The clock component OsLUX regulates rice heading through recruiting OsELF3-1 and OsELF4s to repress Hd1 and Ghd7

Graphical abstract

Highlights

• •OsLUX regulates rice heading by repressing Hd1 and Ghd7.
• •Defect OsLUX causes extremely late heading phenotype dependent on Hd1 and Ghd7 under both photoperiod conditions.
• •The cytoplasm-nuclear shuttling mechanism explains the effect of OsELF3-1 on OsLUX.
• •The complete OsEC (OsELF4s–OsELF3-1–OsLUX) complex is required to regulate heading via binding to the promoters of Hd1 and Ghd7.
• •The repressive strength of OsEC to regulate Hd1 and Ghd7 explained the different flowering effects of OsEC components.

Abstract

Introduction

Circadian clocks coordinate internal physiology and external environmental factors to regulate cereals flowering, which is critical for reproductive growth and optimal yield determination.

Objectives

In this study, we aimed to confirm the role of OsLUX in flowering time regulation in rice. Further research illustrates how the OsELF4s–OsELF3-1–OsLUX complex directly regulates flowering-related genes to mediate rice heading.

Methods

We identified a circadian gene OsLUX by the MutMap method. The transcription levels of flowering-related genes were evaluated in WT and oslux mutants. OsLUX forms OsEC (OsELF4s–OsELF3-1–OsLUX) complex were supported by yeast two-hybrid, pull down, BiFC, and luciferase complementation assays (LCA). The EMSA, Chip-qPCR, luciferase luminescence images, and relative LUC activity assays were performed to examine the targeted regulation of flowering genes by the OsEC (OsELF4s–OsELF3-1–OsLUX) complex.

Results

The circadian gene OsLUX encodes an MYB family transcription factor that functions as a vital circadian clock regulator and controls rice heading. Defect in OsLUX causes an extremely late heading phenotype under natural long-day and short-day conditions, and the function was further confirmed through genetic complementation, overexpression, and CRISPR/Cas9 knockout. OsLUX forms the OsEC (OsELF4s–OsELF3-1–OsLUX) complex by recruiting OsELF3-1 and OsELF4s, which were required to regulate rice heading. OsELF3-1 contributes to the translocation of OsLUX to the nucleus, and a compromised flowering phenotype results upon mutation of any component of the OsEC complex. The OsEC complex directly represses Hd1 and Ghd7 expression via binding to their promoter's LBS (LUX binding site) element.

Conclusion

Our findings show that the circadian gene OsLUX regulates rice heading by directly regulating rhythm oscillation and core flowering-time-related genes. We uncovered a mechanism by which the OsEC target suppresses the expression of Hd1 and Ghd7 directly to modulate photoperiodic flowering in rice. The OsEC (OsELF4s–OsELF3-1–OsLUX)–Hd1/Ghd7 regulatory module provides the genetic targets for crop improvement.

Introduction

Nearly 24 h of endogenous rhythm is generated to optimize internal processes that aim to cope with changes in the external environment and provide a fitness advantage for plants [1]. Currently, the most in-depth study about the plant circadian clock is mainly focused on the model plant Arabidopsis, particularly in photoperiodic flowering. Flowering time (heading date in crops) is critical for reproductive growth and optimal yield determination in cereals, which is controlled by a sophisticated genetic network including circadian oscillators that coordinate internal physiology and external environmental factors to determine flowering [2]. Thus, how clock genes regulate heading date and improve circadian clock regulation network in rice are topics of great interest.

Generally, the biological clock system consists of three main parts, the input pathway of environmental signal perception, the central oscillator for generating rhythm period, and the output pathway for regulating downstream biological processes [3]. The core oscillator of the plant circadian clock consists of a series of transcription-translation feedback loops (TTFLs) [4]. The EC (evening complex) is composed of the LUX aRRYTHMO (LUX), EARLY FLOWERING 3 (ELF3), and EARLY FLOWERING 4 (ELF4). It is a vital component of maintaining rhythmic oscillation in the evening feedback loops, which releases the suppression effect of pseudo-response regulators (PRRs) like PRR9, PRR7, and PRR5 on CIRCADIAN CLOCK ASSOCIATED 1 (CCA1) and LATE ELONGATED HYPOCOTYL (LHY) to make sure CCA1 and LHY are expressed in the morning [5], [6], [7]. Accordingly, CCA1 and LHY inhibited the expression of the EC complex in the morning, making its expression peak at dusk. Subsequently, it was repressed by TIMING OF CAB EXPRESSION 1 (TOC1) at night [5]. Among different plant species, the EC through LUX binding site (LBS) motif (GATT (A) CG) regulates multiple output pathways, including maintaining circadian rhythms and regulating plant growth and development [5], [6], [8], [9], [10]. For example, mutations in LUX, ELF3, and ELF4 caused an arrhythmic and early flowering phenotype in Arabidopsis [8], [11]. LUX orthologs' mutation confers photoperiod-insensitive early heading phenotype in barley and delays flowering in soybean [12], [13]. However, whether the OsEC interacts with the photoperiodic flowering pathway to regulate rice heading through multiple flowering output genes like EC in Arabidopsis is still unknown. Therefore, further research on OsEC-regulated genes could expand our understanding of OsEC function to regulate heading in rice.

Flowering time acts as a master output pathway of the circadian clock, and the different effects on heading are usually controlled by major circadian components. In rice, numerous flowering-related genes which show rhythmic expression is controlled by the circadian clock. OsGIGANTEA (OsGI), a critical clock gene, has a rhythmic expression pattern that might regulate circadian clock genes to produce orchestrated rhythms in global transcriptome expression in rice [14]. Heading date 1 (Hd1) is an ortholog of Arabidopsis CONSTANS (CO), the expression of Hd1 showed a similar diurnal rhythm pattern under both short-day conditions (SDs) and long-day conditions (LDs) and peaked during the night [15]. Hd1 works downstream of OsGI and promotes heading under SDs while delays heading under LDs by either elevating or repressing Hd3a expression [16], which is conservative in the Arabidopsis GI-CO-FT pathway. Grain number, plant height, and heading date 7 (Ghd7) encodes a CCT domain protein, which is a crucial floral repressor in rice under LDs [17], [18]. Ghd7 expression is also controlled by circadian rhythm, and the inducibility peak shifted from dawn under LDs to midnight under SDs. Early heading date 1 (Ehd1) is a critical floral inducer in rice, gate expression induced by blue light in the morning regardless of the day length. However, Ghd7 suppresses the expression of Ehd1 and Heading date 3a (Hd3a) the next morning under LDs. Ehd1 expression inducibility at dawn under both LDs and SDs, and Hd3a expression is acutely induced due to Ghd7 transcripts being too low to sustain its repressor activity against Ehd1 at the next dawn under SDs [19].

In addition, previous reports showed that circadian genes are involved in regulating heading in rice. Like EARLY FLOWERING 3 (ELF3) as a floral repressor in Arabidopsis, OsELF3-1/Ef7 promotes heading via repressing floral repressor Ghd7 under LDs [20], [21], [22]. Besides, OsELF3-1 participates in circadian rhythm regulation by suppressing OsGI, OsPRR95, OsPRR37, OsPRR73, and OsPRR1/OsTOC1 and promotes the expression of OsCCA1/OsLHY [21], [22]. The OsPRRs gene family, the core component of the circadian clock, plays a vital role in regulating photoperiodic flowering in rice, such as OsPRR37/Ghd7.1 delayed heading through negatively regulates the expression of Ehd1 and Hd3a under LDs [23]. OsPRR73 targeted modulation of floral gene Ehd1 and the circadian gene OsLHY to promote heading [24]. One recent research in rice shows that OsLHY sets critical day length for photoperiodic flowering dependent on the OsGI-Hd1 pathway [25]. Although several major circadian clock genes have been identified to regulate flowering in rice, how the circadian clock interacts with the photoperiodic flowering pathway to regulate rice heading remains to be elucidated.

