Circadian clocks are signalling networks that enhance an organism's relationship with the rhythmic environment. The plant circadian clock modulates a wide range of physiological and biochemical events, such as stomatal and organ movements, photosynthesis and induction of flowering. Environmental signals regulate the phase and period of the plant circadian clock, which results in an approximate synchronization of clock outputs with external events. One of the consequences of circadian control is that stimuli of the same strength applied at different times of the day can result in responses of different intensities. This is known as ‘gating’. Gating of a signal may allow plants to better process and react to the wide range and intensities of environmental signals to which they are constantly subjected. Light signalling, stomatal movements and low-temperature responses are examples of signalling pathways that are gated by the circadian clock. In this review, we describe the many levels at which the circadian clock interacts with responses to the environment. We discuss how environmental rhythms of temperature and light intensity entrain the circadian clock, how photoperiodism may be regulated by the relationship between environmental rhythms and the phasing of clock outputs, and how gating modulates the sensitivity of the clock and other responses to environmental and physiological signals. Finally, we describe evidence that the circadian clock can increase plant fitness.
In a world characterized by rhythms of light intensity and temperature cycles, there has been selection for the evolution of an internal clock that optimizes the plant's relationship with the environment. This circadian clock runs with a period close to 24 h, even in the absence of environmental cues, and maintains a relatively constant period within a physiologically relevant temperature range. The plant circadian clock can be considered to be a signalling network that carries time-encoded information and regulates physiological processes. In this review, we show how plants' interactions with the environment are influenced by the circadian clock. We describe how changes in the environment regulate the circadian clock, how circadian clock outputs optimize the plant's responses to the environment and how the circadian clock may improve plant fitness.
STRUCTURE OF ARABIDOPSIS CIRCADIAN CLOCKS
Circadian clocks are usually thought of as being divided into three parts: a central oscillator, which generates the rhythmic behaviour; the input pathways, which carry environmental information to entrain the central oscillator; and the output pathways that regulate physiological processes (Fig. 1; Dunlap 1999). This conceptual model is very useful but overly simplistic as there is crosstalk between different parts of the circadian clock, most strikingly between the output and input pathways, with some circadian-regulated outputs modulating input into the oscillator (for reviews, see, Más 2005; Gardner et al. 2006; McClung 2006).
Here we use the term circadian clock to describe the whole circadian system. In plants, this system is likely to consist of more than one clock. There is good evidence for independent oscillators in each cell. It is also possible that there are cell-specific oscillators and multiple oscillators in individual cells (see Gardner et al. 2006, for a review).
Generation of circadian rhythms
The circadian clock generates rhythms that maintain a robust near-24 h period through a network of multiple feedback loops (Fig. 1). The first model for a plant circadian clock was a single-loop model composed of TIMING OF CHLOROPHYLL A/B BINDING PROTEIN 1/PSEUDO-RESPONSE REGULATOR 1 (TOC1/PRR1), CIRCADIAN CLOCK-ASSOCIATED 1 (CCA1) and LATE ELONGATED HYPOCOTYL (LHY; Alabadíet al. 2001). TOC1, which is expressed with a peak 12 h after dawn (zeitgeiber time 12 or ZT12), is a member of a family of five pseudo-response regulators (TOC1/PRR1, PRR3, PRR5, PRR7 and PRR9; Millar et al. 1995a; Matsushika et al. 2000; Strayer et al. 2000). CCA1 and LHY encode light-induced MYB-like transcription factors that are expressed highly during the early morning (ZT0, Schafferet al. 1998; Wang & Tobin 1998). In the single-loop model, light activates expression of CCA1/LHY, which represses the expression of TOC1. The decrease of TOC1 levels reduces the level of CCA1/LHY expression. At the end of the subjective day, CCA1/LHY levels have fallen sufficiently to allow TOC1 to be expressed (Alabadíet al. 2001). CCA1/LHY binds to a motif called EVENING ELEMENT (EE, AAATATCT) in the TOC1 promoter, repressing TOC1 expression (Harmer et al. 2000; Alabadíet al. 2001). Overexpression of either CCA1 or LHY (CCA1-ox and LHY-ox) leads to arrhythmia, with TOC1 expressed at low levels (Alabadíet al. 2001). In turn, TOC1 indirectly induces the expression of CCA1/LHY, as the recessive loss-of-function toc1-2 mutant has reduced expression of CCA1/LHY (Alabadíet al. 2001).
A number of biological and mathematical studies have demonstrated that the single-loop model requires modification to explain fully circadian behaviour. The cca1-11 lhy-21 double mutant, for example, is a short-period mutant rather than being arrhythmic as the single-loop model predicts (Alabadíet al. 2002; Mizoguchi et al. 2002; Locke et al. 2005b). There is also a requirement for a delay mechanism between TOC1 and CCA1/LHY, because the translation peak of TOC1 does not match the time that the transcription of CCA1/LHY begins (Alabadíet al. 2001; Locke, Millar & Turner 2005a). This is compelling evidence for the existence of additional components in the circadian clock. Recently, Locke et al. (2005b) used mathematical modelling to propose extensions to the single-loop model. The Locke model proposes the existence of two extra hypothetical components: X and Y. X is a proposed delay mechanism between TOC1 and CCA1/LHY, and Y is suggested to regulate TOC1 expression levels and, in turn, is regulated by light, CCA1/LHY and TOC1 (Locke et al. 2005b). The Locke model predicts that Y has one rapid light-induced peak of expression after dawn, which is repressed by CCA1/LHY, and a second circadian-controlled peak in the late afternoon. Data from high-resolution sampling experiments revealed that the expression pattern of GIGANTEA (GI) closely matched the predicted characteristics of Y (Locke et al. 2005b; Mizoguchi et al. 2005). Experimental studies have provided further evidence that GI is a necessary component of the central oscillator, the details of which are discussed later (‘Temperature entrainment and compensation’). Recently, two different groups have independently derived a three-loop model from the Locke model. This three-loop model incorporates a negative feedback loop between CCA1/LHY and PRR7/PRR9 (Locke et al. 2006; Zeilinger et al. 2006; see also Fig. 1).
