Introduction

The circadian clocks of diverse organisms generate robust, 24-h biological rhythms, via gene regulatory circuits that include interlocking loops of transcriptional negative feedback (Young and Kay 2001). Light signalling pathways regulate a component(s) of the feedback loops, entraining the endogenous clock to the environmental day/night cycle. The plant circadian clock regulates leaf position and the expression of many genes, including expression of chlorophyll a/b-binding protein (CAB) genes that peak in the mid-morning and COLD AND CIRCADIAN-REGULATED (CCR) genes peaking late in the day. Rhythmic transcription can be monitored in vivo using bioluminescent luciferase (LUC) reporter genes. Genetic screens have thus identified Arabidopsis clock mutants, such as the 21-h-period mutant toc1-1 (reviewed in Hayama and Coupland 2003). A regulatory loop involving TIMING OF CAB EXPRESSION 1 (TOC1) and CIRCADIAN CLOCK-ASSOCIATED 1 (CCA1) is thought to play a central part in generating circadian rhythms in Arabidopsis (Hayama and Coupland 2003; Mas et al. 2003). TOC1 and its four paralogues, the PSEUDO-RESPONSE REGULATOR (PRR) genes, were separately identified by their homology to the receiver domain of prokaryotic two-component signalling systems (Makino et al. 2000; Strayer et al. 2000). In canonical two-component systems, an environmental sensor regulates an associated histidine kinase, which signals to various effector proteins via a His-Asp phosphorelay mechanism. The receiver domains of TOC1 and PRR proteins have substitutions at conserved residues including the phosphorylated aspartate. PRR transcripts accumulate rhythmically in the order PRR9-PRR7-PRR5-PRR3-TOC1 with peak levels from 2 h (for PRR9) to 10 h (for TOC1, also known as APRR1) after dawn (Makino et al. 2000; Matsushika et al. 2000; Strayer et al. 2000). The sequence of gene expression has been proposed to result from the serial activation of PRR7-PRR5-PRR3-TOC1, initiated by PRR9 (Matsushika et al. 2000). Repression of PRR9 by overexpression of TOC1 suggested that this linear series might be closed to form another circadian feedback loop (Makino et al. 2002). We now characterise prr5 and prr9 mutations and demonstrate that their circadian effects are complementary, light-dependent and inconsistent with the serial activation of PRR genes.

Materials and methods

Plant material and growth and conditions

All experiments were carried out in the Arabidopsis thaliana (L.) Heynh. Columbia (Col) ecotype. SALK lines 6280 and 7551 were generated at the Salk Genomic Analysis Laboratory Institute and identified using the Insert Watch facility (http://www.nasc.nott.ac.uk/insertwatch) at the Nottingham Arabidopsis Stock Centre (NASC, UK). Seeds were obtained from the Arabidopsis Biological Resource Centre (ABRC, Ohio State University, Columbus, OH, USA) and NASC. Homozygous mutants were identified from segregating F3 by PCR amplification of the T-DNA flanking regions. Seedlings were entrained at 22 °C to 12 h light/12 h dark (LD) under cool-white light, 120 μmol m−2 s−1 for 8 days prior to measurements except for the etiolated seedlings, which were grown in darkness for 4 days while entraining to 12 h 18 °C/12 h 24 °C. Circadian period was measured for luminescence rhythms imaged, and leaf movement as described by Doyle et al. 2002. Hypocotyl-growth assays were done as referred to in Doyle et al. (2002).

Constructs

Luciferase constructs reporting the activity of CCA1, CAB2 and COLD AND CIRCADIAN-REGULATED 2 (CCR2) promoters (Doyle et al. 2002), were introduced into prr5, prr9 and Col using Agrobacterium. Multiple independent transformants were analysed for each gene and genotype.

Results

We selected SALK lines 6280 and 7551 that carried a T-DNA insertion in the PRR5 or PRR9 gene in the Columbia-0 background (Col). PRR5 shares the highest sequence identity with PRR9 (Matsushika et al. 2000). PCR analysis showed that the T-DNAs were inserted within the second exon to second intron (data not shown), consistent with publicly available data (http://signal.salk.edu/cgi-bin/tdnaexpress). Reverse transcription (RT)–PCR assays detected no cognate RNA from mutant seedlings, when wild types showed strong, rhythmic expression (data not shown). prr5-1 and prr9-1 are likely null mutations, abbreviated as prr5 and prr9.

We assayed circadian output rhythms and the expression of clock-associated genes in homozygous prr mutant seedlings. Figure 1a, b shows that circadian period relative to Col was 1.0–2.4 h shorter in prr5 and 1.0–1.4 h longer in prr9, for CCA1, CAB2 and CCR2 expression under simulated white (red + blue, RB) light and for rhythms of leaf movement in white fluorescent light. The altered period of leaf movement co-segregated with homozygosity for the prr mutation (data not shown), indicating that the T-DNA insertions caused these recessive phenotypes. CCA1 and TOC1 RNA levels were robustly rhythmic (Fig. 1c), so neither PRR5 nor PRR9 is uniquely required for the expression or rhythmicity of TOC1. The rising and/or falling phases of these transcripts were slightly advanced in prr5 compared to Col and slightly delayed in prr9. Peak levels of TOC1 RNA were lower in prr5 and higher in prr9, supporting a correlation between period and TOC1 expression levels (Mas et al. 2003). Leaf movements in prr5;prr9 double mutants remained robustly rhythmic with a period indistinguishable from Col (Fig. 1a).

