CikA, a Bacteriophytochrome That Resets the Cyanobacterial Circadian Clock

See allHide authors and affiliations

Science  04 Aug 2000:
Vol. 289, Issue 5480, pp. 765-768
DOI: 10.1126/science.289.5480.765


The circadian oscillator of the cyanobacterium Synechococcus elongatus, like those in eukaryotes, is entrained by environmental cues. Inactivation of the gene cikA (circadian input kinase) shortens the circadian period of gene expression rhythms inS. elongatus by approximately 2 hours, changes the phasing of a subset of rhythms, and nearly abolishes resetting of phase by a pulse of darkness. The CikA protein sequence reveals that it is a divergent bacteriophytochrome with characteristic histidine protein kinase motifs and a cryptic response regulator motif. CikA is likely a key component of a pathway that provides environmental input to the circadian oscillator in S. elongatus.

The cyanobacterium S. elongatus PCC 7942 (1) exhibits circadian rhythms of gene expression that can be monitored using luciferase reporter genes (2). These bioluminescence rhythms persist with a period of approximately 24 hours, are temperature compensated, and their phase can be reset by light/dark transitions or by temperature cues (3). The cyanobacterial clock exhibits these characteristics of eukaryotic circadian clocks despite a lack of apparent homology between its protein components and those identified in other groups of organisms (4). For example, the complete genome sequence of Synechocystis sp. strain PCC 6803 is devoid of sequences similar to clock genes of Drosophila, such asperiod, timeless, Clock, andcycle, or the frequency gene ofNeurospora (4, 5). Likewise, no homologs of the cyanobacterial kaiA, kaiB, orkaiC genes, essential for circadian rhythmicity (6), have been detected thus far in eukaryotes. Other cyanobacterial genes that, when mutated, affect relay of temporal information from the clock to downstream genes include a sigma factor (7) and a putative carboxylase (8). A histidine protein kinase, SasA, interacts with the KaiC protein and works with the oscillator either at a point of environmental input or of output transduction to all downstream genes (9). We describe here a new clock-associated gene, cikA, that lies on an input pathway that supplies phase-setting information to theS. elongatus clock.

The cikA gene was identified from a Tn5 transposon insertion mutant (2) that showed subtle alteration in light-responsive regulation of a photosystem II gene, psbAII (10). Expression of a psbAII::luxAB(bacterial luciferase) fusion in the mutant was 50 to 80% of wild type under low light conditions and showed exaggerated induction on exposure to higher light intensity (11). However, a more striking circadian (2, 12) phenotype was noted: the period of bioluminescence oscillation was shortened by approximately 2 hours (22.80 ± 0.45 versus 24.71 ± 0.25, n = 12), and the relative timing of peaks (phase angle) was offset by approximately 6 hours (Fig. 1A).

Figure 1

Circadian phenotypes of reporter strains in whichcikA is inactivated by transposon Tn5 insertion. Bioluminescence (counts per second) is shown from (A) translational luxAB fusion to psbAII(10); (B) transcriptional luxAB fusion to the promoter of kaiB (6); (C) translational fusion of a firefly luciferase gene (luc) topurF (33); and (D) translationalluxAB fusion to kaiA (6). In (A), (C), and (D), black bars on the abscissa indicate dark incubation periods; otherwise, samples were kept in continuous light (onset = time 0). In (B), samples were subjected to one 12-hour dark incubation before release into continuous light. Green, wild type; blue, cikAinsertion mutant. (D) Blue diamonds, Tn5 insertion mutant; blue circles and squares, two orientations of inactivating gentamycin resistance cassettes.

