Control of Circadian Rhythms and Photoperiodic Flowering by the Arabidopsis GIGANTEA Gene

See allHide authors and affiliations

Science  03 Sep 1999:
Vol. 285, Issue 5433, pp. 1579-1582
DOI: 10.1126/science.285.5433.1579


Photoperiodic responses in plants include flowering that is day-length–dependent. Mutations in the Arabidopsis thaliana GIGANTEA (GI) gene cause photoperiod-insensitive flowering and alteration of circadian rhythms. The GI gene encodes a protein containing six putative transmembrane domains. Circadian expression patterns of the GI gene and the clock-associated genes, LHY and CCA1, are altered in gi mutants, showing that GI is required for maintaining circadian amplitude and appropriate period length of these genes. Thegi-1 mutation also affects light signaling to the clock, which suggests that GI participates in a feedback loop of the plant circadian system.

The circadian clock system is a self-sustained biological oscillator with a period of ∼24 hours that operates ubiquitously in animals, plants, and microorganisms (1, 2) and controls a wide range of rhythmic processes (3, 4). In plants, the circadian clock controls daily changes of photosynthetic activities, leaflet movement, cell growth, and expression of several genes, including the chlorophyll a/b binding protein (CAB) genes (2, 5). Several components regulating the circadian system have been genetically defined in Arabidopsis thaliana (6, 7). However, molecular components controlling the plant circadian system are largely unknown, except for the two clock-associated factors, LHY and CCA1 (8, 9).

The ability to perceive changes in day length (7, 10–12), photoperiodism, is essential for organisms to recognize seasonal changes and is associated with the circadian clock system (4,7–9). Arabidopsis is a facultative long-day plant, flowering in a fewer number of days under long photoperiods than under short photoperiods. Certain alleles of the Arabidopsis gimutants are insensitive to photoperiod (11, 13, 14). Here, we report an important role for GI in plant circadian function and molecular nature of the Arabidopsis gi mutations.

Cotyledons and leaves of Arabidopsis show circadian movement, rising and falling in relative position during subjective night and day, respectively (7). Measurement of leaf movement rhythms (15) revealed that both the gi-1and gi-2 mutants have shorter circadian periods than wild-type plants (Fig. 1, A and B, andTable 1).

Figure 1

Representative traces of (A andB) circadian rhythms of leaf movement and (C)cab2::luc expression in wild-type (WT), gi-1, and gi-2 mutant backgrounds under constant white light (LL). Plants were germinated and grown for 6 to 7 days under light and dark cycles as previously described (25) before they were transferred to continuous white light for >110 hours. Leaf movements were recorded (7), and luminescence assays were conducted (25) to obtain period estimates as previously described (6, 27). Individual bioluminescent F3 lines homozygous for thegi-1 (gi-1 2CAC) or gi-2 (gi-2 2CAC) mutations were isolated from crosses to acab2::luc-expressing line (2CAC, C24 ecotype) (6, 28). WT (Col), Columbia ecotype; WT (Col/2CAC),cab::luc introgressed five times into the Columbia ecotype.

Table 1

Circadian periods of leaf movement andcab2::luc expression in the gi mutants. Period length estimates (variance-weighted mean period ± variance-weighted SD) were obtained as described (6, 27).n, number of rhythmic leaves (leaf movement) or seedlings (cab2::luc) contributing to a mean; Col, Columbia ecotype; gi-1 F2 and gi-2F2 indicate the segregating population from which the indicated genotypic subclass was sampled (28); ND, not determined.

View this table:

Expression of the CAB2 gene is under the control of the circadian clock in Arabidopsis. To characterize the effects of the gi mutations on the circadian system, we crossed the clock-controlled reporter construct,cab2::luciferase (cab2::luc) (6), into the gi-1 and gi-2backgrounds. Both mutants alter the circadian period ofcab2::luc expression (Fig. 1C and Table 1) (15). The period is substantially shorter in the homozygousgi-1 mutant than in the wild type, and this effect is dependent on gene dosage (Fig. 2A andTable 1). In contrast, the gi-2 mutation causes the period to lengthen and is fully recessive (Fig. 2B and Table 1). Luminescence cycling in the gi-2 background remained in phase with the wild type for up to 60 hours after transfer to continuous white light (LL), after which the period lengthened (Fig. 1C) (15,16). There was no apparent lag in the period-shortening effects of the gi-1 mutation (Fig. 1C). Both mutations caused an abnormally rapid damping of rhythm amplitude.

