a SD legume) from garden pea and M. truncatula (LD legumes), soybean FT2a and FT5a both might have gained their own functional or regulatory systems distinct from those of the other FT homologues. Arabidopsis FT can form a complex with each of the three FDLs (FDL06, FDL12, and FDL19) (Supplementary Fig. S5). The marked effects of FT overexpression by ALSV vector on floral initiation and stem termination in soybean (Yamagishi and Yoshikawa, 2011) may be conferred by its interactions with FDL06 and FDL19. Thus, FT5a retains a function similar to those of Arabidopsis FT, whereas FT2a is subfunctionalized, in particular, for the interaction with FDL proteins, and may function mainly as a floral inducer in soybean.

Supplementary data

Supplementary data are available at JXB online.

Fig. S1. Flowering time and expression of FT2a and FT5a in WT, 35S:FT2a, and 35S:FT5a transgenic plants in LD condition (16 h light/8 h dark).

Fig. S2. Stem termination at a later growing stage in WT and 35S:FT2a transgenic plants under LD condition.

Fig. S3. Flowering time and stem growth of WT and 35S:FT2a transgenic plants under SD condition (12 h light/12 h dark).

Fig. S4. Flowering time and stem growth of WT and 35S:FT5a transgenic plants under SD condition (12 h light/12 h dark).

Fig. S5 Yeast two-hybrid assays of Arabidopsis FT with three FDLs.

Table S1. Primers for quantitative RT-PCR.

Table S2. Primers for vector construction in yeast-two-hybrid and bimolecular fluorescence complementation vectors.

Click here for additional data file.