Interaction and Regulation Between Lipid Mediator Phosphatidic Acid and Circadian Clock Regulators

Abstract

Circadian clocks play important roles in regulating cellular metabolism, but the reciprocal effect that metabolism has on the clock is largely unknown in plants. Here, we show that the central glycerolipid metabolite and lipid mediator phosphatidic acid (PA) interacts with and modulates the function of the core clock regulators LATE ELONGATED HYPOCOTYL (LHY) and CIRCADIAN CLOCK ASSOCIATED1 (CCA1) in Arabidopsis (Arabidopsis thaliana). PA reduced the ability of LHY and CCA1 to bind the promoter of their target gene TIMING OF CAB EXPRESSION1. Increased PA accumulation and inhibition of PA-producing enzymes had opposite effects on circadian clock outputs. Diurnal change in levels of several membrane phospholipid species, including PA, observed in wild type was lost in the LHY and CCA1 double knockout mutant. Storage lipid accumulation was also affected in the clock mutants. These results indicate that the interaction of PA with the clock regulator may function as a cellular conduit to integrate the circadian clock with lipid metabolism.

INTRODUCTION

The circadian clock is an endogenous molecular oscillator composed of multiple interlocking transcriptional and posttranslational feedback loops that drive cellular processes with a 24-h rhythm (Nagel and Kay, 2012; Hsu and Harmer, 2014). Most clock genes in plants encode transcription factors that regulate one another in a reciprocal manner and also regulate other genes outside the clock, which mediate myriad biological processes. The central loop in Arabidopsis (Arabidopsis thaliana) consists of LATE ELONGATED HYPOCOTYL (LHY) and its closely related CIRCADIAN CLOCK ASSOCIATED1 (CCA1) and their target TIMING OF CAB EXPRESSION1 (TOC1). LHY and CCA1 accumulate in the morning and suppress TOC1 expression by binding to its promoter while TOC1 represses LHY/CCA1 expression in the evening (Alabadí et al., 2001; Gendron et al., 2012; Huang et al., 2012). The circadian clock plays important roles in regulating energy homeostasis and cellular metabolism (Turek et al., 2005; Zvonic et al., 2007). Animals with mutations in clock genes display impaired glucose and lipid metabolism (Turek et al., 2005), and in plants, metabolism of carbohydrates such as starch and lipids is under diel and/or circadian control (Ekman et al., 2007; Graf et al., 2010; Maatta et al., 2012).

In addition to being regulated by the clock, it has become increasingly evident that metabolic processes also serve as a clock input signal, influencing the function and timing of the oscillator. (Tu and McKnight, 2006; Asher and Schibler, 2011; Mora-García et al., 2017; Pritchett and Reddy, 2017). For example, feeding timing regulates the human circadian system and can restart rhythmic expression of some clock genes in a defective clock background (Vollmers et al., 2009; Wehrens et al., 2017). Some of the key metabolic cofactors, such as NAD⁺, flavin adenine dinucleotide, and cyclic AMP affect clock functions in animals (Rutter et al., 2001; Tu and McKnight, 2006; Ramsey et al., 2009; Pritchett and Reddy, 2017). Recent studies in photosynthetic organisms indicate that the function of the clock and photosynthesis are highly interconnected (Pattanayak et al., 2015; Atamian et al., 2016; Müller et al., 2016). The interconnection between the circadian clock and metabolism is thought to strengthen the function of both processes and enables an organism to adapt efficiently to changes in the environment. Recently, it was reported that some lipid metabolic genes were under circadian control in Arabidopsis and differentially expressed in CCA1/LHY double mutants and CCA1-overexpressing lines (Hsiao et al., 2014; Nakamura et al., 2014). Differences were observed in the levels of triacylglycerols (TAGs) and acyl-CoA between wild-type and the clock gene-altered plants (Hsiao et al., 2014). However, the interconnection between lipid metabolism and the circadian clock is not fully understood, and the cellular mediators that act as conduits to integrate the circadian clock with metabolism remain unknown in plants.

Lipids provide not only membrane structure and energy for cellular metabolism, but also a rich source of mediators in organismal growth, development, and response to changing environments. Phosphatidic acid (PA) is a central intermediate in glycerolipid metabolism, providing diacylglycerol (DAG) for the synthesis of membrane glycerolipids and the storage lipid TAG. In addition, PA is a class of lipid mediators affecting a wide range of cellular processes, including signal transduction, cytoskeletal dynamics, and vesicle trafficking in plants, microbes, and animals (Wang et al., 2006; Testerink and Munnik, 2011; Bullen and Soldati-Favre, 2016; Brown et al., 2017). The physical properties of PA, such as its dual deprotonation and capacity to form hydrogen bonds, promote its direct interaction with proteins (Kooijman et al., 2007). The propensity of PA to form a hexagonal II-type phase affects membrane curvature and protein association with membranes (Roth, 2008). PA interacts with different proteins, such as kinases, phosphatases, G-protein regulators, and metabolic enzymes (Fang et al., 2001; Loewen et al., 2004; Zhang et al., 2004; Guo et al., 2011; Kim et al., 2013; McLoughlin et al., 2013; Roy Choudhury and Pandey, 2016). PA binding can affect target protein function, such as altering catalytic activity and/or membrane association (Jang et al., 2012; Hong et al., 2016).

Recent studies indicate that PA also interacts with proteins in nuclei, mediating important processes, such as protein nuclear translocation, transcriptional regulation, DNA replication, and cell proliferation. PA and its metabolizing enzymes are present in the nucleus (Smith and Wells, 1983; Siniossoglou, 2013). For example, DAG kinases (DGKs) that phosphorylate DAG to PA are found in the nucleus in plants and animals, and this activity is essential to cell proliferation (Arisz et al., 2009; Baldanzi et al., 2016; Poli et al., 2017). PA phosphohydrolase (PAH) that dephosphorylates PA to DAG translocates between the endoplasmic reticulum and nucleus (Ren et al., 2010). In animal cells, PA alters nuclear receptor binding to DNA in cancer cells (Mahankali et al., 2015; Henkels et al., 2016). PA was found to interact with a myeloblastosis (MYB) transcription factor and modulates its nuclear translocation and root hair cell differentiation in Arabidopsis (Yao et al., 2013). To determine the nuclear function of PA, we screened an Arabidopsis transcription factor complementary DNA (cDNA) library to identify transcription factors that potentially interact with PA. Here, we report the analysis and function of interaction of PA with the core clock transcription factors LHY and CCA1.

RESULTS

Identification of PA-Interacting Clock Regulators LHY and CCA1 In Vitro

To investigate the mechanism of PA action, we screened an Arabidopsis cDNA library to identify transcription factors that potentially interact with PA by pulling down the proteins with PA-containing liposomes. We modified an Arabidopsis transcription factor cDNA library composed of ∼1500 transcription factors (Mitsuda et al., 2010) to produce recombinant proteins in Escherichia coli. PCR amplification of cDNA inserts from the library using primers upstream and downstream of the insert confirmed the large variety of E. coli clones with an average insert size of ∼1 to 1.5 kbps (Figure 1A). Recombinant proteins expressed as fusion with an N-terminal 6xHis tag were then purified from the colony pool, and a large number of different proteins expressed in the colonies were verified by immunoblotting using an anti-6xHis antibody (Figure 1B). The purified proteins were incubated and pulled down with liposomes containing phosphatidylcholine (PC) only or PC plus PA (PC:PA = 3:1 molar ratio; both total molecular species from egg yolk, hereafter designated as lipid abbreviation, otherwise designated with specific fatty acid composition). PC liposomes were used as a vehicle since PA alone does not form uniform liposomes due to its small head group that forms a cone-shaped structure (Barr and Shorter, 2000). Liposomes with PC plus phosphatidylglycerol (PG) were included as a negative control to test for the possibility of nonspecific electrostatic interactions between proteins and acidic phospholipids. Proteins coprecipitated with the liposomes were identified by protein sequencing with mass spectrometry. The core clock transcription factor LHY was one of the proteins coprecipitated specifically with PA (Supplemental Table 1) and identified by sequencing (Figure 1C; Supplemental Data Set 1).

Figure 1.

Screening of the Library for PA Binding Transcription Factors.

(A) PCR amplification of the cDNA library. PCR was performed with total plasmid DNA from the pooled E. coli colonies as a template and primers binding to upstream and downstream of the cloning sites. PCR products were separated on an agarose gel and visualized by ethidium bromide. Size of the DNA markers is indicated on the left. The experiment was performed at least twice with similar results.

(B) Immunoblotting of the protein expression library. Total proteins from the pooled E. coli colonies were separated on a polyacrylamide gel and immunoblotting was performed with an anti-6xHis antibody conjugated with alkaline phosphatase. Size of the protein markers is indicated on the left. The experiment was performed at least twice with similar results.

(C) Identification of LHY by mass spectrometry. Amino acid sequence of LHY is shown with the peptides underlined that have been sequenced by mass spectrometry (probability > 95% and MASCOT ion score > 40).

