Research Article

Integration of Spatial and Temporal Information During Floral Induction in Arabidopsis

See allHide authors and affiliations

Science  12 Aug 2005:
Vol. 309, Issue 5737, pp. 1056-1059
DOI: 10.1126/science.1114358

This article has a correction. Please see:


Flowering of Arabidopsis is regulated by several environmental and endogenous signals. An important integrator of these inputs is the FLOWERING LOCUS T (FT) gene, which encodes a small, possibly mobile protein. A primary response to floral induction is the activation of FT RNA expression in leaves. Because flowers form at a distant site, the shoot apex, these data suggest that FT primarily controls the timing of flowering. Integration of temporal and spatial information is mediated in part by the bZIP transcription factor FD, which is already expressed at the shoot apex before floral induction. A complex of FT and FD proteins in turn can activate floral identity genes such as APETALA1 (AP1).

One of the major flowering pathways in Arabidopsis, the photoperiod pathway, positively regulates activity of the nuclear protein CONSTANS (CO), which acts upstream of a graft-transmissible signal produced in leaves (13). Experiments with a dexamethasone-dependent, constitutively expressed version of CO have suggested that CO directly activates genes with diverse biochemical functions (4). These include two genes that are known to promote flowering: the transcription factor gene SUPPRESSOR OF OVEREXPRESSION OF CONSTANS 1 (SOC1) and FLOWERING LOCUS T (FT), which encodes a small globular protein related to the floral repressor TERMINAL FLOWER 1 (TFL1) (57). In addition, ACS10 and P5CS2, structural genes for ethylene and proline biosynthetic enzymes, were identified as potential CO targets in these experiments (4).

FT is the major primary target of CO in leaves. Because the results with CO gain-of-function alleles had suggested that the CO-induced signal is complex, we used microarray analyses to identify targets of endogenous CO in leaves. To induce endogenous CO activity, we activated CO protein with light (1). We grew all plants in 8-hour short days. On the day of the experiment, we exposed the experimental group to 16 hours of light and the control group to 8 hours of light followed by 8 hours of darkness. We harvested leaves at the end of the 16-hour period. Differentially expressed genes were identified with a combination of per-gene variance [logit t test, P < 0.025 (8)] and common variance (>1.5× change).

Of 2000 genes that are activated or repressed after exposure to a single long day, merely three genes are responsive to long days in wild-type plants and at the same time differentially expressed between wild-type plants and co mutants in long days. Of these, only one gene does not respond at all to long days in a co background: FT (Fig. 1, A and B). In contrast, SOC1, ACS10, and P5CS2 expression is independent of CO after exposure to a single long day (Fig. 1B), suggesting either that these genes respond only to higher levels of CO or that they respond to CO in tissues other than the leaf. In plants grown in continuous light, FT expression is much higher in leaves than at the shoot apex (Fig. 1C), consistent with leaves being the primary site of FT activation.

Fig. 1.

Expression analyses using microarrays and real-time RT-PCR. (A) Venn diagram of Affymetrix ATH1 microarray data (ArrayExpress experiment E-TABM-2) showing that only a single gene, FT, is differentially affected by the co mutation and the shift from short days (SD) to long days (LD), indicated in bold. (B) Microarray values (averages of three replicates) for FT and three other genes previously identified as potentially direct CO targets (4). (C) Expression patterns of floral regulators, from the AtGenExpress microarray atlas (26). From left to right, tissues are embryonic leaves (cotyledons, 7 days) and rosette leaves 2, 4, 6, 8, 10, and 12 (17 days); petiole, proximal, and distal parts of leaf 7 (3 weeks); vegetative apex (7 days), transition apex (14 days), flowering apex (21 days), and stage-9 flowers. FT is expressed most strongly in the oldest leaves, and within a leaf, in the distal, most differentiated part. SOC1 is expressed more strongly in leaves than in vegetative apices, but induced upon flowering at the shoot apex. Light color indicates an “absence” call by the Affymetrix MAS algorithm in all three replicates. (D) Real-time RT-PCR analysis of FT and AP1 in leaf 3 and shoot apex of 20-day-old wild-type plants.

