Arabidopsis thaliana CENTRORADIALIS homologue (ATC) acts systemically to inhibit floral initiation in Arabidopsis

Authors

  • Nien-Chen Huang,

    1. Molecular and Biological Agricultural Sciences Program, Taiwan International Graduate Program, National Chung-Hsing University and Academia Sinica, Taipei 11529, Taiwan
    2. Institute of Plant and Microbial Biology, Academia Sinica, Taipei 11529, Taiwan
    3. Graduate Institute of Biotechnology, National Chung-Hsing University, Taichung 402, Taiwan
    Search for more papers by this author
  • Wann-Neng Jane,

    1. Institute of Plant and Microbial Biology, Academia Sinica, Taipei 11529, Taiwan
    Search for more papers by this author
  • Jychian Chen,

    1. Institute of Molecular Biology, Academia Sinica, Taipei 11529, Taiwan
    Search for more papers by this author
  • Tien-Shin Yu

    Corresponding author
    1. Institute of Plant and Microbial Biology, Academia Sinica, Taipei 11529, Taiwan
      (e-mail tienshin@gate.sinica.edu.tw).
    Search for more papers by this author

(e-mail tienshin@gate.sinica.edu.tw).

Summary

Floral initiation is orchestrated by systemic floral activators and inhibitors. This remote-control system may integrate environmental cues to modulate floral initiation. Recently, FLOWERING LOCUS T (FT) was found to be a florigen. However, the identity of systemic floral inhibitor or anti-florigen remains to be elucidated. Here we show that Arabidopsis thaliana CENTRORADIALIS homologue (ATC), an Arabidopsis FT homologue, may act in a non-cell autonomous manner to inhibit floral initiation. Analysis of the ATC null mutant revealed that ATC is a short-day-induced floral inhibitor. Cell type-specific expression showed that companion cells and apex that express ATC are sufficient to inhibit floral initiation. Histochemical analysis showed that the promoter activity of ATC was mainly found in vasculature but under the detection limit in apex, a finding that suggests that ATC may move from the vasculature to the apex to influence flowering. Consistent with this notion, Arabidopsis seedling grafting experiments demonstrated that ATC moved over a long distance and that floral inhibition by ATC is graft transmissible. ATC probably antagonizes FT activity, because both ATC and FT interact with FD and affect the same downstream meristem identity genes APETALA1, in an opposite manner. Thus, photoperiodic variations may trigger functionally opposite FT homologues to systemically influence floral initiation.

Introduction

The transition from vegetative to reproductive growth is one of the most critical phase changes in higher plants. Plants have evolved many elegant systems to assess environmental and developmental parameters to ensure proper floral transition. These signals integrate into a number of floral integrators to fine-tune floral initiation. The perception of environmental inputs, such as day-length variation, usually relies on leaves, whereas the floral initiation occurs at the apex. After receipt of the optimal photoperiod, a systemic floral activator, or florigen, is synthesized in leaves. This graft-transmissible signal then trafficks long distances through phloem to shoot apices for floral initiation (Zeevaart, 1976; Bernier et al., 1993). The molecular identity of florigen has recently been addressed. Experiments in Arabidopsis and rice have concluded that FLOWERING LOCUS T (FT) in Arabidopsis or Heading date 3a (Hd3a) in rice is the long-sought-after florigen (Zeevaart, 2006; Turck et al., 2008; Tsuji et al., 2011). The homologues of FT/Hd3a appear to function as general florigens in many other plant species (Kong et al., 2010; Pin et al., 2010; Hecht et al., 2011; Meng et al., 2011; Wigge, 2011).

In addition to florigen, a systemic floral inhibitor, or anti-florigen, may contribute to long-distance floral regulation (Lang et al., 1977; Zeevaart, 2006). In tobacco grafting experiments, a day-neutral variety of tobacco (Trapezond) was used as a recipient for grafting onto different tobacco varieties, including long-day (LD, Nicotiana sylvestris) or short-day (SD, Maryland Mammoth) varieties. When the grafted plants were kept under non-floral induction conditions (i.e., N. sylvestris in SD, and Maryland Mammoth in LD), the floral initiation of recipients grafted onto N. sylvestris but not Maryland Mammoth was suppressed. Thus, a systemic floral inhibitor, or anti-florigen, may be produced in N. sylvestris plants under non-floral induction conditions (Lang et al., 1977). However, the identity of this systemic floral inhibitor remains to be elucidated.

