Temperature compensation of the Arabidopsis circadian clock

Abstract

The circadian clock is an endogenous, self-sustaining oscillator with a period of approximately 24 h. This timekeeping system helps organisms align their physiologies and behaviors with the day-night cycle. Temperature compensation, which allows the clock to maintain a period of about 24 h across physiological temperature ranges, is striking because, the rates of biochemical reaction rates increase with temperature in general. This review summarizes the effects of temperature on the period length of circadian rhythms in plants, especially Arabidopsis thaliana (Arabidopsis). Most core clock genes contribute to the temperature compensation of circadian period length. Temperature compensation involves the active, targeted degradation of specific core clock proteins at lower temperatures. Comparing the molecular mechanisms across different organisms may suggest design principles that allow evolutionarily distinct clock proteins to achieve temperature compensation.

Introduction

Life on Earth, from bacteria to animals, has evolved internal timekeeping systems that generate circadian rhythms in numerous physiological processes and behaviors, enabling adaptation to the day-night cycle¹. The circadian clock shares three common properties: self-sustaining oscillation, entrainment to environmental time cues, and temperature compensation of period length. Self-sustaining oscillation refers to the ability of the circadian rhythm to persist in the absence of temporal information, such as rhythmic light and temperature input from the ambient environment. Entrainment is the ability to synchronize endogenous circadian rhythms with the day-night cycle through environmental light and temperature changes. For example, the circadian clock in Arabidopsis thaliana (Arabidopsis) can synchronize to temperature cycles with only a 4 °C difference². The third property, temperature compensation, refers to the ability to maintain a constant clock speed (τ) at different constant temperatures within the physiological range. Generally, in biochemical reactions, the reaction rate increases with temperature, as described by the Arrhenius equation³. Indeed, in the Belousov-Zhabotinsky chemical oscillatory reaction in vitro, the period length was reduced to approximately one-fourth when the temperature increased by 20 °C⁴. The Q₁₀ value quantifies how many times the reaction rate increases when the temperature rises by 10 °C. In typical biochemical reactions, Q₁₀ values range from 2 to 3⁵. However, the period length of the circadian rhythm exhibits a Q₁₀ of 0.8 to 1.2^5–11. Temperature compensation is evolutionarily conserved across a wide range of organisms, from the photosynthetic microorganism cyanobacteria to mammals^12,13. A comprehensive cross-species review of temperature compensation, long regarded as one of the last great unsolved issues in circadian biology, was recently published¹⁴.

In this review, we summarize studies on temperature compensation in plants, which possess more clock components than other lineages of life.

The three hypotheses for temperature compensation

Temperature compensation was first discovered about 70 years ago by Pittendrigh in his study on the hatching circadian rhythm of Drosophila melanogaster within a temperature range of 16 °C to 26 °C¹². Pittendrigh proposed that temperature compensation could result from the temperature insensitivity of a key reaction determining period length¹². This theory is known as “the critical-reaction hypothesis” (Fig. 1a)¹⁵.

Fig. 1. Three hypotheses for temperature compensation.

a The period of circadian rhythms remains approximately 24 h within the physiological temperature range. The critical reaction hypothesis posits that a specific period-determining reaction or enzyme (indicated by the yellow box) is insensitive to temperature changes. b The balance hypothesis suggests that although all biochemical reactions are generally accelerated by increased temperature, the overall period remains stable due to a compensatory balance among these reactions. c The scaling amplitude hypothesis proposes that while the amplitude of circadian rhythms is temperature-sensitive, changes in amplitude help to maintain a consistent period length.

Hastings and Sweeney later suggested that temperature compensation consists of at least two temperature-sensitive reactions¹⁶. The first reaction is responsible for advancing the clock, while the second reaction acts as a brake, inhibiting the activity of the first reaction. If the rates of both reactions increase with temperature yet their overall balance is maintained, temperature compensation of the period length is achieved¹⁶. This hypothesis is referred to as “the balance hypothesis” (Fig. 1b)¹⁵.

The temperature compensation has been modeled as a scaling amplitude with an increase in reaction rate (changing the size of the limit cycle)^14,17(Fig. 1c). The scaling amplitude with temperature was found in some organisms including plants^15,18–20.