Figure 2

Period length distribution in F2 populations segregating for (A)gi-1 or (B) gi-2 (28). Plants were entrained and assayed as described in Fig. 1. Period bins are labeled with the upper bound.

We cloned the GI gene with a map-based approach. TheGI locus is linked to the THIAMINE REQUIRING 1(TH1) locus on chromosome 1 (17). We thus used the TH1 gene to screen an A. thaliana bacterial artificial chromosome (BAC) library, identifying the BAC clone T23L3 (18). Subsequent probing with T23L3 identified two contiguous BAC clones, T14K14 and T22J18 (19). The gi-1 and gi-2 alleles were generated by x-ray mutagenesis (13), which frequently generates deletion mutants. When wild-type (Columbia), gi-1,and gi-2 genomic DNA were digested with Stu I and probed with theT22J18 clone, two restriction fragments (7.1 and 1.1 kb, respectively) detected in the wild-type and gi-2 lines appeared as a single 8.2-kb band in the gi-1 mutant (19). End sequences of the 7.1- and 1.1-kb fragments (20) matched a genomic sequence (GenBank accession numberY12227) that contains a predicted open reading frame (ORF) of 1167 amino acids. Sequence comparisons in this region between the wild-type,gi-1, and gi-2 lines revealed small deletions within the predicted ORF in both mutants (19).

Using a probe generated by polymerase chain reaction (PCR), we identified a full-length cDNA clone of 3522 base pairs (bp) (GenBank accession number AF105064) that codes for a 1173–amino acid polypeptide (19, 21, 22). The gi-1 mutant has a 5-bp deletion, resulting in a putative premature translation stop with a loss of the 171 COOH-terminal amino acids (19). Thegi-2 mutant has a 7-bp deletion, resulting in a putative premature translation stop with a loss of all but the 142 NH2-terminal amino acids of the polypeptide and an addition of 16 new amino acids at the COOH-terminus.

The flowering time phenotype suggests that gi-1 is a weak allele, still possessing some photoperiodic responses, whereas thegi-2 mutation acts as a strong allele, showing complete insensitivity to photoperiod (11, 14). The sequence data are consistent with these phenotypes. A loss of most of the coding region in the gi-2 mutant is likely to produce a nonfunctional polypeptide. In contrast, the gi-1 mutant lacks only a small portion of the COOH-terminus of GI, and a peptide with partial function or a modified activity may still be produced.GI appears to be a single-copy gene and is expressed in all the organs examined (19). The predicted normal polypeptide contains six potential membrane-spanning domains, strongly suggesting that GI is a membrane protein (19). A database search with the predicted peptide sequence did not reveal any related sequences.

We tested the effects of the gi mutations on GIexpression and on two clock-associated genes, CCA1 andLHY (8, 9). After entrainment under 12 hours of light and 12 hours of darkness (LD), a strong free-running oscillation of GI gene expression persisted in LL in the wild type (Fig. 3A), with peak expression in the subjective evening and with a mean estimated period length of 24.9 ± 0.5 hours (23). In contrast, the gi-1 mutation confers a shorter period (21.3 ± 1.4 hours) than the wild type, whereas in gi-2 the abundance of GI transcript cycles at low amplitude with a longer period length (26.3 ± 3.2 hours) (23). Unlike other clock component genes that negatively regulate their own expression (3, 24), the expression level of the GI gene is much lower in bothgi alleles than in the wild type, suggesting that GI positively regulates its own expression.

Figure 3

Circadian expression of theGI, CCA1, and LHY genes in (A) continuous light or (B) continuous darkness in wild type, gi-1, and gi-2. Plants were entrained in 12 hours of light and 12 hours of dark cycles for 2 weeks before transfer to continuous light (LL) or dark (DD). The open and hatched boxes in (A) represent the subjective day and night, respectively. The solid and hatched boxes in (B) represent the subjective night and day, respectively. Seedlings were sampled every 3 hours, and RNA expression levels were measured by dot blotting. TheBrassica BGB1 clone (29) corresponding to theArabidopsis AtarcA gene was used as a control. Values are normalized to the lowest value of the wild-type samples in each set. The LL and DD experiments were performed three times and twice, respectively. Representative data are shown.