Arabidopsis FT encodes a small protein similar to mammalian phosphatidylethanolamine binding-domain protein (Kardailsky et al., 1999; Kobayashi et al., 1999). The induction of FT is LD dependent and mainly occurs in companion cells of leaves (Takada and Goto, 2003). FT then trafficks to the apex and interacts with FD, a bZIP transcription factor, to upregulate the downstream floral identity genes, APETALA1 (AP1) (Abe et al., 2005; Wigge et al., 2005). The FT family in Arabidopsis contains five other members: TWIN SISTER OF FT (TSF), TERMINAL FLOWER 1 (TFL1), BROTHER OF FT AND TFL1 (BFT), MOTHER OF FT AND TFL1 (MFT), and Arabidopsis thaliana CENTRORADIALIS homologue (ATC). Phylogeneic analysis revealed that FT homologues are grouped into three major clades: FT-, TFL1- and MFT-like clades (Kobayashi et al., 1999; Chardon and Damerval, 2005). Genetic analyses demonstrated that FT-like genes (FT and TSF) are floral activators, whereas TFL1-like genes (TFL1, ATC and BFT) are floral inhibitors (Bradley et al., 1997; Kardailsky et al., 1999; Kobayashi et al., 1999; Mimida et al., 2001; Yamaguchi et al., 2005; Yoo et al., 2010). In addition to involvement in floral regulation, MFT-like genes participate in regulation of seed germination (Yoo et al., 2004; Xi et al., 2010). Interestingly, FT may not be the only florigen. Other FT homologues, Arabidopsis TSF or RICE FLOWERING LOCUS T 1 (RFT1), also act systemically to promote flowering (Jang et al., 2009; Komiya et al., 2009). In addition, TFL1 may move cell to cell to maintain shoot meristem identity (Conti and Bradley, 2007).

In Arabidopsis, at least three members of the FT family (FT, TSF, and TFL1) function in a non-cell autonomous manner (Conti and Bradley, 2007; Corbesier et al., 2007; Jang et al., 2009); this situation raises the possibility that members of the TFL1-like clade may also act non-cell autonomously to regulate flowering. Gene expression analysis of the TFL1-like clade revealed that TFL1 is expressed exclusively in the apex, while BFT and ATC are expressed in vascular tissues (Bradley et al., 1997; Mimida et al., 2001; Yoo et al., 2010). BFT expresses predominately under LD conditions, while ATC is induced under SD conditions (Yoo et al., 2010). Because tobacco grafting experiments suggested that the systemic floral inhibitor is produced under non-floral-induction conditions (Lang et al., 1977), we examined whether ATC may act non-cell autonomously to inhibit floral initiation. Here, we tested whether Arabidopsis ATC, in the TFL-like clade of FT homologues, acts non-cell autonomously to inhibit floral initiation. We found that SD-induced ATC could act non-cell autonomously to inhibit floral initiation, which is probably mediated by antagonism of FT activity. Taken together, we propose that ATC may contribute to anti-florigen activity in Arabidopsis.

Results

Arabidopsis ATC is an SD-induced floral inhibitor

The Arabidopsis FT gene family consists of six members, including floral activators (FT, TSF, and MFT) and inhibitors (TFL1, BFT, and ATC). Arabidopsis transformants carrying P35S-ATC display a late-flowering phenotype, however, the atc-1 mutant, flowers normally under LD conditions (Mimida et al., 2001). Because the expression of ATC is detected in LD conditions but highly accumulated during the dark under SD conditions (Figure 1a, Yoo et al., 2010), we examined whether ATC may contribute to floral inhibition under SD conditions. We obtained two T-DNA insertion mutants, atc-2 and atc-3, which contain a T-DNA insertion at the first exon or the promoter region of ATC, respectively (Figure 1b). Real-time RT-PCR analysis revealed that the expression of ATC was greatly reduced in atc-2 but only a slightly alteration in atc-3 (Figure 1c). In addition, further RT-PCR analysis showed that the expression of ATC in atc-2 mutant was under the detection limit when PCR cycles were increased to 40 cycles (Figure S1), which suggests that atc-2 is a null allele. When these mutants were grown under LD conditions, wild-type plants, atc-2, and atc-3 mutants flowered at a similar time with 8.7 ± 1.0, 9.1 ± 1.8 and 8.9 ± 1.0 leaves, respectively (Figure 2a). However, when the mutants were grown under SD conditions, wild-type plants and atc-3 mutants flowered at a similar time with 50.9 ± 3.1 and 51.8 ± 3.8 leaves, respectively, while atc-2 flowered earlier with 42.5 ± 2.9 leaves (Figure 2b). To further confirm the early flowering phenotype of atc-2, we conducted complementation experiments. We cloned the 8.7-kb ATC genomic DNA fragment, which contains 7272 bp upstream sequences, ATC coding sequences and 345 bp downstream sequences, and introduced it into atc-2 mutant. The flowering time of atc-2 transformants that harbour ATC genomic DNA was complemented to the wild-type level with 49.0 ± 2.7 and 54.4 ± 8.0 leaves under SD conditions (Figure 2b). Thus, ATC functions as a floral inhibitor under SD conditions.

Figure 1.

 Identification of Arabidopsis atc-2 and atc-3 mutants.
(a) Real-time RT-PCR analysis of time course expression of ATC under long-day (LD) or short-day (SD) conditions. Samples were harvested from 15-day-old LD- or SD-grown wild-type plants at different time points. The RNA level of ATC in time point 0 of the LD-grown plants was set to 1 after normalization. (b) Schematic diagram of T-DNA insertion in atc-2 and atc-3 mutants. The number indicates the position of the T-DNA insertion on ATC gene. Exons are represented by black boxes. (c) Real-time RT-PCR analysis of ATC level in wild-type (Col), atc-2 and atc-3 mutants. The level of ATC RNA in Col was set to 1 after normalization.