To validate and test the specificity of the PA interaction, we produced LHY from E. coli and assayed lipid binding with complementary approaches (Figure 2). Filter-blotting assays showed that LHY bound to PA, but not to other membrane phospholipids (Figure 2A, left). Among different PA species tested, LHY displayed binding to 16:0-containing PA (16:0-16:0 PA and 16:0-18:1 PA) but not to 8:0-8:0 PA, 18:1-18:1 PA, or 18:1-lyso PA (Figure 2A, right). Liposomal binding assays also showed that LHY was coprecipitated with liposomes with 16:0-containing PA, but not with liposomes containing 8:0-8:0 PA, 18:0-18:0 PA, 18:1-18:1 PA, 18:1-lyso PA, or PS (Figure 2B). Surface plasmon resonance (SPR) analysis showed that purified LHY bound to 16:0-16:0 PA-containing liposomes much more strongly than to 16:0-16:0 PC-only liposomes (Figure 2C; Supplemental Data Set 2). Because LHY and CCA1 are closely related and functionally redundant (Mizoguchi et al., 2002), we tested whether CCA1 could also interact with PA. Using SPR with liposomes containing 16:0-16:0 PC only or PC plus 16:0-16:0 PA, CCA1 was found to bind to PA-containing liposomes similar to LHY (Supplemental Data Set 2). PA binding kinetic properties were similar between LHY and CCA1 and dissociation constants for LHY and CCA1 binding to 16:0-16:0 PA-containing liposomes were 0.181 and 0.116 µM, respectively (Table 1). Filter-blotting assay demonstrated that CCA1 had specificity similar to that of LHY for both phospholipid classes and PA species (Figure 2D).

Figure 2.

PA and LHY Binding Specificity and Kinetics.

(A) Filter-blotting assays. Nitrocellulose filters spotted with the indicated lipids were incubated with LHY and blotted with an anti-6xHis antibody. Proteins were detected by alkaline phosphatase reaction. Left blot shows total molecular species of phospholipid classes from egg yolk, and right blot indicates PA species with acyl chains at sn-1 and sn-2 positions. The experiment was performed at least twice with similar results. Chl, chloroform control. PC, phosphatidylcholine; PS, phosphatidylserine; PE, phosphatidylethanolamine; PI, phosphatidylinositol; PG, phosphatidylglycerol; PA, phosphatidic acid.

(B) Liposome precipitation assay. LHY was incubated with liposomes containing PC and the indicated PA species. Protein coprecipitated with the liposomes was probed by immunoblotting with anti-6xHis antibody. Bottom blot shows protein input with 1/10 of the amount used for the assay. The experiment was performed at least twice with similar results.

(C) Surface plasmon resonance (SPR) analysis. Protein response was monitored as liposomes with 16:0-16:0 PC or 16:0-16:0 PA in 16:0-16:0 PC (1:3 molar ratio) liposomes were injected onto the sensor chip. Liposome injection was stopped at the time point indicated by arrow head. Values are means of triplicate. The original readout is provided in Supplemental Data Set 2.

(D) PA-CCA1 interaction. Filter-blotting assays were performed as in (A) with CCA1.

Kinetic Comparison of LHY and CCA1 Interaction with PA

Table 1.

Kinetic Comparison of LHY and CCA1 Interaction with PA

Kinetic Constant	LHY	CCA1
Bmax (RU)	2056 ± 63	2650 ± 109
Kon (M−1S−1)	1150 ± 58	1638 ± 48
Koff (S−1)	2.08 × 10−4 ± 6.25 × 10−6	1.90 × 10−4 ± 4.04 × 10−6
KD (M)	1.81 × 10−7 ± 8.15 × 10−9	1.16 × 10−7 ± 6.00 × 10−−9

Surface plasmon resonance (SPR) analysis was performed by monitoring protein response as liposomes with 16:0-16:0 PA were injected onto the sensor chip. Kinetic constants were calculated by Prism v5.01 (GraphPad) and are shown as mean ± SD (n = 3 reactions). RU, response unit.

Isolation of PA-LHY/CCA1 Complexes from Plants

To test whether a PA-LHY complex could be isolated from plants, we used an anti-LHY antibody to immunoprecipitate LHY from Arabidopsis seedlings and analyzed lipids co-precipitated with LHY by mass spectrometry (Figure 3). Successful isolation of LHY was confirmed by immunoblotting that detected the LHY band in the precipitates from wild-type (WT) plants with the antibody (+), but not in the immunoprecipitates from LHY knockout mutants (lhy) or without the LHY antibody (WT⁻, lhy  ⁺, or lhy  ⁻; Figure 3A). A significant amount of PA was detected in the sample immunoprecipitated from WT⁺, when compared with negative controls (WT⁻, lhy  ⁺, or lhy  ⁻; Figure 3B). PA profiling by mass spectrometry revealed that 16C-containing PA species, such as 32:0 PA (16:0-16:0 PA) and 34:2 PA, were precipitated with LHY (Figure 3C). The 16C-containing PA species precipitated are consistent with the in vitro binding data. Other common glycerolipid species, such as monogalactosyldiacylglycerol, digalactosyldiacylglycerol, PC, phosphatidylethanolamine, PG, phosphatidylinositol, and PS were not significantly detected in an LHY-dependent manner (Supplemental Figure 1).

Figure 3.

Mass Spectrometric Confirmation of PA-LHY Binding In Vivo.

(A) Isolation of LHY by immunoprecipitation. LHY was immunoprecipitated using an anti-LHY antibody from 10-d-old Arabidopsis wild-type (WT) and LHY knockout mutant (lhy) seedlings grown on 1/2 MS agar plates and harvested at ZT0, and was probed by immunoblotting using the same antibody. + and − indicate with and without antibody in the immunoprecipitation, respectively. Bottom blot shows protein input (ACTIN) with 1/10 of the amount used for the immunoprecipitation.

(B) Coprecipitation of total PA with LHY. Total PA extracted from LHY immunoprecipitated in (A) was quantified by ESI-MS/MS. Values are mean ± SD (n = 3; for each bar, three independent groups of seedlings sampled at the same time were used for immunoprecipitation, lipid extraction, and MS analysis). Asterisk denotes statistical significance compared with WT⁻, lhy  ⁺, and lhy  ⁻ as determined by one-way ANOVA (Duncan’s multiple range test; P < 0.001).

(C) Coprecipitation of PA species with LHY. PA species from total PA extracted in (B) were quantified by ESI-MS/MS. Values are mean ± SD (n = 3 as in B). Asterisk denotes statistical significance compared with WT⁻, lhy  ⁺, and lhy  ⁻ as determined by one-way ANOVA (Duncan’s multiple range test; P < 0.05).

In addition, we performed another binding assay, in which fluorescence (nitrobenzoxadiazole; NBD)-labeled, 16:0-containing PA was infiltrated into Arabidopsis seedlings expressing LHY-FLAG, CCA1-FLAG, or LUX ARRHYTHMO (LUX)-FLAG, followed by immunoprecipitation using an anti-FLAG antibody and detection of lipids coprecipitated with the respective proteins (Figure 4). Immunoblotting with the anti-FLAG antibody detected LHY, CCA1, and LUX in the immunoprecipitates used for lipid detection, verifying the isolation of the FLAG-tagged proteins from the different plants (Figure 4A). LUX-FLAG was used as non-PA-binding negative control in the immunoprecipitation as LUX is a transcription factor in clock regulation, but it did not bind PA (Figure 4B). The immunoprecipitated LHY, CCA1, and LUX were extracted with chloroform/methanol for lipids, and the lipid extracts were then separated by thin layer chromatography (TLC) and visualized under UV light. NBD-PA spots appeared from LHY and CCA1, but not from wild type or the non-PA binding control LUX (IP, Figure 4C; Figure 4D), indicating the specific interaction of PA-LHY/CCA1. TLC analysis of total lipids extracted from infiltrated plants confirmed that comparable amounts of NBD-PA were infiltrated into the different plants used for co-pull-down (Figure 4C, total). Taken together, our data from different approaches indicate that the PA-LHY/CCA1 interaction is strong and specific.

Figure 4.

Fluorescence Labeling Confirmation of PA-LHY/CCA1 Binding In Vivo.

(A) Isolation of proteins by immunoprecipitation from transgenic plants. Proteins were immunoprecipitated with an anti-FLAG antibody from Arabidopsis lines expressing the FLAG-tagged proteins indicated on the top that were harvested at ZT0 and infiltrated with NBD-PA. Immunoprecipitated proteins were probed by immunoblotting with the anti-FLAG antibody. Position of each protein and size marker are on the left and right, respectively. Bottom blot shows protein input (ACTIN) with 1/10 of the amount used for the immunoprecipitation.

(B) Filter-blotting assays demonstrating the PA-protein interaction. Nitrocellulose filters spotted with total PA (from egg yolk) were incubated with the indicated proteins and blotted with an anti-6xHis antibody. Proteins were detected by alkaline phosphatase reaction. The experiment was performed at least twice with similar results.

(C) Coprecipitation of NBD-PA with LHY and CCA1, but not LUX. Total lipids were extracted individually from the infiltrated plants (“Total”) and from the immunoprecipitates (“IP”). Extracted lipids were separated on TLC and visualized by UV illumination. NBD-PA, authentic NBD-PA standard. The experiment was performed three times with similar results.

(D) Quantification of NBD-PA from “IP” samples. NBD-PA spot density from “IP” samples shown in (C) was quantified using ImageJ software based on the known amount of NBD-PA standard. Values are mean ± SD (n = 3; for each transgenic plant, three independent groups of seedlings sampled at the same time were used for immunoprecipitation, lipid extraction, and TLC quantification). N/D, not detected.