Transcription–translation feedback loops model for the plant circadian clock

Molecular genetic studies in Arabidopsis revealed that the transcription–translation feedback loops composed of core clock genes conserved throughout the plant lineage, are essential for the circadian clock (Fig. 2)^21,22. Two paralogous single MYB transcription factors, CIRCADIAN CLOCK-ASSOCIATED 1 (CCA1) and LATE ELONGATED HYPOCOTYL (LHY) are expressed at dawn, and directly repress expression of PSEUDO-RESPONSE REGULATOR9 (PRR9), PRR7, PRR5, TIMING OF CAB EXPRESSION 1 (TOC1, also called PRR1), EARLY FLOWERING 3 (ELF3), ELF4, LUX ARRHYTHMO (LUX), GIGANTEA (GI), CCA1 HIKING EXPEDITION (CHE), NOX (also known as BROTHER OF LUX ARRHYTHMO, BOA), COLD REGULATED GENE 27 (COR27) and COR28^23–25. During noon to evening, PRR9, PRR7, and PRR5 repress CCA1, LHY, PRR9, PRR7, REVEILLE 8 (RVE8), NIGHT LIGHT-INDUCIBLE AND CLOCK-REGULATED 1 (LNK1), LNK2, LNK3 and LNK4^26–28. Evening Complex, composed of ELF3, ELF4, and LUX represses PRR9, PRR7, LUX, GI, and LNK1^29–33. RVE8 and LNK proteins activate PRR5, TOC1, and ELF4³⁴, whereas COR27 and COR28 repress PRR5 and TOC1³⁵. LIGHT-REGULATED WD 1 (LWD1) and LWD2, TEOSINTE BRANCHED 1, CYCLOIDEA, PCF (TCP)-DOMAIN FAMILY PROTEIN 20 (TCP20) and TCP22 form a complex to activate CCA1³⁶.

Fig. 2. Transcription–translation feedback loops in the Arabidopsis circadian clock.

The protein levels of CCA1, LHY, PRR5, and TOC1 are upregulated at higher temperatures. CCA1, LHY, PRR7, PRR5, TOC1, and GI are degraded at lower temperatures. The Evening Complex (EC) binds to its target DNA regions preferentially at lower temperatures. HOS15-, COP1-, and LKP2-mediated degradation is activated at lower temperatures, whereas XBAT31-mediated degradation is promoted at higher temperatures. The colors indicate different functional categories: blue represents transcription factors, green indicates E3 ubiquitin ligases, and yellow indicates kinases or a SUMO ligase.

In addition to these transcriptional regulations, post-translational regulations are crucial for the clock³⁷. CCA1 and LHY are phosphorylated by CASEIN KINASE 2 (CK2)³⁸. This phosphorylation affects the DNA-binding activity of CCA1 and LHY^39,40. CCA1 and LHY undergo 26S proteasome-dependent degradation at 4 °C⁴¹. CCA1 is also sumoylated, and this post-translational modification attenuates its DNA-binding affinity⁴². PRR5 and TOC1 proteins are targeted for degradation by ZEITLUPE (ZTL), a component of the E3 ubiquitin ligase SCF complex (Skp, Cullin, F-box-containing complex)^43,44. PRR5 and TOC1 are phosphorylated by CASEIN KINASE 1-LIKE4 (CKL4) in vitro⁴⁵. Phosphorylated PRR5 and TOC1 are more effectively recognized by ZTL in vivo^45,46. Upon exposure to blue light, ZTL is stabilized by GI⁴⁷. Furthermore, PRR5 and TOC1 are also targeted for degradation by the ZTL homologue, LOV KELCH PROTEIN 2 (LKP2), at lower temperature⁴⁸, while PRR7 protein degradation is mediated by HIGH EXPRESSION OF OSMOTICALLY RESPONSIVE GENES 15 (HOS15) at lower temperature in the dark⁴⁹. PRR9 is phosphorylated by PHOTOREGULATORY PROTEIN KINASEs (PPKs), also known as MUT9-LIKE KINASEs (MLKs)⁵⁰. Phosphorylation of ELF4 has been shown to alter its period-lengthening activity⁵¹. Lastly, the proteomic study found that ELF3, GI and TIME FOR COFFEE (TIC) proteins are sumoylated⁵².