Figure 2.

 Phenotype of Arabidopsis atc-2 and atc-3 mutants.
(a) Flowering time of atc-2 and atc-3 mutants under long-day (LD) conditions. Number of rosette leaves was used to indicate flowering time. The values are expressed as means [±standard deviation (SD), n = 10]. (b) Flowering time of Arabidopsis atc-2, atc-3 and atc-2 transformants carrying ATC genomic DNA (atc-2/gATC-1 and atc-2/gATC-2) under short-day (SD) conditions. The values are means (±SD, n = 15). (c) Flowering time of atc-2, ft-10 or atc-2 ft-10 double mutant under long-day (LD) conditions. The values are means (±SD, n = 7).

Although atc mutants flower normally under LD conditions, overexpression of ATC in Arabidopsis exhibits a late-flowering phenotype under LD conditions (Mimida et al., 2001), which suggests that the balance between FT and ATC level may determine the start of floral initiation. To examine whether the activation of FT in LD conditions may attenuate the floral inhibitory effect of ATC, we crossed atc-2 with ft-10 (both in a Columbia background) and examined the flowering time of the double mutant under LD conditions. In an F2 population, wild type (ATC, FT) and atc single mutants (atc-2, FT) flowered about 10 to 12 leaves, whereas ft-10 mutants (ATC, ft-10) flowered with 42 leaves (Figure 2c). However, the double mutant (atc-2, ft-10) flowered with 35 leaves, which is earlier than that of the ft-10 single mutants (Figure 2c). These results indicate that ATC can inhibit flowering when the level of FT is reduced. In addition, our results support the notion that the balance in ATC and FT level may determine floral initiation.

ATC acts non-cell autonomously to inhibit flowering

To investigate whether ATC acts non-cell autonomously to inhibit floral initiation, we expressed ATC in a tissue-specific manner and examined the flowering time of the transformants. The full-length cDNA of ATC or TFL1 (ATC contains a 166-bp 5′UTR and 172-bp 3′UTR; TFL1 contains a 25-bp 5′UTR and a119-bp 3′UTR) was cloned and driven by a CaMV35S or AtSUC2 promoter. These constructs were introduced into wild-type plants (Columbia ecotype). Of the transformants we analyzed (10 independent lines for each construct), eight lines of P35S-ATC and seven lines of P35S-TFL1 transformants displayed a late-flowering phenotype (Figure 3a). In P35S-ATC transformants with a severe late-flowering phenotype, the floral apex usually produced leafy-like bracts (Figure S2a). Although cryo-scanning electron microscopy (cryo-SEM) revealed that floral primordia were produced in P35S-ATC transformants (Figure S2b), their growth was usually arrested, and they did not further develop into flowers. Similar to P35S-ATC transformants, 11 lines out of 15 PSUC2-ATC transformants also displayed a late-flowering and abnormal floral-organ phenotype (Figure 3a–c). However, of the 18 transformants we analyzed, the flowering time of PSUC2-TFL1 transformants was similar to that of the wild type (Figure 3a,b). Thus, companion cells that express ATC but not TFL1 are sufficient to inhibit floral initiation. To investigate whether ATC also acts in the apex, we expressed ATC by use of an Arabidopsis MERISTEM LAYER 1 (ML1) promoter, which is expressed in epidermis of shoot and root meristem (Sessions et al., 1999). Similar to P35S-ATC and PSUC2-ATC transformants, Arabidopsis PML1-ATC transformants displayed a late-flowering and abnormal floral-organ phenotype (Figure 3a,b). We further examined the expression of FT in different ATC transformants and found that the mRNA level of FT in P35S-ATC and PSUC2-ATC transformants is similar to that of the wild type (Figure S3), a finding that suggests that floral inhibition by ATC is not mediated by downregulation of FT. These results indicate that companion cells and apex that express ATC are sufficient to inhibit floral initiation.

Figure 3.

ATC acts non-cell autonomously to inhibit flowering.
(a) Flowering time of Arabidopsis transformants carrying different cell type specific expression of ATC or TFL1 under long-day (LD) conditions. Two representative transformants (labelled A and B) are shown. The values are means [±standard devation (SD), n = 28]. (b) Phenotype of cell type specific expression of ATC or TFL1 in Arabidopsis. Left panel: Arabidopsis transformants carrying P35S-ATC or PSUC2-ATC displayed a late-flowering phenotype. Middle panel: Arabidopsis transformants carrying P35S-TFL1 but not PSUC2-TFL1 displayed a late-flowering phenotype. Right panel: Arabidopsis transformants carrying PML1-ATC also displayed a late-flowering phenotype. (c) Cryo-electron microscopy analysis of floral buds of wild-type (left panel), PSUC2-ATC transformant (middle panel) and floral primodia in the apex of the PSUC2-ATC transformant (right panel).

