Diversity of plant circadian clocks: Insights from studies of Chlamydomonas reinhardtii and Physcomitrella patens

ABSTRACT

Arabidopsis thaliana has long been the model plant of choice for elucidating the mechanisms of the circadian clock. Recently, relevant results have accumulated in other species of green plant lineages, including green algae. This mini-review describes a comparison of the mechanism of the A. thaliana clock to those of the green alga Chlamydomonas reinhardtii and the moss Physcomitrella patens, focusing on commonalities and divergences of subsystems of the clock. The potential of such an approach from an evolutionary viewpoint is discussed.

Abbreviations

LD

light-dark cycles

LL

continuous light

DD

continuous darkness

HK

histidine kinase.

Circadian clocks, which are self-sustained oscillations with a period of approximately a day, drive circadian rhythms of a variety of processes in metabolism, growth and development.¹ A circadian system comprises 3 subsystems: 1) input pathways, through which environmental cues such as light and temperature reset the central clock; 2) the central clock, a self-sustained oscillation machinery; 3) output pathways, which transmit circadian timing information from the central clock to overt rhythms.² Many processes of plants show circadian rhythms and various dicotyledonous plants have been used to assess physiological aspects of the clock.^3,4 In the era of molecular genetics, the model dicot species Arabidopsis thaliana has been intensively studied to unravel the clock's mechanism, and many “clock genes,“ which encode components of the central clock machinery, have been identified and analyzed by genetic, biochemical and genomic techniques.⁴ Consequently, molecular models of the A. thaliana circadian system with complex clock gene networks have been postulated.^5,6,7,8,9 These are briefly summarized in Fig. 1. In recent years, many homologs to A. thaliana clock genes were identified in various eudicots and monocots, and studies on them indicate that clock genes and their functions are broadly conserved in angiosperms.^10,11,12 This conservation may also extend, to some extent, even to the marine green alga Ostreococcus tauri, which has a CCA1/LHY homolog (OtCCA1) and a TOC1 homolog (OtTOC1) (see the legend of Fig. 1 for abbreviations of gene/protein names).¹³ The O. tauri clock was significantly compromised when either gene was overexpressed or when OtTOC1 was knocked-down, and these 2 genes form a negative feedback loop.¹³ These results indicate that both genes play central roles in the O. tauri clock. Moreover, the origin of angiosperm clocks can be traced back even to the branching point between higher plants and O. tauri.

Figure 1.

Model of the A. thaliana circadian clock. Arrows and T-shaped bars are regulatory links for the promotion and repression of gene transcription, respectively, except for the T-shaped bar from blue light, which indicates protein degradation. The CIRCADIAN CLOCK-ASSOCIATED 1 (CCA1) and LATE ELONGATED HYPOCOTYL (LHY) genes, which encode paralogous Myb-domain transcription factors, repress transcription of the LUX ARRHYTHMO (LUX) (also termed PHYTOCLOCK 1 (PCL1)) and EARLY FLOWERING 4 (ELF4) genes. LUX, ELF3 and ELF4 form the evening complex (EC), which represses transcription of PRR7 and PRR9, both encoding “pseudo-response regulators” with a receiver-like domain and a CCT (CO/CO-Like/TOC1) domain. PRR7/9 and their paralog PRR5 repress CCA1/LHY transcription, thus closing a feedback loop. This tripartite circuitry appears to be essential for rhythmic expression patterns of its component genes: morning activation of CCA1/LHY immediately leads to repression of EC genes, thereby resulting in de-repression of direct target genes of EC, PRR9/7 from early- to mid-daytime. The serial activation of PRR9/7/5, each of which represses CCA1/LHY, is important for enabling the sharp peaks of CCA1/LHY expression at dawn. On the other hand, CCA1/LHY repress transcription of TOC1 (also termed PRR1), encoding another pseudo-response regulator, and oppositely TOC1 represses CCA1/LHY transcription, forming another loop. Night light-inducible and clock-regulated gene 1 (LNK1) and LNK2 encode transcriptional coactivators necessary for the expression of PRR5 and TOC1, both of which repress LNK1/2 in turn, forming yet another loop. Although there exist more links, e.g., TOC1 and EC each represses the transcription of several other clock genes, they are not depicted in the figure for simplicity. Post-transcriptional/post-translational regulatory processes are also operating in the clock machinery. For example, ZEITLUPE (ZTL), a LOV domain-containing E3 ubiquitin ligase, mediates proteasome-dependent degradation of TOC1. ZTL is stabilized by a blue-light-enhanced interaction with rhythmically expressed GIGANTEA (GI) protein; this rhythmic stabilization of ZTL promotes high-amplitude rhythmicity of TOC1 expression. Other post-transcriptional/post-translational processes, e.g., alternative splicing, protein phosphorylation/dephosphorylation, and rhythmic chromatin regulation, are also involved in the clock mechanism,^48,49 though they are not described in this mini-review.