Expression of LUX ARRHYTHMO (LUX or PHYTOCLOCK1, PCL1), which has a MYB DNA-binding domain, correlates with TOC1 expression (Hazen et al. 2005; Onai & Ishiura 2005). Plants that overexpress LUX become gradually arrhythmic under constant light (LL) and constant dark (DD), whereas the loss-of-function lux-1 is arrhythmic in both conditions. This suggests that LUX has a central role in the generation of rhythms (Hazen et al. 2005; Onai & Ishiura 2005). EARLY-FLOWERING 4 (ELF4), which is also expressed in the same phase as TOC1, might form another feedback loop in the clock. Light-induced expression of ELF4 requires CCA1/LHY, while light-induced CCA1/LHY expression requires ELF4 (Kikis, Khanna & Quail 2005). Mutants of ELF4 have varying period length before reaching arrhythmia after 24 h in LL (Doyle et al. 2002).
Many other components may have a role in the circadian pathway. These include ARABIDOPSIS RESPONSE REGULATORS 3 and 4 (ARR3 and 4; Saloméet al. 2006), TIME FOR COFFEE (TIC, Hall et al. 2003) and a poly(ADP-ribose) glycohydrolase (TEJ, Panda, Poirier & Kay 2002). However, the exact mechanisms of their action, their positions in the pathway and relative importance are still unknown.
Environmental inputs to the circadian clock
The intrinsic period of the circadian clock of Arabidopsis varies between 22 h and 29 h depending on accession and growth conditions (Michael et al. 2003b). To be synchronized with environmental rhythms, the plant circadian clock has a series of mechanisms that feed environmental information into the oscillator, which adjusts its phase and maintains the circadian clock period close to 24 h (Fig. 1, for detailed review Millar 2004; Salomé & McClung 2005b; Somers 2005). As an example, CHLOROPHYLL A/B BINDING PROTEIN 2 (CAB2), which encodes a protein necessary for the light-harvesting complex, has a peak of transcription that changes under different photoperiods but is usually close to the middle of the light period (Millar & Kay 1996). The synchronization between the circadian clock and the 24 h environmental period is mediated by a resetting mechanism that shifts the phase of the clock every cycle in response to environmental cues (Fig. 2). However, precise control of period is also necessary to maintain correct phase of circadian outputs with the environment. For example, the short period mutant toc1-1 can entrain to a 24 h period under 24 °C/20 °C thermocycles but the peak of CAB2 expression occurs earlier in the day than in wild-type plants (Somers et al. 1998b). A similar mismatch is found in the long period mutant ztl-1 (zeitlupe-1), in which GLYCINE-RICH RNA-BINDING PROTEIN 7/COLD, CIRCADIAN RHYTHM AND RNA BINDING 2 (GRP7/CCR2) transcription peaks later under light/dark (LD) cycles than in the wild type (Somers et al. 2000).
Light input pathways
Light affects both the phase and period of the circadian clock. A light pulse provided to plants in DD can change the phase of endogenous rhythms by varying degrees that are dependent on the light signal length and intensity. The time of day that the light signal is given determines whether the phase will advance or delay (Fig. 2a; Millar 2004). The period of the circadian clock under LL decreases as the intensity of light input to the oscillator increases. This is known as Aschoff's rule (Aschoff 1979, cited in Devlin & Kay 2000). Phase adjustment by light, which is dependent on the time of light treatment, provides evidence that the oscillator is correctly responsive to specific cues at appropriate times of day, while Aschoff's rule provides evidence for a continuous readjustment of phase that results in an altered period under LL (Somers 2005).
The main photoreceptors responsible for light input into the clock are the five red-light-sensing phytochromes (PHYA-E) and the two blue-light-sensing cryptochromes (CRY1 and CRY2; Somers, Devlin & Kay 1998a; Devlin & Kay 2000). The complexity of the input pathways to the clock is increased by some of their components being circadian-regulated (Harmer et al. 2000). For example, CRY1 and PHYB expression peak in the middle of the day (ZT6), whereas PHYA and CRY2 expression peak in the latter part of the subjective day (ZT10; Tóth et al. 2001).
The role of PHY and CRY in the circadian clock was established by examining phy- and cry-null mutants under a wide range of light flux. PHYA has a role in setting the pace of the clock under red and blue low light flux, whereas PHYB has a role under high fluence red light (Somers et al. 1998a; Devlin & Kay 2000). The phyA-201 phyB-1 double mutant has longer periods than the wild-type plants in all fluence rates of red light and low fluences of blue light. Similarly, cry1 cry2 (hy4 cry2-1) double mutants have longer periods than the wild-type plants in all fluences of blue light and low fluences of red light (Somers et al. 1998a; Devlin & Kay 2000). The less abundant PHYD and PHYE only appear to have a role when PHYB is mutated (Devlin, Patel & Whitelam 1998; Devlin et al. 1999). PHY mutants have a normal clock function under DD, which excludes a direct role of these components in the oscillator (Devlin & Kay 2000). The effects of PHY mutations under blue light and CRY mutations under red light indicate that there is crosstalk between the signalling pathways (Devlin & Kay 2000). There is no evidence that phototropins (PHOT), which mediate blue light responses in phototropism and stomatal movements are part of the circadian input pathway (Salomé & McClung 2005b).