Reduction of both period and amplitude was observed with all reporters (Fig. 1, A to D) (e.g., periods forkaiB::luxAB, 22.36 ± 0.47 hours versus 25.24 ± 0.35 hours, n = 12; forpurF::luxAB, 22.75 ± 0.24 hours versus 24.86 ± 0.33 hours, n = 12). Nonetheless, expression from the kaiB promoter, indicative of clock gene expression, remained robustly rhythmic with no notable alteration in phase angle (Fig. 1B). The bioluminescence rhythm from apurF::luc reporter (firefly luciferase) was also affected in both amplitude and period (Fig. 1C), indicating that the phenotype is not related to the substrates of bacterial luciferase and that it extends to class 2 genes [purF peaks at subjective dawn and is defined as class 2; the majority of gene expression patterns in the organism peak near subjective dusk and are defined as class 1 (13)]. A gentamycin resistance cassette inserted in both orientations with respect to the cikA open reading frame (ORF) caused phenotypes identical to those of the original Tn5 insertion mutant (Fig. 1D). Note that thekaiA::luxAB reporter showed an altered phase-angle phenotype; thus, in the cikA genetic background, the relative phasing of kaiA and kaiBC expression is uncoupled without dramatically affecting circadian timing (Fig. 1, B and D), as was previously demonstrated for mutation of thecpmA gene (8).

Mutant kai alleles in apsbAI::luxAB reporter background that affect period (6) allowed us to examine the effect of cikA inactivation in both short- and long-period mutants of S. elongatus. The cikA mutation caused a very small, but reproducible, additional period shortening of the psbAI::luxAB reporter rhythm in thekaiB missense mutant B22a, and a dramatic phase-angle change (Fig. 2A). The phase-angle alteration is particularly marked in the kaiC long-period mutant C28a background (Fig. 2B), in which the C28a/cikA double mutant and wild type have a stable period length relationship throughout the run, but their bioluminescence peaks are offset by approximately 9 hours. The additive effect of the combined mutations suggests that CikA and Kai proteins perform independent, nonoverlapping functions.

Figure 2

(A and B) cikAdisruption in period mutants of Synechococcus. Wild-type (green; period 25.11 ± 0.46 hours, n = 12) and period mutant kai backgrounds (rose) carry apsbAI::luxAB fusion (6, 34). (A) Rose, period mutant B22a (21.34 ± 0.10 hours, n = 5); blue, B22a/cikA(20.92 ± 0.34 hours, n = 5). (B) Rose, period mutant C28a, (26.36 ± 0.83 hours, n = 5); blue, C28a/cikA (24.34 ± 0.13, n = 5). (C) Genetic complementation of the cikAphenotype. Ectopic insertion of a copy of cikA (red) restored the period, amplitude, and phasing phenotypes of acikA disruption mutant (blue) to match those of wild type (green), as measured by a psbAI::luxABreporter. Ectopic insertion of the extra copy of cikA into a wild-type background (black). Axes labeled as in Fig. 1.

Genetic complementation confirmed that inactivation of thecikA gene is responsible for the mutant phenotypes, rather than possible polar effects of the transposon insertion on nearby genes. Wild-type amplitude, period, and phase-angle properties were all restored to psbAI::luxABbioluminescence when an ectopic copy of cikA was provided to a cikA mutant strain (Fig. 2C).

Persistence of robust circadian rhythms in the cikA genetic background indicates that the product of this gene is not essential for circadian oscillator function. The global effect on period of more than eight tested genes (14), including representatives of classes that were assigned to distinct output pathways by mutational analyses, suggests that the cikA product is not part of one of these pathways, unless it functions as does SasA in close association with the clock (9). To determine whether CikA provides environmental input to the oscillator, we tested the ability of cikA-inactivated reporter strains to reset the phase of the clock in response to a 5-hour dark pulse (15). During portions of the circadian cycle, wild-type S.elongatus responds to this stimulus by changing the phase of subsequent peaks by 10 to 12 hours after cells are returned to continuous light (Fig. 3). In contrast, cikA mutant strains show little phase resetting in this assay. These data are consistent with CikA functioning in an input pathway to the circadian oscillator.

Figure 3

Phase-resetting of thepsbAI::luxAB bioluminescence rhythm in wild-type (diamonds), cikA (squares), and complementedcikA (triangles) genetic backgrounds in response to a 5-hour dark pulse. At the indicated circadian time on the abscissa, samples received 5 hours of dark incubation, then were returned to continuous light for monitoring of the circadian rhythm (15). The ordinate for each data point indicates the offset of the phase of peaks after the treatment, relative to a control not pulsed with darkness: phase advance (positive values) or phase delay (negative values). To accommodate differences in circadian period between strains, actual time was converted to circadian time (one circadian hour = free running period × 24−1).