Chlamydomonas reinhardtii, sometimes referred as “green yeast,” is a model alga suitable for genetic and biochemical studies.¹⁴ Clock mutants of this alga were isolated more than 40 y ago by monitoring circadian rhythm of phototactic behavior.^15,16 Four long period mutants, isolated and designated as “per” (per-1 to per-4; but almost certainly unrelated to animal per genes), have mutations in 4 independent genomic loci and their period lengthening effects are additive.¹⁶ A short period mutant of circadian phototaxis has also been isolated.¹⁷ Unfortunately, the genes responsible for these mutants have not been identified yet. In recent years, findings related to the molecular components of the Chlamydomonas clock have accumulated rapidly. CHLAMY1 is a RNA-binding protein which binds to UG-repeat containing mRNAs in a night/subjective-night phase-specific manner.¹⁸ Analyses of biochemically isolated CHLAMY1 revealed that it consists of a heteromer of C1 and C3 subunits containing lysine homology domains and RNA recognition motifs, respectively.¹⁹ Since misexpression of genes encoding C1 and C3 induces abnormal circadian rhythms (arrhythmicity and advanced circadian phase, respectively), it is obvious that they are involved in the circadian clock of this alga.²⁰ In 2008, a large number of “roc (rhythm of chloroplast)” mutants were isolated by screening insertional mutants based on circadian phenotypes of a bioluminescence rhythm derived from the luciferase reporter gene transferred into the chloroplast genome.^21,22 The genes responsible for these mutants were termed “ROC” genes, and the proteins encoded by ROC genes were similar to higher plant proteins (Table 1): ROC15 and ROC75 have GARP DNA-binding motifs similar to that of LUX(PCL1); ROC40 contains a Myb domain similar to those of CCA1/LHY; ROC66 contains an N-terminal B-box and a C-terminal CCT motif similar to those of CO/COL family proteins.²² Although their sequence similarities do not extend beyond the domain regions, 4 genes positively identified by a forward genetic screen are similar to the A. thaliana clock/clock-associated flowering genes, indicating that the origin of the C. reinhardtii clock is, at least partially, shared with the A. thaliana clock. On the other hand, ROC55 and ROC114 do not share any significant similarity to clock proteins of not only higher plants but also other clock model organisms.²² In addition to the above-mentioned factors, it is known that Casein kinase 1, N-terminal acetyltransferase 3, and Protein disulfide isomerase 2 are also involved in the C. reinhardtii clock.^23,24,25

Table 1.

Clock (-associated) genes and their homologs in A. thaliana, C. reinhardtii and P. patens. Shown in each row are numbers of the homologs that are present in the 3 species to the clock (-associated) gene(s) in the first column.