Arabidopsis genes affecting temperature compensation

The first experimental approach to investigating temperature compensation in Arabidopsis utilized natural populations or accessions⁹. The study indicated that Q₁₀ values of these natural populations are 0.95 to 1.10. This approach identified ZTL and GI as candidate genes for quantitative trait loci (QTL) associated with temperature compensation, based on recombinant inbred lines derived from the Cvi accession isolated from a warm climate the Cape Verde Islands (15°N to 17°N) and the Ler accession from European location^9,53. The phenotype observed in Cvi is attributed to single amino acid substitutions in ZTL and GI (ZTL:P35T, GI:L718F)⁵³ (Table 1). Independent studies analyzing gi mutants using different alleles and measurement systems suggest that these mutants tend to exhibit shorter periods than the wild-type, particularly at high temperatures, and possesses a higher Q₁₀ compared to the wild-type^54,55.

Table 1.

Gene affecting temperature compensation in Arabidopsis

Component	Allele, treatment	Phenotype	Reference
CCA1, LHY	cca1-1 lhy-20	overcompensation, long at high temperature	57
CCA1, LHY	cca1 lhy Knock down	cancel prr9 prr7 overcompensation	55
CCA1, LHY	cca1-1 lhy-20	cancel rve4 rve6 rve8 undercompensation	57
PRR9	prr9-1 prr7-3	overcompensation, long at high temperature	55
PRR7	prr9-1 prr7-3	overcompensation, long at high temperature	55
PRR7	prr7-11 prr5-11	undercompensation, short at high temperature	48
PRR5	prr5-11	undercompensation, short at high temperature	48
TOC1	toc1-2	undercompensation, short at high temperature	48
GI	gi-11	short period at 12, 22, 27 °C, not 17 °C	54
GI	gi-1	undercompensation, short at high temperature	55
GI	GICvi ZTLCvi	short at low temperature	53
ZTL	GICvi ZTLCvi	short at low temperature	53
RVE8	rve4-1 rve6-1 rve8-1	undercompensation, long at low temperature	57
RVE8	RVE8ox	overcompensation, short at low temperature	57
RVE2	rve2-2	long at low temperature	61
ELF4	elf4-211	long at high temperature	51
LKP2	lkp2-cr4	undercompensation, long at lower temperature	48
CKB4 (CK2 subunit)	CKB4ox	undercompensation, short at higher temperature	39
CK2	CK2 inhibitor treatment	overcompensation, long at high temperature	39
HsfB2b	hsfB2b-1	short at high temperature	56
SIC	sic-1, sic-3	undercompensation, long at low temperature	58
FBH1	FBH1ox	short at high temperature	60
JMJD5	jmjd5-1, jmjd5-2	undercompensation, short at high temperature	66
SIZ1	siz1-2	undercompensation, short at high temperature	67
OTS1, OTS2	ots1-1 ots2-1	undercompensation, long at low temperature	67

The importance of PRR genes in temperature compensation is highlighted by the long-period prr9 prr7 mutant, which displays an overcompensation phenotype where period length increases with rising ambient temperature⁵⁵. Conversely, a mutant of hsfB2b (for HEAT SHOCK FACTOR B2b), exhibiting upregulated PRR7, shows a short period at higher temperatures (27 °C) but not 22 °C⁵⁶. The hsfB2b phenotype suggests a temperature- dependent role of the HSFB2B-PRR7 module in period regulation⁵⁶. The prr9 prr7 phenotype is attributed to the temperature-dependent accumulation of period-lengthening factors due to the loss of PRR9 and PRR7⁵⁵. PRR9 and PRR7 repress the transcription of CCA1 and LHY, which are predicted to function as period-lengthening factors. Suppression of CCA1 and LHY in the prr9 prr7 background cancels the overcompensation phenotype, indicating that the overaccumulation of CCA1 and LHY at elevated temperatures disrupts the balance of core clock gene expression, which is essential for temperature compensation⁵⁵. While the cca1-1 lhy-11 allele shows a Q₁₀ value comparable to the wild type from 19 °C to 25 °C⁴⁸, a different allele, cca1-1 lhy-20 exhibits an overcompensated phenotype between 22 °C to 30 °C⁵⁷. The involvement of CCA1 and LHY in temperature compensation is further supported by a forward genetic approach that successfully identified mutants with impaired temperature responses in PRR7 transcription⁵⁸. The causal gene, SICKLE (SIC), controls intron splicing. The sic mutants show pronounced longer period than the wild type, especially at lower temperature. Intron retention of CCA1, LHY, PRR7, and ELF3 is enhanced in sic mutants. Notably, the retention of intron 1 in LHY, previously observed at cold temperatures, is enhanced in sic mutants^58,59. Epistatic analysis clearly suggests that SIC controls period length of the clock through CCA1 and LHY⁵⁸. The CCA1 promoter is bound by the basic helix-loop-helix (bHLH) transcription factor FLOWERING BASIC HELIX-LOOP-HELIX 1 (FBH1), which shares high protein sequence similarity with other members of FBH⁶⁰. Overexpression of FBH1 reduces CCA1 expression and shortens the period at high temperatures⁶⁰. These findings highlight the importance of CCA1 and LHY in temperature compensation.