Previously reported that OsELF3-1/Ef7 mediated flowering through repression of the Hd1 and Ghd7 [20]. However, it has yet to be shown via direct evidence of how OsELF3-1/Ef7 achieves to suppress Hd1 and Ghd7. In this study, we cloned the ortholog of Arabidopsis LUX using the MutMap method, which acts as a crucial circadian clock component. OsLUX has a positive role in regulating rice heading and might be responsible for the recruitment of OsELF3-1 and OsELF4s to form trimer complex (OsEC) to directly bind to the LBS in the Hd1 and Ghd7 promoter to repress their expression. The OsELF3-1 contributes to the translocation of OsLUX to the nucleus, the cytoplasm-nuclear shuttling mechanism explains the effect of OsELF3-1 on OsLUX, and OsELF4s promote the interaction between OsLUX and OsELF3-1. Mutation in the component of the OsEC complex by the CRISPR system brings about a compromised flowering phenotype with distinct effects, which is due to reducing the suppressive activity of OsEC to Hd1 and Ghd7. Thus, Hd1/Ghd7 repressive complex plays a primary role in suppressing heading. Genetic analyses indicate that Hd1 and Ghd7 act downstream of OsLUX. Collectively, our results provide the OsEC–Hd1/Ghd7 molecular evidence for involving photoperiodic flowering pathway to mediate heading in rice.

Materials and methods

Plant materials and growth conditions

The elh1, elh2, and elh3 mutants were screened by EMS mutagenesis of japonica cv Nipponbare (Nip) and various mutants, including oslux, oself3-1, oself4s, oslux-1 oself3-1, and oself3-1 oself4s were grown under natural long-day conditions (NLDs, Hangzhou, China), and then grown in natural short-day conditions (NSDs, Hainan, China). Nip and elh1 were also cultivated in growth chambers under controlled LDs (CLDs, 14 h light, 30 °C/10 h dark, 25 °C) or controlled SDs (CSDs, 10 h light, 30 °C/14 h dark, 25 °C) in growth chambers with a light intensity of 300 μmol m⁻² s⁻¹ and 70 % relative humidity.

Cloning of OsLUX

We performed a Mutmap method to clone the OsLUX gene according to the previous report [26]. The elh1, elh2, and elh3 mutants were backcrossed with Nip to generate the F₂ segregation population. Randomly selected 40 F₂ plants with extremely late heading for DNA extraction, the equal amount of DNA was mixed to sequence the whole genome, and then clean sequences were acquired for further analysis. The SNP index was calculated for each SNP site, and the high SNP index is probably the candidate mutation site.

Vector construction and transformation

A 5.59-kb Nip genomic DNA fragment containing the OsLUX coding region, 2.98-kb upstream region, and 1.89-kb downstream region was inserted into the pCAMBIA1305 vector at the EcoRI site and transformed into elh1 for genetic complementation. The CRISPR/Cas9 knockout vector construction was performed as described previously [27] to generate various mutants, including oslux, oself3-1, oself4s, oslux-1 oself3-1, oself3-1 oself4s, oslux-1 ghd7, and oslux-3 hd1. For overexpression, the coding sequence (CDS) of OsLUX was inserted into the pCAMBIA2300 vector at the SmaI site to generate the pActin::OsLUX construct and transformed into elh1. The 2.98-kb upstream fragment of OsLUX was amplified from Nip genomic DNA and cloned into the pCAMBIA1305 vector between EcoRI and NcoI sites to generate a GUS reporter gene construct transformed into Nip. The full-length CDSs of OsLUX, OsELF3-1, and OsELF4s were fused at the C-terminal with the pYBA 1132 vector at the NruI site, and OsLUX fused at the N-terminal in pYBA 1132 vector at the EcoRI site used for subcellular localization. OsLUX and OsELF4s were fused with mCherry at the EcoRI site to generate recombinant plasmids used for co-localization. Then, the expression vector was transformed in rice protoplast or Nicotiana benthamiana leaves.

RNA extraction and real-time quantitative RT-PCR (RT-qPCR)

Total RNA was extracted from leaves of 50-day-old seedlings under CLDs and 40-day-old seedlings under CSDs using RNAprep pure Plant Kit (Tiangen Biotech Co. ltd., Beijing, China) according to Kit's instructions. RNA reverse transcribed using a ReverTra Ace® qPCR RT Master Mix with gDNA Remover kit (Toyobo Co. ltd., Osaka, Japan) according to the manufacturer's protocol. RT-qPCR was conducted with SYBR premix Ex Taq Kit (Takara Bio, Inc., Kusatsu, Shiga, Japan) according to the operation instructions. The relative mRNA levels of the investigated genes were normalized to Ubiquitin (Os03g0234350) by a 2^–△△CT calculation method with two biological and three technical replicates.

Yeast two-hybrid (Y2H) assay

OsLUX, OsELF3-1, and OsELF4s CDS amplified from Nip cDNA were used as baits and preys. OsLUX was divided into two truncated fragments (amino acids 1–117 and 118–238), and the truncated fragments of OsELF3-1 (amino acids 1–348, 305–519, and 503–760) were obtained from previously described [28]. The bait and prey fragments were cloned into the pGBKT7 vector between EcoRI and BamHI sites and the pGADT7 vector at the EcoRI site (Clontech, Takara). OsLUX (amino acids 118–238), OsELF4-1, OsELF4-2, and OsELF4-3 were cloned into the pGBKT7 vector, respectively. OsLUX, OsELF3-1, OsELF3-1 (amino acids 1–348), OsELF3-1 (amino acids 305–519), OsELF3-1 (amino acids 503–760), OsELF4-1, OsELF4-2, and OsELF4-3 were cloned into the pGADT7 vector, respectively. The Y2HGold strains containing both BD and AD constructs were incubated on selective media DDO (SD-Leu/-Trp) or QDO (SD-Leu/-Trp/-His/-Ade) for 4 d at 28 °C.

In vitro pull-down assay

The CDS of OsLUX, OsELF4-1, OsELF4-2, OsELF4-3 were cloned into the pCold TF vector to generate the construct OsLUX-pCold TF, OsELF4-1-pCold TF, OsELF4-2-pCold TF, and OsELF4-3-pCold TF, respectively. The CDS of OsELF3-1 were cloned into pGEX-4 T-1 to generate GST-OsELF3-1, and Glutathione S-transferase (GST) empty vector was used as a negative control. Expression of GST OsLUX-pCold TF, OsELF4-1-pCold TF, OsELF4-2-pCold TF, and OsELF4-3-pCold TF in BL21 competent cells were induced with 0.1 mM IPTG at 14 °C for 16 h, and GST-OsELF3-1 induced with 0.5 M IPTG at 12 °C for 16 h, each combined solution added to 30 μL glutathione high capacity magnetic agarose beads (Sigma-Aldrich, St. Louis, MO, USA) followed by incubation at room temperature for one hour with rocking. The beads were washed four times with pull-down buffer (50 mM/L Tris-HCl, pH 7.5, 5 % glycerol, 1 mM/L EDTA, 1 mM/L DTT, 1 mM/L PMSF, 0.01 % Nonodet P-40, and 150 mM/L KCl), and the proteins were separated on SDS-PAGE gels and detected by anti-GST antibody (TransGen Biotech Co. ltd., Beijing, China; lot number HT601, 1:5000 dilution) and anti-His antibody (TransGen Biotech; lot number HT501-01, 1:5000 dilution), respectively.