	A. thaliana	C. reinhardtii	P. patens
CCA1/LHY	2	2 (ROC40)	2
PRR/TOC1	5	2	4
LUX/BOA	2	3 (ROC15/75)	4
ELF3	1	0	3
ELF4/ELF4-like	5	0	1
GI	1	0	0
CHE	1	0	0
LOV-HK	0	1	2
ZTL	3	0	0
LNK	4	0	0
COL	17	3 (ROC66)	3

The Chlamydomonas clock is reset by light with a wide range of wavelengths including violet to red.^26,27 At the molecular level, light responses of several clock-related mRNAs have been demonstrated. The c3 mRNA is strongly induced by blue, yellow, and red light.²⁸ In contrast, the ROC15 and ROC40 mRNAs are reduced.²⁸ aCRY, a Chlamydomonas homolog of animal cryptochrome/photolyase proteins, is able to absorb not only blue but also yellow and red light, and is involved in the responses of these mRNAs.²⁸ At the protein level, ROC15 undergoes phosphorylation and proteasomal degradation by blue, green, and particularly red light.²⁹ ROC114, encoding an F-box protein, is involved in this degradation.²⁹ Indeed, as expected, the roc15 mutant exhibits abnormal phase responses of the clock to light.²⁹ Chlamydomonas Photolyase Homolog 1 (CPH1), another cryptochrome/photolyase homolog which is similar to the higher plant CRYs, also undergoes proteasomal degradation after blue and red light exposure.³⁰ Interestingly, a knockdown strain of CPH1 shows a larger phase response to blue light, indicating that CPH1 is involved in clock resetting as a negative regulator.²⁷ Information about external temperature is integrated into CHLAMY1.³¹ Exposure to low temperature induces hyperphosphorylation of the C1 subunit and accumulation of the C3 subunit due to transcriptional activation of the c3 gene via an E-box cis-element.³¹

Physcomitrella patens is a moss species, which diverged from vascular plant lineages at least 450 Ma.³² The efficiency of gene targeting by homologous recombination in P. patens is as high as those in yeasts,³³ facilitating the use of this moss for the study of gene functions. The P. patens genome³² shows a suite of genes that are similar to the A. thaliana clock genes (Table 1). P. patens has 2 genes that are homologous to CCA1/LHY (PpCCA1a and PpCCA1b). PpCCA1a and PpCCA1b both showed, similar to CCA1/LHY, diurnal expression with peaks at dawn and circadian expression with peaks at subjective dawn in light-dark cycles (LD) and continuous darkness (DD), respectively.³⁴ P. patens also has 4 PRR gene homologs, PpPRR1/2/3/4,³⁵ all of which are expressed in a circadian manner, and whose phases are similar to the profiles of PRR3 and PRR5.³⁵ Holm (2010)³⁶ presented a comparative overview of expression profiles and phylogenetic analyses for the P. patens genes homologous to ELF3, ELF4 and LUX as well as PpCCA1a/1b and PpPRR1/2/3/4. They reported that expression profiles of most homologs are generally similar to their A. thaliana counterparts.³⁶ The double disruptant for PpCCA1a/1b showed circadian gene expression with a shorter period and a dampened amplitude compared to the wild-type (WT) strain,³⁴ similar to the A. thaliana null mutant for CCA1/LHY.³⁷ Interestingly, the receiver-like domain (RLD) of PpRPR2/3/4 exhibits a potential phosphoacceptor motif, aspartic acid-aspartic acid-lysine (DDK), which is distinct from the angiosperm PRRs, in which DDK motifs and probably their phosphoacceptor functions are not preserved.³⁵ Consistently, the PpPRR2 RLD had phosphoacceptor ability in vitro.³⁵ These observations suggest that PpPRR2 functions as a genuine response regulator, and not as a PSEUDO-response regulator, and that its upstream phosphorelay cascade with a counterpart histidine kinase (HK) is also preserved in P. patens.³⁵ This putative “circadian phosphorelay cascade” is predicted to have a regulatory function in the moss clock machinery because phosphorylation of a response regulator via a phosphorelay reaction switches its activity, generally resulting in activation or repression of downstream genes. An attempt to unravel an output pathway in P. patens was also made. The plastid sigma factor is the nuclear-encoded regulatory subunit of the plastid-encoded plastid RNA polymerase, which principally transcribes photosynthesis-related genes in plastids.³⁸ P. patens has at least 3 sigma factor genes, PpSig1/2/5, and of these, only PpSig5 showed circadian expression in LD and DD.³⁸ The plastid gene psbD, which encodes the D2 protein of photosystem II, shows a diurnal expression rhythm in LD in P. patens,³⁸ and its amplitude was lowered when PpSig5 was disrupted.³⁹ Therefore, PpSIG5 is a sigma factor that regulates psbD rhythm, possibly as an intermediate regulator of an output pathway.

