Abstract
The toc1 mutation causes shortened circadian rhythms in light-grown Arabidopsis plants. Here, we report the same toc1 effect in the absence of light input to the clock. We also show that TOC1 controls photoperiodic flowering response through clock function. The TOC1 gene was isolated and found to encode a nuclear protein containing an atypical response regulator receiver domain and two motifs that suggest a role in transcriptional regulation: a basic motif conserved within theCONSTANS family of transcription factors and an acidic domain. TOC1 is itself circadianly regulated and participates in a feedback loop to control its own expression.
The endogenous circadian clock enables organisms to anticipate and adapt to daily variations in the environment and to temporally coordinate internal processes. In animals, fungi, and bacteria, genetic screens for altered circadian rhythms have revealed molecular clock components. The generally conserved core mechanism consists of autoregulatory transcriptional loops in which positive factors act on genes encoding negative factors that in turn feed back to block their own expression (1). Although plant models have proven valuable for understanding circadian input and output pathways, our understanding of processes at the core of the plant circadian system is lacking.
We therefore executed a screen for rhythm mutants inArabidopsis, from which we identified the toc1(timing of CAB expression) mutant (2). The defining phenotype is a shortened period of luciferase-reportedCAB gene expression (∼21 hours, versus ∼24.5 hours in the wild type) under constant light conditions (LL). All clock phenotypes tested are similarly affected by the toc1-1mutation, which is semidominant, as are mutant alleles of diverse clock genes (1–3). Moreover, the effects of the toc1-1mutation are specific to the clock system, with no defects seen, for instance, in clock-independent light responses (3). This is noteworthy because disruption of photoreceptors and phototransduction components that participate in clock entrainment can alter period in LL (4, 5). However, perturbations of these components produce specific, differential effects depending on the quality and quantity of light, whereas the toc1-1 effect is essentially the same in all light conditions (3). To further address this issue, we assayed the bioluminescence rhythm of toc1-1 and wild-type seedlings during extended dark incubation (DD) using a new reporter,ccr2::luc. CCR2(COLD–CIRCADIAN RHYTHM–RNA-BINDING 2) is a clock-controlled gene whose LL expression rhythm is shortened by thetoc1-1 mutation (6). Theccr2::luc reporter (including a luciferase gene fusion) reveals that toc1-1 has a similar effect on the period of gene expression in DD (Fig. 1), consistent with a role for TOC1 outside of light input pathway(s) to the clock.
Bioluminescence rhythms from toc1-1 and wild-type seedlings in constant darkness (DD).ccr2::luc transgenic seedlings (7) were grown in 12:12 LD for 8 days before transfer to DD. Bioluminescence was recorded at the indicated times (7). Traces represent averages of 21 to 23 seedlings from each line. Period estimates (variance-weighted means ± variance-weighted SD) for each line were calculated as described (26, 27): wild type = 27.5 ± 1.16 hours, toc1-1 = 22.3 ± 0.39 hours.
Mutation of TOC1 also affects photoperiodic regulation of floral induction. Wild-type Arabidopsis flowers earlier in long days [16 hours light, 8 hours dark (16:8 LD)] than in short days (8:16 LD), but this differential response is greatly reduced in toc1-1 (3). This phenotype is likely the result of clock-based misinterpretation of photoperiodic information in toc1-1 rather than direct effects oftoc1-1 on floral induction pathways. To test this possibility, we measured the transition to flowering of toc1-1lines grown in LD cycles of 21 hours total duration, where the environmental period (T) more closely matched the period of the endogenous clock (τ) (Fig. 2). Correct photoperiodic response was restored in toc1-1 plants grown in this regime, where toc1-1 plants flowered much later in short days (7:14 LD) than in long days (14:7 LD). The toc1-1 flowering defect therefore can be fully explained by its circadian defect. The cause is not simply incorrect measurement of light or dark intervals: Mutant plants given 7 hours light in a 24-hour period (7:17 LD) also flowered early (Fig. 2). These data, combined with results of experiments measuring gene expression intoc1-1 and wild-type plants entrained to alteredT cycles (3), also suggest a possible mechanism underlying this defect: incorrect modulation of phase and/or waveform of clock-controlled regulatory factors when τ varies greatly fromT.