Fig. 1 a, b.
figure 1

prr mutations conditionally alter the circadian period in Arabidopsis thaliana. prr mutant and wild-type seedlings carrying the LUC reporter genes indicated were grown for 8 days in 12 h L:12 h D cycles, and released into constant light at time 0. a Periods for CCA1 and CAB2 under constant R+B (RB), R or B, and for leaf movements under white light. b Periods for CCR2 in R+B (RB), R, or B; also in darkness, in LD-entrained seedlings (DD) and in seedlings grown in darkness with 12 h 18°C:12 h 24°C temperature cycles (etiolated). Periods for CCR2 in DD were measured from peak times over two cycles. Data are means ± SE, n=15–33 seedlings or (CCR2 etiolated) 8–16 groups of 20–40 seedlings. Leaf movement periods were identical in lines with and without LUC markers (data not shown). c CCA1 (above) and TOC1 (below) RNA levels were assayed by real-time RT–PCR relative to a ubiquitin control (UBQ), in extracts of Arabidopsis plants harvested at the times indicated during an LD cycle. Open circles, Col; filled triangles, prr9; filled squares, prr5. PCR primers and conditions are available from the authors on request. Data are means ± SE of triplicate determinations, normalised to the mean of all measurements for each gene. Filled bar on time axis, dark interval; open bar, light interval. d Hypocotyl elongation was tested (Doyle et al. 2002) in Arabidopsis plants grown for 4 days in darkness (D) or the R (upper) or B (lower) fluence rates indicated. Data are means, n=26–36, SE within symbols, which are as indicated in c. For each physiological assay, 3–4 progeny families or independently transformed lines were tested with similar results in 2–5 independent experiments

Full size image

Mutants with aberrant light signalling can alter the circadian period in a light-dependent manner (reviewed in Hayama and Coupland 2003). We therefore measured the periods of CCA1 and/or CCR2 expression under 10–15 μmol photons m−2 s−1 constant red light (R), blue light (B) or darkness (DD). The prr mutations altered the period of CCA1 expression in B by 2 h, affected the period of CCR2 expression in R and B to a lesser extent, but had no effect on the period of CCA1 expression in R or of CCR2 expression in DD (Fig. 1a, b). Etiolated prr seedlings also had periods that differed from Col by only 0.6 h or less. Both prr5 and prr9 seedlings showed mild long-hypocotyl phenotypes when grown under constant R or B at a range of fluence rates (Fig. 1d).

Discussion

The opposite period phenotypes of prr5 and prr9 show that PRR5 and PRR9 are not functionally equivalent, consistent with the differing phenotypes of plants that overexpress these genes (Matsushika et al. 2002; Sato et al. 2002). The period phenotypes were light-dependent, suggestive of a function in light input to the clock. Strong toc1 mutations abolish circadian rhythms in conditions lacking blue light (Mas et al. 2003), whereas PRR5 and PRR9 had their greatest effect in conditions containing blue light. The opposite period phenotypes of prr5 and prr9 are not due to opposite effects on general phototransduction, because both mutants have similar effects on light-regulated hypocotyl elongation. A phase-specific mechanism can explain the period phenotypes. PRR9 is expressed early in the subjective day, when light treatment advances the circadian phase (Covington et al. 2001). Abrogation of this phase advance in prr9 is consistent with the long period of prr9 under constant light (Fig. 1a); the converse applies to prr5 and toc1.

The rhythms of CCR2 expression had consistently different periods, in multiple transgenic lines, from those of CAB or CCA1 expression under some conditions. The effects of light and prr mutations also differed. Many, if not all, plant cells maintain a circadian system with a qualitatively similar molecular mechanism, so the prr mutations have similar effects on all rhythms. However, there are quantitative differences among rhythms in different cell types (Thain et al. 2002; Michael et al. 2003 and references therein). CCR2 is expressed in a different spatial pattern (for example, in roots) than the other markers, so CCR2 rhythms reflect a set of slightly different cellular clocks.

The small effects of the mutants on TOC1 RNA levels and their opposite effects on period suggest that the PRR genes do not activate each other in a linear daily sequence that ultimately activates TOC1. Such serial activation would result in an epistatic genetic interaction between the prr mutants. The additive interaction that we observed in prr5;prr9 confirms that PRR5 and PRR9 affect the circadian clock by largely independent mechanisms. We conclude that PRR5 and PRR9 participate in the complex interaction of light signalling with the circadian clock but are not required for rhythm generation.