The distribution of clock genes in A. thaliana, C. reinhardtii and P. patens is depicted as a Venn diagram in Fig. 2. In the peripheral regions of the Venn diagram, where factors found only in one or 2 species are indicated, there are several proteins that potentially mediate environmental signals to more centrally located factors. ZTL and GI, present only in A. thaliana, mediate a light signal to the clock by modifying TOC1 stability (Fig. 1). EC is likely involved in a temperature input pathway,^40,41 where ELF3 and ELF4, present only in A. thaliana and P. patens, possibly modify the activity of LUX as a transcription factor. CHLAMY1, which is only found in C. reinhardtii and is known to respond to both light and temperature,^28,31 might mediate these environmental cues to the central region, where ROC40 mRNA is a possible target molecule because its 3′-untranslated region contains an UG repeat, a binding motif of CHLAMY1.²² Factors mediating a light signal that induces proteasomal degradation of ROC15 have not yet been identified.²⁹ ROC114, which is in a peripheral region of the diagram and is involved in ROC15 degradation, could be a potential candidate of such a factor.

Figure 2.

Distribution of clock genes (only factors predicted to have essential functions in the clock mechanisms are included) in A. thaliana, C. reinhardtii and P. patens. Arrows are flows that transmit environmental signals to the clock. Dashed lines have not yet been clarified.

PRRs with a DDK motif are only shared between C. reinhardtii and P. patens (not depicted in Fig. 2), suggesting that the putative circadian phosphorelay cascade is also shared between these 2 species, but not in A. thaliana. Indeed, no clock-related histidine kinase (HK) has been reported in A. thaliana. In O. tauri, which possesses a putative TOC1 homolog with a DDK motif,¹³ a LOV (Light/Oxygen/Voltage) domain-containing HK has an important function for the operation of the clock under blue light.⁴² Similar LOV-HK sequences are also found in the P. patens and C. reinhardtii genomes, but not in the A. thaliana genome (TY & TM, unpublished observations). Therefore, these LOV-HKs are promising candidates for HKs in the circadian phosphorelay cascade. The HK in the putative circadian phosphorelay cascade might mediate an environmental signal to the central region by modifying the activity of the moss PRRs. As for components of the output pathways (output factors), little is known in C. reinhardtii and P. patens. Noordally (2013)⁴³ demonstrated that psbD mRNAs show a robust circadian oscillation in continuous light (LL) in A. thaliana, and this rhythm is nullified in the Sig5 (AtSig5) null mutants, indicating that SIG5 is a sigma factor that specializes as an output factor mediating the timing of circadian information from the clock to psbD in A. thaliana. Therefore, the function of this specialized sigma factor (Sig5) as an output factor seems to be shared between A. thaliana and P. patens. On the other hand, C. reinhardtii has only a single copy sigma-like gene, PROD, and its involvement in circadian regulation has not been clarified.⁴⁴ Thus, commonality/divergence between output factors of the 3 species remains a mystery.