Previous in situ hybridization analysis revealed that ATC is expressed in the vasculature of hypocotyls (Mimida et al., 2001). To further determine the spatial expression pattern of ATC, the promoter region of ATC was amplified and fused with β-glucuronidase (GUS). Histochemical assay of SD-grown PATC-GUS transformants showed GUS activity mainly in vascular tissues of petioles, hypocotyls and roots (Figure 4a,b,d,e) but under the detection limit in the apex (Figure 4c, asterisk). In transverse sections of hypocotyl tissues, GUS activity was mainly detected in the phloem (Figure 4f). The LD-grown PATC-GUS transformants also displayed a similar vasculature pattern of GUS activity (Figure S4), which supports that ATC also expressed in LD condition. However, we did not observe a significant difference of GUS activity in SD or LD-grown transformants. This result suggests that ATC was mainly expressed in vascular tissues but not in the apex.

Figure 4.

 Promoter activity of ATC in vasculature.
Histochemical staining of 17-day-old short-day (SD)-grown Arabidopsis transformants carrying PATC-GUS transgene from (a) aerial part, (b) cotyledon, (c) longitudinal section of shoot apex. Shoot apical meristem is indicated by a red asterisk, and leaf primodia is indicated by arrows. (d) hypocotyls, (e) roots, and (f) transverse section of hypocotyls. X, xylem; P, phloem.

Long-distance movement of ATC in Arabidopsis cleft-grafting experiments

Cell-type-specific expression analysis revealed that ATC may act in both companion cells and the apex to inhibit flowering. However, the promoter activity of ATC is limited in vascular tissues, which suggests that the ATC or the other signal activated by ATC may function as a systemic signal. To examine whether ATC acts systemically to inhibit flowering, we analyzed the long-distance movement of ATC with Arabidopsis seedling-grafting experiments. In atc-2, the mRNA level of ATC was under the detection limit (Figures 1c and S1). We grafted atc-2 seedlings onto de-capped atc-2 or wild-type plants by Arabidopsis seedling-grafting experiment. The scions were cut above the cotyledons and attached to the apex-removed stocks with insect pin (Figure S5). Thus, in our seedling-grafting experiment, the scions and stocks were assembled on epicotyls rather than on hypocotyls of the stock. Two weeks after grafting, real-time RT-PCR analysis of individual scion showed that the ATC RNA was detected in atc-2 scions that grafted onto Col stocks (Figure 5a, SC1-5), but not from atc-2 scions grafted onto atc-2 stocks (Figure 5a, atc2). A finding that suggests that ATC RNA moves long distances from wild-type stocks to atc-2 scions. In atc-2 scions grafted onto Col stocks, the degree of ATC RNA in atc-2 scions is usually 1–5% of that in the Col stocks, a finding that suggests that a small proportion of ATC RNA moves a long distance to the apex (Figure 5a). We also conducted Arabidopsis inflorescence-grafting experiments (Huang and Yu, 2009) with P35S-ATC or PSUC2-ATC transformants. RT-PCR revealed that the transgenic ATC RNA moves a long distance from transformant stocks to the wild-type scions (Figure S6). Thus, our grafting experiments indicated that ATC RNA is a non-cell autonomous RNA. To examine whether ATC protein is present in scions, western blotting with antibodies against ATC detected a 20-kD ATC protein in P35S-ATC transformants but not atc-2 plants (Figure 5b); this result indicates that atc-2 is a null allele. After atc-2 seedlings were grafted onto P35S-ATC plants, western blot analyses detected ATC protein in atc-2 scions at 18 or 28 days after grafting (Figure 5b). To further explore that the floral inhibition by ATC is graft transmissible, we grafted wild-type or atc-2 plants onto different stocks. Our results showed that the distribution of flowering time of scions was shifted to late flowering after they were grafted onto P35S-ATC stocks than that scions grafted onto wild-type or atc-2 stocks (Figure 5c, Table S1), a situation that indicates that floral inhibition by ATC is graft transmissible.

Figure 5.

 Long-distance movement of ATC in Arabidopsis seedling-grafting experiments.
(a) Real-time RT-PCR analysis of individual scion samples grafted onto wild-type or atc-2 stock. The scion samples were harvested from atc-2 scion grafted onto atc-2 stock (atc2), or atc-2 scions grafted onto wild-type stocks (SC1-SC5), and wild-type stock of SC1 (ST). The grafting experiments were conducted under short-day (SD) conditions. The relative expression is calculated by the cycle of threshold (−ΔΔCT). (b) Western blot analyses of ATC protein detected in atc-2 scions grafted onto P35S-ATC stocks. Upper panel: Arabidopsis P35S-ATC transformants, atc-2, or atc-2 scions grafted onto P35S-ATC stocks. The samples were collected from atc-2 scions 18 (SC1 and SC2) or 28 days (SC3 and SC4) after grafting. Lower panels: Coomassie blue (CB) staining was the loading control. (c) Floral inhibition by ATC is graft transmissible. Flowering time of individual scions (asterisk) grafted onto atc-2, Col, or P35S-ATC transformants stocks under long-day (LD) conditions. The mean of the flowering time is indicated by arrowhead. The distribution of flowering time of scions was shifted after being grafted onto P35S-ATC transformants.