Although the role of the photoreceptors in the clock is well described, little is known about the transduction of the light signal to the oscillator. Light has many putative entry points to the oscillator as transcription of CCA1/LHY, GI, ELF4 and PRR9 are induced by light (Wang & Tobin 1998; Martínez-Garcia, Huq & Quail 2000; Kim et al. 2003; Farréet al. 2005; Kikis et al. 2005; Locke et al. 2005b). One component believed to link light signalling and the oscillator was PHYTOCHROME-INTERACTING FACTOR 3 (PIF3), a transcription factor that interacts with both PHYA and PHYB (Ni, Tepperman & Quail 1998, 1999). PIF3 has been shown to form a complex with the G-box element of CCA1/LHY and the far-red (active) form of PHY (Pfr; Martínez-Garcia et al., 2000). Furthermore, red light induction of CCA1/LHY is attenuated in some PIF3-antisense lines (Martínez-Garcia et al. 2000). These data suggest that PIF3 is involved in the light-regulation of CCA1/LHY and, consequently, the oscillator. In addition, PIF3 interacts with TOC1 in yeast two-hybrid assays (Yamashino et al. 2003). However, the manipulation of PIF3 levels (PIF3 overexpression, PIF3 antisense and pif3 null lines) had no effect on the period or phase of the clock, which suggests that PIF3 does not have an important role in the regulation of the plant circadian clock (Monte et al. 2004; Oda et al. 2004; Viczián et al. 2005).
Light may also signal to the oscillator through DE-ETIOLATED 1 (DET1), a negative regulator of PHY and CRY signalling. The det1-1 mutant has a reducedcircadian period, possibly through the inhibition of LHY degradation (Millar et al. 1995b; Song & Carré 2005). CONSTITUTIVELY PHOTOMORPHOGENIC 1 (COP1) is believed to function in a similar light transduction pathway as DET1. The mutant cop1-6 reduces the period of the clock (Millar et al. 1995b; Ma, Zhao & Deng 2003; Song & Carré 2005). FAR-RED ELONGATED HYPOCOTYL 1 (FHY1) and FHY3 are two genes related to PHYA signalling that are required for phase shifting of leaf movement rhythms in response to far-red light (Yanovsky et al. 2001). Furthermore, mutation of SUPPRESSOR OF PHYA-105 (SPA1), which is also involved in PHYA signalling, leads to a small reduction in the free-running period of TOC1 and CCA1 expression (Ishikawa, Kiba & Chua 2006). Finally, mutation of SENSITIVITY TO RED LIGHT REDUCED 1 (SRR1), which is related to PHYB signalling, also causes a shortening of the circadian period of leaf movement and TOC1 and CCA1 expression. SRR1, however, may act through a pathway independent of PHYB because the short-period phenotype is still observed under DD (Staiger et al. 2003).
Another route for light input mediated by PHY and CRY may be ZEITLUPE (ZTL). ZTL targets TOC1 for degradation in a light-dependent manner (Más et al. 2003b). ZTL is part of a protein family that contains three characteristic protein domains: a central F-box, a LOV (LIGHT, OXYGEN, VOLTAGE) domain, which is similar to the chromophore-binding domain of PHOT, and six C-terminal KELCH repeats (Nelson et al. 2000; Somers et al. 2000; Schultz et al. 2001). Furthermore, the LOV domain changes conformation when exposed to blue light in vitro (Imaizumi et al. 2003). However, all described mutant alleles of ZTL have a long-period phenotype not only in LL but in DD as well. This suggests that light might not be an important regulator of ZTL (Kevei et al. 2006). Another member of the ZTL family, LOV/KELCH PROTEIN 2 (LKP2), might regulate TOC1 degradation in a similar fashion. However, lkp2 has no circadian phenotype, which suggests that either LKP2 is not associated with clock function or there is a degree of redundancy (Salomé & McClung 2005b; Somers 2005). Both ZTL and LKP2 overexpression lead to arrhythmia, which could be caused by increased degradation of TOC1 (Schultz et al. 2001; Más et al. 2003b). A third member of the ZTL family, FLAVIN-BINDING KELCH REPEAT F-BOX 1 (FKF1) does not have a role in the clock but is important in the photoperiodic control of flowering (Imaizumi et al. 2003) and is discussed later (‘Circadian clock and seasonal responses’).
As temperature affects the rate of biochemical reactions, one might expect temperature to affect the pace of the circadian clock. However, circadian period does not change significantly in response to a range of physiologically relevant temperatures (from 12 °C to 27 °C in Arabidopsis; Edwards et al., 2005). This demonstrates that there are mechanisms to compensate temperature-induced changes in reaction rates. Characterization of circadian clock gene expression at different temperatures showed that LHY expression decreased in high temperatures. This was counterbalanced by increases in TOC1 and GI levels. In contrast, in low temperatures, expression levels of CCA1 increased slightly, which was counterbalanced by small reductions of GI levels. In addition, the null mutant gi-11 and the loss-of-function mutants cca1-11 and lhy-21 did not compensate for temperature changes (Gould et al. 2006). These data suggest that temperature-induced changes in GI, CCA1 and LHY levels compensate for temperature changes, resulting in maintenance of the period of the circadian clock. Distinct roles for CCA1 and LHY in temperature compensation were established by studying the effects of low and high temperature treatments in loss-of-function lines. Changes in the period of the circadian clock in response to temperature were most severe in high temperatures for lhy-21 (27 °C), while the effects were greatest at low temperatures (12 °C; Gould et al. 2006) for cca1-11. The FLOWERING LOCUS C (FLC) null-mutant flc-3 has short periods in 27 °C but not in 22 °C or 15 °C, which suggests that FLC is required for period control specifically in warm temperatures (Edwards et al. 2006). The period-shortening phenotype was even more pronounced if FRIGIDA (FRI-SF2), a positive regulator of FLC, was also mutated. In 27 °C, the FRI-SF2 flc-3 double mutant had the same expression levels of TOC1, CCA1/LHY and GI as the wild type. However, LUX levels in the FRI-SF2 flc-3 double mutant were higher in 27 °C but not in 22 °C, which suggests that FLC and FRI have a role in temperature compensation through the repression of LUX in 27 °C (Edwards et al. 2006).
