Ambient temperature response establishes ELF3 as a required component of the core Arabidopsis circadian clock

Abstract

Circadian clocks synchronize internal processes with environmental cycles to ensure optimal timing of biological events on daily and seasonal time scales. External light and temperature cues set the core molecular oscillator to local conditions. In Arabidopsis, EARLY FLOWERING 3 (ELF3) is thought to act as an evening-specific repressor of light signals to the clock, thus serving a zeitnehmer function. Circadian rhythms were examined in completely dark-grown, or etiolated, null elf3-1 seedlings, with the clock entrained by thermocycles, to evaluate whether the elf3 mutant phenotype was light-dependent. Circadian rhythms were absent from etiolated elf3-1 seedlings after exposure to temperature cycles, and this mutant failed to exhibit classic indicators of entrainment by temperature cues, consistent with global clock dysfunction or strong perturbation of temperature signaling in this background. Warm temperature pulses failed to elicit acute induction of temperature-responsive genes in elf3-1. In fact, warm temperature-responsive genes remained in a constitutively “ON” state because of clock dysfunction and, therefore, were insensitive to temperature signals in the normal time of day-specific manner. These results show ELF3 is broadly required for circadian clock function regardless of light conditions, where ELF3 activity is needed by the core oscillator to allow progression from day to night during either light or temperature entrainment. Furthermore, robust circadian rhythms appear to be a prerequisite for etiolated seedlings to respond correctly to temperature signals.

The rotation of Planet Earth creates predictable daily environmental fluctuations of light and dark along with concomitant oscillations in temperature. The circadian clock is an endogenous timekeeper that anticipates these predictable changes in the environment, confers rhythmic behavior to biological processes, and optimally phases biological activities to specific times of the day. Circadian clocks are widespread in nature, and processes under their control range from sleep–wake cycles in humans to daily expression of photosynthetic genes in plants. Clocks also time seasonal responses, such as the flowering transition in many plant species.

Core molecular oscillators in eukaryotes incorporate interlocked transcription–translation feedback loops. In the model plant Arabidopsis thaliana, three such loops are critical to generation and maintenance of circadian rhythms. The loop first discovered is composed of the pseudoresponse regulator TOC1 (1) and two partially redundant Myb-like transcription factors, CCA1 (2) and LHY (3). Morning expression of CCA1 and LHY represses TOC1 expression by binding to its promoter (4), and circadian accumulation of TOC1 in the evening helps to induce CCA1 and LHY. A second morning-phased loop includes two TOC1-related proteins, PRR7 and PRR9 (5, 6). CCA1 and LHY induce PRR7 and PRR9 expression, whereas the two PRRs subsequently repress CCA1 and LHY. In a third evening-phased loop, GI activates TOC1 expression, whereas GI itself is negatively regulated by CCA1/LHY and TOC1 (7, 8).

Light and temperature environmental cues, or zeitgebers, set clock feedback loops to the correct time of day (9). Among factors believed to be important for entrainment is EARLY FLOWERING 3 (ELF3) (10), which encodes a highly conserved plant-specific nuclear protein (11). In the Arabidopsis elf3 mutant, normally rhythmic CCR2 and CAB2 expression is arrhythmic under continuous light (LL) (12), although rhythms have been reported in continuous darkness (DD) (12, 13). ELF3 protein normally accumulates in the evening (11), and, in the context of the clock, ELF3 is thought to allow progression through this light-sensitive phase (14). This hypothesis is bolstered by the observations that the elf3 clock seems to arrest following approximately 11 h in LL (14) and that light resetting of the clock is reduced when ELF3 is present in excess (13). Recently, ELF3 has been described as a substrate adaptor that promotes interaction between the E3 ubiquitin ligase COP1 and GI to influence Arabidopsis flowering time (15).

Genetic components governing clock entrainment by temperature remain largely undefined in Arabidopsis. Exceptions are PRR7 and PRR9, which are partially redundant and required for clock temperature entrainment (16). Because both PRR7 and PRR9 also participate in clock responses to light (17), temperature entrainment and light entrainment appear to share components. However, the relative contributions of light and temperature to entrainment of the Arabidopsis clock are not well understood, except for the possibility of two oscillators capable of distinguishing light and temperature cues (18).