BiFC assay

OsLUX and OsELF4s cDNAs were cloned into 35S-SPYNE and 35S-SPYCE vectors, respectively, named OsLUX-YN, OsELF4-1YC, OsELF4-2YC, and OsELF4-3YC. In addition, the complete CDS of OsELF3-1 was cloned into pCAMBIA1305-mcherry. The constructs were transformed into Agrobacterium strain GV3101 (pSoup) and then co-transfected into Nicotiana benthamiana leaves of 3-week-old. The mcherry was used as a control.

Cell fractionation assays

The cellular components were isolated according to the previous description [29]. The cDNA fragments of the OsLUX and OsELF3-1 genes were fused to the N terminus of HA and C terminus of GFP, respectively. Transfection of the HA and GFP expression vector into rice protoplasts. Collecting the cultured cells at 36 h later and then suspended in lysis buffer (20 mM Tris-HCl, PH7.5, 20 mM KCl, 2 mM MgCl₂, 25 % Glycerol, 250 mM sucrose, 50 mM DTT) at 4 °C, and centrifuged at 12000 rpm for 10 min, the resulting supernatant is total protein. The resultant supernatant was first centrifuged at 1500 g for 10 min, then centrifuged at 10,000 g for 10 min. The resultant supernatant was the soluble cytosol fraction, and the crude pellet was the nucleus fraction. The crude nucleus fraction was resuspended in 1 ml of washing buffer (20 mM Tris-HCl, PH 7.5, 25 mM MgCl₂, 25 % Glycerol, 0.2 % Triton X-100). It was centrifuged at 15,000 g for 10 min and repeated the operation several times. After the washing step was finished, the pellet was diluted with 500 μL resuspend buffer (20 mM Tris-HCl, PH 7.5, 10 mM MgCl₂, 250 mM sucrose, 0.5 % Triton X-100, 5 mM β-Mercaptoethanol) and transferred to another tube with the buffer (20 mM Tris-HCl, PH 7.5, 1.7 M sucrose, 10 mM MgCl₂, 0.5 % Triton X-100, 5 mM β-Mercaptoethanol), and then centrifuged at 16,000 g for 45 min, the final pellet was the nucleus protein and resuspended with lysis buffer.

Luciferase complementation assay (LCA)

The CDSs of the OsLUX and OsELF4s were cloned into the N terminus of luciferase (LUC) in the pCAMBIA-nLUC vector to generate OsLUX-nLUC, OsELF4-1-nLUC, OsELF4-2-nLUC, OsELF4-3-nLUC constructs, respectively. Similarly, OsLUX-cLUC and OsELF3-1-cLUC were developed by the pCAMBIA-cLUC vector, which fused to the C terminus of LUC. The Agrobacterium strain GV3101 (pSoup) carried with nLUC and cLUC recombinant plasmids were co-transfected into Nicotiana benthamiana leaves through equally mixed. After 48 h infiltration, Nicotiana benthamiana leaves were injected with 1 mmol/L d-luciferin potassium substrate for the qualitative detection of luciferase activity by a charge-coupled device (CCD) imaging system. Leaf discs were incubated with 200 μL of 1 mM μL⁻¹ d-luciferin potassium in a 96-well plate for quantitative detection of luciferase activity with a GloMax 96 microplate luminometer (Promega, Madison, WI, USA).

Yeast one-hybrid assay

The CDS of OsLUX and the promoter of OsPRRs were cloned into pB42AD and pLacZi reporter vectors, respectively. The EGY48 strains containing both pB42AD and pLacZi constructs were incubated on SD/-Ura/-Trp plates, and then grew on SD/-Ura/-Trp with 1 × BU salts, 1 % raffinose, 2 % galactose, and 80 mg/L X-Gal. The blue colonies indicate the interaction.

Electrophoretic mobility shift assay (EMSA)

The BL21 (DE3) strain (Tsingke Biotech Co. ltd., Beijing, China) was carried with a GST-OsLUX recombinant construct and expressed with 0.1 mM IPTG (isopropyl-1-thio-d-galactopyranoside) at 14 °C for 12 h, and then using Beaver Beads^TM GSH (Beaver Biosciences Inc, Suzhou, China; catalog no. 70601–100) to purify the GST-OsLUX fusion protein. The probes containing LBS motifs in Hd1 and Ghd7 promoter were synthesized and labeled with biotin at the 3′-end by the EMSA Probe Biotin Labeling Kit (Beyotime Institute of Biotechnology, Shanghai, China). The EMSA experiment was conducted using the Chemiluminescent EMSA Kit (Beyotime) according to the manual provided by the manufacturer. The labeled probes or unlabeled oligonucleotides were incubated with GST-OsLUX fusion protein in 10 μL mixtures at 23 °C for 30 min. The mixtures were separated with 6 % polyacrylamide gels and visualized using a chemiluminescence imaging system (Bio-Rad Laboratories, Inc, Segrate, Italy).

Chromatin immunoprecipitation-quantitative PCR (ChIP-qPCR) assay

The ChIP-qPCR assay was performed as described [30]. About 3 g leaf tissue of 35S::3 × flag:OsLUX seedlings with 30-day-old under NLDs was used for crosslinking fixation. Chromatin was fragmented to 200–700 bp by sonication. The antibody used for ChIP-grade antibody against FLAG (Sigma-Aldrich, St. Louis, MO, USA; catalog no. F1804). The input and precipitated DNA samples were used as a template for performing qPCR. The values were normalized to input samples, and IgG was used as a negative control.

Luciferase transient transcriptional activity assay

The CDS of OsLUX was inserted into the pGreenII 62-SK vector at the BamHI site, and the promoters of Hd1 and Ghd7 were cloned into the reporter vector pGreenII 0800-LUC at the HindIII site. The Agrobacterium strain GV3101 (pSoup) carried with effector and recombinant reporter plasmid were co-transformed into Nicotiana benthamiana cells. A Dual-Luciferase® Reporter Assay System (Promega) was used to measure the firefly and renilla LUC signals according to the manufacturer's instructions. The data was indicated as mean ± SD, and each assay was performed with three biological and three technical replicates. In addition, full-length OsLUX, OsELF3-1, and OsELF4s cDNA were inserted into pYBA 1132 to generate the GFP–OsLUX, GFP–OsELF3-1, and GFP–OsELF4s constructs, and then co-transfected into Nicotiana benthamiana leaves with Hd1–LUC or Ghd7–LUC to achieve luciferase luminescence images.