ATC may antagonize FT activity

In Arabidopsis, the interaction between FT and FD upregulates the expression of AP1 to promote flowering (Abe et al., 2005; Wigge et al., 2005). To examine whether ATC and FT regulate similar floral pathways antagonistically, we tested the interaction between FD and ATC by bimolecular fluorescence complementation (BiFC) assay. The coding regions of ATC and FD were fused with BiFC vectors VenusN and SCFP3AC, respectively (Waadt et al., 2008). For a control, the coding region of FT was fused with VenusN. The resulting constructs were co-expressed in Nicotiana benthamiana leaves by Agrobacterium-mediated infiltration. Five days after infiltration, confocal microscopy revealed a green fluorescent signal in the nucleus of epidermal cells that co-expressed ATC and FD or FT and FD (Figure 6a,b). In contrast, co-expression of ATC or FT with the C-terminal truncated form of FD (FDΔC), a FD mutant defective in FT interaction (Abe et al., 2005), produced no green fluorescent signal in epidermal cells (Figure 6c,d). Therefore, both FT and ATC can interact physically with FD, probably with the C-terminal domain of FD protein. To analyze whether the FD interaction may determine the subcellular localization of ATC or FT, we co-expressed FD with ATC or FT. When RFP-ATC, RFP-FT and GFP-FD was infiltrated into the leaves of Nicotiana benthamiana, RFP-ATC or RFP-FT was localized in both cytosol and nucleus (Figure 6e,f), whereas GFP-FD and GFP-FDΔC was exclusively localized in the nucleus (Figure 6g,h). However, when RFP-ATC or RFP-FT was co-infiltrated with GFP-FD, red and green fluorescent signals were localized exclusively in the nucleus (Figure 6i,k). In contrast, when RFP-ATC or RFP-FT was co-expressed with the non-interacting form GFP-FDΔC, red fluorescent signals were detected both in cytosol and nucleus, and green fluorescent signals were detected only in the nucleus (Figure 6j,l). Thus, ATC and FT interacts with the same transcription factor FD, and the FD interaction may stabilize the ATC/FT and FD complex in the nucleus.

Figure 6.

 Interactions between ATC/FT and FD proteins in Nicotiana benthamiana leaves. (a–d) BiFC analyses showing that both ATC and FT interacting with FD protein. The leaves of N. benthamiana were co-infiltrated with (a) VenusN-ATC and SCFP3AC-FD, (b) VenusN-FT and SCFP3AC-FD, (c) VenusN-ATC and SCFP3AC-FDΔC (C-terminal deletion of FD), or (d) VenusN-FT and SCFP3AC-FDΔC. Images were obtained 5 days after agro-infiltration. (e–h) Subcellular localization of (e) RFP-ATC, (f) RFP-FT, (g) GFP-FD, and (h) GFP-FD ΔC proteins in N. benthamiana leaves. Images were obtained 3 days after agro-infiltration. (i–l) Subcellular localization of co-infiltration of (i) RFP-ATC and GFP-FD, (j) RFP-ATC and GFP-FD ΔC, (k) RFP-FT and GFP-FD, or (l) RFP-ATC with GFP-FD ΔC in N. benthamiana leaves. The images were obtained 5 days after agro-infiltration.

In Arabidopsis, the multiple floral signals converge on two major floral integrators, FT and SUPPRESSOR OF OVEREXPRESSION OF CONSTANS 1 (SOC1). FT and SOC1 subsequently activate downstream floral meristem identity genes, LEAFY (LFY) and AP1, respectively (Liu et al., 2009). To test whether ATC regulates LFY or AP1, we examined the expression of AP1 and LFY in P35S-ATC transformants. Real-time RT-PCR analyses revealed that the RNA level of AP1, but not that of LFY, was reduced significantly in P35S-ATC transformants (Figure 7). Thus, ATC may downregulate the expression of AP1.

Figure 7.

 Accumulation of AP1 and LFY RNA in Arabidopsis wild-type and P35S-ATC transformants.
Real-time RT-PCR analysis of RNA level of (a) AP1 or (b) LFY in floral buds of Arabidopsis wild-type or P35S-ATC transformants.

Discussion

In tobacco, classical grafting experiments showed that photoperiodic floral regulation is governed by two counteracting signals, florigen and anti-florigen. Florigen is synthesized when plants are grown under floral induction conditions, whereas anti-florigen is produced when plants are grown under non-floral-induction conditions (Zeevaart, 1976, 2006; Lang et al., 1977; Turck et al., 2008). Recent evidence has shown that the members of the FT family, FT and TSF in Arabidopsis or Hd3a and RFT1 in rice, function redundantly as florigen (Corbesier et al., 2007; Tamaki et al., 2007; Jang et al., 2009; Komiya et al., 2009). Because the FT family contains both floral activators and suppressors, a single amino acid variation can interconvert the activity of floral activation or inhibition (Hanzawa et al., 2005). It is not surprising that ATC, an Arabidopsis FT paralog but with an opposite function, may act in a non-cell autonomous manner to inhibit floral initiation. Unlike FT, the expression of ATC is induced under SD conditions (Figure 1a; Yoo et al., 2010), a finding that is consistent with the production of anti-florigen is under non-floral induction conditions. The activity of the ATC promoter is mainly in the vasculature (Figure 4), whereas ectopic expression of ATC in companion cells or the apex is sufficient to inhibit floral initiation, suggesting that ATC itself may act as part of the systemic floral inhibitor (Figures 1, 2 and 3). Indeed, our grafting experiments detected both RNA and protein of ATC in atc-2 scions grafted onto wild-type or transformant stocks (Figure 5). That floral inhibition by ATC is graft transmissible further supports that the translocated ATC is functional (Figure 5). Thus, our results indicate that the movement of ATC may act as a part of an anti-florigen. Whether other molecules also contribute to the activity of anti-florigen remains to be investigated.