Outputs of the circadian clock and the environment
Little is known about how the temporal information in the circadian clock is transduced to bring about rhythms in physiological and biochemical events, but at least three regulatory mechanisms are likely to be involved: the control of the protein abundance, the regulation of enzymatic activity and alterations in metabolite levels. The circadian clock may regulate the abundance of proteins involved in metabolic pathways through the control of gene transcription, RNA stability, translation rates or regulation of protein degradation. Alternatively, regulatory proteins with rhythmic activity may modulate the activity of proteins of ametabolic pathway. Finally, there might be rhythms in metabolites necessary for the metabolic pathway to proceed effectively (Harmer et al. 2000; Harmer, Panda & Kay 2001).
Approximately 2–16% of the genes expressed in Arabidopsis have circadian rhythms in the steady-state levels of transcript abundance (Harmer et al. 2000; Schaffer et al. 2001; Edwards et al. 2006). Among these rhythmically expressed transcripts, many encode proteins that are involved in a large number of the Arabidopsis metabolic pathways. Furthermore, many of the clock-regulated transcripts encode proteins with regulatory roles, such as kinases and phosphatases, so it is possible that the level of the circadian control over plant metabolism is even greater than suggested by the percentage of transcripts under circadian control (Harmer et al. 2000). Some mechanisms involved in the circadian control of transcript abundance have been determined. The EE confers evening-phased expression to genes (Harmer et al. 2000; Alabadíet al. 2001; Michael & McClung 2002). A skeleton promoter of four tandem repeats of the EE is sufficient to result in rhythmic evening-phased gene expression, probably because of rhythmic CCA1/LHY expression and binding to the promoter, repressing expression (Harmer & Kay 2005). It is also possible that EE acts to promote transcription during the morning (Harmer & Kay 2005). Another CCA1/LHY-binding promoter motif, CBS (AAAAATCT), confers dawn-phased rhythmic expression through the positive action of CCA1/LHY (Wang et al. 1997; Piechulla, Merforth & Rudolph 1998; Michael & McClung 2002).
There may be a link between the phase at which transcript abundance peaks and the function of its product. Some circadian-regulated genes, such as those involved in photosynthesis, flavonoid synthesis, cell elongation, nitrogen fixation and mineral assimilation, may have a direct impact on the response of the plant to the environmental rhythms (Harmer et al. 2000; Schaffer et al. 2001; Edwards et al. 2006). Transcripts encoding proteins involved in photosynthesis or required for the synthesis and binding of photosynthesis pigments tend to be expressed during the time of maximum light intensity (between ZT4 and ZT8, Harmer et al. 2000). Likewise, transcripts required for phenylpropanoid synthesis, whose products protect the plant against ultraviolet (UV) radiation (Landry, Chapple & Last 1995), peak before dawn (ZT20), possibly to prevent photodamage during the day (Harmer et al. 2000). Furthermore, the induction of stress-related genes in the late afternoon (ZT8) may anticipate water-deficit stress because of extended stomatal opening in the afternoon and cold in the early evening (Harmer et al. 2000; Kreps et al. 2002). Transcript abundance may oscillate not only to anticipate plant responses caused by rhythmic changes in the environment, as in the examples above, but also to rhythmically modulate the way in which a plant responds to environmental stimuli. In the following section, we discuss the rhythmic alterations in the plant's responses to the environment.
CIRCADIAN MODULATION OF ENVIRONMENTAL RESPONSES
One of the consequences of circadian control is that stimuli of equal strength applied at different times of the day can result in a different intensity of response (Fig. 3). This phenomenon is called gating. One example of gating is the diurnal variation in the inhibition of stem elongation by wind (Gaal & Erwin 2005). When wind perturbation was given to Cosmos bipinnatus at different times of the day, the most intense effect on growth was observed when wind was applied during the day (Gaal & Erwin 2005). Similarly, inhibition of stem growth in the legume Phaseolus vulgaris by mechanical stimulation was also greatest at the beginning of the day (Fig. 4a; Anderson-Bernadas et al. 1997). These data suggest that there is rhythmic sensitivity to mechanical stimuli in a range of plant species and that the circadian clock is a likely controller of rhythmic sensitivity to extracellular signals.
Gating of a signal may allow plants to better process and react to the wide range and intensities of environmental signals to which they are constantly subjected. The changes of sensitivity to environmental signals also may allow plants to respond only when it is advantageous. The circadian clock may gate a signalling pathway through a direct mechanism (Fig. 3a), in which outputs of the clock are part of the signalling pathway, or an indirect one, in which the clock regulates a specific gating pathway whose role is to modulate other signalling pathways (Fig. 3b) or both (Fig. 3c). In all cases, gating is a consequence of circadian control and can therefore operate via the regulation of the abundance of signalling intermediates, the control of the activity of signalling molecules or the availability of metabolites involved in the pathway (see ‘Outputs of the circadian clock and the environment’; Harmer et al. 2000). As examples, PHY and CRY abundance are controlled by the clock, which may result in the gating of light input to the clock (Harmer et al. 2000; Tóth et al. 2001; Sharrock & Clack 2002). Alternatively, gating of light induction of GI expression is mediated by the binding of CCA1/LHY to the EE in its promoter (Locke et al. 2005b).
