Gibberellin Acts Positively Then Negatively to Control Onset of Flower Formation in Arabidopsis

See allHide authors and affiliations

Science  09 May 2014:
Vol. 344, Issue 6184, pp. 638-641
DOI: 10.1126/science.1250498

One Hormone, Two Phases

The switch from vegetative growth to flowering in the plant Arabidopsis involves two phases—inflorescence branching and flowering. Yamaguchi et al. (p. 638) examined how the phytohormone gibberellin regulates each phase differently. First, gibberellin levels increase and stimulate production of key flowering factors, one of which degrades gibberellin. As gibberellin levels then fall, the next phase of flowering factors is released from gibberellin repression. By regulating inflorescence branching separately from flowering, this system determines overall seed yield.


The switch to reproductive development is biphasic in many plants, a feature important for optimal pollination and yield. We show that dual opposite roles of the phytohormone gibberellin underpin this phenomenon in Arabidopsis. Although gibberellin promotes termination of vegetative development, it inhibits flower formation. To overcome this effect, the transcription factor LEAFY induces expression of a gibberellin catabolism gene; consequently, increased LEAFY activity causes reduced gibberellin levels. This allows accumulation of gibberellin-sensitive DELLA proteins. The DELLA proteins are recruited by SQUAMOSA PROMOTER BINDING PROTEIN–LIKE transcription factors to regulatory regions of the floral commitment gene APETALA1 and promote APETALA1 up-regulation and floral fate synergistically with LEAFY. The two opposing functions of gibberellin may facilitate evolutionary and environmental modulation of plant inflorescence architecture.

Synchronization of the developmental transitions that lead to reproductive competence is important for species survival. Plants form new lateral organs iteratively throughout their life from the flanks of the shoot apical meristem (fig. S1) (1, 2). The type of the lateral organ produced depends on the phase of the life cycle. In Arabidopsis, rosette leaves are produced during the vegetative phase. During the reproductive phase, an inflorescence forms. Not all lateral organ primordia of the inflorescence are competent to become flowers. Those that are not instead give rise to branches subtended by cauline leaves during the first inflorescence phase (3, 4). The duration of the branch-producing first inflorescence phase determines inflorescence architecture and is critical for optimal seed set.

To gain insight into the regulation of the transition from branch to floral fate in the lateral primordia of the inflorescence, we analyzed public genome-wide binding and expression data and identified genes that are direct targets of the LEAFY (LFY) transcription factor (57) (table S1). LFY promotes flower formation (8, 9). We identified for further study the EUI-LIKE P450 A1 (ELA1) gene (fig. S2), which encodes a cytochrome P450 (10). ELA1 expression was very low in vegetative tissues but increased when flowers formed (fig. S3). On the basis of in situ hybridization and reporter studies, ELA1 was initially expressed on the abaxial side of incipient flower primordia and later along their entire circumference (Fig. 1A and fig. S3). ELA1 expression was dependent on the presence of functional LFY and on the presence of LFY-bound cis elements in the ELA1 regulatory region (Fig. 1B and fig. S4). To determine whether ELA1 is required for the switch to flower formation in long-day photoperiods, we obtained three mutant alleles (fig. S5). Consistent with its very low expression in vegetative tissues, ela1 mutants did not alter the duration of the vegetative phase (fig. S5) (11). By contrast, loss of ELA1 function significantly delayed flower formation (Fig. 1C and fig. S5). Thus, ELA1 promotes floral fate in lateral organ primordia of the inflorescence.

Fig. 1 ELA1 is a direct LFY-regulated target and promotes flower formation.

(A) ELA1 in situ hybridization in wild-type (top) and lfy-1 (bottom) inflorescences. (B) GFP reporter expression driven from wild-type (top) or LFY binding site–mutated (bottom) ELA1 promoter. (A and B) Arrows: young floral primordia. (C) Top view of wild-type (top) and ela1-1 mutant (bottom) inflorescences. The average number of cauline leaves and branches formed was significantly (P < 10−5, two-sided Student’s t test) higher in ela1-1 (4.6 ± 0.1) than in the wild type (2.8 ± 0.1). Means ± SEM. Asterisks: cauline leaves subtending branches. Scale bars: 50 μm (A) and (B), 5 mm (C).