The expression of ATC is induced under SD conditions, however, a limited amount of ATC is detected when plants was grown under LD conditions (Figure 1; Yoo et al., 2010). Interestingly, the atc null mutant did not show an obvious flowering phenotype under LD conditions. It is possible that the highly expressed FT under LD conditions attenuates floral inhibition by ATC. Indeed, Arabidopsis transformants that overexpressed ATC displayed a late-flowering phenotype when grown under LD conditions (Figure 3). In addition, the double mutant atc-2, ft-10 flowered earlier than the ft-10 single mutant (Figure 2). Thus, Arabidopsis may adjust the balance of FT and ATC by photoperiodic variation to fine-tune floral initiation. In Arabidopsis, TFL1 may act as a transcriptional repressor to antagonize FT activity (Hanano and Goto, 2011). Because of significant similarity between the TFL1 and ATC sequences, it is likely that ATC also acts antagonistically with FT to regulate downstream floral identity genes. Indeed, our BiFC analysis showed that both ATC and FT interact with FD and regulate the same downstream floral meristem identity genes with opposite effects (Figures 6 and 7). In addition to Arabidopsis, it has recently been shown that floral initiation in cultivated beet (Beta vulgaris) is controlled by the interplay of two antagonistic FT paralogs, BvFT1 and BvFT2 (Pin et al., 2010). The floral repressor BvFT1 is expressed under conditions that do not induce flowering, whereas the floral activator BvFT2 is expressed under conditions that induce flowering (Pin et al., 2010). Interestingly, both BvFT1 and BvFT2 are expressed mainly in the leaves, a situation that suggests that these two genes may act systemically to regulate flowering. Thus, the antagonistic FT paralogs may be systemic florigenic signals to integrate different environmental cues. However, it remains to be investigated whether the FT paralogs in other plant species also participate in florigen/anti-florigen activity.

Our Arabidopsis cleft-grafting experiments showed that ATC RNA moved long distances from the wild-type stock to atc-2 scion (Figure 5). Although real-time RT-PCR analysis showed that only a limited amount of ATC RNA was detected in the scions, it is possible that the amplification during RNA translation may produce a sufficient amount of ATC protein to inhibit flowering. Alternatively, both ATC RNA and protein may travel long distances to regulate flowering. Indeed, this idea is consistent with our analyses that ATC protein was detected in atc-2 scions at 18 days after grafting (Figure 5b). Interestingly, our cleft-grafting experiments contradict previous Arabidopsis micro-grafting experiments that failed to detect the movement of FT RNA (Corbesier et al., 2007). The contrasting results may be explained by different grafting methods. RNA translocation is highly dynamic and depends on the sink-source strength between stock and scion (Haywood et al., 2005). In Arabidopsis micro-grafting experiments, the scions were attached to the hypocotyls of the stocks (Corbesier et al., 2007). As the scions grow, the source leaves on the scions may dilute the systemic RNA signal derived from the stock. In contrast, in our cleft-grafting system, the scions were attached on the epicotyls of the stocks (Figure S5). Therefore, the source leaves on the stocks may produce sufficient source strength to transport RNA signal. Interestingly, recent experiments have demonstrated that FT RNA may facilitate the RNA trafficking of movement-defective virus in Arabidopsis and Nicotiana benthamiana (Li et al., 2009). In addition, our recent findings with Arabidopsis cleft-grafting experiments also showed that FT RNA is a mobile florigenic signal (Lu et al., 2012). More experiments are required to clarify the variations in cleft- or micro-grafting experiments and reveal the underlying mechanism. In addition, it has been shown that the cis-elements on non-cell autonomous RNA are required for RNA movement (Huang and Yu, 2009). Whether the RNA-mobile elements participate in ATC RNA trafficking remains to be investigated.

The FT family encodes small soluble proteins that are similar to mammalian phosphatidylethanolamine binding-domain proteins (Kardailsky et al., 1999; Kobayashi et al., 1999) Although our results showed the presence of ATC protein in the atc-2 mutant grafted onto P35S-ATC transformant stock, we did not determine whether ATC protein can move cell to cell or for a long distance. In our preliminary co-bombardment assay with P35S-GFP-ATC and P35S-RFP in Arabidopsis leaves, green but not red fluorescent signals were detected in cells that surrounded the bombarded cells (Figure S7), a result that indicates that GFP-ATC protein most probably moves from cell to cell. Thus, similar to FT and TFL1 protein, ATC protein probably also moves systemically in Arabidopsis. In addition to the FT family, an increasing number of non-cell autonomous proteins have been identified. These proteins play critical roles during development in multi-cellular organisms. In Arabidopsis, the endodermis specification of root meristem is governed by a mobile transcription factor, SHORT-ROOT (SHR). SHR moves from synthesized cells to adjacent cells, where it interacts with SCARECROW (SCR). The interaction of SHR and SCR sequesters SHR in the nucleus and restricts SHR from further movement (Cui et al., 2007). Similar to SHR nucleus relocation, FD interaction may stabilize the ATC/FT and FD in the nucleus (Figure 4). In addition, recent findings have demonstrated that the interaction between FD and Arabidopsis TFL1 or rice Hd3a may target the complex into the nucleus (Hanano and Goto, 2011; Taoka et al., 2011). However, because the expression of FD is restricted in the apical meristem (Abe et al., 2005; Wigge et al., 2005), FD may be required to finalize the long-distance trafficking of FT/ATC protein in the apex.