RVE8 encodes a Myb-type transcription activator that shares the target genes with CCA1 and LHY in the clock⁵⁷. For example, while RVE8 activates PRR5 and TOC1, CCA1 and LHY repress them. The triple loss of function mutant rve4 rve6 rve8 shows an undercompensation phenotype, and the overexpression line of RVE8 shows overcompensated rhythm. The undercompensated phenotype of rve4 rve6 rve8 is rescued by the cca1 lhy double mutations, suggesting that a precise balance between the activator- and repressor-type Myb transcription factors is crucial for temperature compensation⁵⁷. A recent study found that the period length of rve2 is longer than that of the wild type at lower temperatures, showing an undercompensated phenotype⁶¹.

A comparative survey identified the double mutant prr5 toc1 as exhibiting undercompensation⁴⁸. The single mutants toc1 and prr5 also decrease compensation, suggesting these genes additively contribute to temperature compensation. The period length of the double mutant prr7 prr5 is undercompensated, whereas prr9 prr7 shows overcompensation, suggesting that PRR7 controls temperature compensation through distinct mechanisms in combination with other PRRs.

Assessing temperature compensation for arrhythmic mutants, such as those of ELF3, ELF4 and LUX- components of the Evening Complex- is not possible. However, a mutation in ELF4 (elf4-211) that possesses a single amino acid substitution (S45L), shows a long-period phenotype exclusively at higher temperatures, suggesting the involvement of ELF4 in temperature compensation⁵¹. Although the arrhythmic elf3 phenotype could not be used to evaluate the role of ELF3 in compensation, the number of glutamines in the polyglutamine repeats in ELF3 correlates with period length⁶². A recent study suggested that proper ELF3 activity in Chinese cabbage is required for temperature compensation⁶³. lux mutants have been known to be arrhythmic at around 22 °C^64,65; however, at 17 °C, lux-2 shows circadian rhythmicity, indicating a dispensable role of LUX for the clock at lower temperatures⁶⁶. Additionally, mutations in the Jumonji domain-containing protein JMJD5 results in a short period at higher temperatures, showing an undercompensated phenotype⁶⁶.

A reverse genetic study demonstrated that the Small Ubiquitin-related Modifier (SUMO) pathway is involved in temperature compensation⁶⁷. Sumoylation, a reversible post-translational modification, is catalyzed by SUMO ligases and removed by SUMO proteases. The SUMO ligase mutant siz1 (for SAP and MIZ1 domain-containing ligase1) shows a short-period phenotype at higher temperatures (22 °C and 27 °C), but not at low temperatures (12 °C)⁶⁷. Conversely, the double mutant of the two genes encoding SUMO protease, OVERLY TOLERANT TO SALT1 (OTS1) and OTS2 (ots1 ots2), shows a long-period phenotype at lower temperatures (12 and 22 °C), but not at 27 °C⁶⁷. CCA1 is one of the known substrates, whose sumoylation reduces its DNA-binding activity⁴². Sumoylation of ELF3, GI, and TIC had been reported; however, the biological significance of these modifications remains unknown⁵².

Collectively, these studies indicate that most genes involved in the transcription–translation feedback loops are essential for temperature compensation in the Arabidopsis circadian clock, suggesting that a functional clock is required for its ability.

Effect of temperatures on core clock proteins

In general, chemical reactions are accelerated by increasing temperature, implying that all biochemical reactions within the Arabidopsis clock are influenced by ambient temperature. Indeed, temperature-dependent regulation of core clock proteins has been reported, with some mechanisms linked to temperature compensation.