Our investigation of the circadian clock in photocycle-entrained elf3-1 revealed arrhythmic TOC1 expression in this null mutant over a range of free-running conditions, including DD, a condition where this mutant was previously found to exhibit rhythms (12, 13). Therefore, circadian gene expression in elf3-1 was examined in dark-grown, or etiolated, seedlings exposed to thermocycles in place of photocycles to determine if light signals alone cause clock arrhythmia. Rhythmic gene expression was not sustained in thermocycle-entrained etiolated elf3-1 seedlings released into continuous temperatures; furthermore, the clock in this mutant was unable to entrain to thermal cues. ELF3 was required to restrict normal expression of temperature-sensitive genes like PRR7 and PRR9 to the day and to establish enhanced sensitivity of these genes to warm temperature cues during the night. These results indicate that ELF3 serves as part of the core molecular oscillator instead of solely modulating clock sensitivity to environmental cues.

Results

Circadian Rhythms in Dark-Grown Arabidopsis Seedlings Require ELF3.

Rhythmic gene expression was evaluated in the null elf3-1 mutant with a transcriptional fusion of firefly LUC⁺ to the TOC1 promoter (TOC1::LUC⁺). WT and elf3-1 seedlings were grown in photocycles [light/dark (LD)] for 6 days and then released at ZT0 (time of dark-to-light transition) on day 7 into free-running conditions, either LL or DD. As expected, TOC1::LUC⁺ expression in WT was robustly rhythmic on release into either LL or DD, and cycles persisted for at least 5 days (Fig. 1 A and B). These seedlings exhibited statistically significant rhythms under both free-running conditions with relative amplitude error (RAE) values ≤0.6 for all individuals (Fig. S1 A and B and Table S1). The WT period on release into LL and DD was 24.47 ± 0.09 h (±SEM) and 26.52 ± 0.49 h, respectively. TOC1 expression in LD-entrained elf3-1 seedlings released into LL was arrhythmic, as was TOC1 expression in elf3-1 seedlings released into DD (Fig. 1 A and B). Estimated period values were not returned for most elf3-1 seedlings, and in the limited instances in which period values were returned for elf3-1, these were accompanied by an RAE >0.6 or a period outside a physiologically relevant range of 15–30 h in almost all cases (Fig. S1 A and B and Table S1). Poor rhythms for elf3-1 in DD were not confined to TOC1 expression, because comparable arrhythmic expression was observed from the clock-driven output promoters CCR2::LUC and FKF1::LUC⁺ (Figs. S1B and S2 and Table S1). Thus, elf3 mutant seedlings do not exhibit circadian rhythms under continuous conditions following entrainment with photocycles.

Fig. 1.

ELF3 is required for sustained rhythms in Arabidopsis seedlings. Mean bioluminescence from the TOC1::LUC⁺ reporter in WT (•) and elf3-1 (□) individuals (n = 15) entrained in LD for 6 days and then released at ZT0 on day 7 into LL (A) or DD (B). TOC1::LUC⁺ expression after HC entrainment in DD for 6 days and then release on day 7 into continuous 22 °C (C) or 18 °C (D) while maintaining DD. Representative data are shown for three experimental replicates.

The profound disruption of circadian rhythms in LD-entrained elf3-1 regardless of free-running condition led us to evaluate the extent of the clock defect in this mutant. To eliminate light effects on the oscillator, rhythmic gene expression was examined in dark-grown, or etiolated, WT and elf3-1 seedlings, where light exposure was limited to a brief 3-h white-light pulse to promote seed germination. Following entrainment in thermocycles differing by 4 °C [hot/cold (HC)], etiolated WT seedlings exhibited sustained rhythmic TOC1::LUC⁺ expression for at least 5 days in continuous 22 °C warm or 18 °C cool conditions (Fig. 1 C and D). The mean estimated free-running period in WT was 26.70 ± 0.22 h and 25.97 ± 0.15 h in warm and cool conditions, respectively (Fig. S1B and Table S1). Rhythms from all WT spots were good enough to be assigned RAE values ≤0.6 (Fig. S1 C and D). Similar rhythmic TOC1 expression was also observed in etiolated seedlings free running under warm conditions following entrainment in thermocycles differing by 10 °C (Fig. S1 C and D and Table S1). Therefore, the 4 °C change in ambient temperature was an effective zeitgeber comparable to the larger temperature differentials described elsewhere (16). Importantly, the strong rhythmic expression of TOC1 indicates that the clock system in the etiolated seedlings is similar to that in light-grown seedlings instead of the circadian clock in root tissue, where TOC1 does not appear to cycle (19). Expression from the FKF1::LUC⁺ reporter was also strongly rhythmic in etiolated WT seedlings (Fig. S1C and Table S1). On the other hand, WT seedlings harboring the CAB2::LUC reporter did not consistently exhibit robust rhythms across several independent experiments, likely because CAB2 expression is light-dependent (20).