ELA1 functions in catabolism of bioactive gibberellins that are not hydroxylated at the carbon 13 position, such as gibberellin A4 (GA4) (10, 11) (fig. S6). Through up-regulation of ELA1, LFY may reduce the levels of GA4 during flower formation. Indeed, analysis of GA4 levels by mass spectrometry after liquid chromatography revealed that this hormone is elevated in lfy null mutants and reduced in transgenic plants overexpressing LFY (Fig. 2A and fig. S7). Presence of bioactive gibberellin leads to degradation of DELLA proteins and triggers transcriptional repression of gibberellin biosynthesis genes, including GIBBERELLIN 20 OXIDASE 2 (GA20ox2) (12, 13). Consistent with their elevated levels of GA4, lfy inflorescences had reduced levels of the DELLA protein REPRESSOR OF GA1-3 (RGA) and of GA20ox2 mRNA. In contrast, LFY-overexpressing inflorescences had elevated levels of RGA protein and GA20ox2 mRNA, relative to wild-type plants (Fig. 2, B and C, and fig. S7). Moreover, removal of ELA1 activity from 35S:LFY plants restored GA20ox2 mRNA to wild-type levels (fig. S7). In agreement with their altered gibberellin levels, the Gene Ontology term “response to gibberellin stimulus” was significantly enriched (adjusted P < 0.0005) among genes differentially expressed in lfy mutants or LFY-overexpressing plants (fig. S7). Finally, LFY-overexpressing plants displayed phenotypes characteristic of gibberellin-deficient mutants, such as reduced height and increased chlorophyll content (14), that were partly rescued by application of exogenous gibberellin (fig. S7). The data indicate that LFY directs a reduction of gibberellin levels in inflorescences.

Fig. 2 LFY causes a reduction of gibberellin levels in the inflorescence.

(A) The amount of GA4 in lfy null mutant and in 35S:LFY-overexpressing relative to control inflorescences was determined by liquid chromatography–tandem mass spectrometry. ***P < 10−3 based on two-sided Student’s t test. (B) (Top) Antibody against GFP in a Western blot of lfy null mutant, wild-type (WT), or 35S:LFY inflorescences expressing pRGA:RGA-GFP. (Center) Ponceau S–stained membrane. (Bottom) Band intensity in three biological replicates (P = 0.007 for lfy versus WT, P = 0.008 for 35S:LFY versus WT; two-sided Student’s t test). (C) mRNA abundance of the GA biosynthesis gene GA20ox2 in lfy, wild-type, and 35S:LFY inflorescences. Means ± SEM.

In Arabidopsis, gibberellin promotes the transition from vegetative development to the first inflorescence phase of reproductive development (1519). However, our observations suggest that gibberellin may actually inhibit the transition to flower formation. This result was unexpected because, in general, mutants that delay onset of the first inflorescence phase also delay flower formation, whereas those that accelerate onset of the first inflorescence phase also accelerate flower formation (20) (Fig. 3A and fig. S8). We therefore examined the effect of altered gibberellin levels or response on these two transitions. Under long-day conditions, mutants deficient in gibberellin biosynthesis or gibberellin response produced more rosette leaves (delayed onset of the first inflorescence phase) (19) but fewer branches and cauline leaves (accelerated flower formation) (Fig. 3B, and fig. S8). Likewise, plants treated with the gibberellin biosynthesis inhibitor paclobutrazol (21) had more rosette leaves but fewer branches and cauline leaves (fig. S8). By contrast, plants treated with exogenous gibberellin formed fewer rosette leaves (15) and more branches and cauline leaves (Fig. 3C and fig. S8). We then decreased the gibberellin response selectively in lateral organ primordia of the inflorescence, by expressing a negative gibberellin-response regulator (the stabilized DELLA protein rgl1) from the LFY promoter. This had no effect on the duration of the vegetative phase but significantly accelerated flower formation (Fig. 3D and fig. S9). These results demonstrate that, whereas gibberellin promotes the transition from vegetative to inflorescence development, it inhibits flower formation.

Fig. 3 Gibberellin inhibits the switch to flower formation.