CCA1 binds more strongly to the TOC1, PRR7, PRR9, and LUX promoters at high temperatures (27 °C) than at 22 °C³⁹. CK2 phosphorylation reduces the binding affinity of CCA1 to the core clock gene promoters at 27 °C, whereas phosphorylation of CCA1 enhances its binding affinity to the Lhcb1*3 promoter, a CCA1 target gene, in vitro⁴⁰. These results indicate that the DNA-binding activity of CCA1 is temperature-sensitive. CK2 activity exerts a more pronounced effect at high temperatures³⁹. CK2 interacts with CCA1; however, it remains unclear whether this interaction is modulated by ambient temperature changes. The amount of CCA1 and LHY proteins decreases at low temperatures (4 °C) via a 26S proteasome-dependent degradation pathway⁴¹. Two PEST sequences (rich in proline [P], glutamic acid [E], serine [S] and threonine [T]) in CCA1 are responsible for the degradation at 4 °C, a temperature at which the rhythmicity of the Arabidopsis clock is largely lost⁴¹. Another study demonstrated that the amount of LHY protein driven by its own promoter decreases at 12 °C compared to 27 °C, likely due to reduced translational efficiency or enhanced protein degradation⁶⁸. Accumulation of CCA1 and LHY at 27 °C compared to lower temperatures was thought to cause a lengthening of the period; however, the period of the wild type is slightly shorter at 27 °C than at lower temperatures⁶⁸. This may imply that CCA1 and LHY act to slow down the clock speed, which tends to accelerate at higher temperatures due to enhanced biochemical reaction rates. As mentioned, sumoylation weakens the DNA-binding activity of CCA1, and mutants of SUMO ligase or proteases attenuate temperature compensation⁴². However, temperature-dependent sumoylation of CCA1 is unknown.

RVE genes regulate temperature compensation antagonistically to CCA1 and LHY. Two RVE proteins, RVE8 and RVE4, localize to the nucleus at low temperatures (below 10 °C), but are predominantly cytoplasmic at 22 °C⁴¹. The cytoplasm-to-nucleus localization change of RVE proteins occurs within 10–15 min; however, how this temperature-dependent regulation of cellular RVE proteins localization controls temperature compensation remains to be determined.

Temperature-dependent quantitative control of two PRR proteins, PRR5 and TOC1, is required for temperature compensation (Fig. 3)⁴⁸. The period length of the prr5 toc1 double mutant is shorter than that of the wild type, especially at higher temperatures, indicating an attenuated temperature compensation phenotype. This phenotype suggests that PRR5 and TOC1 function as period-lengthening factors at higher temperatures. Consistent with this, PRR5 and TOC1 proteins accumulate at higher temperatures and are degraded at lower temperatures. LKP2, a component of the E3 ubiquitin ligase SCF complex, interacts with PRR5 and TOC1 at lower temperatures in vivo. In lkp2 mutants, PRR5 and TOC1 protein levels are upregulated, and these mutants exhibit a longer period than the wild type at lower temperatures, likely due to the accumulation of PRR5 and TOC1 proteins. Interestingly, ZTL, a homologue of LKP2, interacts with PRR5 and TOC1 regardless of ambient temperature. Therefore, the temperature-dependent quantitative regulation of PRR5 and TOC1 by LKP2 plays a crucial role in maintaining temperature compensation⁴⁸. In other words, the degradation of PRR5 and TOC1 proteins, which are considered period-lengthening factors, at lower temperatures prevents unnecessarily prolonged circadian periods in Arabidopsis at low temperatures. The differential temperature dependency between LKP2 and ZTL, and how LKP2 specifically recognizes PRR5 and TOC1 in a low-temperature-dependent manner, remains unknown. Since LKP2 and GI interact in vivo⁴⁷, GI may contribute to temperature compensation combined with LKP2.

Fig. 3. Temperature-dependent regulation of PRR5 and TOC1 protein levels contributes to circadian period maintenance.

The degradation of PRR5 and TOC1 proteins, which are considered period-lengthening factors, by LKP2 at lower temperatures contributes to preventing unnecessarily prolonged circadian periods in Arabidopsis at low temperatures.