PA Inhibition of LHY Binding to Target DNA In Vitro

LHY is a transcription factor that constitutes the core clock loop by binding to the TOC1 promoter and repressing its transcription. To probe the effect of PA binding on LHY function, we determined if PA alters LHY binding to its DNA targets (Figure 5). We used an electrophoretic mobility shift assay to examine in vitro effect of PA on LHY binding to the TOC1 promoter. A previous study reported a region in the TOC1 promoter that was essential for circadian oscillatory expression of TOC1 and was a binding target of LHY (Alabadí et al., 2001). To test for LHY binding, we used a DNA fragment that contained the intact sequence of the region (TOC1_pro-W) and the same fragment with the LHY binding site mutated (TOC1_pro-M). The LHY-DNA interaction was confirmed by the observation that LHY caused a dose-dependent upward gel shift of the TOC1_pro-W (LHY, Figure 5A), but LHY did not result in such a gel shift of the TOC1_pro-M control (Figure 5A, TOC1_pro-M). In addition, we used a PA binding protein, glyceraldehyde-3-phosphate dehydrogenase1 (GAPC1; Kim et al., 2013), as another control and inclusion of GAPC1 had no effect on the TOC1_pro-W DNA banding pattern (GAPC1;  Figure 5A).

Figure 5.

PA Inhibition of LHY-DNA Interaction.

(A) LHY interaction with TOC1 promoter. Electrophoretic mobility shift assay (EMSA) was performed with the proteins indicated on the top and a 48-bp oligonucleotide of TOC1 promoter region (TOC1_pro-W) or the same region with LHY binding site mutated (TOC1_pro-M). Black triangles indicate increasing amount (0.1, 0.2, and 0.5 mg) of proteins added. Arrows indicate positions of protein-bound (top) and free (bottom) DNA bands. The experiment was performed at least twice with similar results. GAPC1, GLYCERALDEHYDE-3-PHOSPHATE DEHYDROGENASE1.

(B) PA inhibition of LHY-DNA interaction in vitro. EMSA was performed as in (A) with TOC1_pro-W, 0.5 mg LHY, and the effectors indicated on the top. Black triangle indicates increasing amount (0, 0.01, 0.1, and 1 mg) of total PA (from egg yolk) added. Arrows indicate positions of protein-bound (top) and free (bottom) DNA bands. The experiment was performed at least twice with similar results. PC, phosphatidylcholine (1 mg); PG, phosphatidylglycerol (1 mg); NaH₂PO₄, sodium phosphate (0.1 M); HCl, hydrochloric acid (0.1 N); 18:1-18:1 and 16:0-18:1, PA molecular species (1 mg each).

Next, PA was added to the electrophoretic mobility shift assay to test whether PA could alter LHY binding to TOC1_pro-W fragment. The LHY-DNA interaction was impeded in the presence of PA as determined by PA dose-dependent loss of the gel shift (PA, Figure 5B), but not in the presence of other phospholipids PC or PG (PC and PG, Figure 5B). PA has a monoesterified phosphate group and is slightly acidic. However, introducing another acidic phospholipid PG or the acid sodium phosphate (NaH₂PO₄) or hydrochloric acid (HCl) did not cause the loss of gel shift (Figure 5B). Furthermore, consistent with the PA species specificity of LHY binding shown in Figures 2A and 2B, the LHY binding 16:0-18:1 PA markedly reduced the LHY-DNA interaction while the non-LHY binding 18:1-18:1 PA did not (Figure 5B, right), indicating that a direct PA-LHY interaction is required for the PA inhibition of LHY-DNA interaction.

In addition, we tested whether PA affects LHY binding to other known LHY target genes, such as EARLY FLOWERING4 (ELF4) and PSEUDO-RESPONSE REGULATOR9 (PRR9). Without PA, LHY caused gel shift of ELF4pro and PRR9pro, but the presence of PA led to no gel shift (Figure 6A), indicating that PA inhibits LHY-DNA interaction. As TOC1_pro-W was also a target of CCA1 (Alabadí et al., 2001), we tested the CCA1-TOC1_pro interaction sensitivity to PA. Like LHY, PA also interfered with the CCA1-TOC1_pro-W interaction in a dose-dependent manner (Figure 6B). By contrast, PA did not interfere with binding of LUX, which did not bind to PA, to its target sequence LUX_target (LUX binding site of PRR9pro; Figure 6C; Helfer et al., 2011). Taken together, these data indicate that PA inhibits the binding of both LHY and CCA1, but not other clock transcription factors, to their target promoters.

Figure 6.

Effect of PA on Various Clock Protein-DNA Interactions.

(A) PA effect on LHY binding to ELF4_pro and PRR9_pro. Electrophoretic mobility shift assay (EMSA) was performed with 0.5 mg LHY, 1 mg of total PA (from egg yolk), and oligonucleotides of ELF4 and PRR9 promoter regions. Arrows indicate positions of protein-bound (top) and free (bottom) DNA bands. The experiment was performed at least twice with similar results.

(B) PA effect on CCA1-DNA interaction. EMSA was performed with TOC1_pro-W and 0.5 mg purified CCA1. Black triangle indicates increasing amount (0, 0.01, 0.1, and 1 mg) of total PA (from egg yolk) added. Arrows indicate positions of protein-bound (top) and free (bottom) DNA bands. The experiment was performed at least twice with similar results.

(C) PA effect on LUX-DNA interaction. EMSA was performed with an oligonucleotide of LUX target sequence (LUX binding site of PRR9 promoter region), 0.5 mg purified LUX, and 1 mg total PA (from egg yolk). Arrows indicate positions of protein-bound (top) and free (bottom) DNA bands. The experiment was performed at least twice with similar results.

Effects of Genetically Altered PA Levels on LHY Binding to Target DNA in Plants

To test if changing PA levels in planta could affect LHY interactions with its target genes, we performed chromatin immunoprecipitation (ChIP), followed by quantitative PCR (qPCR) using Arabidopsis mutants with altered levels of PA. Multiple reactions contribute to PA metabolism in the cell (Figure 7A). PAH dephosphorylates PA into DAG, and the double knockout mutant of PAH1 and PAH2 (pah1 pah2) has substantially increased PA levels (Eastmond et al., 2010). By contrast, PHOSPHOLIPASE D (PLD) hydrolyzes other phospholipids to produce PA, and the double knockout mutant of two major PLDs (pldα1 pldΔ) was reported to substantially decrease cellular PA levels in Arabidopsis (Guo et al., 2012). We verified these mutants using PCR by detecting transfer DNA (T-DNA) insertion (Supplemental Figure 2A) and/or the loss of transcripts of mutated PLD, PAH, LHY, or CCA1 genes (Supplemental Figure 2B). To validate these mutant effects on PA levels, we analyzed the PA content in seedlings of the wild type, pah1 pah2, and pldα1 pldΔ. Total PA levels were elevated in pah1 pah2 by nearly 2-fold and reduced in pldα1 pldΔ to ∼40% (Figure 7B, total). However, there was no significant increase in the LHY binding PA species (32:0 PA and 34:1 PA) in pldα1 pldΔ, whereas pah1 pah2 showed an increase in the 16C-containing 34:1 PA level (Figure 7B).

Figure 7.

ChIP-PCR Analysis of LHY-DNA Binding in Mutants with Altered PA Production.

(A) Multiple reactions in PA production and removal in plants. Enzymes catalyzing each reaction are on arrows, with the number of isotypes identified in Arabidopsis in parenthesis. DAG-PPi, diacylglycerol pyrophosphate; PIP₂, phosphatidylinositol 4,5-bisphosphate; PLA, phospholipase A; LPAT, lysophosphatidic acid acyltransferase; NPC, nonspecific phospholipase C; PI-PLC, phosphatidylinositol-phospholipase C; LPP, lipid phosphate phosphatase; PAK, phosphatidic acid kinase.

(B) PA levels in wild-type (WT) Arabidopsis and PLD and PAH mutants. Total lipids were extracted from 10-day-old seedlings and PA was quantified by ESI-MS/MS. Values are mean ± SD (n = 5; for each plant line, five independent groups of seedlings sampled at the same time were used for lipid extraction and MS analysis). Asterisk denotes statistical significance compared with WT as determined by Student’s t test (P < 0.01).

(C) Immunoblotting of LHY protein expression. Nuclear proteins were extracted from 10-day-old Arabidopsis plants indicated at the time point used for the ChIP (ZT0). LHY was probed by an anti-LHY antibody. Histone H3 is included as a nuclear marker protein for a loading control. The experiment was performed at least twice with similar results.

(D) Verification of precipitated DNA by PCR. Chromatin immunoprecipitation (ChIP) was performed using an anti-LHY antibody from 10-day-old Arabidopsis plants indicated on the top. Input DNA (ID) and DNA precipitated with antibody (+) or without antibody (−) were PCR-amplified using primers specific to the promoter region (TOC1_pro) or 3′ UTR region (TOC1_{3  ′  UTR}) of TOC1. Note that the 3′ UTR region is not detected due to the DNA shearing. The experiment was performed at least twice with similar results.

(E) Quantification of precipitated DNA by qPCR. ChIP was performed as in (D). DNA precipitated with the antibody was quantified by qPCR using primers specific to TOC1 promoter region. Data are shown as % of PCR product amplified from the input DNA. Values are mean ± SD (n = 5; for each plant line, five independent groups of seedlings sampled at the same time were used for IP and qPCR). Asterisk denotes statistical significance compared with WT as determined by Student’s t test (P < 0.01). N/D, not detected.

Thus, we used pah1 pah2 and pldα1 pldΔ for the ChIP and compared them with the wild type. We verified by immunoblotting with nuclear proteins that the level of LHY protein at the time point used for the ChIP was comparable among wild-type, pldα1 pldΔ, and pah1 pah2 plants, except that cca1-1 lhy-20 had no LHY (Figure 7C). PCR analysis confirmed the successful isolation of the TOC1 promoter region binding to LHY by the ChIP in all plants except for cca1-1 lhy-20 (Figure 7D). qPCR analysis showed a decrease in TOC1_pro coprecipitated with LHY in pah1 pah2, compared with the wild type (Figure 7E), indicating a reduction of LHY-TOC1_pro interaction in pah1 pah2 plants with an increased level of PA. However, the level of TOC1_pro coprecipitated with LHY was comparable between the wild type and pldα1 pldΔ (Figure 7E). The lack of effect of pldα1 pldΔ, which had a lower total PA level than the wild type, on TOC1_pro coprecipitated with LHY could be explained by the lack of significant change in the LHY binding PA species (32:0 PA and 34:1 PA) in pldα1 pldΔ (Figure 7B).