In contrast to WT, etiolated elf3-1 seedlings exposed to equivalent HC entrainment conditions did not show obvious circadian TOC1::LUC⁺ and FKF1::LUC⁺ expression on release into either continuous warm or cool conditions (Fig. 1 C and D). Cryptic rhythms were not present in elf3-1, because curve fit analysis failed to detect coherent rhythms in the traces from the mutant background (Fig. S1 C and D and Table S1). Identical results were obtained with seedlings entrained in thermocycles differing by 10 °C (Fig. S1 C and D and Table S1). Thus, ELF3 is required for circadian rhythms in seedlings deprived of light exposure, indicating that the elf3 arrhythmic phenotype is independent of light cues.

ELF3 Activity Is Needed for Temperature Entrainment of the Circadian Clock.

The arrhythmic behavior in etiolated elf3-1 seedlings may have arisen from two possible sources: (i) ELF3 forms an integral component of the core oscillator, and its absence impairs clock function so that rhythms are never present, or (ii) ELF3 is a temperature zeitnehmer, in which case the clock is entrained in thermocycles but arrests after exposure to continuous temperature. To explore these possibilities, the oscillator in etiolated elf3-1 seedlings was examined for evidence of proper entrainment.

Frequency demultiplication is a hallmark of clock entrainment and function (21). In environmental cycles with time periods (T) less than 24 h, circadian rhythms “frequency demultiply,” or skip an environmental cycle(s), so that overall repeated patterns still occur approximately every 24 h. In contrast, organisms with nonfunctional oscillators display repeated patterns that typically reflect the shorter environmental T cycle (22–24). To determine whether etiolated elf3-1 seedlings exhibit any signs of temperature entrainment, seedlings were entrained in HC (T = 24 h) and then moved to 12-h thermocycles (T = 12 h; 6 h 22 °C/6 h 18 °C). Seedlings with properly entrained circadian clocks were expected to retain the T = 24-h entrainment. Accordingly, TOC1::LUC⁺ expression was examined immediately on transfer of WT and elf3-1 seedlings to T = 12 h. In WT, the original T = 24-h entrainment period was evident after the transfer as a strong peak every 24 h at a phase matching that in normal HC entrainment (Fig. 2A). In addition, a prominent trough in expression was apparent on the first day after transfer, and to a lesser degree 24 h later, which matched the nadir in TOC1 expression observed in free-running conditions following T = 24-h entrainment, and the transition between the trough and major peak showed anticipation typical of clock-driven expression. A driven rhythm was apparent as a weaker peak starting ∼16 h after the change in entrainment conditions. The persistence of the 24-h timing in WT indicates correct clock entrainment during the preceding thermocycles. The elf3-1 seedlings given the same treatment lacked all these hallmarks of entrainment (Fig. 2B). On transfer to the T = 12-h thermocycles, TOC1::LUC⁺ expression in elf3-1 seedlings immediately adopted a pattern reflecting the changed external temperature cycles and showed no indication of the prior exposure to the T = 24-h conditions. Notably, the waveform in the mutant background was invariant, exhibiting a single square peak and trough consistent with expression driven by external temperature changes instead of an endogenous circadian rhythm. The behavior of elf3-1 seedlings indicates that the oscillator in this mutant could not generate strong circadian rhythms; therefore, the circadian clock itself appeared to be impaired without ELF3 activity instead of the oscillator arresting in the changed environmental conditions.

Fig. 2.

ELF3 is required for clock temperature entrainment in dark-grown seedlings. Mean TOC1::LUC⁺ expression from HC-entrained etiolated WT (A; •) and elf3-1 (B; □) etiolated seedlings transferred at ZT0 into T = 12 thermocycles. Representative data from two independent experimental replicates are shown, where six groups of ∼50 etiolated seedlings for each genotype were imaged for 72 h in T = 12 thermocycles in complete darkness.

Clock Sensitivity to Warm Temperature Is Not Altered by Overex-pression of ELF3.