Gating of light responses
We have briefly described gating of light input to the circadian oscillator and of light-induced expression of GI. Light-induced transcription of CAB2 and shade-avoidance responses are other examples of light signals gated by the circadian clock. The promoter of the CAB2 gene is regulated by both light and the circadian clock. The intensity of the acute light-induced increase of CAB2 expression is dependent on the time of the day the light stimulus is applied (Fig. 4c; Millar & Kay 1996). The highest response of CAB2 expression to light in DD was observed around 30 h and, the lowest, 18 h and 42 h after transfer to DD, which coincide, respectively, with the maximal and minimal levels of CAB2 transcription in DD (Millar & Kay 1996; McWatters et al. 2000). In null mutants of EARLY-FLOWERING 3 (ELF3), the intensity of CAB2 response to light is similar to the maximal response found in wild-type plants, which suggests that in the wild type, ELF3 represses this pathway in a rhythmic manner (Fig. 4d; McWatters et al. 2000). In addition, the null mutant elf3-1 is arrhythmic for many circadian outputs after 10 h in LL, but not in DD, which suggests that gating of light signalling is important in the maintenance of free-running circadian rhythms under LL (Hicks et al. 1996; McWatters et al. 2000; for a review, see Carré 2002). Many of the elf3-1 phenotypes, such as pale leaves and long hypocotyl and petioles, are associated with defective light responses (Covington et al. 2001; Hicks, Albertson & Wagner 2001; Liu et al. 2001). A two-hybrid screen suggested that ELF3 could act through interaction with the C-terminal domain of PHYB (Liu et al. 2001). However, elf3-1 and phyB-1 mutations are additive in hypocotyl elongation, which suggests that ELF3 may function independently of PHYB (Reed et al. 2000). Mutation of TIC also causes alterations in the gating of light-induced CAB2 expression, but this is observed at a different time of the cycle. While ELF3 acts with maximal effect during the early parts of the night, TIC acts in the middle to late part of the night (Hall et al. 2003). Recently, FAR-RED ELONGATED HYPOCOTYL 3 (FHY3) was associated with the gating of PHY signalling into the circadian clock (Allen et al. 2006).
Shade-avoidance responses, such as increases in stem and petiole elongation in response to shading, are important strategies for maximizing light harvesting by plants that grow in highly populated areas (Schmitt et al. 2003). Light quality alters when sunlight passes through, or reflects from the leaves of other plants. As far-red (FR) is poorly absorbed by plants, when other plants are nearby or covering a plant, the FR portion of the spectra is enriched compared with the red (R). The ratio between R and FR (R : FR) in daylight is usually ≈ 1.15 and drops to 0.05–0.7 underneath canopies of vegetation (Smith 1982). Consequently, a low R : FR signals shading by other plants. Plants can estimate R : FR of incoming light through the interconversion of PHY between a red-absorbing, biologically inactive form (Pr) and a far-red-absorbing, biologically active form (Pfr). The R : FR ratio of the incoming light is perceived by the plant as a ratio between the Pr and Pfr forms of PHY. Low R : FR will reduce the amount of the active Pfr form of the PHY, which leads to a series of photomorphogenic events, the shade-avoidance responses, usually at the expense of leaf and storage organ development (for a review, see Franklin & Whitelam 2005). Shade-avoidance responses are also modulated by the circadian clock (Fig. 4b; Salter, Franklin & Whitelam 2003). PIF3-LIKE 1 (PIL1), a component necessary for hypocotyl elongation in shade-avoidance responses, is one of the genes most highly up-regulated in response to a low R : FR (Salter et al. 2003). PIL1 transcript levels oscillate with a circadian period under low, but not under high R : FR, with a peak at dawn. This coincides with the peak of PHYA protein levels (Sharrock & Clack 2002). Hypocotyl growth in response to low R : FR, on the other hand, occurred maximally when the stimulus was given at dusk (Fig. 4b; Salter et al. 2003), which coincides with the maximum rate of cell expansion (Dowson-Day & Millar 1999; Harmer et al. 2000).
Circadian anticipation of dawn in C3 and C4 plants promotes stomatal opening, which allows CO2 uptake and fixation as soon as sufficient light is available to drive photosynthesis. Stomata stop opening around midday and start closing long before dusk (Webb 1998). These responses were traditionally considered a consequence of the water status of the leaf, but recent work demonstrates that, at least in well-watered plants, they are due to circadian control of the guard cell (Dodd et al. 2005b). Under LD cycles, the stomata of arrhythmic CCA1-ox plants did not anticipate dawn and continued to open for the entire light period. Furthermore, CCA1-ox plants had no pre-dusk closure, which demonstrates that these responses are a consequence of circadian control (Dodd et al. 2005b). Consequently, the CCA1-ox lines used more water, which suggests that circadian control of stomatal movements provides advantage by increasing water-use efficiency (Dodd et al. 2005b).
The intensity of the stomatal response to light depends on the time of the day the stimulus is given, which provides evidence of circadian gating of stomatal responses (Martin & Meidner 1971; Gorton, Williams & Assman 1993; Webb 1998). Stomatal aperture increases in response to white, red and blue light. The stomatal responses to blue light are mediated by PHOT (Kinoshita et al. 2001, 2003) and, possibly, the carotenoid zeaxanthin (Frechilla et al. 1999; Talbott et al. 2003). However, the basis of stomatal responses to red light is less clear. PHY-mediated stomatal opening could be observed in npq1-2, a zeaxanthin mutant, but not in the wild type (Talbott et al., 2003). The establishment of a link between PHY and stomatal movements has been controversial (Karlsson 1988), and many of the responses to red light have been attributed to a photosynthetic component of stomatal regulation.
Sensitivity to red, blue and white light is maximal in the early to the middle of the subjective day and less effective during the subjective night (Dodge, Marsh & Tallman 1992; Gorton et al. 1993). In contrast, dark is more effective at closing the stomata during the subjective night in LL (Martin & Meidner 1971). Green light was not thought to be a biologically active signal in plants, but there is accumulating evidence that green light can reverse the effects of blue light. This may be physiologically significant because foliage cover removes more blue than green light (Klein 1992). Blue light induced stomatal opening, which is maximal in the morning, can be reversed by green light (Talbott et al. 2006). Guard cells are most sensitive to green light in the morning and are almost insensitive to green light during the rest of the day (Talbott et al. 2006).