Duration of the vegetative phase (number of rosette leaves formed) and of the first inflorescence phase (number of cauline leaves and associated branches formed) in long-day photoperiods. White bar (control) and black bar (experimental) genotype or treatment. Means ± SEM. ***P < 10−6 based on two-sided Student’s t test. NS: not significant (P = 0.25, two-sided Student’s t test) (A) Typical effect of a mutant [fd-1 (20)] that prolongs both phases. (B) Mutants with decreased gibberellin (GA) response (gai) or levels (ga1-3). (C) Treatment with exogenous GA. (D) Reduction of the GA response specifically in incipient flower primordia (pLFY:rgl1). Throughout, matched WT genotypes were used [WT (L), wild type in the Ler accession; WT (C); wild type in the Columbia accession].

Reduced gibberellin may increase the competence of lateral organ primordia to adopt a floral fate by enhancing their responsiveness to LFY. To test this hypothesis, we used a constitutively expressed version of LFY fused to the rat glucocorticoid receptor hormone–binding domain (GR), which enables control of LFY activity by dexamethasone (5). We also took advantage of the observation that the direct LFY target APETALA1 (AP1) is only expressed in flower primordia (3), which makes AP1 expression a good proxy for floral competence. Simultaneous treatment of plants expressing LFY-GR with dexamethasone and the gibberellin biosynthesis inhibitor paclobutrazol caused increased production of AP1 in young flower primordia (Fig. 4A and fig. S9), relative to treatment with either paclobutrazol or dexamethasone alone. Via its inhibition of gibberellin biosynthesis, paclobutrazol promotes accumulation of DELLA proteins (21, 22). In agreement with this, dexamethasone treatment of plants expressing both a constitutively expressed DELLA protein fused to GR (RGA-GR) (23) and LFY-GR also caused an increase in AP1 expression (Fig. 4A and fig. S9). The slightly lower expression of AP1 in RGA-GR LFY-GR plants treated with dexamethasone compared with LFY-GR plants treated with dexamethasone and paclobutrazol may be explained by the ability of paclobutrazol to stabilize multiple DELLA proteins or by degradation of RGA-GR due to the presence of endogenous gibberellin. Although DELLA proteins are best known for their roles in transcriptional repression, they can also activate transcription (13, 22, 24). Using chromatin immunoprecipitation followed by quantitative polymerase chain reaction (ChIP-qPCR), we found that a tagged version of RGA expressed from its own promoter (12) associated with several regulatory regions in the AP1 locus (Fig. 4B and fig. S9). Moreover, dexamethasone activation of RGA-GR LFY-GR plants in the presence of protein synthesis inhibitor also caused increased AP1 induction (fig. S9). We conclude that the DELLA protein RGA potentiates LFY activity and directly promotes the transcription of AP1.

Fig. 4 Gibberellin-sensitive DELLA transcriptional co-regulators promote floral fate.

(A) qRT-PCR determination of AP1 induction relative to mock treated plants. Treatments: dexamethasone (Dex), gibberellin biosynthesis inhibitor paclobutrazol (PAC). (B) ChIP-qPCR of pRGA:RGA-HA and pSPL9:rSPL9-FLAG inflorescences using antibodies against hemagglutinin (HA) and FLAG. WT: ChIP in nontransgenic wild-type plants. NC: negative control locus. (Bottom) AP1 promoter with PCR fragments amplified. Asterisks, SPL binding motifs; triangles, LFY binding motifs. (C) qRT-PCR determination of AP1 induction relative to mock treated plants. Treatments: Dex, PAC, ethanol vapor induction of ETOH:MIR156f (26) to deplete miR156-sensitive SPL proteins, or GA to deplete DELLA proteins (12). (D) pRGA:RGA-HA association with the AP1 locus “d” region in the presence (-ETOH) or absence (+ETOH) of miR156-sensitive SPL proteins. ChIP controls were as in (B). (A) to (D) Means ± SEM. (E) Immediate switch to flower formation in plants with increased LFY (35S:LFY), SPL9 (miR156 resistant pSPL9:rSPL9-FLAG) and DELLA (PAC treatment) levels. The number of cauline leaves formed in 35S:LFY pSPL9:rSPL9-FLAG PAC plants (0.3 ± 0.1) was significantly less (P < 10−16, two-sided Student’s t-test) than that of 35S:LFY plants (2.5 ± 0.2). (F) LFY and SPL9/DELLA synergistically induce AP1.