To examine the contribution of ELF3 to temperature entrainment of the Arabidopsis clock, oscillator sensitivity to resetting by a step up in ambient temperature was determined in WT and an ELF3 overexpression (ELF3OX) line by constructing a tPRC. A tPRC represents the extent of phase change induced by a resetting temperature signal given at different times across a normal circadian day. The ELF3OX line, which contains ELF3 transcript at ≈50 times endogenous levels (11), was tested, because, unlike elf3-1, this line exhibits robust circadian rhythms that are only of a moderately long period in light-grown seedlings (13). Etiolated ELF3OX seedlings also exhibited robust rhythms after temperature entrainment comparable to those observed in WT (Table S1). If ELF3 acts to repress temperature cues during normal entrainment, the ELF3OX would be more resistant to resetting by a temperature change. An analogous phenotype for light resetting of the oscillator in ELF3OX has been described previously (13).

The apparent phase change of TOC1::LUC⁺ expression after a step up from 18 °C to 22 °C was determined in etiolated WT and ELF3OX seedlings that experienced a day of free run at 18 °C after 6 days in 10 °C thermocycles. Seedlings were transferred to 22 °C at 3-h intervals starting at subjective dawn [circadian time (CT) 0]. Phase differences relative to the CT0 sample were calculated so that phase advances and delays were positive and negative values, respectively (Fig. 3). In WT, the 4 °C shift up in temperature caused phase delays, with the magnitude of change increasing as the subjective day progressed to subjective night, with a maximal delay of just under 12 h achieved at CT18. A phase advance of nearly 6 h occurred after CT18, and following this breakpoint, advances were maintained for the remainder of the subjective night. The shape of this tPRC from etiolated seedlings was analogous to that of light-grown Arabidopsis seedlings, where the clock was reset by light pulses (13), and opposite to that of the resetting pattern elicited by cold pulses (16). The response of the oscillator in ELF3OX was nearly identical to that of WT (Fig. 3), indicating that ELF3 did not have a substantial effect on resetting of the clock by warm cues. Therefore, ELF3 on its own does not appear to modulate temperature information feeding into the Arabidopsis oscillator directly.

Fig. 3.

WT and ELF3OX seedlings are similarly responsive to resetting by warm temperature. Six-day-old WT (▲, solid line) and ELF3OX (□, dashed line) etiolated seedlings entrained in HC were transferred to 22 °C after free running for 24 h at 18 °C every 3 h starting at CT0. Phase advances and delays are positive and negative values, respectively. Each point and error bars represent the mean ± SEM of three experimental replicates. Data from a single 24-h period are double-plotted to aid in identification of the features in the tPRC.

ELF3 Is Required for Appropriate Response to Ambient Temperature Cues.

The circadian clock and other physiological responses adjust to environmental temperature changes through altered gene expression (25–27). Expression of the integral clock component GI is elevated by warm temperature (27 °C) (28). The activities of PRR7 and PRR9 are required for temperature entrainment of the circadian clock (16), and their expression is induced following ambient temperature increases (see below). Outside of the core clock, PIF4 is required for high-temperature signaling and growth responses, and a shift to 28 °C triggers its induction (29). These temperature-responsive genes were examined to evaluate whether ELF3 directly contributes to temperature responses or if defects in elf3 are more consistent with aberrant clock function.

The acute response of PRR7, PRR9, PIF4, and GI to high temperature was measured in dark-grown WT and elf3-1 seedlings. Each genotype was grown in typical HC conditions for 7 days; while under these conditions, sets of seedlings received a 3-h 28 °C pulse at either ZT4 (midday) or ZT16 (midnight). As observed in light-grown WT seedlings (7, 17, 24, 30), abundance of PRR7, PRR9, PIF4, and GI transcripts in WT exhibited diurnal patterns of higher expression at ZT4 and limited levels at ZT16 (Fig. 4). A 3-h warm pulse at ZT16 given to WT induced PRR7 transcript 20-fold over the untreated control, which was an order of magnitude greater than at ZT4 (Fig. 4A and Table 1). Stronger night-phased responses were also observed for PIF4 and GI: Inductions were 7.8- and 17.2-fold, respectively, after a warm treatment at ZT16, but changes were <2-fold at ZT4 (Fig. 4 C and D and Table 1). Clearly, induction by warm temperature was set to a specific time of day, and the response was strongest during the night in etiolated seedlings.

Fig. 4.

High-temperature-induced gene expression is altered in elf3-1. PRR7 (A), PRR9 (B), PIF4 (C), and GI (D) transcript levels in etiolated seedlings before and after a 3-h 28 °C pulse, as indicated at the bottom. In each panel, genotypes are depicted in sets of four, which were composed of WT (black bars), elf3-1 (open bars), ELF3OX (light grey bars), and CCA1OX (dark grey bars). Background indicates time of day: White is ZT4, and gray is ZT16. All samples are normalized to the WT 22 °C sample at ZT4 in each panel. Values and error bars are mean ± SEM of three experimental replicates.