Results

elh1/oslux-1 mutant exhibits extremely late heading phenotype

To identify new rice genes regulating heading, we screened an extremely late heading mutant, elh1, from our rice mutant library generated by ethyl methanesulfonate-induced in Nipponbare (Nip). The mutant displayed 63.2 days later heading than Nip under NLDs at Hangzhou and 76.6 days later under NSDs at Hainan (Fig. 1A-B). In addition, we grew the elh1 mutants and Nip plants in controlled chambers under CLDs and CSDs. Under CSDs, the heading date of the elh1 mutants (214.7 ± 19.2 d) was delayed by 160.7 d compared with the Nip (54.0 ± 2.7 d). Under CLDs, Nip plants headed at 80.3 ± 5.0 d; however, the elh1 mutants exhibited no heading for more than 260 d (Fig. 1B). The extremely late heading phenotype of the elh1 mutants under both photoperiodic conditions suggests that this mutant significantly affected heading. The leaf emergence rates of elh1 had no difference with Nip (Fig. S1A-B), indicating that the delayed heading in the elh1 mutants was due to prolonged floral transition but not retarded growth under both CLDs and CSDs. Moreover, under Hangzhou NLDs, we found the elh1 mutants showed shorter panicles than the Nip, but it's the other way around under NSDs at Hainan. In addition, the primary branches of elh1 mutants were significantly increased while the grain size and thousand-grain weight are markedly smaller and lighter than Nip (Fig. S1, table. S1).

Fig. 1.

The role of OsLUX in rice heading. (A) Phenotypes of the mature period with Nip and elh1 under NLD conditions in Hangzhou. Bar = 20 cm. (B) Heading data of Nip and elh1 under NLD, NSD, CLD, and CSD conditions. The asterisk indicates statistically significant differences with P < 0.001. (C) The identification of oslux alleles. The Top section indicated distributions of SNP indexes along chromosomes in elh1. OsLUX gene structure and allelic mutation of oslux-1, oslux-2, and oslux-3 are shown in the bottom section; red characters indicate the mutant site. (D) Heading phenotypes of Nip and the elh1-complementation lines under NLDs, Scale bar = 20 cm. (E) The phenotype of Nip and two independent knockout lines under NLDs, Scale bar = 20 cm. (F) mutation sites of the CRISPR-edited sgRNA target sites of OsLUX, red underline indicate PAM sequence. (G) The heading date investigation of complementation lines and knockout lines in both NLDs and NSDs. The asterisk indicates statistically significant differences with P < 0.001. (H, I) Diurnal expression patterns of OsLUX from LD to LL and LD to DD, respectively. (J, K) Diurnal expression patterns of OsLUX from SD to LL and SD to DD, respectively. The blue and white boxes indicate dark and light periods, respectively. ZT, Zeitgeber time, ZT = 0 represents the start-time of lights on. The error bars represent standard deviations.

OsLUX encodes a transcription factor that belongs to the MYB family

We then performed cloning of the elh1 mutant gene using the MutMap method. The elh1 mutant first back-crossed with the Nip, and all F₁ progeny displayed normal heading. The F₂ population exhibited 3:1 segregation of normal heading and extremely late heading (176:55, χ² = 0.17, P = 0.68), indicating that elh1 is a single recessive mutation. The MutMap analyses indicated a candidate region at the end of chromosome 1, including three candidate SNPs: one of which is in the intergenic regions, one in the intron of LOC_Os01g72650, and one (Chr1: 42,874,923) in the exon of LOC_Os01g74020 which resulted in a stop-gain mutation (CAG to TAG) at position 376 (Fig. 1C). Co-segregation analysis showed that all 55 extremely late heading plants carried this mutation, whereas the 176 normal heading plants showed a 2:1 ratio of heterozygous and homozygous Nip genotypes (117:59). LOC_Os01g74020 putatively encodes an MYB family circadian clock gene OsLUX (also known as OsPCL1), an ortholog of Arabidopsis LUX [31], [32], and OsLUX is highly orthologous among both monocot and dicot species. OsLUX is a single-copy gene closely related to maize ZmLUX (Fig. S2A) and encodes a transcription factor with a conserved MYB DNA binding domain in the C terminus (Fig. S2B).

To verify the function of LOC_Os01g74020, genetic complementation assay was carried out with a 5.59-kb genomic fragment from Nip to introduce into the elh1/oslux-1 mutant, and positive transgenic plants were able to fully rescue the extremely late heading phenotypes under both NLDs and NSDs (Fig. 1D, 1G). When a pUbi::OsLUX construct was transferred to the elh1 mutant, the transgenic plants ectopically expressed OsLUX showed a normal heading phenotype (Fig. S3A-D). The oslux mutants generated by CRISPR/Cas9 exhibited a similar phenotype to the elh1 mutant under both NLDs and NSDs (Fig. 1E-G), suggesting that the mutation in LOC_Os01g74020 is responsible for the extremely late heading phenotype of elh1. In addition, we evaluated two other mutant alleles of LOC_Os01g74020 (elh2/oslux-2 and elh3/oslux-3, Fig. 1C). The elh2/oslux-2 also causes a stop-gain mutation (TGG to TGA) at position 369, which had an exceptionally late heading phenotype similar to elh1. In contrast, the elh3/oslux-3 causes amino acid substitution at position 469 (CTC to TTC), which resulted in Leu to Phe, and exhibited 34.4 days and 36.9 days later heading compared to Nip under NLDs and NSDs, respectively (Fig. S4A-B). Taken together, these results confirmed the positive role of OsLUX in rice flowering time regulation.

OsLUX has a constitutive and rhythmic expression pattern

To understand the spatio-temporal expression pattern of OsLUX, we first performed RT-qPCR to examine the mRNA levels of OsLUX under NLDs from root, stem, leaf blade, and leaf sheath at the vegetative stage and the different lengths of young panicles at the reproductive stage. The results revealed that OsLUX was constitutively expressed among different tissues preferentially accumulated in leaf blades, which is in accord with a β-glucuronidase (GUS) activity assays shown by expressing the pOsLUX::GUS (Fig. S5A-B). Next, we examined the diurnal expression pattern. The results showed that OsLUX transcripts were increased during the light period and reached a peak at ZT = 12, subsequently reduced to the lowest level at dawn under LDs and SDs. However, the rhythm and amplitude of OsLUX transcripts were slightly decreased when transferred to continuous light (LL) conditions (Fig. 1H, 1 J). Still, they declined sharply when shifted to continuous dark (DD) conditions (Fig. 1I, 1 K).

Moreover, the expression waveform or amplitude of OsLUX was slightly higher in elh1 than Nip, probably due to the negative feedback regulation of LUX, as shown in Arabidopsis [6], [9]. In fact, we found the native OsLUX transcript levels were strongly reduced in 35S::3 × flag:OsLUX (Fig. S5C), which showed that OsLUX participates in a feedback loop. Together, these results revealed that OsLUX acts as a circadian gene necessary for photoperiodic response to regulate heading in rice.

OsLUX is involved in maintaining the circadian rhythm and regulating heading in rice

The flowering time/heading date is strictly regulated by the circadian rhythm clock. Thus, we examined the rhythm expression of the clock genes in rice. The results showed that the expression levels of OsPRR37, OsPRR73, OsPRR59, OsPRR95, and OsPRR1 were up-regulation in elh1 and resulted in a higher amplitude of the rhythmical pattern compared to Nip under normal photoperiod. Still, in LL conditions, the expression patterns of these genes showed an arrhythmic oscillation (Fig. S6A-E, 6G-K). In contrast, the transcriptional levels and amplitude of OsCCA1 were significantly lower than Nip under regular photoperiod. When transferred to LL conditions, the amplitude of OsCCA1 in Nip and elh1 decreased sharply (Fig. S6F, 6L). To explore whether OsLUX bind to the promoter of these circadian clock gene, we performed a yeast one-hybrid assay. We found the promoter of clock genes with potential LBS motifs except for OsCCA1/OsLHY. The results showed that OsLUX activated the LacZ expression of OsPRR1, OsPRR59, and OsPRR95 but not OsPRR37 and OsPRR73 (Fig. S7D), which indicated that OsLUX might directly bind to the promoter of OsPRR1, OsPRR59, and OsPRR95 to regulate their expression. Collectively, these results showed that OsLUX is involved in maintaining the circadian clock of rice.