DELLA proteins lack DNA binding domains and are thought to be recruited to target loci by sequence-specific transcription factors (22). The regulatory regions of the AP1 locus occupied by RGA were similar to those occupied by a known transcriptional activator of AP1, the microRNA miR156 target SQUAMOSA PROMOTER BINDING PROTEIN–LIKE 9 (SPL9) (25) (Fig. 4B). SPL9 can physically interact with DELLA proteins (18). We therefore next examined the possibility that RGA promotes AP1 transcription in association with SPL proteins. The AP1 induction by dexamethasone plus paclobutrazol treatment of LFY-GR was reduced when we simultaneously depleted SPL proteins with an ethanol-inducible version of MIR156 (26) (Fig. 4C), which suggested that SPL proteins contribute to this effect. In agreement with this interpretation, concurrent activation of constitutively expressed SPL9 fused to GR (SPL9-GR) and LFY-GR also led to increased AP1 induction (Fig. 4C). The synergistic effect of SPL9-GR and LFY-GR on AP1 induction most likely requires the presence of DELLA proteins, because it was strongly reduced after application of exogenous gibberellin (Fig. 4C). In addition, SPL proteins were necessary for recruitment of RGA to the AP1 regulatory region both in plants and in yeast one-hybrid assays (Fig. 4D and fig. S9). Finally, increased activity of LFY, SPL9, and DELLA proteins caused flower formation immediately after termination of the vegetative phase (Fig. 4E and fig. S9). We conclude that SPL9 recruits DELLA proteins such as RGA to the AP1 locus, where they induce AP1 expression and promote the transformation of lateral primordia into flowers (Fig. 4F).

We reveal a mechanism for sequential coupling of the biphasic transition to reproductive competence in Arabidopsis thaliana. This mechanism is based on an increase and a subsequent decrease in hormone levels. Elevated gibberellin promotes termination of the vegetative phase and increases expression of genes encoding transcription factors such as the SPLs and LFY (1618, 27, 28) (fig. S9). The subsequent reduction in gibberellin levels allows reaccumulation of transcriptional co-regulators, the DELLA proteins (12), which potentiate the ability of SPL9 (directly) and LFY (indirectly) to induce AP1 and to trigger the onset of flower formation (fig. S9). LFY initiates the reduction in gibberellin levels—which results in increased DELLA accumulation—at least in part by inducing expression of the gibberellin catabolism enzyme ELA1.

Our findings may help explain the previously paradoxical observation that gibberellin acts positively in the switch to reproductive development in most plants but negatively in some woody plant species, such as grapevine (29). In addition, our data make gibberellin a prime candidate for a “branching” factor predicted by mathematical modeling of inflorescence architectures (30). Finally, our results suggest that the degree of inflorescence branching, which determines seed yield and, thus, reproductive success, could be adjusted by altering gibberellin accumulation before the inflorescence forms or the rate of gibberellin catabolism thereafter.

Supplementary Materials

Materials and Methods

Figs. S1 to S9

Tables S1 and S2

References (3165)

References and Notes

  1. Acknowledgments: We are grateful to Wagner laboratory members and K. Gallagher, R. S. Poethig, and J. D. Wagner for critical comments. We thank R. Austin for the LFY binding motif functional depth analyses, T. Sun for pRGA:RGA-GFP and GA pathway mutant seeds, X. Chen for 35S:AlcR pAlcA:MIR156f and pRGA:RGA-HA seeds, M. Schmid for RGL1 and rgl1delta17 constructs, S. Poethig for pSPL9:rSPL9-FLAG and 35S:SPL9-GR seeds, H. Yu for ga1-3 rgl2-1 rga-2 35S:RGA-GR seeds, G. Angenent for pAP1:AP1-GFP seeds, and the Arabidopsis Biological Resource Center for insertion mutant seeds and the F6A4 bacterial artificial chromosome clone. This work was supported by NSF grants IOS 0849298 and 1257111 to D.W. and training grant T32-HD007516 (Developmental Biology) support to C.M.W.
View Abstract

Stay Connected to Science

Navigate This Article