The finding that FT is the major target of CO in leaves is in agreement with our observation, also based on global expression profiles, that FT is the major output of CO at the shoot apex and that early floral markers such as SOC1 and FRUITFULL (FUL) are similarly dependent on both CO and FT (9). Thus, the initial signal acting downstream of CO in leaves might be less complex than previously thought.

The bZIP protein FD is required for FT activity. To understand how FT activity is transduced, we searched for FT interactors. In a yeast two-hybrid screen, we isolated two closely related bZIP transcription factors, At2g17770 and At4g35900 (fig. S1), the Arabidopsis orthologs of tomato SIP8/SPGB, which interacts with Arabidopsis FT and TFL1, as well as the tomato TFL1 homolog SELF-PRUNING (10). The available collections of transferred DNA (T-DNA) insertion lines do not contain any At2g17770 alleles, but there are four different alleles but of At4g35900 (11). The late-flowering phenotype of these lines (Fig. 2A and Table 1) is rescued by a minigene, indicating that At4g35900 promotes flowering (fig. S2). The only late-flowering mutant described for this region of the genome is fd-1 (12), and complementation crosses showed FD and At4g35900 to be allelic. We therefore designated our reference insertion allele of At4g35900 as fd-2. By their phenotype and genetic interactions, FD and FT have been placed in the same class of floral regulators (12, 13). A special property of ft compared with several other late-flowering mutants is that ft leafy (lfy) double mutants never form flowers, indicating partially redundant functions of FT and LFY (14). fd-2 lfy-12 double mutants also lack flowers (Fig. 2, B and C). One of the outputs of combined LFY and FT activities is activation of the floral identity gene AP1. Again consistent with a role of FD in the FT pathway, activation of AP1 is delayed in fd-2 mutants (Fig. 3, B and C). Finally, FD is required for FT activity, such that the early flowering caused by FT overexpression is partially suppressed by fd-2 (Fig. 2A and Table 1).

Fig. 2.

Phenotypes of plants grown in long days. (A) Partial suppression of 35S:FT by fd-2. From left: 30-day-old 35S:FT, 35S:FT fd-2, wild-type, and fd-2 plants. (B) lfy-12 mutants produce abnormal flowers (f). (C) lfy-12 fd-2 plants produce leaves (l) instead. (D) An 18-day-old 35S:FT (left) compared with 35S:FD 35S:FT plant. (E) Cotyledons (c) and two normal leaves (l) are seen in a 10-day-old 35S:FT tfl1 ft plant (left). The first two leaves of 35S:FT-VP16 tfl1 ft are curled up and surround a single terminal flower (right). Plants in each panel were grown simultaneously. Scale bars, 1 cm (A), 5 mm [(B) to (D)], and 1 mm (E).

Fig. 3.

In situ hybridization patterns of FD and AP1 at the shoot apex. Plants transferred from short to long days were sampled at the end of each day. “0d” indicates last short day. A complete series until day 7 after transfer is shown in fig. S3. Day 4 after transfer (“4d”) was the first day on which a robust AP1 signal was apparent in the wild-type plant. (A) FD expression at the shoot apex of wild-type plants. Note transient expression in floral anlagen (a), with rapidly declining expression in floral primordia (p). (B) AP1 expression in wild-type plants. AP1 is detected in floral primordia and subsequent stages. (C) Delayed and weaker activation of AP1 expression in fd-2 plants. A range of floral stages with corresponding variation in AP1 expression was found in each sample. An average example is shown for each day and genotype.

Table 1.

Flowering times of mutant and transgenic plants. The 95% confidence interval is 2 × the standard error of the mean.

Genotype Leaves 95% confidence interval Range n
Experiment 1 (long days)
Wild type 16.3 ±0.7 15-18 12
fd-2View inline 23.5 ±0.6 18-28 54
Experiment 2 (long days)
Wild type 15.7 ±0.8 10-19 23
35S:FT 5.3 ±0.2 4-8 75
fd-2 22.8 ±1.5 20-28 10
fd-2 35S:FT 10.7 ±0.7 8-13 15
Experiment 3 (long days)
Wild type 15.6 ±1.0 13-18 14
35S:FD 10.0 ±0.4 8-12 24
Experiment 4 (short days)
Wild type 50.5 ±0.4 48-53 30
35S:FD 45.6 ±0.8 43-50 21
Experiment 5 (long days)
Wild type 13.6 ±0.5 12-15 12
FD:FT (T1 plants) 5.3 ±0.3 5-6 10
  • View inline* At4g35900 insertion allele SALK_013288.