Table 1.

Loss of ELF3 substantially alters induction of gene expression by a warm temperature treatment

Time*	ZT4/ZT16
	Genotype
Gene	WT	elf3-1	ELF3OX	CCA1OX
	Fold change ± SEM†
PRR7	1.7 ± 0.1/20.4 ± 2.3	1.1 ± 0.1/0.9 ± 0.2	2.0 ± 0.3/8.9 ± 1.4	1.4 ± 0.1/1.5 ± 0.2
PRR9	2.3 ± 0.3/2.5 ± 0.9	0.7 ± 0.2/0.4 ± 0.1	4.3 ± 1.3/3.3 ± 0.4	2.2 ± 0.5/1.0 ± 0.1
PIF4	1.3 ± 0.3/7.8 ± 0.7	1.6 ± 0.5/1.3 ± 0.2	1.5 ± 0.2/7.4 ± 0.4	1.5 ± 0.3/2.2 ± 0.7
GI	1.5 ± 0.2/17.2 ± 1.7	0.6 ± 0.3/0.7 ± 0.1	1.8 ± 0.3/12.7 ± 3.6	1.0 ± 0.1/1.4 ± 0.5

*Etiolated seedlings grown in thermocycles were given a 3-h 28 °C pulse at ZT4 (midday) or ZT16 (midnight).

†Fold change of indicated transcripts normalized to untreated control samples. Values are averages of three independent experimental replicates.

Gene expression in elf3-1 was substantially different from that in WT. PRR7 expression in this mutant showed no significant high-temperature induction regardless of the time of day (Fig. 4A and Table 1). This was attributable to constitutively elevated PRR7 expression, because PRR7 transcript level in untreated elf3-1 seedlings matched that of WT individuals receiving the ZT16 high-temperature pulse (Fig. 4A). Basal PRR9 expression in elf3 mutant seedlings was also substantially elevated, especially at ZT16, where the expression in this mutant was ∼30-fold higher than levels reached in WT following the high-temperature pulse (Fig. 4B). High-temperature treatment of elf3-1 led to a reduction in PRR9 expression at both times, but these transcript levels remained substantially higher than the maximal response achieved in WT under any condition. Like PRR7 and PRR9, basal expression of both PIF4 and GI in elf3-1 was drastically higher than in WT at ZT16 (Fig. 4 C and D) and GI expression in elf3-1 was repressed somewhat in response to the elevated temperature. Collectively, these data indicate that normal repression of PRR7, PRR9, PIF4, and GI during the night was not achieved in elf3-1 and, as a result, response to warm temperature cues was dramatically altered in the elf3 mutant. Thus, ELF3 appeared critical to normal gating of the expression for these warm temperature-responsive genes.

Because ELF3 function appeared to be required for nighttime repression of these warm-responsive genes, expression of PRR7, PRR9, PIF4, and GI in response to warm treatment was evaluated in ELF3OX and the arrhythmic CCA1 overexpression (CCA1OX) line (2). ELF3OX was included to determine what, if any, immediate effect ELF3 had on acute high-temperature gene induction. The arrhythmic CCA1OX line served to test what effect an arrhythmic clock had on temperature responses and, therefore, to distinguish elf3-specific phenotypes from general effects induced by clock dysfunction. In ELF3OX, the expression profiles of PRR7, PRR9, PIF4, and GI largely mirrored that of WT, where these genes showed high expression at ZT4 but not at ZT16 (Fig. 4). Also a 28 °C pulse given at ZT16 was more effective at inducing expression than one given in the morning (Table 1). The absence of a notable phenotype in the rhythmic ELF3OX suggests that ELF3 does not directly influence response to temperature signals. The phenotype of CCA1OX mirrored that observed in elf3-1. In particular, basal transcript levels for PRR9, PRR7, PIF4, and GI were elevated compared with WT, and a warm treatment at either ZT4 or ZT16 had little positive effect on expression (Fig. 4 and Table 1). The similarity between CCA1OX and elf3-1 phenotypes points toward the possibility that without a functional circadian clock, Arabidopsis seedlings were unable to respond to temperature cues properly. Therefore, ELF3 appeared vital for circadian clock function; without this protein, the circadian clock was unable to produce robust rhythms, which ultimately precipitated a defect in temperature response.

Discussion

Temperature is an important environmental cue that sets circadian clocks to local time in ectothermic organisms. Rhythms dependent on the Arabidopsis clock were not sustained in thermocycle entrained etiolated elf3-1 seedlings, indicating that defects caused by loss of ELF3 extend to conditions wherein light is absent. Furthermore, the circadian clock showed no evidence of temperature entrainment in elf3-1. These findings demonstrate that the circadian system as a whole is compromised in the absence of ELF3 activity and show that ELF3 acts as more than a zeitnehmer that represses light input to the clock. Therefore, ELF3 appears to be an integral component of the core oscillator that is needed for circadian rhythms.

In contrast to our findings, previous models placed ELF3 as a zeitnehmer responsible for gating light input to the central oscillator based on the perceived light-dependent arrhythmic phenotype of the elf3 mutant. Hicks et al. (12) and Covington et al. (13) reported that light-entrained elf3 seedlings transferred to DD exhibit circadian rhythms for the CAB2::LUC and CCR2::LUC reporters, respectively. Hicks et al. (12) also described CAB2::LUC rhythms in etiolated elf3 seedlings. Expression from the CAB2 promoter rapidly damps in the dark (20), rendering CAB2::LUC problematic for analysis of circadian rhythms in the dark. Typically, WT seedlings exhibit a single discernable CAB2 expression peak, appearing within the first 24 h in DD, after which rhythmic expression becomes difficult to detect (12, 20, 31). The elf3 mutants show a comparable first peak in DD (12, 14). The question is whether this single CAB2::LUC peak in elf3 represents bona fide oscillator activity or, instead, reflects the last driven cycle of a clock that is running down after receiving a final zeitgeber before free run (32). We favor the latter interpretation and for this reason our characterization of elf3-1 focused mainly on the behavior of TOC1::LUC⁺ expression, because this reporter in WT was strongly rhythmic in the dark following light or temperature entrainment (Fig. 1). With this more appropriate reporter construct, rhythms were absent from elf3-1 under all tested conditions (Fig. 1 and Figs. S1 and S2). Similar arrhythmic behavior was observed from CCR2::LUC and FKF1::LUC⁺ in the mutant, thereby reinforcing TOC1::LUC⁺ as an accurate readout of the elf3-1 phenotype.

In addition, without close control of ambient temperature, the influence of inadvertent temperature cues on LUC expression cannot be ruled out. Indeed, CCR2 promoter is temperature-responsive (33, 34). Arabidopsis seedlings respond to 4 °C steps in temperature, but the lower limit of responsiveness for temperature cues has not been defined in this species. Strikingly, temperature steps of 1 °C entrain the Kalanchoë circadian oscillator (35). Given the possibility of similar sensitivity for Arabidopsis, we designed our experimental system to avoid zeitgeber information arising from unintended ambient temperature cycles or cues that could drive expression rhythms and mask the underlying state of the oscillator in elf3. Whether this precaution was also included in the previous studies is difficult to know.

Taking these factors into account, the absence of robust circadian expression from the distinct CCR2, TOC1, and FKF1 promoters in etiolated elf3-1 seedlings (Fig. 1 and Figs. S1 and S2) is compelling evidence of severe clock dysfunction in this background. Furthermore, dark-grown elf3-1 seedlings exposed to temperature cycles exhibited obviously arrhythmic expression of PIF4, PRR7, PRR9, and GI that was similar to the well-characterized arrhythmic CCA1OX line (Fig. 4). Nozue et al. (24) have also demonstrated that elf3-1 seedlings no longer show time of day control of hypocotyl elongation consistent with an arrhythmic clock. The simplest interpretation of this collection of arrhythmic phenotypes is that ELF3 serves a fundamental role in the core Arabidopsis circadian oscillator under all conditions.

A formal possibility exists that ELF3 is both a temperature zeitnehmer (36) and a light zeitnehmer (14). As a dual zeitnehmer, ELF3 could blunt daytime cues (light and warm temperature) to the clock during the transition to night to avoid inappropriate resetting during this part of its cycle. If this is the case, altered ELF3 activity should modify the clock’s sensitivity to temperature cues; however, ELF3OX seedlings were no more or less responsive to resetting by a shift to warm temperature compared with WT (Fig. 3). Absolute gene expression levels caused by a warm temperature shift were also similar between WT and ELF3OX (Fig. 4), rather than showing a drastically diminished response in ELF3OX as would be predicted if a primary function of ELF3 is to inhibit high-temperature cues (Fig. 3). Furthermore, elf3 mutant seedlings lacked obvious hypersensitivity to warm treatments; instead, these plants were essentially nonresponsive to acute warm pulses that trigger strong induction of gene expression in WT (Fig. 4). By these measures, the principal role of ELF3 in etiolated seedlings appears not to be that of a temperature zeitnehmer.

