Regulation of flowering time in Arabidopsis



FIBR homepage


Flowering pathway

Gibberellin pathway



The Gibberellin transduction pathway




GAI gene and sequence:

GAI is a putative transcriptional regulator, closely related to the SCARECROW family of transcription factors. The gene encodes a predicted protein of 532 amino-acids (Peng et al.,1997), of about 59 kDa (Dill et al., 2004).

It contains an N-terminal DELLA domain, a middle VHIID domain and a C-terminal RVER domain. The VHIID domain may be a feature for transcriptional regulation (Silverstone et al., 1998). The sequence also contains two leucine zippers, as well as two putative Nuclear Localization Signal  (NLS) regions (Peng et al.,1997). However, these NLS regions are actually not necessary for nuclear localization, nor for GAI function (Fleck and Harberd, 2002).

GAI is identical to RGA2, and highly similar to RGA (Peng et al.,1997; Silverstone et al., 1998)

GAI is expressed in the whole plant:

GAI expression seems to be ubiquitous (Silverstone et al., 1998). The expression level is higher in the inflorescence, a bit lower in the rosette, and very low in the siliques. The GAI expression is also low in young seedlings (Wen and Chang, 2002).

In a more recent experiment, the expression level of GAI was measured by real-time RT-PCR in different tissues and at different times during development: GAI is expressed in all tissues tested (rosettes, roots, shoots, cauline leaves, flowers, siliques and seeds) and at all times during development. There is no obvious correlation between the level of expression and the organ or the stage of development considered (Tyler et al., 2004).

GAI protein:

  • GAI is a low-abundance protein: Immunology experiments were unable to detect the GAI protein in whole rosettes of WT plants before flowering (Fleck and Harberd, 2002).

  • GAI is a nuclear protein: The NLS motif of GAI has been identified as a putative nuclear localization signal. Silverstone et al. (1998) did not detect the presence of GAI in the nucleus, but, using GAI::GFP protein fusions, Fleck and Harberd (2002) showed the presence of the GAI protein and its localization in the nucleus.

  • The NLS domain is not necessary for the nuclear localization of GAI: But in the mutant gai-1, the gai-1 protein (lacking the NLS motif) is still localized in the nucleus (Fleck and Harberd, 2002).

  • GA treatment degrades the GAI protein: the native GAI protein as well as the mutant gai-1 protein are both present in the nucleus. Usding GFP protein fusions, Fleck and Harberd (2002) reported that GA treatment did not affect the nuclear localization, either when the gene expression was controlled by the 35S promoter or by the original GAI promoter. This was unlike what had been observed in the case of the RGA protein. However, more recent papers (Dill et al. (2004); Tyler et al., 2004) show that GAI is indeed degraded in response to GA treatment.

GAI is a redundant, negative regulator of flowering:

  • GAI function is redundant: As the plants missing the GAI function (gai-d mutants) have a WT phenotype, this means that GAI is a redundant gene in the GA transduction pathway (Peng and Harberd, 1993). RGA, which has a very similar sequence, may be mostly redundant with GAI (Peng et al.,1997).

  • GAI is a negative regulator of flowering time: Plants over-expressing GAI in the WT background are late-flowering, and have the same phenotype as the gai-1 gain-of-function mutants (Fleck and Harberd, 2002).

There are two distinct groups of gai mutants:

gai-1 is a gain-of-function mutation:

The gai mutant characterized by Wilson and Somerville (1995), was later re-named gai-1 (Peng et al.,1997).

The gai-1 mutants have many features of GA-deficient mutants: they are dwarf, have dark green leaves and do not flower in short days. Even when continuous light is applied, they still flower later than the WT.  However, contrary to ga1-3, the gai-1 mutant phenotype cannot be complemented by GA treatment (Koorneef  et al., 1985; Wilson et al., 1992; Peng and Harberd, 1993; Wilson and Somerville, 1995; Peng et al.,1997; Peng and Harberd,1997; Peng et al.,1999(2)).

The gai-1 mutation is semi-dominant, and may be a gain-of-function mutant (Koorneef  et al., 1985; Peng and Harberd, 1993).

  • gai is still expressed in the gai-1 mutants: In the gai-1 gain-of-function mutant, the gene is still expressed (Peng et al.,1997).