Despite protein sequence similarity to PRR5 and TOC1, PRR9 and PRR7 are unlikely to be targeted for degradation by ZTL and LKP2. A recent study indicates that HOS15, a DWD (DDB1-binding WD40) motif-containing protein, controls PRR7 protein turnover at low temperature in the dark⁴⁹. HOS15 also controls GI protein amount at low temperatures⁶⁹. CONSTITUTIVE PHOTOMORPHOGENIC1 (COP1), a central RING-finger E3 ligase in the photomorphogenesis, controls degradation of GI at low temperatures⁷⁰. Recently, two proteins have been reported to affect GI protein abundance, although it remains unclear whether these regulations are temperature-dependent. First, the Na⁺/H⁺ antiporter SALT OVERLY SENSITIVE 1 (SOS1) stabilizes GI under high salinity conditions⁷¹. Second, the Regulatory-associated protein of target of rapamycin 1B (RAPTOR1B) interacts with GI and influences its protein amount⁷².

Among the core clock genes, PRR9, PRR7, and GI mRNA expression had been known to be induced by higher temperatures^2,73. Evening Complex, consisting of ELF3, ELF4, and LUX, represses the expression of these genes. Thus, PRR9, PRR7, and GI expression is constitutively higher in these mutants even at lower temperature². Further studies revealed that the target DNA-binding of Evening Complex in vivo is enhanced at lower temperatures³³. The DNA-binding of Evening Complex in vitro is also enhanced at lower temperature, suggesting that Evening Complex acts as a direct thermosensor⁷⁴. ELF4 amount strengths the binding in this in vitro assay system⁷⁴. In planta, the long distance ELF4 transport from shoot to root is accelerated at lower temperatures⁷⁵. In addition to ELF4, the other Evening Complex component ELF3 is affected by temperature, as ELF3 condensation is changed by ambient temperature^76–78. ELF3 protein undergoes targeted degradation via XBAT31 (one of XA21 BINDING PROTEIN3 homologue) at high temperatures⁷⁹. In summary, qualitative and quantitative changes in core clock proteins in response to ambient temperature have been observed (Fig. 2). However, determining whether and how these changes contribute to temperature compensation remains a significant challenge.

Computational approaches for temperature compensation in plants

A mathematical approach that incorporated temperature dependence into the Arabidopsis transcription–translation feedback loops, by applying the Arrhenius equation to model parameters characterizing transcription, translation, and degradation rates, revealed that temperature entrainment and temperature compensation are distinct from each other³. They also found that an increase in mRNA and protein degradation rates with temperature consistently leads to a reduction in period length, whereas an increase in mRNA and protein synthesis with temperature can either increase or decrease the period length. Thus, theoretically, mRNA and protein degradation rates have a greater impact on temperature compensation than synthesis rates³. This theoretical proposal may align with the experimental finding that degradation of core clock proteins is involved in temperature compensation. However, the degradation rates of Arabidopsis core clock proteins such as CCA1, LHY, PRR7, PRR5, and TOC1 are not higher at elevated temperatures. Rather, these proteins are actively degraded at lower temperatures^41,48,49. This enhanced degradation activity at low temperatures cannot be readily explained by the Arrhenius law, leaving the mechanism by which these degradation processes are triggered and maintained at lower temperatures an intriguing mechanistic question.

Mathematical modeling approaches based on ordinary differential equation (ODE) models have been applied to investigate the network structure of the Arabidopsis circadian clock^80–84. Recently, a comparative analysis was conducted to understand how temperature compensation is achieved from the perspective of network architecture¹⁰. The authors analyzed eight previously published ODE models, including the first ODE in which CCA1/LHY negatively regulates TOC1, which in turn activates CCA1/LHY, and a repressor-enrich model composed of four classes of repressors (CCA/LHY, PRR9/PRR7, PRR5/TOC1, and ELF4/LUX). They noted that the highly repressive interactions between the components, along with autoregulation loops and three-node feedback loops, enhances temperature compensation. Considering the entire network structure as the basis for temperature compensation may be consistent with experimental findings showing that the Q₁₀ values of undercompensated mutants do not exceed 2.0^48,57, implying that individual core clock components contribute to temperature compensation in a cooperative manner.

Another computational analysis emphasized the importance of CRYPTOCHROME (CRY) signaling for temperature compensation⁶⁸. The study also simulated overcompensation phenotype of prr9 prr7 and predicted increased LHY expression with temperature. Notably, the study predicted that the temperature-dependent stability of LHY mRNA would be lost in cry1 cry2 mutants, a prediction that was subsequently validated experimentally⁶⁸. mRNA modifications might be involved in this LHY mRNA accumulation, as the mRNA stability of CCA1, a LHY homologue, decreased in cry1 cry2 mutants due to a reduction in N⁶-methyladenosin (m⁶A) modification in CCA1 mRNA⁸⁵.

Possibility of temperature insensitive biochemical reaction

While no specific temperature-insensitive chemical reaction has yet been identified in plants that supports the “critical reaction hypothesis,” two temperature-compensated chemical reactions have been reported in the circadian clocks of other lineages of life. The first example is that cyanobacterial clock proteins KaiA, KaiB, and KaiC, along with ATP, induce an autonomous circadian phosphorylation and dephosphorylation rhythm in vitro⁸⁶. The low KaiC ATPase activity, which hydrolyzes approximately 12 ATP molecules per day, remains temperature compensated between 25 °C and 45 °C⁸⁷. Quasielastic neutron scattering, a direct technique for analyzing protein dynamics on picosecond to nanosecond time scales, has shown that the global motions of KaiC accelerate with increasing temperature (Fig. 4a)⁸⁸. Interestingly, this temperature-dependent increase in global motions is impaired in KaiC mutants possessing temperature-sensitive ATP hydrolytic activity. These results suggest that the temperature-dependent global motions of KaiC are crucial for maintaining its temperature-compensated ATPase activity⁸⁸.

Fig. 4. The temperature-sensitive processes or properties involved in temperature- compensated enzymatic reactions.

a Temperature-induced increases in global protein motions contribute to the temperature compensation of cyanobacterial KaiC ATPase activity. b Temperature-dependent changes in binding affinities to the substrate (PER2) and the product (phosphorylated PER2) underlie the temperature compensation of mammalian CK1δ kinase activity.

Secondly, the in vitro phosphorylation activities of mammalian CKIε and CKIδ on a synthetic peptide derived from mouse core clock protein PER2 are also temperature compensated⁸⁹. Further biochemical studies have identified two key mechanisms underlying the temperature compensation of CKIδ phosphorylation: (1) lower substrate affinity for the CKIδ-ATP complex at higher temperatures, and (2) higher product affinity for the CKIδ-ADP complex at higher temperatures (Fig. 4b)⁵. Additionally, an evolutionarily conserved CKI-specific domain has been identified as a determinant of higher product affinity at elevated temperatures. Collectively, these findings indicate that biochemical reactions previously observed as temperature compensated exhibit temperature sensitivity at the intramolecular level. Arabidopsis possesses 13 genes encoding CKL proteins. Chemical inhibition of these CKLs results in period lengthening, suggesting their involvement in the circadian clock^45,90. The CKI-specific domain is also conserved in Arabidopsis CKLs, raising intriguing questions about whether the kinase activities of these CKLs are temperature-insensitive and contribute to circadian temperature compensation in plants.

Very recently, a conserved RuvB-like (RUVBL) AAA+ type ATPase has been proposed as a clock component across eukaryotes including plant lineage⁹¹. Similar to cyanobacterial KaiC, mammalian RUVBL2 consumes about 13 ATP molecules per day in vitro, a rate far lower than typical ATPases. Genetic mutants of mice lacking RuvB like AAA ATPase 2 (RUVBL2) exhibit short-period, long-period, and arrhythmic phenotypes, suggesting that RUVBL2 is a key factor for period determination and clock function. Whether the activity of RUVBL2 is temperature-compensated, and if so, the underlying mechanism of this temperature compensation, remains an intriguing open question, considering its potential role as a clock pacemaker akin to KaiC.

Although possible homologues of CKI and RUVBL2 have been found, and the “critical reaction hypothesis” for temperature compensation in plants has not been ruled out, several studies suggest that temperature-dependent regulation of core clock proteins contributes to temperature compensation. In particular, the period-lengthening factors CCA1, LHY, PRR5, and TOC1 are more active at higher temperatures and are essential for temperature compensation. A combination of known and yet-to-be-identified temperature-dependent regulatory mechanisms of core clock proteins may underlie temperature compensation in plants adapted to a wide range of ambient temperatures.

Author contributions

A.E.M. and N.N. wrote the manuscript.

Data availability

No datasets were generated or analysed during the current study.

Competing interests

The authors declare no competing interests.