We named the gene cikA, for circadian input kinase, on the basis of mutant phenotypes and inference from sequence analysis. The most striking features in the deduced protein sequence (16) are histidine protein kinase motifs that conform to all conserved blocks for that family (Fig. 4A, blocks H, N, D/F, and G) (17). The carboxyl terminus is similar to the receiver domains of response regulators, most notably PhoB (Fig. 4C) (18). Although other key residues of this motif are present, the invariant Asp in this family, which is the residue phosphorylated by a cognate histidine protein kinase in each case, is absent from the sequence (Fig. 4C) (16). Thus, if the CikA histidine protein kinase domain transfers a phosphoryl group to its receiver domain, another residue must become phosphorylated. Alternatively, phosphotransfer may not be the role of this segment of the protein; perhaps it interacts with other regulatory partners, and this contact is modulated by autophosphorylation within the H box.

Figure 4

(A) Graphic representation of the 754–amino acid CikA protein, indicating relative size and distribution of identifiable motifs: chromophore binding domain of phytochromes (CB); H, N, D/F, and G boxes of histidine protein kinases; and a receiver domain of response regulators (RR). (B) Comparison of chromophore binding domains of CikA, PhyE from Arabidopsis thaliana (GenBank accession no. X76610), Cph1 (Kazusa DNA Institute CyanoBase ORF slr0473), and slr1969 (Kazusa DNA Institute CyanoBase ORF slr1969) from Synechocystis sp. strain PCC 6803, and RcaE from Fremyella diplosiphon (GenBank accession no. U59741). Black diamond, PhyE residue 322 bilin chromophore ligand. Residues conserved: in all sequences, white letters on black; in four out of five, white on gray; in three out of five, black on gray. Numbers at the beginning of each line indicate position in the respective protein sequence. Asterisks mark each tenth residue in alignment. (C) Comparison of receiver domains of CikA, slr1969 from Synechocystis sp. strain PCC 6803, and PhoB from Escherichia coli (GenBank accession no. P08402). Black diamond, residue expected to be Asp in response regulator receiver domains. Black background, identical residues; gray background, chemically similar residues. For alignments (B and C), we used a ClustalW 1.8 alignment tool accessed through the BCM Search Launcher (35). Alignment in (B) was modified by hand on the basis of information from 54 phytochrome-like sequences with assistance from C. Lagarias (36).

The amino-terminal sequence reveals that the protein belongs to the expanding family of bacteriophytochromes (19), similar toSynechocystis sp. strain 6803 Cph1 (20),Fremyella diplosiphon RcaE (21),Deinococcus radiodurans BphP (22), andArabidopsis thaliana PhyE (23) (Fig. 4B). This raises the possibility that CikA is a photoreceptor. However, unlike other known phytochromes and bacteriophytochromes, CikA lacks the conserved Cys residue expected as a bilin ligand for phytochromes (24, 25). It also lacks the His residue reported to be the bilin ligand for D. radiodurans BphP, which corresponds to His 323 in the PhyE sequence (Fig. 4B) (22). This suggests several possibilities for CikA structure and function: it does not bind a bilin chromophore, it binds a chromophore (bilin or another cofactor) noncovalently, or it binds a chromophore by a novel attachment.

The similarity of CikA to phytochromes provides the first potential evolutionary parallel between cyanobacterial and eukaryotic circadian systems. Phytochromes play several distinct roles in relaying light information to the circadian clocks of plants (26). Although the white collar proteins ofNeurospora, important for light-dependent processes and for circadian clock function (27), bear similarity to phytochromes, the correspondence is through shared PAS domains. No direct link can be drawn between the white collar proteins and CikA, which lacks a PAS domain and resembles a different part of the phytochrome sequence—the chromophore binding domain.

A subsequent direct screen for transposon mutants that affect phase resetting has identified five independent cikA mutants, and no other loci, as causing clear resetting phenotypes (28). This further supports a key role for CikA in providing environmental input to reset the cyanobacterial circadian clock.

  • * Present address: Botanisches Institut, University of Cologne, Gyrhofstrasse 15, 50931 Cologne, Germany.

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


View Abstract

Navigate This Article