The Arabidopsis CCA1 and LHY genes encoding the MYB-related transcription factors that may be central to normal circadian function (8, 9) show circadian expression patterns. The circadian expression patterns of both genes were altered in the gi mutants (Fig. 3A). Consistent with thecab2::luc results (Table 1 and Fig. 1C), thegi-1 mutation shortened the free-running period (23) of both cycling transcripts (22.7 ± 1.2 and 21.8 ± 0.9 hours for CCA1 and LHY, respectively), relative to wild type (24.5 ± 0.4 and 24.6 ± 0.4 hours for CCA1 and LHY, respectively), and also reduced amplitude. A greater reduction in amplitude was observed for transcript levels of both genes in the gi-2 background (Fig. 3A). This resulted in period estimates (25.7 ± 2.6 and 25.3 ± 2.2 hours for CCA1 and LHY, respectively) that are less precise but consistent with the effects ofgi-2 on cab2::luc expression (Table 1and Fig. 1C).

In sum, the gi-1 mutation shortened period lengths of leaf movement, cab2::luc luminescence, and RNA transcript abundance rhythms. In contrast, gi-2 caused a shortening of the leaf movement period but caused a gradual lengthening of the luminescence and RNA transcript abundance rhythms. Independent circadian oscillators might separately control different outputs, and GI may affect input pathways feeding into these clocks. Alternatively, the input pathways to a common clock may differ between tissues (for example, leaf blade and petiole), and GI may contribute to these pathways differently in each location.

Of the three components of a circadian system [an input pathway (or pathways), central oscillator, and output pathways (1, 3)], GI is unlikely to be a central oscillator component because the putative null mutation (gi-2) does not abolish rhythmicity but alters period and reduces amplitude. In wild-typeArabidopsis, the free-running period of a circadian clock lengthens with decreasing light intensity in LL (25). The rate of period length increase with decreasing light is less ingi-1 than in the wild type (Fig. 4). This results in an increased relative difference in period lengths between the wild type and mutant at fluence rates between 2 to 5 μmol m−2 s−1, consistent with the notion that gi-1 increases the sensitivity of the circadian system to the controlling effects of light on period length. Only at the lowest intensity tested (0.6 μmol m−2 s−1) was there a lengthening of period ingi-1, demonstrating that the mutation has shifted the threshold at which light becomes limiting (16,25). This change in the fluence-dependent period lengthening in gi-1 suggests that GI functions, at least in part, in controlling light signaling to the clock. When we examined the circadian expression pattern of the GI gene in continuous darkness (DD) after entrainment in LD (Fig. 3B), the effect of thegi-2 mutation on the amplitude and sustainability was less severe than that under LL (Fig. 3A). The period-shortening effect ofgi-1 on cycling of the GI transcript was also less severe in DD than in LL (Fig. 3B) (26). This light-dependent conditional effect of the gi mutations further supports the notion that GI functions in a light input pathway.

Figure 4

Effect of red-light fluence rate on free-running period length of cab2::luc expression in the wild type and the gi-1 mutant. Seedlings were entrained as described (25), then transferred for >110 hours to continuous red light (600 to 700 nm) at the fluence rates indicated. Period estimates were obtained as previously described (6, 27). Error bars indicate ±SEM (n = 8 through 20). Representative data are shown from two independent experiments with similar results.

However, the cyclic expression of the GI transcript shows that it is also under circadian control. In the absence of GI in the gi-2 mutant, the cyclic expression of the GIgene as well as known clock-associated genes and simple output genes (for example, CAB) shows a reduced amplitude and a gradual increase of period length. Together, these results suggest a model in which GI defines an outer feedback loop that is required to maintain circadian amplitude and proper period length. In this view, yet unidentified circadian clock components would form the core of an oscillator but would be unable to sustain a sufficient amplitude or a proper period length under continuous light to act as a robust timer: An outer feedback loop involving GI could be a mechanism to sustain the necessary amplitude and period length.

  • * These authors contributed equally to this work.

  • To whom correspondence should be addressed. E-mail: hgn{at}


Stay Connected to Science

Navigate This Article