Other stimuli such as indole-3-acetic acid (IAA), K+ anions and fusiccocin are less effective at inducing stomatal opening during the night in the C3 plant Commelina communis than during the day (Snaith & Mansfield 1985, 1986). The signalling molecule abscisic acid (ABA) promotes stomatal closure but is less effective when given in the early to middle part of the subjective day (Correia et al. 1995). The rhythmic sensitivity to ABA favours CO2 uptake in the morning before water becomes a limiting factor in the late afternoon. Circadian gating also probably allows stomata to respond appropriately to signals in a phase-specific manner. For example, in C3 and C4 plants, high levels of leaf auxin, in the absence of circadian gating, would promote stomatal opening at night, resulting in water loss with no gain in carbon fixation (Webb 1998).
Gating of responses to low temperature
In cold acclimation, plants acquire tolerance to freezing temperatures after exposure to non-freezing low temperatures (LT). When Arabidopsis plants are exposed to LT, the expression of a family of transcription factors called C-REPEAT BINDING FACTOR (CBF1-3, also known as DEHYDRATION-RESPONSIVE ELEMENT-BINDING PROTEIN 1b, DREB1c and DREB1a, respectively) are induced rapidly (Vogel et al. 2005). The CBF genes, in turn, induce transcription of more than 100 genes known as the CBF regulon, which confers not only enhanced freezing and chilling tolerance but also salt and drought tolerance (Liu et al. 1998; Fowler & Thomashow 2002; Maruyama et al. 2004; Vogel et al. 2005). However, if the CBF regulon is constitutively activated, plants have a reduced growth rate, reduced height, delayed flowering and low seed yield (Liu et al. 1998; Kasuga et al. 1999; Gilmour et al. 2000). In order to avoid those deleterious effects, the CBF regulon is very tightly regulated. During continuous exposure to LT, for example, the expression of the regulon is reduced (Zarka et al. 2003). CBF1-3 expression at both basal levels and in response to LT is regulated by the circadian clock (Fowler, Cook & Thomashow 2005). When Arabidopsis plants grown at 24 °C were exposed to 4 °C under LL, the maximum CBF1-3 increase in transcription was observed when the stimulus was given in the early morning (ZT4, Fig. 4e). Similar results were observed in the expression pattern of the RELATED TO ABA-INSENSITIVE 3/VIVIPAROUS 1 (RAV1) transcription factor associated with CBF1-3. The expression of ZINC FINGER (C2H2 TYPE) FAMILY PROTEIN 12 (ZAT12), a zinc-finger transcription factor that represses the CBF1-3 regulon, was highly induced at an opposite phase of CBF1-3 (ZT16). In CCA1-ox lines, no variation in the cold-response was present (Fowler et al. 2005).
The mechanisms through which the circadian clock gates responses to LT are unknown, but Fowler et al. (2005) suggest that the LT sensor is more sensitive at ZT4 or that the CBF gene promoters have regulatory elements to respond to both low temperature and the circadian clock. Alternatively, the plant circadian clock may gate responses to LT through the regulation of the cold signal transduction pathway. Exposure of Arabidopsis to LT evokes increases in the concentration of cytosolic-free calcium ([Ca2+]cyt) (Knight et al. 1991; Plieth 1999) that, in turn, can induce the expression of CBF-related genes, such as DESICCATION-RESPONSIVE PROTEIN 29A (RD29A; Henriksson & Trewavas 2003). RD29A also responds to ABA and cyclic adenosine diphosphate ribose (cADPR; Viswanathan & Zhu 2002; Wu et al., 2003). [Ca2+]cyt responses to LT are gated by the circadian clock in LD and LL, with a maximum response at the middle of the day (Fig. 4f; Dodd et al. 2006). Both basal and LT-induced levels of RD29A expression are circadian-regulated and they correlate with the gating of LT-induced [Ca2+]cyt increases (Dodd et al. 2006).
CIRCADIAN [Ca2+]cyt OSCILLATIONS ENCODE ENVIRONMENTAL INFORMATION
In plants, changes in [Ca2+]cyt are involved in the transduction of many signals, such as ABA, auxin, CO2, blue and red light, heat, cold, and salt stress, touch and pathogen attack (Hetherington & Brownlee 2004). The stimulus-induced increases in [Ca2+]cyt have great variation in dynamics, including sustained oscillations, some of which can encode information (Evans, McAinsh & Hetherington 2001; Dodd et al. 2006). In addition to stimulus-induced oscillations of [Ca2+]cyt, which have oscillation periods in the order of minutes, [Ca2+]cyt also oscillates with a period of 24 h in LL, LD and DD, which is indicative of circadian control (oscillations in DD occur in tobacco but not in Arabidopsis; Johnson et al. 1995; Wood et al. 2001; Love, Dodd & Webb 2004). The peak of the 24 h oscillations of [Ca2+]cyt happens during the subjective day, but the exact timing depends on the photoperiod during entrainment. In short days (8 h light/16 h dark, 8L/16D) [Ca2+]cyt peaks 6–8 h after dawn, and the peak coincides with dusk, whereas in long days (16L/8D) the peak is a few hours later but [Ca2+]cyt has returned to ‘resting’ concentrations by dusk (Love et al. 2004). Thus, there is potential for photoperiodic information to be encoded in the phase of circadian and LD oscillations of [Ca2+]cyt (Love et al. 2004). The amplitude of circadian oscillations of [Ca2+]cyt increased as the light intensity increased, suggesting that information about photon flux density could also be encoded in circadian oscillations of [Ca2+]cyt (Love et al. 2004).
It is striking that there is circadian regulation of [Ca2+]cyt (Johnson et al. 1995), circadian gating of cold-induced increases in [Ca2+]cyt (Dodd et al. 2006) and that many of the abiotic and biotic stimuli, whose responses are known to be gated by the circadian clock, can signal through alterations in [Ca2+]cyt (e.g. ABA, IAA, blue and red light; Hetherington & Brownlee 2004). This may suggest that [Ca2+]cyt sits at the centre of a network integrating temporal and environmental information. Alternatively, it suggests that gating of stimuli may occur early in the signalling cascades. Until more is known about the mechanisms by which the circadian clock regulates [Ca2+]cyt and participates in gating, we can only speculate as to the architecture of this network (for reviews, see Dodd, Love & Webb 2005a; Gardner et al. 2005).
CIRCADIAN CLOCK AND SEASONAL RESPONSES
The circadian clock may act as a reference that allows plants to measure changes in the timing of external events. This is especially useful in latitudes where day length varies through the year. The ability to detect changes in day length permits anticipation of seasonal changes and induction of responses such as cold acclimation and flowering. The detection of changes in day length and the associated responses are called photoperiodism. Many physiological processes, such as bud dormancy, tuber and bulb formation, frost tolerance and flowering, are dependent on photoperiodism (Thomas 1998).
The regulation of flowering time is a good model of how photoperiodic mechanisms might work in Arabidopsis (see Hayama & Coupland 2004; Baurle & Dean 2006 for reviews). The perception of day length in Arabidopsis depends on the expression of CONSTANS (CO) and FLOWERING LOCUS T (FT) and their relative phase to the LD cycle (Roden et al. 2002; Yanovsky & Kay 2002; Imaizumi et al. 2003). CO transcription is regulated by two components: GI, which provides a direct input from the circadian oscillator; and FKF1, which integrates both circadian and photoperiodic signals (Imaizumi et al. 2005; Mizoguchi et al. 2005). CO, in turn, directly induces FT, a flowering elicitor (Samach et al. 2000). GI appears to affect flowering time by regulating the phase and period of the oscillator but may also have clock-independent effects on flowering (Mizoguchi et al. 2005). Under short days, transcripts of FKF1 and CO peak in the dark phase. In contrast, under long days, FKF1 peaks during the light phase. Light activates FKF1, which targets CYCLING DOF FACTOR 1 (CDF1) for degradation. CDF1 represses CO expression. Thus, when CDF1 is degraded, CO expression levels are increased and CO transcript abundance, like FKF1, peaks before dusk (Suárez-López et al. 2001; Imaizumi et al. 2005). Control of CO also occurs at the post-transcriptional level as lines that transcribe CO constitutively still have rhythms in CO protein levels (Valverde et al. 2004). CO protein degradation is mediated by PHYB in the morning. In contrast, PHYA and CRYs counteract PHYB action in the afternoon, stabilizing CO proteins (Valverde et al. 2004). Therefore, CO proteins are only accumulated when both CO transcripts levels are high and the translated protein is stabilized by PHYA and CRYs. Finally, high levels of CO proteins activate the expression of FT, which promotes flowering (Hayama & Coupland 2004). It is not clear how the PHYA and CRYs counteracts PHYB action in flowering-time control, but gating of light signals is likely to be important in this process. In order to stabilize CO levels, PHYA must be present late in the afternoon. However, even though PHYA expression peaks at this time, PHYA protein is photolabile and only accumulates during the night (Tóth et al. 2001; Sharrock & Clack 2002).
HOW DOES THE CIRCADIAN CLOCK INCREASE FITNESS?
The complexity of interactions between the circadian clock and the environment suggests that a gain of fitness might be conferred by these signalling mechanisms. Circadian clocks have arisen many times during evolutionary history, with a certain degree of convergence, which may indicate that a functional circadian clock with certain characteristics results in selective advantage (Dunlap 1999; Young & Kay 2001). However, there are surprisingly few clear measurements of the fitness benefits conferred by a correctly operating oscillator. When the period of the circadian clock (τ) is similar to the period of the environment (T), it is said that both rhythms are resonant. The selective advantages conferred by the clock, however, are most clearly seen in non-resonant cycles, when τ mismatches T. As an example, tomato plants grew higher, with greener and larger leaves, under 12L/12D cycles (T = 24 h) than the plants grown under 6L/6D cycles (T = 12 h) or 24L/24D cycles (T = 48 h), even though the total amount of light in all treatments was equal (Highkin & Hanson 1954). More striking are the experiments using cyanobacteria or Arabidopsis-period mutants under cycles with different T (Ouyang et al. 1998; Woelfle et al. 2004; Dodd et al. 2005b). When strains of cyanobacteria with different τ(22 h, 25 h or 30 h) were grown together under 11L/11D cycles (T = 22 h), the short-period mutant strain (τ = 22 h) outgrew the other strains. In contrast, under 15L/15D cycles (T = 30 h), the dominating strain was the one with a longer period and, similarly, the dominating strain in 24 h cycles (T = 24 h) was the wild-type strain (τ = 25 h; Ouyang et al. 1998). Circadian resonance was also found to be important in Arabidopsis (Dodd et al. 2005b). Both the short-period mutant toc1-1 (τ ≅ 21 h) and the long-period mutant ztl-1 (τ ≅ 28 h) had enhanced fitness traits (biomass, photosynthesis and competitive advantage) when grown in environmental rhythms that were matched to their endogenous rhythms. The mutant toc1-1 performed best in an environmental rhythm of 10L/10D (T = 20 h) and ztl-1 did best in 14L/14D (T = 28 h). Similarly, the wild type did better in 12L/12D (T = 24 h) than in 20 h or 28 h cycles. Furthermore, arrhythmic CCA1-ox performed worse than the wild types grown under 12L/12D (Dodd et al. 2005b). CCA1-ox lines also produce less viable seed under very short days (Green et al. 2002), but the interaction of the circadian clock with the flowering-time apparatus makes the interpretation of such a result difficult.
The circadian clock is likely to increase the fitness of plants through many mechanisms, including (1) temporal compartmentation of metabolic processes; (2) anticipation of daily environmental changes; (3) optimization of the turnover rate of proteins; (4) anticipation of seasonal environmental changes; and (5) gating of environmental signals. We discuss each of these possibilities in turn.