To evaluate how OsLUX regulates the rice flowering pathway, the expression levels of flowering-time-related genes were also monitored in Nip and elh1 mutants via RT-qPCR under LDs and SDs. We first examined two floral integrators, Hd1 and Ehd1. Hd1 transcripts were up-regulated in elh1 mutants compared with Nip (Fig. 2A-B). However, Ehd1 transcripts were almost undetectable in elh1 mutants (Fig. 2C-D), indicating that OsLUX represses Hd1 and promotes Ehd1 expression. Subsequently, we examined the expression levels of Hd3a and OsMADS14, which lie downstream of Hd1 and Ehd1. These genes showed similar expression patterns to Ehd1 (Fig. 2E-H), indicating that OsLUX plays a vital role in flowering initiation. Finally, we investigated the transcripts of OsGI and Ghd7, which are upstream regulators of Hd1 and Ehd1, respectively. Notably, their transcripts significantly increased in elh1 mutants (Fig. 2I-L). Our results were similar to a previous report that the overexpression of OsGI caused extremely late heading phenotype through up-regulate Hd1 transcription levels under both LDs and SDs [33]. These results suggest that OsLUX promotes heading by suppressing OsGI, Hd1, and Ghd7.

Fig. 2.

Diurnal expression of flowering-time-related genes under LDs (14 h light, 10 h darkness) and SDs (10 h light, 14 h darkness) in Nip and elh1. The blue and white filled boxes indicate dark and light periods, respectively. The samples were collected every 4 h under both LDs and SDs. The relative expression levels of Hd1 (A, B), Ehd1 (C, D), Hd3a (E, F), OsMADS14 (G, H), OsGI (I, J), and Ghd7 (K, L) normalized with rice Ubiquitin under LDs and SDs, respectively. The Zeitgeber time (ZT). ZT = 8 is set to the time of lights on.

OsLUX forms a heterotrimer complex through the recruitment of OsELF3-1 and OsELF4s

To further investigate the molecular mechanism underlying OsLUX regulates heading, we explored whether rice forms an EC complex similarly to Arabidopsis. In rice, there are three orthologs of AtELF4: OsELF4-1 (LOC_Os11g40610), OsELF4-2 (LOC_Os03g29680), and OsELF4-3 (LOC_Os08g27860), and two orthologs of AtELF3: OsELF3-1 (LOC_Os06g05060) and OsELF3-2 (LOC_Os01g38530). OsELF3-1 is involved in circadian rhythm regulation and promotes heading [21], [28], while OsELF3-2′s predominant role is immunity regulation [34]. Consequently, we focused on whether OsELF3-1 mediates flowering through the rice EC complex.

We conducted yeast two-hybrid (Y2H) assays to test the physical interactions between OsLUX with OsELF3-1 and OsELF4s. Due to the autoactivations activity of OsLUX, truncated fragments of OsLUX (amino acids 118–238) with no transcriptional activation activity (Fig. S7). Through Y2H assay, we found that the C-terminal domain (amino acids 503–760) and the middle region (amino acids 305–519) of OsELF3-1 could mediate a physical interaction with OsLUX and OsELF4s, respectively. However, OsLUX and OsELF4s failed to interact with each other (Fig. 3A). The interaction was also supported by pull-down (Fig. 3B) and luciferase complementation assays (LCA; Fig. S8). We performed another LCA and found that the LUC signal can be generated between OsLUX-Cluc with OsELF4s-Nluc due to the existence of mCherry-OsELF3-1 rather than mCherry alone (Fig. 3C-E). Collectively, these results confirmed that OsELF3-1 acts as a bridge to allow the formation of the ternary complex (OsELF4s-OsELF3-1-OsLUX).

Fig. 3.

OsELF3-1 interacts with OsLUX and OsELF4s in vivo and in vitro. (A) Yeast two-hybrid assay of interactions between OsELF3-1 with OsLUX and OsELF4s. OsELF3-1 and its truncated fragments as prey with bait constructs, including OsELF4s and OsLUX truncated fragments, the bait and prey constructs were cotransformed into Y2H Gold yeast cells and then grown on DDO medium (SD–Leu/–Trp) and QDO medium (SD–Leu/–Trp/-His/-Ade). (B) A pull-down assay demonstrated that OsELF3-1 directly interacts with OsLUX and OsELF4s. pCold-TF-OsLUX and pCold-TF-OsELF4s were pulled down by GST-OsELF3-1 immobilized on glutathione Magnetic Agarose beads and analyzed by immunoblotting (IB) with anti-His antibody. (C) The luciferase complementation assays (LCA) validation of OsELF3-1 is required for ternary complex formation with OsLUX and OsELF4s. mCherry was used as the negative control.

OsELF3-1 contributes to the translocation of OsLUX to the nucleus

The nuclear localization of the OsEC complex is vital for the transcriptional regulation of target genes. To investigate the subcellular localization of OsLUX, a transient expression assay was performed using tobacco cells, showing that the p35S::OsLUX:GFP fusion protein was located in both the cytosol and the nucleus, which the same results were confirmed with p35S::GFP:OsLUX transient expression in tobacco cells and rice protoplasts (Fig. 4A; Fig. S9). Consistent with previous reports [21], [35], we also conducted the subcellular localization of OsELF3-1 and OsELF4s and found that they only localized to the nucleus (Fig. S9). To exam whether OsELF3-1 affects the translocation of OsLUX to the nucleus, we transiently co-expressed p35S::OsLUX:mCherry and p35S::GFP:OsELF3-1 in rice protoplasts. The fusion signal was detected only in the nucleus but not in the cytoplasm, whereas, OsLUX-GFP co-expressed with mCherry were co-localized to both the cytoplasm and nucleus (Fig. 4A). This observation shows that OsELF3-1 alters the subcellular localization of OsLUX and promotes OsLUX translocation into the nucleus. The cell fractionation and immunoblotting assays in rice protoplasts also confirmed that OsLUX was detected in the cytosol and nucleus fraction (Fig. 4C), and the OsELF3-1 was present in the nucleus fraction (Fig. 4D). We co-expressed OsLUX and OsELF3-1 in rice protoplasts and found both OsLUX and OsELF3-1 had accumulated in the nucleus fraction (Fig. 4E).

Fig. 4.