Factors shared among the 3 species are LUX(ROC75/15), PRRs, CCA1/LHY(ROC40) and ROC66(COL). The three species have diverged considerably with great evolutionary distances between them. Despite this, they all preserve these factors, suggesting that these factors have indispensable functions in the operation of the clock. Consistently, LUX, PRRs and CCA1/LHY form the circuitry that seems to be essential in the generation of circadian oscillation.⁶ In P. patens, PpCCA1a/1b feed back to PpCCA1b transcription and they also regulate PpPRR1 transcription,³⁴ consistent with the regulatory interactions in the A. thaliana tripartite circuitry. However, in C. reinhardtii, the regulatory relationship between ROC75 and ROC40 may not be similar to that between LUX and CCA1/LHY (TM, unpublished observation); in addition, expression profiles of ROC genes are different from those of the A. thaliana counterparts.²² These observations indicate that rewiring of the network structure took place after the divergence between C. reinhardtii and A. thaliana, while conserving the network components. ROC66 is supposed to be an essential clock gene because it considerably prolongs the period length of the clock (29.9 h) when mutated in C. reinhardtii.²² In A. thaliana, the function of COL genes in the central clock machinery is unclear, although shortening of the period length of circadian rhythms of leaf movement and CAB gene expression were observed in COL1 overexpressors.⁴⁵ P. patens has 3 COL genes but their functions have not been elucidated.^46,47 The position of ROC66 in the Venn diagram and the phenotype of the ROC66 mutant suggest that some COL genes might also be important clock genes in A. thaliana and P. patens.

As suggested by McClung.¹² Comparisons of genomic data alone will only shed light on common factors between different species, and therefore, a small number of factors will become distinct when the genomes of phylogenetically distant species are compared. On the other hand, when the results of forward genetics are compared, as done with C. reinhardtii and A. thaliana in this mini-review, not only common factors but also species-specific factors are highlighted. Comparisons of clock genes, largely between these 2 species, suggest a possibility that clock components receiving environmental signals (directly or indirectly) may have diversified during evolution (Fig. 2). This idea is plausible because ambient cues affecting the performance of the clock could vary between species with different habitat conditions, though such proof awaits further investigations. Forward genetics has not been applied to P. patens, because monitoring of rhythms from small tissue colonies, which is essential for screening clock mutants, is still technically difficult (SA, unpublished observation). We are developing a method for monitoring the circadian expression of the luciferase reporter gene based on a particle bombardment-mediated transient gene expression assay (Fig. 3). This will facilitate, if combined with genomic analyses, effective screenings for new clock gene loci in P. patens. Another aspect that this mini-review highlights is that the data obtained with primitive groups, such as C. reinhardtii and P. patens, could be important because such ancient groups may still have prototypic gene circuitries, which were lost in higher plant lineages (such as the putative circadian phosphorelay cascade). If the clocks of more diverged species are investigated and systematically compared, the evolution, diversity and even origins of plant clocks will be clearer. Moreover, this may help to clarify the inherent complexity of the plant clock mechanisms.

Figure 3.

Bioluminescence rhythms from protonema tissues that transiently express a luciferase reporter construct. A DNA construct carrying a fusion of the promoter region of PpCCA1b and the firefly luciferase (luc+) gene was transferred to the moss protonema tissues by particle bombardment-mediated transformation,^34,50 entrained to 12-h:12-h light dark (LD) cycles, and released for bioluminescence monitoring into conditions shown below each graph. The peaks of the rhythm occurred immediately before the onsets of light periods in LD cycles, largely due to the rapid decline of bioluminescence caused by light. On the other hand, the peak times are shifted toward the beginnings of the subjective days in continuous darkness (DD), reflecting innate timing based on free-running of the clock. Additionally, the amplitudes of the rhythms in DD were always significantly lower than those in LD cycles for unknown reasons. The open and filled squares on the horizontal axes are light (˜50 µmol/m²/sec) and dark periods, respectively. The gray squares are subjective days.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed

Acknowledgments

This work was supported by JSPS KAKENHI (25440179 to TM, 26292051 to TY, 2529105 to MS, 24570007 to SA), Kato Memorial Bioscience Fondation (to TM) and Kurata Memorial Hitachi Science and Technology Foundation (to TM) and Naito Foundation (to TY).