Photoperiodic response of toc1-1 plants in LD cycles of different period. Wild-type seedlings [ecotype Landsberg (Laer)] or seedlings homozygous for the toc1-1 (C24) allele introgressed into Laer (toc1-1/Laer) were grown in various photocycles as described (3, 28) (n = 12 to 22 for each condition). Initiation of flowering was defined as rosette leaf number at flower bolt height of 1 cm. Open bars, long days (14:7); solid bars, short days (7:14); striped bars, short days (7:17). Error bars denote SEM.
To further investigate its role in the Arabidopsis circadian system, we isolated the TOC1 gene. Genetic mapping delimitedTOC1 to a 78-kb interval on chromosome 5 (7). Sequencing of toc1-1 and wild-type DNA from this region identified a missense mutation in thetoc1-1 allele of one candidate gene (Fig. 3). We confirmed this gene as TOC1 by identifying a nucleotide change in a second mutant allele, toc1-2 (Fig. 3). Thecab2::luc rhythm defect of toc1-2homozygotes is similar to that of toc1-1 homozygotes (LL period ≈ 22 hours) (8); however, the toc1-2allele appears to be recessive (1). The toc1-2mutation changes the last nucleotide in exon 1, and one result is preferential splicing at a site 13 nucleotides 3′ to the normal donor site with a concomitant reduction in correctly spliced transcript (Fig. 3, A and E) (7). Quantitative reverse-transcription polymerase chain reaction (RT-PCR) revealed that correctly spliced transcripts are 6% of the total in toc1-2, versus 96% of the total in the wild type (Fig. 3E). The incorrectly spliced transcript would encode a truncated protein of only 59 residues. Taken together, these results suggest that toc1-2 is a hypomorph, the loss-of-function phenotype is short period, and the semidominanttoc1-1 allele is therefore likely an antimorph.
Structure of TOC1 gene and gene products. (A) Gene organization of TOC1. Black boxes represent exons. Positions of start and stop codons and mutations in TOC1 are indicated. (B) Predicted protein structure of TOC1. (C) Comparison of receiver domains of TOC1 (GenBank accession no. AF272039), TL1 (AF272040), representativeArabidopsis response regulator homologs ARR2 (AB016472) (type B) and ARR4 (AF057282) (type A), and CheY (M13463) (9, 10,29). Residues identical to TOC1 are in black boxes; similar residues are in gray boxes. An asterisk indicates the residue mutated in toc1-2. (D) Comparison of CCT motifs of TOC1, TL1, Arabidopsis CO (CAA64407), Arabidopsis COL1 (Y10555), Arabidopsis COL2 (L81119), andBrassica CO9a (Bn CO: AF016011). Sequences are highlighted as in (C). An asterisk indicates the residue mutated intoc1-1; # and + indicate residues mutated inco5 and co7, respectively (15). (E) RT-PCR analysis of TOC1 transcripts. Autoradiograph from polyacrylamide gel electrophoresis (PAGE) analysis of products from quantitative RT-PCR (7) is shown at the left. A cartoon represents RT-PCR products from the two splice products. The lower band derives from the correctly spliced product containing the complete open reading frame. Nucleotide sequences surrounding the splice junctions (slashes) from each species from thetoc1-2 allele are shown at the right. Nucleotides included in the species represented in the upper band are underlined. The mutated nucleotide and amino acid residues changed in toc1-2are boxed in gray. Single-letter abbreviations for amino acid residues: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.
The predicted TOC1 protein contains several notable features. At the NH2-terminus, TOC1 contains a motif similar to the receiver domain of response regulators from two-component signal transduction systems (Fig. 3, B and C). Typically in these systems, the first component, a sensor kinase, perceives changes in environmental conditions and propagates signals by autophosphorylation of a conserved histidine residue (9). This phosphate is subsequently transferred to an aspartate within the receiver domain of a response regulator, which then effects the response, often a change in the transcriptional state of target genes (9). However, TOC1 would be unique among characterized response regulators in that two of three invariant residues required for normal response are substituted. In CheY, an Escherichia coli response regulator, Asp57 (D1 in Fig. 3C), is the normal site of phosphorylation by its cognate sensor kinase (CheA), and Asp13 and Lys109 (D2and K in Fig. 3C) are also crucial for normal phosphorelay and function (9). Residues corresponding to D1 and D2 are changed to Glu and Asn, respectively, in TOC1. About 14 response regulator homologs have been identified inArabidopsis, which can be classified as type A or type B on the basis of primary structure (Fig. 3C) (10). TOC1defines a new class because of the atypical primary structure of its receiver-like domain. We have identified several TOC1-LIKE(TL) sequences in the Arabidopsis genome (Fig. 3, C and D) (11) that are defined by this atypical motif and another feature, a distinctive COOH-terminal motif first identified within the CONSTANS family of plant transcription factors. The CONSTANS (CO) locus controls photoperiodic flowering response (12). Structurally, the CONSTANS-LIKE (COL) family is typified by the presence of this highly conserved COOH-terminal domain, herein termed the CCT (CO, COL, and TOC1) motif, and two NH2-terminal zinc finger domains (12–14). The CCT motif is ∼45 amino acids, rich in basic residues, and contains a putative nuclear localization signal in the NH2-terminal half (14).toc1-1 is mutated at a residue within the second half of the CCT motif, and proximal residues are mutated in two coalleles (Fig. 3D) (15). Immediately downstream of the CCT motif is a region rich in acidic residues, a feature common to a number of transcriptional activators (16). The last identifiable feature within TOC1 is an almost perfect direct repeat of 47 amino acids in the center of the protein, of unknown function.
Messenger RNA levels of TOC1 cycled robustly in light-dark cycles, peaking in the late day (Fig. 4A). Changes in TOC1 levels did not coincide with dawn or dusk transitions, nor wereTOC1 transcript levels acutely light-induced (17). In the absence of entraining signals, robust cycling persisted throughout an LL time course (Fig. 4B). Notably, the period of the TOC1 circadian rhythm was shortened in thetoc1-1 mutant, demonstrating that TOC1 products feed back to affect their own expression (Fig. 4B). The TOC1rhythm was bimodal in LD conditions, and aspects of this were evident in the LL time course (Fig. 4, A and B). One explanation is that TOC1 expression is always biphasic. Another possibility is that the phase of TOC1 expression is distinct in two different populations of cells or tissues, as has been shown for circadian genes in other organisms (18). In DD, the rhythm of mRNA expression dampened after one cycle, although mean levels were maintained (Fig. 4C).
Expression patterns of TOC1.Northern analysis was performed on RNA extracted from (A) wild-type seedlings grown in 12:12 LD (open bars denote light intervals, solid bars denote darkness; ZT, zeitgeber time), (B) toc1-1 and wild-type seedlings entrained to 12:12 LD and then released into LL at time 0, and (C) wild-type seedlings entrained to 12:12 LD then released into DD at time 0. Ordinate values are TOC1probe signal as a proportion of ribosomal DNA probe signal normalized to highest value within the same data set. All experiments were performed two or three times with similar results (7). (D and E) Subcellular localization of YFP::TOC1. BY-2 protoplasts were transfected with a plasmid directing expression of a YFP::TOC1 fusion protein (7) and kept in the dark for 24 hours before examination. (D) Superimposition of bright-field and fluorescent images (7); (E) close-up of two nuclei. Scale bars, 5 μm (D), 2 μm (E).
As predicted from the pervasive effect of TOC1 mutation, atoc1::luc transcriptional reporter was expressed in all organs of seedlings and mature plants (17, 19). We used translational fusions to yellow fluorescent protein (YFP) to investigate the subcellular localization of TOC1. In transiently transfected tobacco cells incubated in the dark, YFP::TOC1 appeared exclusively in the nucleus in a distinctive speckled pattern (Fig. 4, D and E). This pattern suggests TOC1 participation in functionally important complexes such as transcriptosomes, spliceosomes, or proteosomes (20). The same pattern was observed in transfected cells kept in the light (17), arguing against light-dependent differential partitioning of TOC1, such as we have shown for the circadian photoreceptor dCRY, which forms similar patterns in insect nuclei (21).
Although it is premature to unequivocally place TOC1in a central pacemaker, the identification of TOC1 molecular targets and partners will elucidate its specific role. Interestingly, PAS (PER, ARNT, SIM) domains found in several clock proteins show sequence similarity to two-component sensor domains (4, 22). Other clock-associated proteins are similar to sensor kinases and exhibit kinase activity (23, 24). More typical sensor kinase homologs have also been identified in Arabidopsis(25). However, considering the atypical nature of its receiver domain, TOC1 may not be involved in an orthodox phosphorelay.
Note added in proof: Several Arabidopsissequences were recently recognized as encoding atypical response regulator homologs, including one corresponding to TOC1(30).