1Temporal compartmentation of metabolic processes. The circadian clock allows organisms to separate the occurrence of two incompatible mechanisms in time. Temporal organization can protect biochemical pathways that are photoinhibited or easily photodamaged by light (Pittendrigh 1993). The most striking example of temporal compartmentation found in plants is Crassulacean acid metabolism (CAM). CAM plants fix CO2 through two mechanisms that are separated in time but not space (for reviews, see Borland & Taybi 2004; Hartwell 2005). In CAM plants, the stomata open during the night when PHOSPHOENOLPYRUVATE CARBOXYLASE (PEPC) fixes CO2, producing malic acid, which is stored inside the vacuole. During the day, the malic acid is released from the vacuole and is decarboxylated. The released CO2 accumulates inside the mesophyll, causing stomatal closure, while ribulose bisphosphate carboxylase oxygenase (Rubisco) refixes the CO2, as in C3 plants (Hartwell 2005). As the stomata are closed during the day and opened during the night, water loss is reduced. In order to prevent futile cycles of carbon, the timing of the metabolic pathways involved in CAM biochemistry is regulated. Several orthologues of the Arabidopsis circadian clock genes are found in the CAM plant Mesembryanthemum crystallinum, but the mechanisms by which the temporal control of CAM is brought about are still under investigation (Boxall et al. 2005). CAM photosynthesis is energetically expensive, but it results in increased water-use efficiency. Moreover, the accumulation of CO2 because of malic acid decarboxylation inside the mesophyll inhibits photorespiration. In the balance between the energetic costs of CAM and the benefits provided by it, there is an increase of selective advantage in environments where water availability and photorespiration are key selective pressures (Borland & Taybi 2004).
2Anticipation of daily environmental changes. Rhythms of leaf movements, driven either by differential growth changes in the upper and lower part of the Arabidopsis petiole or by turgor changes in the pulvini of legumes (for a review, see Webb 2003), may increase photosynthetic rates by enabling tracking of the sun during the morning (Pastenes, Pimentel & Lillo 2005). More importantly, when sunlight is at its greatest intensity, in the middle of the day, legume leaves stay almost vertical, thus reducing sunlight absorption and avoiding photodamage, overheating and increased water loss (Pastenes et al. 2005). Water loss is also prevented by the anticipation of dawn and dusk by the stomatal guard cells (see ‘Gating of stomatal responses to the environment’; Dodd et al. 2005b). The synthesis of phenylpropanoids before dawn allows the preparation of ameliorative mechanisms before the onset of photodamage (Harmer et al. 2000). The anticipation of dawn is another key process for the plant as the photosynthetic apparatus can assemble prior to light availability. A correctly functioning clock allows plants to accumulate more chlorophyll and fix more carbon, which results in 45% more biomass, faster growth and lower mortality (Dodd et al. 2005b).
3Optimization of the turnover rate of proteins. The timing of protein synthesis might also optimize protein turnover when there is a periodic increase of protein damage. During light-harvesting, many proteins and pigments are photodamaged. The time of highest expression of proteins related to the light-harvesting complex and other related proteins coincides with the timing of higher photon flux in the day (Millar & Kay 1996; Harmer et al. 2000). The timing of expression of those proteins and pigments results in almost uniform amounts of those components during the light phase (Prombona & Argyroudi-Akoyunoglou 2004). Indeed, plants are chlorotic under photoperiods with light phases longer than 18 h (Withrow & Withrow 1949) and have lower chlorophyll content in non-resonant conditions (Dodd et al. 2005b).
4Anticipation of seasonal environmental changes. Besides the advantages of freezing tolerance development, bulbs and tuber formation, bud dormancy and other phenomena related to winter avoidance and tolerance, photoperiodism also allows the synchronous flowering of plant species, which enhances outbreeding and therefore increases genetic recombination. At the beginning of spring, photoperiodism can also allow small plants to exploit seasonal niches. In the temperate woodlands, some fast-growing plants can use the days before the leaf canopy is formed to grow, flower and set seed (Thomas 1998). In addition, different Arabidopsis ecotypes have a wide range of free-running circadian periods, which are positively correlated with the latitude of their collection (Michael et al. 2003b). It is perhaps counter-intuitive that there may be selection pressure for clock periods greater than 24 h at high latitudes, because circadian periods greater than the environmental rhythm can be disadvantageous to plants (Highkin & Hanson 1954; Dodd et al. 2005b). However, circadian clocks with periods not exactly matched to the period of the environment are associated with more flexible and accurate entrainment mechanisms on long days (Roennenberg, Daan & Merrow 2002). Longer circadian periods, which deviate from environmental rhythm periods, may be expected in latitudes where day length varies greatly to optimize photoperiodic perception. Furthermore, a longer rhythm is required for the correct phasing of metabolic pathways under days with very long light phases (Michael et al. 2003b).
5Gating of environmental signals. Gating might contribute to the optimization of environmental signal processing. Previously, it has been suggested that gating of stomatal responses to ABA and auxin increases water-use efficiency by allowing responses that favour CO2 uptake in the day while minimizing water loss at night (Webb 1998). The gating mutant elf3-1 is less viable than the wild type, in both long days (16L/8D) and very short days (4L/20D) and performs even worse than the circadian arrhythmic lines CCA1-ox and LHY-ox in the same conditions (Green et al. 2002). These data are difficult to interpret because early flowering time, and other pleiotropic effects, may affect seed quality independently of the circadian clock, but they do provide evidence that the gating of responses to the environment is a fundamental component in plant and environment interactions and is essential for the success of land plants.
The circadian clock structure and function reflect the characteristics of the rhythmic environment in which they evolved. Entrainment mechanisms that continuously adjust the phase of the circadian clock in relation to the environment were selected because the day length is continuously changing. Photoperiodism was selected in response to seasonal changes that could be predicted by measuring changes in day length. The circadian clock is the result of the selective pressure from a rhythmic environment that continuously challenges organisms with predictable environmental changes.