  • In contrast to FT, FD is highly expressed at the shoot apex (Fig. 1C). Specifically, FD RNA was observed in leaf and floral anlagen. Its levels decrease soon after floral primordia start to express AP1 (Fig. 3, A and B, and fig. S3). Expression in floral anlagen appears to be slightly higher than in leaf anlagen, and a modest increase after floral induction is also seen with microarrays, both in wild-type plants and ft mutants (fig. S4).

    It thus appears that FT, which is induced most strongly in leaves, integrates environmental signals that allow for correct timing of flower initiation, whereas FD, which is most strongly expressed at the shoot apex, provides spatial specificity for FT action. Because FD is already expressed at the shoot apex before floral induction, FD should be less limiting for flowering than FT. Indeed, the early flowering of lines that overexpress FD is weaker than that of 35S:FT plants (Table 1). In addition to early flowering, 35S:FD plants are small with curled leaves (Fig. 2D). Both phenotypes largely disappear in short days, when little FT activity is present (Table 1). That FT and FD are mutually limiting is further supported by the synergistic phenotype of 35S:FT 35S:FD plants (Fig. 2D).

    FT and FD are sufficient to activate the expression of floral marker genes. The strong leaf phenotype of 35S:FD plants, which is reminiscent of plants that overexpress AP1 or FUL (4, 15), suggested that combined FT and FD activities are largely sufficient for triggering the floral transcriptional program. Consistent with this hypothesis, AP1 and FUL are ectopically activated in the leaves of 35S:FD plants, but only during long days (Fig. 4, A and B).

    Fig. 4.

    Response of floral meristem-identity genes to FD and FT activity. (A) Expression of floral meristem-identity genes in leaves of 26-day-old, long-day-grown wild-type [left; two replicates, light and dark blue (mostly too small to be visible at this scale)] and 35S:FD plants (right; three replicates, yellow, orange, and red), as measured by real-time RT-PCR. LFY does not respond to 35S:FD. (B) Photoperiod-dependent activation of a FUL:GUS reporter in leaves of 35S:FD. (C) Activation of FT and AP1 in leaves of 35S:FD plants in response to transfer to long days. Abscissa indicates chronological time, starting at 16:00 hours on the first day (end of short day). There appears to be a slight delay in AP1 activation relative to FT activation. (D) AP1:GUS reporter activity is only induced in the part of a leaf from a 35S:FD plant that has been exposed to long days. Dashed box indicates covered portion of the leaf. An uncovered leaf is shown below. (E) Expression of AP1 and FUL in 18-day-old 35S:FT tfl1 ft (left; two replicates, light and dark blue) and 35S:FT-VP16 tfl1 ft plants (right; two replicates, yellow and red). (F) Mapping of an FD-response element in the AP1 promoter to a region that encompasses binding sites for LFY (18), SPL3 (27), and a C-box, a motif that can be bound by bZIP proteins (28). (G) Immunoprecipitation of chromatin from 35S:FD plants with antibodies to FT, followed by real-time PCR amplification of control HSF1 promoter (blue) or AP1 promoter (yellow) sequences (11). PCR cycle difference between precipitate and supernatant is plotted; a lower number of cycles indicates relative enrichment in the immunoprecipitate compared with the supernatant. Numbers on top indicate fold enrichment of AP1 compared with HSF1 sequences, by assuming PCR efficiency of 1.8-fold amplification per cycle. Results are averages of at least three replicates. Tissue was harvested at 23:15 hours on the second day after transfer from short to long days.

    Two additional experiments demonstrated that ectopic activation of floral markers requires the combined presence of FT and FD. First, we manipulated the temporal expression of FT in 35S:FD plants by transferring them from short to long days. In this experimental set-up, activation of AP1 closely follows that of FT (Fig. 4C), indicating that AP1 is an early target of combined FT and FD activity. Second, we manipulated the spatial pattern of FT expression in 35S:FD plants containing an AP1:GUS reporter by partially covering leaves with aluminum foil. FT itself is cell-autonomously activated in response to light-dependent activation of the upstream regulator CO (1, 2, 16, 17). Upon transfer of 35S:FD AP1:GUS plants from short to long days, which induces FT expression in leaves, AP1:GUS is strongly activated in the light-exposed but not the shaded part of a leaf (Fig. 4D).