  • The gai-1 mutants are mostly insensitive to GAs: Unlike the ga1-3 mutants, the gai-1 mutants are insensitive to GA treatment (Koornneef et al., 1985; Wilson et al., 1992; Peng and Harberd, 1993; Wilson and Somerville, 1995). However, the double mutants gai-1 ga1-3 are more severely dwarfed than the gai-1 single mutants, and if exogenous GAs are applied, the phenotype reverts to that of the gai-1 single mutants. This means that the double mutants gai-1 ga1-3 are not completely insensitive to GAs (Koornneef et al., 1985).

  • The GA biosynthesis is up-regulated in gai-1 mutants: Endogenous GA levels are higher in the gai-1 mutants than in the WT. Also, several genes belonging to the GA biosynthesis pathway are more expressed in the gai-1 mutants than in the WT (Talon et al., 1990; Peng and Harberd, 1993; Peng et al.,1997; Peng et al.,1999(2); Thomas et al., 1999).

  • Plants over-expressing GAI have a phenotype similar to gai-1 mutants: they are dwarf, with dark green leaves, and are unresponsive to GAs. This phenotype is semi-dominant, like gai-1 (Fleck and Harberd, 2002). This differs from RGA, because over-expression of RGA does not confer any obvious phenotype (Dill et al., 2001).

Conclusion: GAI is a negative regulator of the GA response and the gai-1 mutants are not affected in GA synthesis, but in GA signal reception or transduction. 

The gai-1 allele contains a 51 bp deletion, which leads to a 17 amino-acid deletion in the N-terminal DELLA domain. This deletion confers altered properties to the gai-1 mutant protein: it is able to constitutively repress the GA transduction pathway, and resists to inactivation by the GA signal. This leads to the observed dwarf phenotype and reduced GA response (Peng et al., 1997; Dill et al., 2004). Besides, as the expression of GA biosynthesis genes is enhanced in the gai-1 mutants, the negative feedback regulation of this pathway may also be impaired.

gai loss-of-function mutants:

  • The gai-t6  mutant has a WT phenotype: A Ds transposon insertion within the gai-1 mutant allele results in a plant that has an overall WT phenotype (Peng et al., 1997; Peng et al., 2002). This new allele, called gai-t6,  features both the 51 bp N-terminal deletion of the gai-1 allele and a Ds insertion causing a C-terminal deletion in the GRAS domain (Dill et al., 2004). Actually, the gai-t6 mutants flower a bit earlier than the WT, as they have a shorter juvenile phase (Dill and Sun, 2001). However, these mutants display an increased resistance to Paclobutrazol (PAC), which indicates lower requirements for GAs. Therefore, GAI is a negative regulator of GA response (Peng and Harberd, 1993; Wilson and Somerville, 1995; Peng et al., 1997). The lack of phenotype in the gai-t6 single mutant suggests that another gene has a redundant function with GAI, likely RGA.

  • Other gai-t mutants: Peng et al. (2002) also describe the gai-t5 and gai-t7 mutants. As the gai-t6 mutants, they have been obtained by Ds transposon mutagenesis on gai-1 plants and are presumed to be loss-of-function mutants. All these mutants are homozygous for the tt1 allele. These 3 mutants contain GA levels similar to the WT, and have a phenotype similar to the WT. No Ds element was detected in the gai-t5 mutant, but the gai-1 sequence is rearranged. gai-t7 contains one Ds transposon, in its original position. However, it also contains a 2Kb insert within the gai-1 sequence. In the gai-t mutants, the gene is still transcribed, but, due to the rearrangements of the sequence, multiple transcripts are produced, all larger than the WT, and are presumed to be non-functional (Peng et al., 2002).

  • gai-d mutants have a WT phenotype: After irradiation mutagenesis on the gai-1 mutant, several gai-derivative (gai-d) mutants were isolated, with a phenotype similar to the WT (Peng and Harberd, 1993). However, only gai-d5 and gai-d1 show an increased resistance to PAC as does gai-t6 (Peng et al., 1997). The gai-d1 mutant features the original 51 bp deletion of gai-1, and an additional mutation: a premature stop codon that results in a truncated polypeptide. gai-d2, gai-d5 and gai-d7 are frame-shift mutations that also result in truncated polypeptides (Peng et al., 1997; Peng et al., 2002).


Redundancy between RGA and GAI: the DELLA protein genes

Regulation of GAI by GAs

Interaction between GAI and SOC1

Interaction between DELLA proteins and miR159

Interaction between SLY1 and DELLA proteins

Interaction between SPY and DELLA proteins