Perturbations of Altered PA Levels on Clock Outputs

The PA inhibition of the LHY-TOC1_pro interaction in pah1 pah2 prompted us to determine PA effects on clock function in planta. We tested pah1 pah2 and a PAH1-complemented pah1 pah2 mutant with PA levels close to the wild type (Craddock et al., 2015) for alterations in the oscillation of TOC1 expression and vertical leaf movement under circadian conditions (Figure 8; Table 2). Reverse transcription quantitative PCR (RT-qPCR) analysis performed with 7-day-old seedlings for 72 h with 3-h intervals demonstrated that the rhythmicity of TOC1 expression was significantly altered in pah1 pah2; its period was 1.3 h longer than that of the wild type (Figure 8A; Table 2), but not in PAH1-complemented pah1 pah2, indicating that PA regulates the clock function. Vertical leaf movement of 5-day-old wild-type and PAH1-complemented pah1 pah2 seedlings monitored every hour for 5 d persisted with a period of ∼24 h, but in contrast, pah1 pah2 exhibited a severe reduction in amplitude and a longer period of leaf movement by 1 h (Figure 8B; Table 2). In cca1-1 lhy-20 seedlings used as a control, TOC1 expression was arrhythmic, and the period of leaf movement was much shorter than that of the wild type, as expected (Figures 8C and 8D). T-DNA insertion in pah1 pah2 and the introduction of PAH1 into PAH1-complemented pah1 pah2 mutant were confirmed by PCR (Supplemental Figure 2A).

Figure 8.

Perturbation of Circadian Outputs by PAH Mutations.

(A) and (C)  TOC1 expression under circadian condition. Plants were entrained to 12-h light/12-h dark cycles for 7 days, and TOC1 expression was analyzed under constant light by RT-qPCR. Values are mean ± SD (n = 6; for each plant line, six independent seedlings sampled at each time point indicated were used for RNA extraction and RT-qPCR analysis) normalized to the wild type (WT) at 0 h. Period length is in parenthesis. PAH-com, PAH1-complemented pah1 pah2.

(B) and (D) Vertical leaf movement under circadian condition. Plants were entrained to 12-h light/12-h dark cycle for 5 days, and leaf movement was monitored under constant light. Values are mean ± SD (n = 12 for C and 8 for D; for each plant line, 12 or 8 independent seedlings photographed at each time point indicated were used for image analysis) normalized to initial leaf position. Period length is in parenthesis. PAH-com, PAH1-complemented pah1 pah2.

Period Estimates of Circadian Rhythms Shown in Figures 8 and 9

Table 2.

Period Estimates of Circadian Rhythms Shown in Figures 8 and 9

Circadian Output	Plant Line or Drug Treatment	Period (h)	P	RAE	N
TOC1 expression	WT	24.4 ± 1.1	1.000	0.42 ± 0.063	6
(Figure 8A)	pah1 pah2	25.7 ± 0.8	0.041*	0.37 ± 0.039	6
	PAH-com	24.5 ± 1.5	0.898	0.39 ± 0.052	6
Leaf movement	WT	23.7 ± 0.4	1.000	0.29 ± 0.049	12
(Figure 8B)	pah1 pah2	24.7 ± 0.9	0.002**	0.28 ± 0.036	12
	PAH-com	23.8 ± 1.2	0.787	0.34 ± 0.051	12
TOC1 expression	None	24.9 ± 1.1	1.000	0.25 ± 0.087	>20
(Figures 9A and 9C)	1-butanol	23.9 ± 1.4	0.016*	0.26 ± 0.063	>20
	2-butanol	24.6 ± 1.5	0.475	0.28 ± 0.066	>20
	DGK inhibitor	24.0 ± 1.3	0.023*	0.23 ± 0.019	>20
	DMSO	24.7 ± 1.5	0.633	0.23 ± 0.026	>20
Leaf movement	None	23.8 ± 0.9	1.000	0.30 ± 0.049	12
(Figures 9B and 9D)	1-butanol	22.6 ± 0.5	<0.001**	0.28 ± 0.033	12
	2-butanol	23.3 ± 0.8	0.164	0.32 ± 0.084	12
	DGK inhibitor	22.5 ± 1.2	0.007**	0.31 ± 0.044	12
	DMSO	23.2 ± 1.5	0.247	0.28 ± 0.057	12

Brief description of how each experiment was performed is provided in the figure legends. P, Student’s t test paired with control (wild type [WT] or None); RAE, mean of relative amplitude error; N, the number of rhythmic samples. N/D, not determined; asterisk denotes significant difference when compared with controls (*P < 0.05; **P < 0.01). See Methods for details of statistical analysis.

To further test the PA effect on circadian phenotypes, we used small molecules to suppress PA formation in plants. We exposed wild-type plants to 1-butanol, which decreases PLD-mediated PA production by diverting the phosphatidyl moiety to form phosphatidylbutanol, and a DGK inhibitor R59022, which suppresses the PA formation from DAG (Gardiner et al., 2003; Peters et al., 2010). For TOC1 expression analysis, transgenic Arabidopsis with luciferase gene driven by TOC1 promoter (TOC1:LUC) was used to monitor its luciferase reporter activity following the drug applications (Alabadí et al., 2001). The period of TOC1 expression was shortened by both 1-butanol and DGK inhibitor by ∼1 h (Figure 9A; Table 2), but their respective controls, 2-butanol and DGK inhibitor solvent DMSO had no significant effect (Figure 9C; Table 2). In addition, 1-butanol and DGK inhibitor, but not their respective controls, shortened the period of vertical leaf movement by ∼1.3 h (Figures 9B and 9D; Table 2). To validate whether the PA levels were indeed changed by the chemical treatments, we quantified total PA and the 16C-containing (32:0 PA and 34:1 PA), LHY/CCA-binding PA species by mass spectrometry. The levels of at least one of the two LHY/CCA-binding PA species were reduced by the drugs when compared with their controls (Figure 9E). The treatment of 1-butanol, but not 2-butanol, decreased the level of total PA, 32:0 PA, and 34:1 PA, whereas the DGK inhibitor decreased 34:1 PA but had no effect on the total or 32:0 PA (Figure 9E). The opposite effects of PAH mutation and the drug applications on the clock outputs suggest that altered PA levels regulate the circadian clock.

Figure 9.

Perturbation of Circadian Outputs by Chemical Manipulation of PA.

(A) and (C) Effect of 1-butanol and DGK inhibitor on circadian expression of TOC1. Transgenic plants with TOC1:LUC were entrained to 12-h light/12-dark cycles for 7 days then treated with 1% 1-butanol, 100 mM DGK inhibitor (R59022), 1% 2-butanol, or 1% DMSO at ZT12. Luciferase reporter activity was monitored under constant light. Values are mean ± SD (n > 20; for each treatment, more than 20 independent seedlings photographed at each time point indicated were used for image analysis) normalized to background signal. Period length is in parenthesis.

(B) and (D) Effect of 1-butanol and DGK inhibitor on circadian leaf movement. Wild-type (WT) plants were entrained to 12-h light/12-dark cycles for 5 days then treated with 1% 1-butanol, 100 mM DGK inhibitor (R59022), 1% 2-butanol, or 1% DMSO at ZT12. Leaf movement was monitored under constant light. Values are mean ± SD (n = 12; for each treatment, 12 independent seedlings photographed at each time point indicated were used for image analysis) normalized to initial leaf position. Period length is in parenthesis. In (C) and (D), the untreated control (None) was reproduced from (A) and (B), respectively, for ease of comparison.

(E) PA levels in Arabidopsis treated with butanol and DGK inhibitor. Total lipids were extracted from 7-day-old seedlings treated as in (A) and (C) for 1 h. PA was quantified by ESI-MS/MS. Values are mean ± SD (n = 5; for each treatment, five independent groups of seedlings sampled at the same time were used for lipid extraction and MS analysis). Asterisk denotes statistical significance compared with controls as determined by Student’s t test (P < 0.01).

Effects of Altered Expression of Clock Regulators on Membrane and Storage Lipids

Levels of some plant fatty acids and glycerolipids have been reported to change diurnally (Sasaki et al., 1997; Ekman et al., 2007; Maatta et al., 2012), but whether lipid metabolism is under circadian control remains largely unexplored in plants. We used the clock mutant cca1-1 lhy-20 to test the effect of the clock regulators on the daily lipid changes in Arabidopsis (Figure 10; Supplemental Data Set 3). Seedlings were grown under 16-h light/8-h dark cycles and monitored for changes in their membrane glycerolipid levels for up to 24 h to assay how lipid levels change during the full cycle of LHY/CCA1 expression in the wild type and how they are affected in cca1-1 lhy-20. Out of more than a hundred lipid species detected by mass spectrometry, several lipid species, PG (36:4), PS (38:5), and PA (34:4 and 36:6), showed daily cycling with a period of ∼24 h in the wild type but no apparent cycling in cca1-1 lhy-20 (Figures 10A to 10D; Supplemental Data Set 3). The levels of 36:4 PG and 38:5 PS appeared to cycle in the wild type with a phase slightly delayed from that of the PA species (Figures 10A and 10B compared to Figures 10C and 10D).

Figure 10.