Several contradictions have been observed for light responses in the elf3 mutant; for example, this mutant shows excessive hypocotyl elongation under white-light and monochromatic light conditions (10, 37), which suggests that ELF3 promotes light signaling; it also has enhanced acute induction of CAB2 expression by light (14), which indicates that ELF3 represses light signals. Similarly, ELF3 physically interacts with phyB in vitro (11), consistent with ELF3 acting downstream of phyB, but hypocotyl and flowering time phenotypes of elf3 mutants are additive with those of a phyB mutant (10, 37). Additive phenotypes are also found with elf3 combined individually with phyA or hy4, a cryptochrome 1 mutant background (37). Clearly, the relationship between ELF3 and light signaling components is more complicated than a model in which ELF3 is a direct repressor within a linear phototransduction pathway.

A model that reconciles these contradictions in the elf3 phenotype is one in which ELF3 activity is required by the core oscillator to produce circadian rhythms. Reed et al. (37) suggested that aberrant circadian rhythms may be the basis for the apparent light phenotypes of elf3. The gating role of the plant oscillator attenuates or intensifies cellular activities to create permissive or nonpermissive times of the day for responses, including light, hormone, and, as shown here, temperature signaling (38, 39). In arrhythmic clock mutants, the oscillator is likely locked in one particular phase, such as morning, and the activities and responses observed in such a mutant reflect the phase state of the arrested oscillator. We propose that the clock in elf3 exists in a constitutively day state, not unlike the arrested oscillator predicted previously (14). In fact, the hyperresponsiveness of the CAB promoter in an elf3 mutant appears to stem from this clock-gated pathway being locked into a permissive state for light signals (14). Because PRR7, PRR9, PIF4, and GI are normally daytime-phased by the oscillator, their unchanging high expression in etiolated elf3 is in line with the notion of a dysfunctional clock stuck in a state more like day than night.

The excessive hypocotyl elongation observed in elf3 is consistent with the idea of an arrested clock that no longer properly gates growth. Recently, the arrhythmic CCA1OX line was used to show that loss of circadian-regulated expression of the growth-promoting transcription factors PIF4 and PIF5 is the molecular basis for the long hypocotyl phenotype in this clock mutant (24). We propose the same molecular explanation for the ELF3 hypocotyl growth phenotype. PIF4 and PIF5 together promote hypocotyl elongation, and their action is limited by phyB-mediated light-dependent degradation, effectively limiting growth to just before dawn. Because PIF4 expression is constitutively high in elf3-1, PIF4 protein is also likely elevated and could promote the exaggerated hypocotyl growth typical of elf3 mutants, as it does in seedlings in which PIF4 is overexpressed (40). Loss of clock control over PIF4 in elf3-1 is consistent with previous hypocotyl measurements for this mutant (10): Hypocotyl elongation is greatest under short day conditions, where longer nights support elevated PIF4 abundance, and hypocotyls are shorter in long days, where the longer light periods serve to repress accumulation of PIF4. Therefore, a fully arrhythmic clock readily explains the apparent light-associated hypocotyl growth phenotypes for elf3.

The elf3 circadian phenotype suggests a role for ELF3 in the central oscillator as a contributor to the X and Y activities suggested by Locke et al. (41). Loss of X activity is expected to diminish expression of CCA1/LHY, and both CCA1 and LHY are expressed at very low levels in elf3-1 (42). Conversely, these genes are expressed more highly in ELF3OX. Y function is modeled to promote TOC1 expression, and GI has been proposed as a component of Y activity. TOC1::LUC⁺ is expressed at elevated levels in elf3-1 (42), as is GI (43). The low levels of CCA1/LHY expression could, in part, explain the markedly high TOC1 and GI expression (5, 6), but another potential contributor may be the effect of the elf3 mutation on GI accumulation. ELF3 promotes degradation of GI by acting as a scaffold for the association of GI with the E3 ubiquitin ligase COP1 (15); consequently, ELF3 is formally a repressor of Y activity. Thus, experimental data suggest that ELF3 may contribute to both X and Y activity. Interestingly, PRR7 and PRR9 expression is also elevated in elf3, and what role ELF3 may have in repressing PRR7 and PRR9 is presently unclear, because these regulatory details are missing from the present mathematical model of the clock. Clearly, further work is needed to determine whether ELF3, in fact, serves in this dual capacity and to define the biochemical means by which this protein acts. Nevertheless, ELF3 is most likely an essential part of the core oscillator, allowing passage into the nighttime phase of the circadian cycle under all conditions.