Experimental procedures

Plant materials and growth conditions

Arabidopsis thaliana seeds were obtained from Arabidopsis Biological Resource Center (http://www.arabidopsis.org/) and were grown in growth chambers under LD (16 h light/8 h dark) or SD (8 h light/16 h dark) conditions, with a 22°C/20°C day/night cycle with white fluorescent light (light intensity 100 μmol m−2 sec−1). The atc-2 and atc-3 mutant lines were obtained from ABRC (Salk_021699 and SAIL_1264_D07, respectively).

Plasmid construction

Arabidopsis ATC gDNA was amplified by PCR with gATC-For and gATC-Rev primers. The Arabidopsis full-length ATC or TFL1 cDNA was amplified by RT-PCR with ATC-For and ATC-Rev primers or TFL1-For and TFL1-Rev primers, respectively (primer sequences are listed in Table S2). The ATC coding region was generated by PCR with ATC-CDS-For and ATC-CDS-Rev primers. All constructs were confirmed by sequencing analysis, then cloned into pCAMBIA1390 binary vectors containing CaMV35S, SUC2 or ML1 promoters.

Arabidopsis transformation

Arabidopsis was transformed by the Agrobacterium-mediated floral dip method. The T1 transformants were selected on Murashige and Skoog (MS) medium containing 40 μg ml−1 hygromycin. For each construct, at least 10 representative independent transformants were isolated and used for grafting experiments or phenotype observation.

Arabidopsis cleft grafting

Plants were grown at 22°C under LD condition for 1–2 weeks. The scions were cut at about 0.1 cm from the top of the hypocotyl under microscopy, and leaves larger than 0.3 cm were removed. A fine insect pin was inserted into the scion hypocotyl and then pierced into the hypocotyl of de-capped stock plants. The grafted unions were kept humid for 7 days and then returned slowly to normal growth conditions. Two weeks after seedling grafting, the scion tissues were harvested and subjected to RT-PCR to analyze the movement of endogenous ATC transcripts. The flowering time of scions was measured by the appearance of the first floral bud after seedling grafting.

Inflorescence grafting was as previously described (Huang and Yu, 2009). The inflorescence from scions was cut in a V shape and inserted into the stock, which was vertically cut in the middle of the inflorescence. The grafted union was secured by a piece of polyethylene tubing and kept under high humidity for 1 week, then returned to normal growth conditions for another week. RNA samples were extracted from both stock and scion tissues 2 weeks after grafting.

RNA extraction and RT-PCR analyses

Total RNA was extracted by the Trizol® reagent method (Invitrogen, http://www.invitrogen.com). Five micrograms of total RNA was used to synthesize the first-strand cDNA with oligo(dT)20 and Superscript III reverse transcriptase (Invitrogen). The PCR reaction was carried out with 2 μl of first-strand cDNA. ATC-FOR and ATC-REV were used to detect endogenous ATC transcripts and ATC-For and NOS term-Rev primers to distinguish transgenes from endogenous ATC transcripts (primer sequences are in Table S2). AP1-For, AP1-Rev and LFY-For, LFY-Rev were used to amplify AP1 and LFY, respectively. The PCR reactions were performed at 94°C for 1 min; 94°C for 30 sec, 60°C for 30 sec, 68°C for 1 min for 35 cycles; and 68°C for 7 min. An aliquot of 5 μl PCR products was separated on 1.2% agarose gels, and the DNA fragments were visualized by staining with SYBR® Green I nucleic acid gel stain (Invitrogen).

Protein extraction and immunoblot analyses

Proteins were extracted from 0.2 g tissues. After tissues were ground in liquid nitrogen, the plant powder was mixed with 100 μl extraction buffer (0.3 mm Tris–HCl pH8.5, 8% sodium dodecyl sulphate (SDS), 1 mm ethylene diamine tetraacetic acid (EDTA), and 1× complete protease inhibitor cocktail) and boiled in 100°C water for 5 min. After centrifugation at 4°C for 5 min at full speed, the supernatant were subjected to immunoblot analysis.