The co-localization and BiFC of OsELF3-1 between OsLUX and OsELF4s. (A) Analysis of the subcellular localization of OsLUX-GFP using rice protoplasts and Nicotiana benthamiana cells transient expression, and co-localization of OsELF3-1 between OsLUX and OsELF4s using rice protoplast. Ghd7-mCherry is used as a nuclear marker. (B) OsELF3-1 interacts with OsLUX and OsELF4s confirmed by bimolecular fluorescence complementation (BiFC) assay, OsLUX-YN and OsELF4s-YC were co-expressed with mCherry or mCherry-OsELF3-1 in tobacco epidermal cells. mCherry as a control, Bar = 10 μm. (C-E) The cell fractionation and immunoblotting assays in rice protoplasts. (C) OsLUX, (D) OsOsELF3-1, (E) both OsLUX and OsOsELF3-1 in cell fractions extracted from rice protoplasts. T, total protein, C, cytosol protein, N, nucleus fraction. The UGPase (cytoplasm marker) and H3 (nucleus marker) were used in immunoblotting.

Bimolecular fluorescence complementation (BiFC) assay was conducted to detect the subcellular location of the OsLUX–OsELF3-1–OsELF4s interaction, and the fluorescence signals were observed only in the nucleus when expressing OsLUX–YFP^N/OsELF4s–YFP^C due to the existence of mCherry–OsELF3-1 compared to mCherry alone (Fig. 4B). OsLUX failed to interact with OsELF4s, which triggered us to consider the effects of OsELF4s on OsLUX–OsELF3-1 interaction. In Arabidopsis, ELF4 promotes the nuclear localization of ELF3 [36]. Our results found that OsELF3-1 contributes to the translocation of OsLUX to the nucleus (Fig. 4A, 4C-E), so we want to know if OsELF4s might promote the interaction between OsLUX and OsELF3-1. We used LCA assays to test this hypothesis; the relative luciferase activity observed for the OsELF3-1–OsLUX interaction was significantly increased when OsELF4s were co-expressed with OsELF3-1 and OsLUX (Fig. S10). Taking the results from co-localization, BiFC, and cell fractionation immunoblotting assays together, we conclude that OsELF3-1 contributes to the translocation of OsLUX to the nucleus and the localization of the OsEC complex in the nucleus, and complete OsEC activity is necessary for regulating output genes.

Defect in OsEC causes compromised heading

To verify the genetic relationship between the OsLUX–OsELF3-1–OsELF4s trimer components, CRISPR/Cas9 was used to generate null mutants for each of the three genes in the Nip background, and oslux-1 oself3-1 and oself3-1 oself4s double mutants were also made. Consistent with previous reports [20], the mutation in OsELF3-1 resulted in 22.1-d and 18.2-d later heading than Nip under NLDs and NSDs, respectively (Fig. 5A-B). Under NSDs at Hainan, we obtained the oslux-1 oself3-1 double mutant through screening F₂ plants from the hybridized elh1 and oself3-1. The oslux-1 oself3-1 double mutant resembled the elh1 mutant, exhibiting an extremely late heading phenotype. (Fig. 5G-H;). As for the elh1 mutants, a recent study reported that oself4-1 delayed heading in Beijing NLDs [35], while in our study, the mutation in OsELF4-2 (but not OsELF4-1 and OsELF4-3) resulted in late heading in Hangzhou NLDs (Fig. 5C-D; Fig. S11A-D). Under Hainan NSDs, the elf4-1 has no difference in heading, but the elf4-2 and elf4-3 mutants flowered slightly later than Nip (Fig. 5D; Fig. S11D). The late heading phenotype of oself3-1 oself4-2 plants resembles the oself3-1 single mutant under NSDs (Fig. 5I-J). Our results suggest that OsELF4s, which have a minor effect on rice heading, may function together to form a ternary complex with OsELF3-1 and OsLUX to regulate photoperiodic flowering. Like EC mutants display the early flowering in Arabidopsis, a compromised flowering phenotype results upon mutation of any component of the OsEC complex.

Fig. 5.

The phenotype of the oself3-1, oself4-2, and oslux-1 oself3-1, oself3-1 oself4-2 double mutants under different photoperiod conditions. (A and B) The phenotype (A) and days to heading (B) of Nip and oself3-1 lines. (C and D) The phenotype (C) and days to heading (D) of Nip and oself4-2 lines. The photos in A and C were taken under NLDs in Hangzhou, and the data from B and D were obtained under both NLDs and NSDs. (E and F) The mutant site of CRISPR-edited oself3-1 (E) and oself4-2 (F). (G and H) The phenotype (G) and days to heading (H) of Nip, oself3-1, oslux-1, and oslux-1 oself3-1 under NSDs in Hainan. (I and J) The phenotype (I) and days to heading (J) of Nip, oself3-1, oself4-2, and oself3-1 oself4-2 under NSDs in Hainan. (K and L) The genotype of mutants at OsELF3-1 and OsLUX sites (K) and OsELF3-1 and OsELF4-2 sites (L). The blue characters indicate the PAM sequence, and the tangerine character indicates the mutant site. All statistical data represent mean ± SD. *** indicates statistically significant differences with P < 0.001 from student’s t-test, * indicates P < 0.05. Bar = 20 cm.

OsEC complex binds to the promoters of Hd1 and Ghd7 and suppresses their expression

Hd1 and Ghd7 have been reported as crucial regulators of photoperiodic flowering in rice. The expression levels of Hd1 and Ghd7 were up-regulated, and similar expression patterns under both LDs and SDs in elh1 mutants (Fig. 2A-B, K-L). In Arabidopsis, LUX binds to the LBS motifs in the promoters of the output genes [9], [37]. We found putative LBS motifs in the promoter of Hd1 and Ghd7 via using the online tool PlantCARE (Plant cis-Acting Regulatory Elements, https://bioinformatics.psb.ugent.be/webtools/plantcare/html/) [38], which implied that Hd1 and Ghd7 might be the direct target genes of OsLUX. To validate this hypothesis, Chip-qPCR analysis was performed using p35S::FLAG:OsLUX expressing seedlings with anti-FLAG antibodies under NLDs. Four primer pairs were designed within 2-kb upstream of the transcription initiation site, which corresponds to four putative LBS motifs (S1, S2, S3, and S4, Fig. 6A), and significant enrichment of OsLUX on the S4 region in vivo was found instead of S1 S2 and S3 (Fig. 6B). We also conducted an electrophoresis mobility shift assay (EMSA) further to confirm this result, glutathione S-transferase (GST)-tagged OsLUX fusion protein bound to the S4 site in the Hd1 promoter with a retarded band, when fusion protein incubated together with the unlabeled oligonucleotide and the shifted band was substantially weakened with unlabeled in a dosage-dependent manner (Fig. 6D), indicating that OsLUX bind to the S4 site in particular. Moreover, a transcriptional activity assay was carried out further to confirm the suppression effect of OsLUX to Hd1. The p35S::OsLUX expression construct was co-transformed with pHd1::LUC into Nicotiana benthamiana leaves, and the results showed the relative LUC activity in p35S::OsLUX was significantly lower than the negative control (Fig. 6C, F). Similarly, OsLUX also directly binds to the LBS in the Ghd7 promoter to suppress its expression (Fig. 6A, E, G). Genetic studies further corroborated the biological function of OsLUX in Hd1 and Ghd7 regulation. We obtained the oslux-3 hd1 double mutant through generating a null mutation of Hd1 by CRISPR/Cas9 (Fig. 7E) against the background of elh3, and the oslux-1 ghd7 double mutants via screening F₂ plants from the F₁ (elh1 × ghd7, Fig. 7F). Knock out Hd1 and Ghd7 neutralized the repression of oslux on rice heading, which the oslux-3 hd1 and oslux-1 ghd7 double mutants restored the extremely late heading phenotype of the elh3 and elh1, respectively (Fig. 7A-B, 7C-D). The genetic data suggest that Hd1 and Ghd7 act downstream of OsLUX to regulate rice flowering.