Materials and Methods

Plant Material and Growth Conditions.

A. thaliana (Columbia) was used unless otherwise noted. Seeds were surface-sterilized and plated on 1× Murashige and Skoog basal salt medium with 0.8% agar and 3% (wt/vol) sucrose (MS plates) (44). After stratification in the dark at 4 °C for 3 days, plates were transferred to a Percival (percival-scientific.com) incubator set to the indicated light or temperature conditions. Light entrainment was a 12 h light/12 h dark photoperiod at a continuous temperature of 22 °C. Light was supplied by cool-white fluorescent bulbs at 50 μmol·m⁻²·sec⁻¹. For etiolated seedlings, imbibed and stratified seeds were exposed to 3 h of white light at the same fluence rate, wrapped in two layers of foil, and transferred to a dark incubator. Unless otherwise noted, etiolated seedlings were grown in thermocycles of 12 h 22 °C/12 h 18 °C. The TOC1::LUC⁺ (45), CAB2::LUC (20), and FKF1::LUC⁺ (46) reporters were in WT or the elf3-1 mutant (10). CCR2::LUC⁺ was in Wassilewskija (Ws) with elf3-3 (Ws) (47). The TOC1::LUC⁺ reporter was crossed into the ELF3OX line from Liu et al. (11) and into the CCA1OX line from Wang and Tobin (2).

Bioluminescence Assays.

Seedlings were sprayed with ∼1 mL of 5 mM firefly luciferin (Biosynth, biosynth.com; Gold Biotechnology, goldbio.com) prepared in 0.01% (vol/vol) Triton X-100 (Sigma Aldrich) 24 h before imaging. While in continuous conditions, seedlings were imaged every hour (etiolated) or 2.5 h (light-grown) with a Berthold NightOwl II imaging system (berthold.com). Bioluminescence data from the images was extracted with Berthold WinLight software and curve-fitting analysis of bioluminescence traces with fast Fourier transform-nonlinear least squares (FFT-NLLS) using the Biological Rhythms Analysis Software System 3.0 (48).

Temperature Phase Response Curve.

For each experimental replicate, nine MS plates were sown with six spots each of WT and ELF3OX seeds. Each spot was composed of ≈50 seeds. Seeds were given a 3-h white-light pulse to induce uniform germination and were then grown in the dark under thermocycles of 12 h 22 °C/12 h 12 °C. After 6 days, all plates were transferred to 18 °C at ZT0, and starting 24 h later, one plate was transferred to 22 °C every 3 h. Bioluminescence from TOC1::LUC⁺ was recorded for 7 days after transfer. Period and phase were calculated with FFT-NLLS (48), and the first 24 h were excluded from analysis to measure stable phase change more accurately. The phase was divided by the period and multiplied by 24 to determine the circadian time (CT, in hours) phase. CT phase values were averaged for three experimental replicates and compared with the average CT0 phase. Phase delays were calculated as negative numbers, whereas phase advances were calculated as positive numbers.

Expression Analysis by Quantitative RT-PCR.

Seedlings grown for 7 days on Whatman filter paper in the dark under the indicated thermocycles were transferred at ZT4 or ZT16 in the dark via the filter paper to plates preequilibrated to the pulse temperature (28 °C) or the current growth chamber temperature (22 °C or 18 °C). Three hours after transfer, seedlings were harvested under a green safe light and immediately placed in liquid N₂. RNA was extracted using a Qiagen RNeasy Plant Mini Kit (qiagen.com). First-strand cDNA was synthesized using 2 μg of total RNA with an Invitrogen SuperScript III First Strand Synthesis Kit (invitrogen.com). Quantitative RT-PCR was performed as described previously (45). Transcript levels of target genes were calculated using the equation 2^{[C_t (Control) − C_t (Experimental)]}, where C_t is the mean threshold cycle for each sample. Isopentenyl pyrophosphate/dimethylallyl pyrophosphate isomerase (At3g02780) was used as the normalization control (49). Primers sequences are shown in Table S2.