    How, then, does FT control FD activity? FT might affect posttranslational modification of FD, which in turn may alter its DNA binding capacity or transcriptional activity. Alternatively, FT itself might provide transcriptional activation potential. To test the latter scenario, we added the potent VP16 activation domain to FT and introduced a 35S:FT-VP16 construct into ft tfl1 double mutants. These plants flower even earlier than 35S:FT ft tfl1, and have a phenotype similar to that of 35S:FT 35S:LFY (5, 6), with a single terminal flower between the first two leaves (Fig. 2E). AP1 and FUL are expressed at much higher levels in 35S:FT-VP16 than in 35S:FT plants, confirming that the VP16 domain alters FT activity (Fig. 4E).

    That adding a transcriptional activation domain to FT changes its activity supports a model in which FT acts in the nucleus as part of a transcriptional complex with FD. To generate additional support for this scenario, we mapped the FD response element in the AP1 promoter to a 130–base pair region (Fig. 4F), which also encompasses the previously identified binding site of LFY, another AP1 activator (18). The LFY binding site is not required for ectopic activation of the AP1 reporter, confirming an LFY-independent effect of FD on the floral transcriptional program. To test whether an FT/FD complex directly interacts with the FD response element, we performed chromatin immunoprecipitation, using 35S:FD plants and FT antibodies (11). AP1 promoter sequences are enriched in long days, in which FT expression is induced, but not in short days (Fig. 4G). An enrichment of AP1 sequences is not seen with leaf material of wild-type plants. Thus, FT protein is recruited to the FD-response element in the AP1 promoter in an FD-dependent manner.

    Roles of FT and FD in mediating floral inductive signals. It has long been known that day length is detected in leaves, which in turn release a systemic signal dubbed “florigen” (19, 20). Our data suggest a model in which FT directly regulates the activity of FD at the shoot apex by formation of a transcriptional complex, even though induction of FT RNA expression in response to photoperiod occurs predominantly in leaves. Consistent with FT being the primary target of CO in leaves, CO overexpression causes activation of the FT promoter in the phloem of leaves (2, 21). The most parsimonious hypothesis to integrate these findings is that induction of FT RNA in the leaf phloem leads to the presence of FT protein at the shoot apex. Conventional means are insufficient to detect FT protein at the shoot apex, nor can FT RNA be detected at the shoot apex on microarrays (Fig. 1C). More sensitive real-time reverse transcription polymerase chain reaction (RT-PCR) analyses indicate a merely threefold increase of FT RNA at the shoot apex of plants grown in the long-day compared with the short-day condition; this contrasts with the approximately 70-fold difference in leaves (Fig. 1D). In addition, transgenic experiments show that FT activity at the shoot apex needs to be tightly controlled in order to prevent precocious flowering: When FT is expressed directly in the FD domain with the use of FD regulatory sequences, early flowering ensues, demonstrating exquisite sensitivity of the shoot apex to local FT levels (Table 1).

    Our data do not necessarily imply that FT RNA or FT protein themselves are the long-distance signal. They are similarly consistent with a relay mechanism in which FT induces a mobile signal in leaves, which in turn leads to FT activation in secondary tissues near or at the shoot apex. Such an intermediate step may be required to translate the circadian activation of FT in leaves into a more constant and finely tuned signal at the shoot apex. Secondary activation of FT expression at the shoot apex may also explain why this tissue has an important function in perceiving the effects of vernalization, the transient exposure to winterlike temperatures that promotes flowering in many plants (22, 23).

    Finally, FT and FD are unlikely to be the only factors that mediate the effects of upstream regulators such as CO on flowering. Rather, we envision an intricate system of flowering promoting and repressing activities that act at the shoot apex to provide a robust flowering response. This system is likely to include the multitude of MADS box genes affecting flowering, such as SOC1 and others, as well as the targets of the miR156 and miR172 microRNAs (2325). How the FD/FT module interacts with these other factors and their role in transmitting the mobile flower-inducing signal are important avenues for further research.

    Supporting Online Material

    Materials and Methods

    Figs. S1 to S4


    References and Notes

    View Abstract

    Stay Connected to Science

    Navigate This Article