Effect of LHY and CCA1 on Lipid Production.

(A) to (E) Time-dependent change in levels of 36:4PG (A), 38:5PS (B), 36:6PA (C), 34:4PA (D), and total PA (E) in the wild type (WT) and lhy cca1. 10-day-old plants were grown in 16-h light/8-h dark cycles, and total lipids were extracted every 3 h for up to 24 h. Membrane glycerolipids were quantified by ESI-MS/MS. Data points (WT closed; lhy cca1 open) are mean ± SD (n = 5; for each plant line, five independent groups of seedlings sampled at each time point indicated were used for lipid extraction and MS analysis), through which the best-fit lines (WT solid; lhy cca1 dashed) are shown with the coefficient of determination (R²) for WT resulted from the polynomial regression analysis. Note that R² for lhy cca1 was not significant (<0.5) from any regression analysis performed (linear regression shown here). Asterisks denote statistical significance compared with lhy cca1 at each time point as determined by Student’s t test (P < 0.05). DW, dry weight of seedlings. All individual values are provided in Supplemental Data Set 3.

(F) Seed oil content. Triacylglycerols extracted from dry seeds were transmethylated and the resulting fatty acid methyl esters were quantified by gas chromatography. Values are % of dry seed weight and mean ± SD (n = 5; for each plant line, five independent groups of dry seeds harvested at the same time were used for lipid extraction and GC analysis). Asterisks denote statistical significance compared with WT as determined by Student’s t test (P < 0.01). OE, transgenic plant overexpressing LHY.

Total PA levels were not significantly different between the wild type and cca1-1 lhy-20 at all time points measured, except for the first few hours (Figure 10E). The loss of cyclic changes in lipid levels in the clock mutant suggests that the lipid level changes observed in the wild type are influenced by the circadian clock, rather than just diurnal (light-dark) response.

In addition, we examined whether LHY and CCA1 mutations affected storage lipid accumulation in Arabidopsis seeds. The rationale was that PA is a central intermediate for the biosynthesis of glycerolipids, including TAG, and that the misregulation of PA due to the defect of the circadian clock might affect the production of TAG, the desired product of many oilseed crops. Also, a previous report indicated that overexpression of CCA1 increases TAG accumulation in Arabidopsis seeds (Hsiao et al., 2014). Compared with seeds of wild-type plants that were grown side-by-side with mutant plants, the content of major seed oil TAG was decreased in cca1-1 lhy-20 by ∼18%, whereas no significant difference in seed oil content was observed in lhy, possibly due to the overlapping function of CCA1 (Figure 10F). The seed oil content of three LHY-OE (overexpression) lines displayed an increase by 12% to 17% over that of wild-type seeds (Figure 10F). These data indicate that perturbations of the circadian clock affect the production of both membrane and storage lipids.

To probe how LHY might affect lipid metabolism, we analyzed the expression of lipid metabolic genes using a database (diurnal.mocklerlab.org; Mockler et al., 2007) for time-course gene expression in lhy1-1 (LHY-OE line). Some members of the gene families involved in the PA metabolism as shown in Figure 7A were differentially expressed in LHY-OE compared with the wild type in a short-day time course (8-h day/16-h night; Supplemental Figure 3). In LHY-OE plants, DGK1 and PLDΔ showed oscillatory expression with greater amplitude, and LPP2 was expressed with reduced amplitude. LPAT2, PAH1, and PI-PLC2 were expressed in antiphase with those in the wild type (Supplemental Figure 3). Importantly, some of these, including PLDΔ, PAH1, and DGK1, contain cis EE (evening element) motifs in their promoter sequences, which is bound by both LHY and CCA1. CCA1 was reported to bind to the EE of PAH1 (Nagel et al., 2015). The gene expression change in response to altered LHY levels indicates that LHY/CCA1 regulate genes involved in PA metabolism as a clock output process.

DISCUSSION

This study shows that the lipid mediator PA directly interacts with key clock regulators LHY and CCA1, inhibits their interaction with target DNA, and alters circadian outputs. The PA-protein interaction was initially identified using in vitro approaches but further confirmed using LHY and PA in plant cells. In addition, our genetic and pharmacological data show that altered PA metabolism affected clock function, as indicated by altered circadian oscillation of TOC1 expression and leaf movement. Elevation of PA by pah1 pah2 mutation lengthened the period, whereas suppression of PA production shortened the period. Furthermore, the data indicate that the circadian clock regulates lipid metabolism and affects seed oil production. These results suggest a reciprocal regulatory mechanism between the circadian clock and the lipid mediator PA. The interaction between the central glycerolipid metabolite PA and the core clock regulators may represent a molecular conduit to integrate the circadian clock with lipid metabolism (Figure 11).

Figure 11.

Proposed Model for Reciprocal Regulation of Circadian Clock and Lipid Metabolism.

PA binds to LHY and CCA1 and perturbs the circadian clock by suppressing their binding to TOC1 promoter. Reciprocally, the clock oscillators regulate the expression of genes involved in lipid metabolism to tune membrane and storage lipids including PA.

The preference of LHY/CCA1 binding for particular PA species suggests that acyl chain composition of PA is an important determinant for PA interaction with specific target proteins. This binding specificity is in contrast with some other PA binding proteins. ABSCISIC ACID INSENSITIVE 1 binds to 18:1-18:1 PA but has no detectable binding with 16:0-16:0 PA (Zhang et al., 2004), whereas sphingosine kinases bind to 16:0-18:1 PA and 18:1-18:1 PA equally well, but not 18:0-18:0 PA or 18:2-18:2 PA (Guo et al., 2011) and GAPCs have no specific preference for PA species tested (Kim et al., 2013). The PA species preference could underlie a basis for cellular control of lipid-protein interactions and ultimately feedback regulation of lipid metabolism by the clock. In addition, the PA levels used in the study and binding kinetics determined are within the cellular range of PA levels. For example, cellular PA in Arabidopsis leaves was estimated to be around 50 to 100 µM (Zhang et al., 2004; Wang et al., 2006). The dissociation constant for LHY and CCA1 binding to 16C-containing PA liposomes as determined by SPR in this study was around 0.1 to 0.2 µM. Taking into account the finding that not all PA species bind to LHY/CCA1, the level of 16C-containing PA that binds LHY/CCA1 would still be in the physiological range. In addition, the cellular PA level is highly dynamic and transiently increased under various stress condition (Vu et al., 2012). Moreover, PA that interacts with LHY/CCA1 is expected to be associated with nuclei, and further study is needed to determine the level and molecular species of PA associated with nuclei.

The cellular level of PA is affected by the activity of multiple enzymes and what PA species are changed by what PA-metabolizing enzymes is largely unknown. Our data showed that pah1 pah2 had an increase in 34:1 PA and altered the circadian outputs, suggesting that PAH1 and PAH2 affect the PA that interacts with the clock regulators. This is partially supported by a previous report that 16:0 and 18:1 fatty acid contents are increased in both root and leaf of pah1 pah2 (Eastmond et al., 2010). Results from butanol treatments suggest that PLDs might play a role in the clock regulation, but the two active PLDs, PLDα1 and PLDΔ, are not involved because the double mutant pldα1 pldΔ did not show alterations in LHY-TOC1_pro interaction or clock outputs. In addition, DGK inhibition altered TOC1 expression and leaf movement. Furthermore, de novo synthesis of PA could not be ruled out to affect the clock regulation, as our database analysis showed that the expression of LPAT2 and LPP2 was misregulated when LHY was overexpressed. These data suggest that LHY/CCA1 binding PA species could result from multiple enzymatic reactions. However, some of the altered PA-metabolizing genes/enzyme are involved in producing PA for the interaction, whereas others may be part of the target lipid-metabolizing genes regulated by the clock. Future studies will be needed to delineate the function of the multiple enzymes involved in PA metabolism, including analyzing various mutants with altered PA levels and their molecular species.

The PA species that binds LHY/CCA1 also weakens the LHY/CCA1 interaction with target DNA, suggesting that PA binding is involved in the PA perturbation of LHY/CCA1-DNA binding. The data that non-LHY/CCA1 binding PAs have no effect on LHY/CCA1-DNA binding suggest that the PA impediment is not simply due to the negative charge of PA that may interfere with the protein-DNA binding. To test the relevance of PA effect in plants, we used mutant plants with altered PA levels to perform ChIP-PCR, and the data were consistent with the finding that higher levels of PA impede LHY/CCA1-DNA interaction. It is intriguing to note that the elevation of PA by pah1 pah2 mutation lengthened the period, whereas the period in CCA1/LHY-null mutant was shortened compared with that of the wild type. The effect of pah1 pah2 on period length is consistent with an early report that an Arabidopsis transgenic line expressing additional copies of TOC1 locus lengthened the free-running period of CHLOROPHYLL-A,B BINDING PROTEIN 2 (CAB2) expression under constant light condition (Más et al., 2003).

Because TOC1 is the direct target that LHY and CCA1 suppress, one would expect that increased TOC1 expression would decrease LHY/CCA1 repression, so that TOC1-overexpression would mimic the effect of lhy cca1 (i.e. shortening the period length). One explanation for this apparent inconsistency is that the effect of an effector (such as PA and TOC1) exerted in the presence of CCA1/LHY versus that in the absence of CCA1/LHY may not be directly comparable. For example, in the absence of CCA1/LHY, other transcription factors may interact with the gene targets that CCA1/LHY would normally bind, resulting in inappropriate crosstalk. Thus, the activity of LHY/CCA1 suppressed by increased PA and the absence of LHY/CCA1 in the null mutant background might not have the same effect on the clock period.

It is also notable that pah1 pah2 lacks overall increase in TOC1 expression despite LHY/CCA1 binding to TOC1 promoter being perturbed by the increase in PA level. This discrepancy has also been exemplified by the lack of substantial amplitude difference from the wild type in (1) LHY/CCA1 expression in a TOC1 mutant line with a semi-dominant allele (Alabadí et al., 2001), (2) TOC1 expression in LHY or CCA1 single mutant (Mizoguchi et al., 2002), and (3) CAB2 expression in a transgenic line expressing additional copies of TOC1 locus (Más et al., 2003), all of which display changes in circadian period. One explanation would be that ELF4 expression increased by PA could attenuate increase in TOC1 expression in pah1 pah2 since ELF4, a known repressor of TOC1, is repressed by LHY/CCA1. As such, the robust architecture, tight reciprocal regulation, and interlocking feedback loops within the circadian clock impose a level of complexity that can prevent simple prediction of how the clock will respond to perturbation.

Meanwhile, the results indicate that the circadian clock regulates lipid metabolism. An animal study using mouse showed that clock dysfunction resulted in hyperlipidemia and obesity (Turek et al., 2005), suggesting that circadian regulation is critical for the accumulation of cholesterol and neutral lipids. Another study showed that RHYTHM OF CHLOROPLAST 40, a clock transcription factor with N-terminal MYB repeat found in LHY and CCA1, is required for nitrogen starvation-induced TAG accumulation in green algae (Goncalves et al., 2016). Our results show that genetic ablation or overexpression of the clock regulator LHY and CCA1 alters seed oil contents, and that diel changes for several phospholipid species are compromised in cca1-1 lhy-20 in Arabidopsis. PA is a central intermediate for TAG biosynthesis, and its misregulation due to the defect of the circadian clock could affect the accumulation of its final product. The levels of total and some species of PA tend to be reduced in cca1-1 lhy-20 when LHY and CCA1 are highly expressed (e.g. at 0 and 24 h), suggesting that the PA pool in tissues of cca1-1 lhy-20 may not provide sufficient PA for the normal production of TAG. Also, it is conceivable that the circadian clock regulates lipid content and composition through regulating genes involved in lipid metabolism. These results, together with others, indicate that a number of genes involved in glycerolipid metabolism display circadian profiles, and the clock regulators LHY and CCA1 regulate their expression (Hsiao et al., 2014; Nakamura et al., 2014). The clock effect of lipid metabolism may provide insight into understanding of lipid-related plant performance and underlie a molecular basis for fluctuations in lipid content and composition affected by seasons and latitudes.

In addition, the interaction between the stress-responsive lipid second messenger PA and the clock regulators may suggest that PA functions to integrate clock function with plant stress responses because many biotic and abiotic stress-responsive genes are controlled by the clock and have an EE in their promoters. For example, expression of C-REPEAT BINDING FACTORS (CBFs), the major players in cold acclimation, is clock-regulated, peaking at midday, and their promoters are bound by some clock proteins, including CCA1 (Dong et al., 2011). CCA1 activates CBFs and promotes freezing tolerance, whereas PRRs repress CBFs and inhibit freezing tolerance in Arabidopsis (Nakamichi et al., 2009; Dong et al., 2011). PA is a potent cellular mediator produced under various stress conditions, including cold (Hong et al., 2016). Therefore, PA may serve as a metabolic conduit that connects the circadian clock and plant responses to environmental changes.

In summary, this study suggests a mechanistic interconnection between the circadian clock and lipid metabolism. We propose that the circadian clock regulates lipid metabolism through controlling the temporal expression of lipid metabolic genes, whereas the glycerolipid intermediate PA modulates the clock function (Figure 11). The PA-LHY/CCA1 interaction acts as a retrograde mediator of the clock in response to metabolic and/or environmental cues. Such regulation may allow plant cells to sense intracellular levels of lipid metabolites and monitor the activity of lipid metabolism to coordinate flux through lipid pathways with other physiological demands in response to daily changes in the environment. In addition, data on the effect of circadian clock on lipid content and composition potentially have significant application to seed oil production as affected by growth conditions, such as planting seasons, day length, and latitudes.

METHODS

Cloning, Protein Expression, and Purification of Transcription Factor (TF) cDNA Library

The cDNA library was obtained from Dr. Mitsuda (National Institute of Advanced Industrial Science and Technology; Mitsuda et al., 2010). The DNAs were cloned into pDONR221^TM then pET-53-DEST^TM (Novagen) using Gateway Recombination Kit according to manufacturer’s instructions (Life Technologies) and were transformed into Max Efficiency DH5α Competent Cells (Life Technologies). The resulting expression clones were then transformed into Escherichia coli Rosetta (DE3) for protein expression. Colonies were pooled and incubated in Luria-Bertani media at 37°C to mid-log phase, and protein expression was induced at 15°C for 6 h. Harvested cells were broken up by sonication in lysis buffer (50 mM NaH₂PO₄, pH 8.0, 300 mM NaCl, 10 mM imidazole). Following centrifugation at 10,000g for 20 min, pellet containing insoluble proteins was solubilized in lysis buffer containing 8 M urea and centrifuged at 10,000g for 20 min. The originally soluble and the urea-solubilized proteins were purified using Ni-nitrilotriacetic acid (NTA) Agarose (QIAGEN) with native and denaturing buffers, respectively, according to manufacturer’s instructions. The urea-solubilized proteins were spontaneously refolded during the purification by omitting urea in washing/eluting buffers (Holzinger et al., 1996). The two eluates were combined and dialyzed overnight at 4°C in Tris-buffered saline (TBS) buffer (50 mM Tris, pH 7.6, 150 mM NaCl).

Screening for PA-Binding TFs

Fifty micrograms of the purified TF proteins were incubated for 1 h with liposomes prepared as previously described (Kim et al., 2013). Protein-liposome complex was harvested by centrifugation at 16,000g for 30 min and washed three times to remove unbound proteins using binding buffer (25 mM Tris-HCl, pH 7.5, 125 mM KCl, 1 mM DTT, 0.5 mM EDTA). The liposome-bound proteins were resolved by SDS-PAGE and visualized by Coomassie blue. Each whole lane containing protein bands was carefully excised from the gel, and the proteins were in-gel digested with trypsin (Sigma-Aldrich) at 37°C overnight, according to the manufacturer’s instructions. The digested peptides were run on the LC-tandem MS using an LTQ-Orbitrap Velos mass spectrometer (Thermo Fisher Scientific) coupled with a U3000 RSLCnano HPLC (Thermo Fisher Scientific). The samples were dried down to concentrate and resuspended in 10 μL of 5% acetonitrile/0.1% formic acid mixture. Five microliters was injected for liquid chromatography-tandem mass spectrometry (MS/MS) on 2-h gradient separation and data acquisition. The database search was performed with peptide mass fingerprint data using MASCOT (v2.4) database search engine (Matrix Science) against the NCBI database for Arabidopsis (Arabidopsis thaliana). The criteria for a significant protein identification were both at least two unique peptides per protein identified and each peptide showing a probability > 95% (MASCOT ion score > 40). Proteins found from PA + PC liposome but not from control liposomes (PC only or PG + PC) were considered as PA binding candidates.

SPR Analysis

SPR analysis was performed using Biacore 2000 system according to the manufacturer’s instructions with some modifications. Liposomes were suspended in running buffer (10 mM HEPES, pH 7.4, 150 mM NaCl, 50 μM EDTA) at a final concentration of 100 μM. Purified proteins (LHY and CCA1) were dialyzed overnight at 4°C and diluted in running buffer at a final concentration of 100 nM. A sensor chip preimmobilized with NTA was used to capture the 6xHis-tagged proteins. For each experiment, running buffer containing 500 μM NiCl₂ was injected to saturate the chip with nickel. The 6xHis-tagged proteins were immobilized on the chip via Ni²⁺/NTA chelation. Response unit was monitored as liposomes were injected over the surface of the chip. After each run, the NTA chip was regenerated by stripping nickel from the surface with regeneration buffer (10 mM HEPES, pH 8.3, 150 mM NaCl, 0.35 M EDTA). The sensorgrams were plotted by Microsoft Office Excel (2007), and kinetic constants were calculated by Prism v5 (GraphPad Software).

Filter-Lipid Blotting Assay and Immunoblotting

Ten micrograms of lipids dissolved in chloroform were spotted on a piece of nitrocellulose membrane (0.45 μm pore; Whatman) and air-dried for 30 min. The same volume of chloroform was spotted on the same membrane as a solvent control. The membrane was incubated with TBS containing Tween 20 (TBST) buffer (10 mM Tris-HCl, pH 7.4, 140 mM NaCl, 0.1% [v/v] Tween 20) containing 0.5% (w/v) fatty acid-free BSA for 1 h to block the membrane. The membrane was then washed three times with TBST buffer and incubated with 20 μg of purified protein for 2 h. After washing three times with TBST buffer to remove unbound proteins, proteins on the membrane were probed with an anti-6xHis antibody conjugated with alkaline phosphatase (Sigma-Aldrich; cat# A5588) and were visualized with Alkaline Phosphatase Conjugate Substrate (Bio-Rad) according to the manufacturer’s instructions.

Protein samples were dissolved in SDS-PAGE loading buffer, boiled for 5 min, and loaded on 10% (v/v) polyacrylamide gel. After running the gel at 100 V for ∼1 h, proteins were electrophoretically transferred onto a polyvinylidene fluoride membrane using Semidry Trans-Blot apparatus (Bio-Rad) at 20 V for 20 min. The membrane was blocked in TBST buffer containing 5% (w/v) nonfat milk for 1 h, followed by washing three times with TBST buffer. The membrane was incubated for 1 h with primary antibodies (anti-LHY antibody from Abiocode [cat# R3095-2] or anti-6xHis antibody from Sigma-Aldrich). After washing three times with TBST buffer, the membrane was incubated with secondary antibodies from mouse (Sigma-Aldrich; cat# A1293) or rabbit (Sigma-Aldrich; cat# A7539) conjugated with alkaline phosphatase for 1 h. Proteins were visualized by Alkaline Phosphatase Conjugate Substrate (Bio-Rad) according to the manufacturer’s instructions.

Immunoprecipitation and Analysis of Protein-Lipid Complex from Plants

Immunoprecipitation was performed using 10-d-old Arabidopsis seedlings grown on 1/2 Murashige and Skoog (MS) agar plates, an anti-LHY antibody (Abiocode; cat# R3095-2; Figure 3A) or anti-FLAG antibody (GenScript; cat# A00187; Figure 4A) and Protein A-Sepharose (Sigma-Aldrich), according to the manufacturer’s instructions. For NBD-PA labeling (Figure 4C), 10-d old seedlings were vacuum-infiltrated with 0.5 mg/mL sonicated NBD-PA (Avanti polar lipids) and 0.05% (v/v) Silwet L-77 for 5 min, followed by agitation for 80 min. Plant tissues were ground with liquid nitrogen and incubated in protein extraction buffer (50 mM Tris-HCl, pH 7.3, 50 mM NaCl, 5% glycerol, 1 mM DTT) containing a protease inhibitor cocktail (Sigma-Aldrich) on ice for 30 min. The homogenate was filtered through four layers of Miracloth (Calbiochem). Following brief sonication for membrane disruption, supernatant after centrifugation at 12,000g for 10 min at 4°C was used as a protein extract. Ten milligrams of Protein A-Sepharose was swollen with 0.2 mL of buffer A (20 mM NaH₂PO₄ pH8.0, 150 mM NaCl) for 1 h and washed with buffer A. The protein extract (100 μg total proteins determined by Bradford assay) was mixed with 5 μg of antibody, the swollen Protein A-Sepharose, and buffer A to the final volume of 1 mL, and gently agitated overnight at 4°C. The mixture was washed five times with buffer A by centrifugation at 2500g for 1 min. For protein-lipid complex analysis, lipids were extracted from the resulting pellet with chloroform/methanol (2:1) mixture, dried under gentle stream of nitrogen gas, and resuspended with chloroform. The lipid extracts were analyzed by mass spectrometry, as described below, for Figure 3B or by TLC for Figure 4C. TLC was performed on a silica plate (Silica Gel 60 F254; Merck) using chloroform/methanol/ammonium hydroxide (65:35:5) mixture as a developing solvent. NBD-PA was visualized on a UV illuminator. NBD-PA spot was quantified using Image J software (v1.48).

Gel Mobility Shift Assay

DNA probes used were a 48 bp of TOC1 promoter region (−734∼−687 from start codon) containing the EE and the same region with the last six nucleotides of EE changed to CTGCAG (Alabadí et al., 2001; TOC1pro-W and TOC1pro-M, respectively), a 45 bp of ELF4 promoter region (−336∼ −292), a 43 bp of PRR9 promoter region (−251∼209), and a 60 bp of LUX target sequence (−424∼ −365 in PRR9 promoter). Nucleotide sequence of all probes used is in Supplemental Table 2. The probes were labeled with fluorescein (6-FAM) at 5ʹ end. Single-stranded complementary oligonucleotides were self-annealed in TAE buffer (10 mM Tris, pH 8.0, 50 mM NaCl, 1 mM EDTA) by incubating at 95°C for 5 min and slowly cooling down to room temperature for 1 h. One picomole of the resulting double-stranded DNA was incubated in TAE buffer with purified protein (LHY, CCA1, LUX, or GAPC1) and various effectors (e.g., PA) at room temperature for 15 min. Ten percent (v/v) native polyacrylamide gel was used for separation of DNA. After prerunning without sample at 100 V for ∼1 h to remove ions and equilibrate the gel, the reaction mixture was added with gel loading buffer (30% [v/v] glycerol without bromophenol blue) and separated at 100 V at 4°C. DNA was visualized by an Azure c600 imager (Azure Biosystems) at 495 nm excitation and 520 nm emission wavelengths.

ChIP

ChIP was performed using EpiQuik Plant ChIP Kit (Epigentek) according to the manufacturer’s instructions. Proteins and DNA were crosslinked in vivo by vacuum infiltration of 1% formaldehyde into 10-day-old Arabidopsis seedlings, followed by quenching with 0.125 M glycine. Tissues were then ground in liquid nitrogen and suspended in 1× CP3C solution. Tissue debris was removed by filtering through two layers of Miracloth (22∼25 μm pore; Calbiochem) and centrifuging at 1900g for 20 min. The pellet was re-suspended in CP3D solution containing β-mercaptoethanol (BME) and centrifuged at 13,000g for 10 min to pellet nuclei. The nuclear pellet was resuspended in CP3E solution containing BME, layered on 300 μL cushion of CP3E solution containing BME, and centrifuged at 17,500g for 45 min. The resulting chromatin pellet was resuspended in CP3F solution containing protease inhibitor cocktail. DNA was randomly sheared by sonication for 1 min (4 × 15-s burst) and checked for size (200∼1000 bps) on an agarose gel. After centrifuging at 17,500g for 10 min, the supernatant (“input DNA”) was incubated for 1 h in the strip wells provided that were precoated with 2-μg antibodies. Unbound proteins/DNA were removed by washing the strip wells, and bound proteins were digested by proteinase K at 65°C for 15 min. Crosslinked DNA was reversed with CP6 solution at 65°C for 90 min and purified through the spin column provided. The final DNA product was used as a template for qPCR, as described below.

RT-qPCR

Total RNA was extracted from plant tissues using TRIzol Reagent (Life Technologies) per the manufacturer’s instructions. RNA was quantified by Nanodrop 2000 spectrophotometer (Thermo Scientific) and checked for integrity by agarose gel electrophoresis. cDNA was synthesized by SuperScript III reverse transcriptase (Life Technologies) with 1 μg of RNA and 0.5 μg of oligo(dT)₁₈ primers, according to the manufacturer’s instructions. The reaction was at 50°C for 30 min with preincubation at 65°C for 5 min and enzyme inactivation at 70°C for 15 min. The cDNA was amplified with a Taq polymerase using gene-specific primers (sequence provided in Supplemental Table 2) through the following thermal cycling conditions: preincubation at 95°C for 2 min, 40 cycles of 95°C for 30 s, 55°C for 30 s, and 68°C for 1 min, and final extension at 68°C for 5 min. The PCR progress was monitored by adding SYBR Green dye using StepOnePlus Real-Time PCR System (Applied Biosystems), and data were processed by StepOne Software (v2.0.2). The gene expression was normalized with UBIQUITIN 10 as an internal standard.

Arabidopsis Growth Conditions and Mutant Lines

All Arabidopsis lines used are in the Col-0 background. Arabidopsis seeds were surface-sterilized with 70% (v/v) ethanol then 20% (v/v) bleach followed by washing with water and were stratified at 4°C for 2 d. Plants were germinated and grown in 1/2 MS media with 1% (w/v) sucrose at 22°C under light cycles of 16-h light/8-h dark or 12-h light/12-h dark with a photosynthetic photon flux density of 120 to 150 μmol/m²/s (30-W fluorescent bulbs). At the end of the dark period, plants were transferred to constant light or dark for circadian conditions. PLDα1 PLDΔ-double knockout (KO) pldα1 pldΔ (SALK_053785/SALK_023247) was generated by crossing single mutants (Guo et al., 2014). PAH1 PAH2 double KO pah1 pah2 (SALK_042970/SALK_047457) and that complemented with PAH1, which were generous gifts from Dr. Eastmond, were generated as described previously (Eastmond et al., 2010; Craddock et al., 2015). KO mutants of lhy (SALK_149287) were obtained from the Arabidopsis Biological Resource Center (Ohio State University) and LHY-CCA1 double-KO cca1-1-lhy20 was generated as described below. All KO mutants were confirmed as homozygotes by antibiotics selection and PCR-based genotyping. To generate LHY-overexpression lines, LHY-coding sequence was amplified from 10-d-old seedlings by PCR using pFAST-LHY primers (sequence provided in Supplemental Table 2). The PacI/SalI-digested DNA fragment was cloned into a binary vector p35S-FAST and introduced into Arabidopsis (Col-0) via agrobacterium-mediated transformation, as described previously (Guo et al., 2014). LHY-overexpression lines were identified by resistance to kanamycin and verified by PCR and DNA sequencing, and a total of 15 positive lines were generated.