Fig. 6.

The OsEC complex suppresses Hd1 and Ghd7 expression via binding to the LBS element in their promoter. (A) Diagram of the Hd1 and Ghd7 promoter region (2-kb upstream) shows LBS (S1-S6) locations and fragments used for ChIP-qPCR. (B) ChIP-qPCR analysis of Hd1 and Ghd7 promoter fragments with 35S::FLAG::OsLUX plants, ChIP was performed with antibodies specific for FLAG and IgG as a control. (C) Diagram of the various constructs used in the transcriptional activity assay. (D and E) EMSA showing the OsLUX protein bound directly to Hd1 (D) and Ghd7 (E) promoter, unlabeled probes (five-, Ten-, thirty- and fiftyfold) or mutated probe (thirtyfold) were used in competition assays, the top arrow indicates the shifted band, and bottom arrow indicates the free probe. (F and G) The relative LUC activity values indicate the transcriptional activity. LUC, firefly LUC protein, and REN, Renilla LUC protein, as a control). The error bars shown mean ± SD, * P < 0.05, **P < 0.01.

Fig. 7.

The phenotype and heading data of the oslux-3 hd1 and oslux-1 ghd7 under NSDs in Hainan. (A and B) The phenotype (A) and days to heading (B) of Nip, hd1, oslux-3, and oslux-3 hd1 under NSDs. (C and D) The phenotype (C) and days to heading (D) of Nip, oslux-1, and oslux-1 under NSDs. (E and F) The genotype of mutants at Hd1 and OsLUX sites (E) and Ghd7 and OsLUX sites (F). The blue characters indicate the PAM sequence, and the tangerine characters indicate the mutant site. All statistical data represent mean ± SD. *** indicates P < 0.001. Bar = 20 cm.

To determine the effects of the OsEC complex on transcriptional repression of Hd1 and Ghd7, we performed a LUC transient transcriptional activity assay. When a p35S::OsLUX expression construct was co-transformed into Nicotiana benthamiana leaves with pHd1::LUC, the luciferase luminescence was significantly suppressed compared with the negative control (Fig. S12A-C), indicating that OsLUX acts as a transcription repressor to bind to suppress Hd1 activity. When a p35S::OsELF3-1 construct was added, the luciferase luminescence was reduced (Fig. S12A-C), showing that the OsELF3-1-OsLUX dimer complex could enhance the suppressive activities of OsLUX to Hd1. Moreover, when the p35S::OsELF4s were further added, we found that the luciferase luminescence was further reduced (Fig. S12A-C), suggesting that the OsELF4s–OsELF3-1–OsLUX complex has the most potent suppression effect on Hd1. There is a parallel case that the OsEC complex has a similar suppression effect on Ghd7 (Fig. S12D-F). Taken together, these results indicate that the OsEC complex directly binds to Hd1 and Ghd7 promoters and suppresses their expression.

Discussion

The circadian clock regulates numerous growth and development processes throughout the whole growth cycle of plants. Among the circadian clock-regulated events, one of the most characterized is the internal circadian rhythms, and the external environment functionally works coordinately to regulate photoperiodic flowering. As a core clock regulator, LUX has been reported to play multiple roles in the circadian clock and output pathway that regulates plant growth, photoperiodic flowering, freezing tolerance, and defense response [9]. This study revealed that OsLUX, which encodes an MYB-like DNA-binding transcription factor, acts as a positive regulator of rice heading through modulating the expression of flowering-time-related genes.

OsLUX plays a critical regulatory role in rice heading

In Arabidopsis, LUX functions as a flowering repressor, and the mutation in LUX exhibits an arrhythmic and early flowering phenotype [6], [11], [15], [39]. In this study, we isolated three mutant alleles of OsLUX (elh1, elh2, and elh3), both elh1 and elh2 containing a stop-gain mutation showed late heading for more than two months under both NLDs and NSDs. The elh3 mutant possesses amino acid substitution in the MYB DNA-binding domain, which delayed heading for almost one month compared to Nip under both conditions (Fig. 1A-C; Fig. S4). So, based on our findings, we conclude that the OsLUX is indispensable for heading in rice. First, OsLUX affects the expression of multiple central clock genes, including OsPRR37, OsPRR73, OsPRR1, OsPRR59, OsPRR95, and OsLHY/OsCCA1, which are expressed with peaks at different times of day (Fig. S6). Consistent with LUX repressing the expressions of Arabidopsis orthologs of these rice clock genes [6,], these OsLUX-regulated clock genes control flowering through the photoperiodic output pathway, such as OsPRR37 and OsPRR73 negatively regulate the expression of Ehd1 to modulate rice heading under LDs [23], [24], OsLHY/OsCCA1 possesses dual flowering effect to depend on OsGI–Hd1 pathway under both LDs and SDs [25]. In addition, the highly late heading phenotypes of oslux were similar to overexpression of OsGI, which up-regulates Hd1 during the light period and down-regulates Hd3a under both SDs and LDs [33].

Unlike in Arabidopsis, barley (Hordeum vulgare), and Medicago truncatula, the mutation in LUX homologs causes the photoperiod-insensitive early flowering by elevating the expression of FT [39]. In contrast, oslux caused an extremely late heading phenotype under both SDs and LDs. This situation is likely to different molecular mechanisms regulating FT expression. In Arabidopsis, the CO-FT pathway is the rhythmic output of the biological clock to regulate flowering. MtLUX mutation in Medicago truncatula caused early flowering due to the up-regulation of the MtFTa1 but not in a CO-like dependent manner [40].

In contrast to Medicago truncatula, we found that OsLUX regulates the diurnal expression of Hd1 (a CO homolog) in rice. Although the situation is similar to Arabidopsis CO, Hd1 activates Hd3a (an FT homolog) expression in SDs but suppresses it in LDs [41], indicating the conservative function of LUX in flowering-time control but a different regulatory mechanism in rice. In our study, OsLUX promotes heading by suppressing the expression of OsGI and Hd1 under the whole photoperiod instead of the light period only under both SDs and LDs. In fact, the OsLUX-OsGI-Hd1 regulatory module is not enough to thoroughly explain the extreme late-heading phenotype due to the elevated and undetected transcripts of Ghd7 and Ehd1 in elh1, respectively (Fig. 2C-D, 2 K-L), these results indicate that OsLUX modulates rice heading mainly through suppressing the expression of Hd1 and Ghd7. Therefore, we uncover the underlying role of OsLUX in the circadian clock and its positive effect on heading in rice.