For SDS-PAGE and immunoblotting, 60 μg protein was separated on 12% NuPAGE® Novex® Bis-Tris Mini Gels (Invitrogen) with 1× NuPAGE® 2-(N-morpholino)ethanesulfonic acid (MES) buffer, then transferred to polyvinylidene difluoride (PVDF) membrane for immunoblot analyses or stained with Coomassie blue as a loading control. Immunoblotting analyses were performed with the Ultra-Sensitive ABC Peroxidase Staining Kit (Thermo Scientific, http://www.thermoscientific.com). The membrane was incubated in blocking reagent [5% non-fat dry milk in 1× TBS (20 mm Tris–HCl pH 7.4, 150 mm NaCl)] at room temperature for 30 min. The 1:1000 diluted rabbit anti-ATC antibodies were used as primary antibody and biotinylated affinity-purified goat anti-rabbit IgG were used as secondary antibody. After incubation with ABC reagent, the substrate solution (6 mg diaminobenzidine tetrachloride, 8 mg NiCl2, and 10 μl H2O2 in 1× TBS) was added for detecting protein signals.

Histochemical analyses

ATC promoter (7272 bp upstream of ATG) was amplified by PCR with gene-specific primers gATC-For and ATC pro-Rev (sequences are in Table S2). The clone was confirmed by sequencing and fused with a GUS reporter gene. The Arabidopsis transformants carrying PATC-GUS were selected on MS medium containing 50 μg ml−1 kanamycin. For GUS activity assay, transformants were incubated with GUS staining solution (50 mm sodium phosphate pH 7.0, 10 mm EDTA, 0.5 mm potassium ferricyanide, 0.5 mm potassium ferrocyanide, 1 mm X-Gluc, 0.01% Triton X-100) at 37°C for 16 h. Plants were treated with 95% ethanol to remove chloroplast and photographed under a Leica Z16 Apo microscope.

Real-time RT-PCR analyses

Total RNA was extracted from the floral buds of Arabidopsis by the Trizol® reagent method (Invitrogen). After DNase I treatment, 5 μg total RNA was used to synthesize first-strand cDNA by use of Superscript III reverse transcriptase (Invitrogen) with reaction volume 20 μl. The volume was then brought up to 500 μl, and an aliquot of 5 μl was used for real-time PCR reactions. For each reaction, 200 nm gene-specific primers were used with the AB 9500 Real-Time PCR System (Applied Biosystems, http://www.appliedbiosystems.com). PCR parameters were 95°C for 10 min, followed by 40 cycles of 95°C for 15 sec and 60°C for 1 min. Triplicate reactions were conducted for each sample. The expression of ubiquitin-conjugating enzyme (UBC) was used as a normalization control. The quantification of RNA level was calculated by use of fold change (Figures 1 and 7), and the cycle of threshold (CT) method (Figure 5). To calculate ΔΔCT, ΔΔCT = (ΔCT of each stocks or scions) – (ΔCT of atc-2 scion grafted onto atc-2), whereas ΔCT represents the CT of ATC normalized to UBCCT = CT of ATCCT of UBC). The specific primers used were: AtATC-Q-For, AtATC-Q-Rev, AtAP1-Q-For, AtAP1-Q-Rev, AtLFY-Q-For, AtLFY-Q-Rev, UBC-Q-For, and UBC-Q-Rev (sequences are listed in Table S2).

Bimolecular fluorescence complementation (BiFC) analyses

cDNA of ATC and FT was cloned into VenusN binary vector HygII-VYNE(R); cDNA of FD or FD ΔC (40 a.a. deletion in the C-terminus of FD) was PCR-amplified and cloned into SCFP3AC vector KanII-SCYCE(R) under control of a CaMV35S promoter. All the constructs were introduced into Agrobacterium tumefaciens strain AGL1. For BiFC analysis, Agrobacterium cells carrying individual BiFC constructs were cultured media containing 50 μg ml−1 of kanamycin, 10 mm MES, pH 5.7, and 20 μm acetosyringone at 28°C overnight. Agrobacterium cells were pelleted and resuspended in infiltration solution (10 mm MgCl2, 10 mm MES, pH 5.7, 200 μm acetosyringone) to OD600 1.0. The bacteria solution was left at room temperature for 1 h. Co-infiltration was conducted with a 1:1 mix of HygII-VYNE(R)-ATC or HygII-VYNE(R)-FT with KanII-SCYCE(R)-FD bacteria solution. The solution was infiltrated into leaves of 3-week-old Nicotiana benthamiana by syringe. Five days after infiltration, plants were photographed under a confocal laser scanning microscope (Zeiss LSM 510 Meta).

Cryo-scanning electron microscopy

Floral tissues were loaded on a cryo-specimen holder and frozen in liquid nitrogen, and transferred to a sample preparation chamber at −160°C for 5 min, with temperature increased to −85°C for 15 min. After being coated with platinum (Pt) at −130°C, samples were transferred to a cryostage in an SEM chamber and observed at −160°C with a cryo-scanning electron microscope (FEI Quanta 200 SEM/Quorum Cryo System PP2000TR FEI) at 20KV.

Acknowledgements

We thank the Arabidopsis Biological Resource Stock Center (ABRC) for providing Arabidopsis seeds, Dr Jörg Kudla for providing the BiFC vectors. The authors declare no conflict of interest. This work was supported by grants NSC 96-2311-B-001-021-MY3 from the National Science Council, Taiwan.

Ancillary