To generate Arabidopsis lines with FLAG-tagged LHY/CCA1/LUX, cca1-1 lhy20 and cca1-1 CCA1:LUC were generated by crossing cca1-1 to either lhy20 and CCA1:LUC (Pruneda-Paz et al., 2009), respectively. For 35S:LUX-FLAG₃-His₆, pENTR-LUX (Helfer et al., 2011) was recombined with pB7-6xHis-3xFLAG-C-term (HFC; Huang and Nusinow, 2016) by LR Clonase II (Invitrogen) to generate pB7HFC-LUX. For LHY:LHY-FLAG₃-His_6, first the LHY coding sequence was amplified from cDNA using LHY-FLAG primers (sequence provided in Supplemental Table 2) and recombined into NotI/AscI-digested pENTR-dTOPO using InFusion (Takara Clonetech) to produce pENTR-dTOPO LHY with no stop codon. The pENTR-dTOPO LHY construct was recombined into pB7HFC by LR Clonase II to generate pB7HFC-LHY. To generate a pB7HFC-LHY line that was driven by the endogenous LHY promoter, we amplified 1807 bp from the 3′ UTR of the adjacent gene to the 5′ UTR of LHY using pB7HFC-LHY primers containing PmeI/SpeI sites (sequence provided in Supplemental Table 2). pB7HFC-LHY was digested with SpeI and PmeI, and the LHY-promoter fragment was replaced with the Cauliflower mosaic virus 35S promoter using InFusion recombination, generating pB7HFC-LHY:LHY. pENTR-CCA1 (Pruneda-Paz et al., 2009) was recombined into pB7HFC by LR Clonase II to generate pB7HFC-CCA1. To produce a CCA1 promoter-driven construct, we amplified 2623 bp from the end of the 3′ UTR of the adjacent gene to an internal BglII site using pB7HFC-CCA1 primers (sequence provided in Supplemental Table 2). pB7HFC-CCA1 was cut with PmeI and BglII and the amplified promoter fragment was recombined in using InFusion. All vectors were sequenced prior to transformation to verify sequence integrity. pB7HFC-LUX, pB7HFC-LHY:LHY, pB7HFC-CCA1:CCA1 were transformed by floral dip into lux-4 CAB2:LUC (Hazen et al., 2005), lhy-20 CCA1:LUC (Pruneda-Paz et al., 2009), or cca1-1 CCA1:LUC background, respectively. Transformants were selected for single insertion via Basta resistance and evaluated for rescuing circadian period mutant phenotypes. All lines were brought to homozygosity before use.

Measurements of Leaf Movement and Luciferase Activity

Five-day-old Arabidopsis seedlings grown on a 1/2 MS plate under the light/dark cycle were individually transferred to a 24-well plate by carefully excising an agar square (∼1 cm²) surrounding the seedling and transferring it onto the wall of each well, with the cotyledons perpendicular to the plate. In a growth chamber with continuous light, the plate was placed in vertical orientation with a white background to facilitate imaging. Plant images were taken every hour for 5 d using Optio-750Z digital camera (Pentax) with the interval shooting mode. Values for vertical coordinates of leaf edge and hypocotyl apex were obtained for each image using the pixel tracking function of Image J software (v1.48), and plotted by Microsoft Office Excel (2007).

The TOC1:LUC construct was previously described (Alabadí et al., 2001). 7 Arabidopsis seeds sterilized were sown on 1/2 MS medium + 1% sucrose in each well of 6-well dishes. Following stratification at 4°C for 2 d, the plants were grown in 12-h light/12-dark cycles at 22°C for 7 days and transferred to a growth chamber equipped with PIXIS 1024B CCD camera (Princeton Instruments) and LED lights (70 µmol/m2/s, wavelengths 400, 430, 450, 530, 630, and 660 at intensity 350; Heliospectra LED lights). Plant images were taken every 60 min after a 180 s delay for 4 min with 1 × 1 binning using µManager software (Edelstein et al., 2010, 2014). After 36 h (Zeitgeber time [ZT] = 12 h) the plants were removed from the chamber, sprayed with the drugs and returned to continue imaging. The images were processed using Metamorph imaging software (Molecular Devices).

Lipid Extraction, Profiling, and Oil Content Analysis

Lipids were extracted and analyzed by electrospray ionization-MS/MS-based method as described previously (Welti et al., 2002). Plant tissues were incubated with isopropanol containing 0.01% (w/v) butylated hydroxytoluene (BHT) at 75°C for 15 min to prevent lipoxidation and lipolysis. Polar lipids were extracted from the tissues with chloroform and water by agitating the mixture for 1 h. Lipids were further extracted three times with a 2:1 (v/v) mixture of chloroform and methanol containing 0.01% (w/v) BHT. All lipid extracts were combined and washed twice by mixing with 1 M KCl (water for the second wash) and a brief centrifugation. The lower organic phase was dried under a gentle stream of nitrogen gas and re-dissolved in chloroform. The resulting lipids were applied to electrospray ionization-MS/MS (API-4000) detection system with a mixture of internal lipid standards and solution B (95% [v/v] methanol, 14.3 mM ammonium acetate), and data were processed by Analyst software (v1.5.1). After the lipid extraction, remaining tissues were air-dried and measured for dry weight.

To analyze seed oil content, seeds were collected from Arabidopsis wild-type and mutant plants grown side-by-side under light cycles of 16-h light/8-h dark or 12-h light/12-h dark. Mature seeds (3 to 5 mg/sample) were incubated with 2 mL methanol containing 5% (v/v) H₂SO₄ and 0.01% (w/v) BHT in a glass tube with Teflon-lined screw cap at 90°C for 90 min for transmethylation. Fatty acid methyl esters (FAMEs) were extracted with hexane and water. After a brief centrifugation, the upper phase with FAMEs was applied to a gas chromatography detection system (Shimadzu GC-17A) with heptadecanoic acid (C17:0) as an internal standard for quantification. The gas chromatography system was supplied with a hydrogen flame ionization detector and a capillary column SUPELCOWAX-10 (30 m; 0.25 mm) with helium carrier at 20 mL/min. The oven temperature was maintained at 170°C for 1 min and then increased in steps to 210°C, raising the temperature by 3°C/min. FAMEs from TAG were identified by comparing their retention times with known standards.

Statistical Analyses

For analysis of circadian rhythm data, relative amplitude error and period length for TOC1 expression and leaf movement shown in Figures 8 and 9 were analyzed by fast Fourier transformed nonlinear least squares (Plautz et al., 1997) using the Biological Rhythms Analysis Software System 3.0 available at http://www.amillar.org. 6 to more than 20 biological replicates (individual seedlings) were tested and for each, the seedling with relative amplitude error below 0.5 was considered to be significantly rhythmic. The number of these samples is denoted as “N” in Table 2. Periods of the rhythmic samples were subjected to statistical analysis performed with GraphPad software t test calculator, in which case only period changes with P < 0.05 or 0.01 following Student’s t test paired with controls were considered significant and marked with asterisks in Table 2. For other quantitative data, one-way ANOVA, Student’s t test, or correlation analyses were used as specified in each figure.

Accession Numbers

Sequence data from this article can be found in The Arabidopsis Information Resource (http://www.arabidopsis.org) under the following accession numbers: LHY, At1g01060; CCA1, At2g46830; TOC1, At5g61380; ELF4, At2g40080; PRR9, At2g46790; LUX, At3g46640; PLDα1, At3g15730; PLDΔ, At4g35790; PAH1, At3g09560; PAH2, At5g42870; Ubiquitin10, At4g05320.

Supplemental Data

• Supplemental Figure 1. Amount of lipids detected from the immunoprecipitation.

• Supplemental Figure 2. PCR confirmation of arabidopsis mutant lines.

• Supplemental Figure 3. Expression of PA metabolic genes in the wild type versus LHY-OE.

• Supplemental Table 1. PA-binding transcription factors identified by mass spectrometry.

• Supplemental Table 2. Oligonucleotides and PCR primers used (5′→3′).

• Supplemental Data Set 1. Proteomic analysis of PA binding transcription factors.

• Supplemental Data Set 2. Original readout for SPR analysis.

• Supplemental Data Set 3. Individual values for diurnal lipid levels.

Dive Curated Terms

The following phenotypic, genotypic, and functional terms are of significance to the work described in this paper:

• GAPC1 Gramene: AT3G04120

• GAPC1 Araport: AT3G04120

 

• LHY Gramene: AT1G01060

• LHY Araport: AT1G01060

 

• CAB2 Gramene: AT1G29920

• CAB2 Araport: AT1G29920

 

• DGK1 Gramene: AT5G07920

• DGK1 Araport: AT5G07920

 

• LPAT2 Gramene: AT3G57650

• LPAT2 Araport: AT3G57650

 

• LPP2 Gramene: AT1G15080

• LPP2 Araport: AT1G15080

 

• proteins CHEBI: CHEBI:36080

 

• PAH1 Gramene: At3g09560

• PAH1 Araport: At3g09560

 

• PAH2 Gramene: At5g42870

• PAH2 Araport: At5g42870

 

• LPP CHEBI: CHEBI:138807

ACKNOWLEDGMENTS

We thank Dr. Nobutaka Mitsuda (National Institute of Advanced Industrial Science and Technology, Tsukuba, Japan) for generous donation of the Arabidopsis TF cDNA library and Dr. Peter Eastmond (University of Warwick, Wellesbourne, UK) for pah1 pah2 and complement seeds. This work was supported by grants from the National Science Foundation (IOS-0818740 and MCB-1412901 to X.W.) and the U.S. Department of Energy (DE-SC0001295 to X.W.). We thank Anne Helfer and Steve A. Kay for generously supplying LUX-HFC lux-4. M.L.S. acknowledges support from the NSF GRFP (DGE-1745038). D.A.N. acknowledges the support of NIH R01 GM067837 to Steve A. Kay for the generation of LUX-HFC.

AUTHOR CONTRIBUTIONS

S.-C.K. designed and performed all experiments, except for ones done by coauthors, obtained and verified DNA and plant materials, collected and analyzed numerical data, generated and modified visual images, and wrote and revised the article. D.A.N. aided in luciferase reporter assay, experimental design, generating reagents, and writing. M.L.S. generated Arabidopsis lines with FLAG-tagged clock proteins and edited the article, and J.L.-P.P. produced clock double mutants and LHY:LHY-expressing lines. X.W. proposed and supervised the study and edited the article.

Author notes

www.plantcell.org/cgi/doi/10.1105/tpc.18.00675

The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantcell.org) is: Xuemin Wang ().