OsLUX recruits OsELF3-1 and OsELF4s to directly suppress Hd1 and Ghd7 expressions via binding to the LBS in their promoters

Among the EC components, merely LUX possesses a direct DNA binding activity and is responsible for numerous effects such as control flowering, hypocotyl elongation, and defense response through directly binding to the LBS elements alone or with other interaction factors [9], [42]. One previous ChIP-seq data reported that the G-box elements are highly enriched at EC binding peaks, indicating that multiple G-box binding transcription factors may co-binding with the EC to regulate the output pathway [37]. In this study, our data showed that OsELF3-1 acts as a bridge to the ternary complex formation (OsELF4s–OsELF3-1–OsLUX) supported by Y2H, pull-down, BiFC, and LCA in vitro and in vivo (Fig. 3). In Arabidopsis, loss of function in any EC components (lux, elf3, and elf4) results in a similar early flowering phenotype [10]. However, the mutation in the component of OsEC (OsLUX, OsELF3-1, and OsELF4s) complex in rice showed the various extent of late heading phenotype, particularly in oslux, exhibited an extremely late-flowering phenotype, indicating the conservative function but a distinct molecular mechanism for OsEC to regulate rice heading. In Arabidopsis, at least 800 LBSs were found in the genome [9], which supports the importance of LUX to maintain clock function and regulate multiple output pathways via direct control of multiple downstream target genes. Although one recent study has reported that the OsEC1 (OsELF4a–OsELF3-1–OsLUX) complex was found to coordinately regulate salt tolerance and heading in rice through targeted regulation of OsGI [35]. However, it is unknown whether OsLUX directly targeted the regulation of multiple genes located downstream of OsGI in the photoperiodic flowering pathway like in Arabidopsis. Thus, further research on OsEC-regulated genes could expand our understanding of OsEC's function in regulating heading in rice.

Interestingly, our study found OsLUX directly and explicitly binds to the promoters of Hd1 and Ghd7, supported by EMSA and Chip-qPCR assay (Fig. 6A-E). The LCA assays suggest that the OsLUX suppresses the target genes alone or coordinates with OsELF3-1 and OsELF4s (Fig. 6F-G, Fig. S12A-F). Our data support the pivotal role of OsLUX in mediating heading function through direct regulation of expression of key flowering genes in output pathways.

The complete OsEC complex is required for control rice heading

Hd1 and Ghd7 are the essential genes of two flowering signal pathways with partial crosstalk [43]. Previously, Saito et al. (2012) reported that OsELF3-1/Ef7 modulates rice heading though negatively affecting Hd1 and Ghd7, while the molecular mechanisms are not fully domesticated. Based on our findings and previous research, in this study, the OsELF4s–OsELF3-1–OsLUX–Hd1/Ghd7 regulatory model was proposed to illustrate the flowering mechanism of the OsEC complex (Fig. 8). The OsELF3-1–OsLUX complexes were co-located in the nucleus, whereas the OsLUX showed a dual cytosolic–nuclear localization (Fig. 4A). OsLUX is imported into the nucleus depending on interaction with OsELF3-1, discovered only in the nucleus (Fig. 4A-B). These results suggest that OsELF3-1 retains bound OsLUX in the nucleus and helps OsLUX enter the nucleus to improve its repression functions. Besides, we found that OsELF4s could enhance the interaction between OsLUX and OsELF3-1 (Fig. S10), consistent with the role of ELF4 in increasing nuclear localization of ELF3 in Arabidopsis [36]. Collectively, OsELF4s and OsELF3-1 mediate photoperiodic flowering pathways by increasing OsLUX nuclear accumulation to positively affect its function.

Fig. 8.

A working model shows how the OsELF4s-OsELF3-1-OsLUX complex functions to regulate rice heading. In WT, the OsELF3-1 contributes to the translocation of OsLUX to the nucleus. The OsELF4s interact with the heterodimer OsELF3-1-OsLUX to form a trimer complex to suppress the transcription of the vital flowering-time-related genes Hd1 and Ghd7 through directly binding to the “LBS” elements of their promoter, hence regulating heading in rice. In oslux, defect in OsLUX causes the suppression function of the OsEC complex to be entirely forfeit, which enhances the repression effect of Hd1-Ghd7 on Ehd1 and Hd3a under both LDs and SDs, thus, resulting in a late heading phenotype.

Given the dual functional regulation of Hd1 under distinct photoperiod conditions, we wondered how the OsEC complex mediates rice heading through targeted repression of the expression of Hd1 and Ghd7 under both LDs and SDs. Recently reported that Ghd7 forms a dimer complex with Hd1 to suppress the expression of Ehd1 and Hd3a during light periods to delay heading under LDs [43]. Hd1 promotes heading under SDs due to the too low abundance of Ghd7 to generate Hd1-Ghd7 repressive complexes, and a similar pattern was observed in ghd7 and dth8 under LDs [17], [43], [44], [45], [46]. In our study, the phenotype of highly late heading was caused by a functional deficiency in OsLUX under both SDs and LDs. The expression levels of Hd1 and Ghd7 were significantly up-regulated as well as a similar expression pattern in elh1 under both SDs and LDs (Fig. 2A-B, K-L), which means more Hd1 and Ghd7 proteins abundance may be accumulated in elh1 since OLUX released its disincentive effects on Hd1 and Ghd7 under both photoperiod conditions. Thus, Hd1-Ghd7 repressive complexes play a predominant role in delays heading in elh1 under both SDs and LDs. This is why the almost unobservable Ehd1 and Hd3a transcripts in elh1 under SDs are similar to those under LDs.

Both OsELF3-1 and OsELF4-1 overexpressing plants showed early heading under LDs and late heading under SDs [28], [35]. Our finding may explain the dual phenotype resulting from the strength of OsEC target repression, where OsELF4-1 and OsELF3-1 increased the repressive strength of OsEC for targeted regulation of Hd1 and Ghd7 in their overexpressing plants (Fig. S12A-F). Therefore, in over-expression plants, the down-regulated Hd1 and Ghd7 transcripts further relieved the suppression of Ehd1 and Hd3a, then caused an early heading at LDs. Under SDs, Hd1 exerts its promotion function in the background of a low level of Ghd7, and the low transcription levels of Hd1 will cause a delayed heading [43], [44], [45]. On the contrary, defects in OsELF4s and OsELF3-1 weakened the repressive strength of OsEC to regulate Hd1 and Ghd7. In oself4s mutants, OsELF4s failed to improve the interaction between OsLUX and OsELF3-1, while OsLUX-OsELF3-1 dimer still plays a predominant role in target repression. Thus, a minor effect on heading in oself4s was observed, which delayed heading slightly late than Nip (Fig. 5C-D; Fig. S11A-D). As for oself3-1, the repression strength of OsEC is significantly weakened, maybe due to OsLUX being unable to enter the nucleus completely. Consequently, in this study, the mutation in OsELF3-1 caused a more significant flowering effect that exhibited 22.1-d and 18.2-d later heading than WT plants under NLDs and NSDs, respectively (Fig. 5A-B). Defect in OsLUX resulted in the suppression function of the OsEC complex being entirely forfeit, which utterly released the targeted repression of Hd1 and Ghd7. Thus, an extreme late heading phenotype in the elh1 results from the nearly invisible level of Ehd1 and Hd3a under both LDs and SDs (Fig. 1A-B, Fig. 2C-F). Together, the complete OsEC complex is required to control rice heading via repress Hd1 and Ghd7 expression.

Conclusion

Our findings show that the MYB family transcription factor OsLUX is involved in the rice heading by directly regulating rhythm oscillation and core flowering-time-related genes. We uncovered a mechanism by which OsEC (OsELF4s–OsELF3-1–OsLUX) target suppresses the expressions of Hd1 and Ghd7 directly to modulate rice photoperiodic flowering. The OsEC–Hd1/Ghd7 regulatory module provides the genetic targets for crop improvement by selecting various combinations of OsEC–Hd1/Ghd7 components, which achieved optimization of photoperiod and planting area.

Compliance with Ethics Requirements

This article does not contain any studies with human or animal subjects.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary material

The following are the Supplementary data to this article: