Abstract
From screening a population of Arabidopsis overexpression lines, two Arabidopsis genes were identified, EFO1 (EARLY FLOWERING BY OVEREXPRESSION 1) and EFO2, that confer early flowering when overexpressed. The two genes encode putative WD-domain proteins which share high sequence similarity and constitute a small subfamily. Interestingly, the efo2-1 loss-of-function mutant also flowered earlier in short days and slightly earlier in long days than the wild type, while no flowering-time or morphological differences were observed in efo1-1 relative to the wild type. In addition, the efo2-1 mutation perturbed hypocotyl elongation, leaf expansion and formation, and stem elongation. EFO1 and EFO2 are both regulated by the circadian clock. Expression and genetic analyses revealed that EFO2 suppresses flowering largely through the action of CONSTANS (CO) and FLOWERING LOCUS T (FT), suggesting that EFO2 is a negative regulator of photoperiodic flowering. The growth defects in efo2-1 were augmented in efo1 efo2, but the induction of FT in the double mutant was comparable to that in efo2-1. Thus, while EFO2 acts as a floral repressor, EFO1 may not be directly involved in flowering, but the two genes do have overlapping roles in regulating other developmental processes. EFO1 and EFO2 may function collectively to serve as one of the converging points where the signals of growth and flowering intersect.
Introduction
During the life cycle of Arabidopsis thaliana, leaves are continuously produced from small clusters of actively dividing cells (leaf primordia) on the flanks of the shoot apical meristem. This ends when the shoot apex starts to initiate the floral transition in response to endogenous and exogenous cues. In recent years, molecular characterization of genetic mutants in Arabidopsis has revealed a large number of positive and negative floral regulators that are involved in monitoring the developmental age and physiological cues (autonomous and gibberellin pathways) as well as the response to environmental factors (vernalization and photoperiod pathways).
In Arabidopsis, a facultative long-day plant, the transition from vegetative to reproductive growth is promoted by an increase in day length. Studies on Arabidopsis have contributed greatly to our understanding of the molecular mechanisms that plants utilize to sense photoperiod in the determination of floral transition (Hayama and Coupland, 2003; Imaizumi and Kay, 2006). The photoperiod signal is primarily interpreted by an endogenous circadian clock to mediate flowering-time control. The floral transition is only one of numerous biological processes influenced by the circadian clock; others include daily changes of photosynthetic activities, leaf movement, cell growth, and various developmental processes (Harmer et al., 2000; Dodd et al., 2005). The prominent role of the clock in determining plant fitness is also reflected in the fact that the transcription of a large portion of the Arabidopsis genome changes in a rhythmic manner and thereby forms the molecular basis of achieving its regulatory function (Harmer et al., 2000; McClung, 2006). Photoreceptors transduce the light signal and entrain the circadian clock, and the signal information is processed and maintained through an autoregulatory feedback loop (Alabadi et al., 2001). CONSTANS (CO), which encodes a B-box zinc finger protein (Putterill et al., 1995), appears to act specifically in the signal output of the clock regulating photoperiodic flowering and to promote flowering in response to an inductive long-day photoperiod in Arabidopsis. The CO transcript is subject to the regulation of the circadian clock and exhibits a rhythmic expression pattern. A number of factors have been found to be involved in the transcriptional regulation of CO to mediate photoperiodic flowering (Hayama and Coupland, 2003; Imaizumi and Kay, 2006).
A growing body of evidence indicates that the post-transcriptional regulation of CO by light plays a crucial role in the response of photoperiodic flowering. Light is required to repress the degradation of the CO protein, and the light-regulation of CO stability appears to be collectively mediated by photoreceptors (Valverde et al., 2004). SPA1 and COP1, key components in light signalling, are known to be directly involved in regulating CO degradation (Laubinger et al., 2006; Jang et al., 2008; Liu et al., 2008). CO, in turn, induces the expression of FLOWERING LOCUS T (FT), a Raf kinase inhibitor-like gene, in a dose-dependent manner (Kardailsky et al., 1999; Kobayashi et al., 1999). Recent findings indicate that FT can travel a long distance in the phloem from the leaves to the shoot apex and is a mobile signal in promoting flowering (Corbesier et al., 2007; Jaeger and Wigge, 2007; Mathieu et al., 2007). FT and SUPPRESSOR OF OVEREXPRESSION OF CO 1 (SOC1), a MADS-box transcription factor (Lee et al., 2000; Onouchi et al., 2000) are the two best characterized integrators of different floral pathways, and SOC1 transcription is, at least in part, regulated by FT. The external coincidence mechanism underlying photoperiodic flowering, a long known yet unconfirmed theory, has been confirmed largely due to the progress in understanding the regulation of CO and FT activities in the context of the photoperiod response (Hayama and Coupland, 2003; Imaizumi and Kay, 2006).
The reductionism approach in forward and reverse genetic screens has been very effective in deciphering various developmental processes and physiological responses in Arabidopsis. Many flowering-time genes have been discovered in the genetic screens. As a complementary approach, a population of Arabidopsis transgenic lines, each overexpressing one of thousands of transgenes from several plant species under the control of the constitutive cauliflower mosaic virus (CaMV) 35S promoter, was used to identify genes that regulate flowering time. Two genes, EFO1 (EARLY FLOWERING BY OVEREXPRESSION 1) and EFO2, encoding putative WD-domain proteins with high sequence similarity to each other, that promote flowering when overexpressed in Arabidopsis, are reported here. Intriguingly, the efo2 loss-of-function mutant is also early flowering. It is shown that EFO2 acts as a repressor in photoperiodic flowering, and EFO1 and EFO2 have overlapping functions in other developmental processes.
Materials and methods
Screening the Arabidopsis overexpression population
Full-length cDNA or genomic clones (when cDNA clones were not available) from Arabidopsis and several crop species were placed under the control of the constitutive CaMV 35S promoter in CRS338, a vector constructed at Ceres which contains a bar gene (phosphinothricin acetyltransferase) as a selectable marker. The clones were transformed into Wassilewskija (Ws) plants using the Agrobacterium-mediated floral-dipping method. Five independent transgenic lines were randomly selected from the T1 plants carrying each clone and propagated. An equal amount of seed from 500 different lines representing 100 distinct transgenes were pooled and constituted a ‘superpool’.
Approximately 2000 seeds from each superpool were evenly sown on soil. The flats were placed at 4 °C in the dark for 3 d and then transferred to a growth chamber in a 12/12 h light/dark photoperiod regime. After 8 d, the plants were sprayed with 0.2% Finale™ (5.78% glufosinate ammonium) (Sanofi Aventis, Paris, France) every 3 d for a total of three applications. Plants that flowered earlier than the wild type (WT) Ws and the other overexpressing lines in the pots were identified. Genomic DNA was isolated from the candidates, and transgenes were amplified by PCR and identified by sequencing. To validate the early flowering phenotype of candidates, the five independent lines corresponding to the transgene identified in each candidate were examined in the T2 and T3 generations. Transgenic plants were differentiated from non-transgenic plants based on their resistance to glufosinate by placing leaf tissues on agar plates containing half-strength Murashige and Skoog (MS) salts (Murashige and Skoog, 1962). The copy number of the transgene for a line was determined and homozygous and heterozygous plants for each line were identified based on the analysis of the segregation pattern for the presence and absence of the herbicide resistance marker gene, bar, in the progeny.
Generation of mutants
The mutant alleles efo1-1 (SALK_060638) and efo2-1 (SALK_108846) were obtained from the Arabidopsis Biological Resource Center at The Ohio State University. The mutants were identified in the Salk collection using a PCR-based method (Alonso et al., 2003). The mutants contained a T-DNA insertion located 251 bp downstream of the ATG in efo1-2 and 605 bp in efo2-1. Both mutants were in a Columbia (Col) ecotype genetic background, and were backcrossed to the WT Col twice, and then homozygous mutants were identified for phenotypic and genetic analyses. The co-9, ft-10, and soc1-2 mutants in the Col background were kindly provided by Dr Richard Amasino (University of Wisconsin, Madison).
To generate the efo1 efo2, efo2 co, efo2 ft, and efo2 soc1 double mutants, efo2-1 was crossed to efo1-1, co-9, ft-10, and soc1-2, respectively, and the mutants homozygous for both loci were identified from the F2 progeny using a PCR-based method. The primers employed were EFO1-LP (5′-GTACCAATAACGTGAATGCATTTC-3′) and EFO1-RP (5′-CTAACCGTTTCTTCTAACGTCCC-3′) for detection of the WT EFO1 allele; SALK-LBa1 (5′-TGGTTCACGTAGTGGGCCATCG-3′) and EFO1-RP for efo1-1; EFO2-LP (5′-ACAGATCGTATGCTCTCACAACTC-3′) and EFO2-RP (5′-TCTCTTTTACGTAACAACGCTGTCTC-3′) for EFO2; SALK-LBa1 and EFO2-RP for efo2-1; CO-LP (5′-CCAGTTTCCATGGATGAAATG-3′) and CO-RP (5′-CCCCTTCTTTCAGATACCAGC-3′) for CO; SAIL-LB3 (5′-TAGCATCTGAATTTCATAACCAATCTCGATACAC-3′) and CO-RP for co-9 (Balasubramanian et al., 2006). The ft-10 and soc1-2 alleles were genotyped as described by Yoo et al. (2005).
Plant growth conditions
Plants were grown on 60% Sunshine (Sun Gro Horticulture, Bellevue, WA) and 40% vermiculite in Conviron growth chambers at 22 °C and 70% relative humility with a light intensity of 120 μE m-2 s-1 or on agar plates with 60 μE m-2 s-1. Phenotypic analysis was conducted in either long days (LDs, 16/8 h light/dark) or short days (SDs, 8/16 h light/dark) as indicated.
Measurement of flowering time
Flowering time was measured chronologically and by leaf number. The days to bolting were recorded as the number of days required from planting to the time when the first floral bud on the primary inflorescence became visible (∼2 mm). The number of rosette leaves produced by the apical meristem prior to bolting was counted. Flowering time for each genotype was measured at least twice.
Analysis of hypocotyl length
Arabidopsis seeds were surface-sterilized and placed on agar plates containing half-strength MS salt (PhytoTech), 0.5% sucrose (Sigma), and 0.7% agar (PhytoTech). The plates were kept at 4 °C in the dark for 3 d and then placed vertically in growth chambers. Seedlings were grown for 9 d and then scanned with an Epson Perfection 4870 flatbed scanner (EPSON America, Long Beach, CA) using the WinRhizo program (Regent Instruments, Canada). The hypocotyl length of each seedling was measured using the ImageJ program (1.37v, NIH, Bethesda, MA).
Gene expression analysis
To analyse gene expression in overexpressing lines and mutants, plants grown on agar plates as described above in SDs were collected and immediately frozen in liquid nitrogen. Total RNA was isolated using the RNeasy Mini Kit (Qiagen, Valencia, CA). Genomic DNA was removed during the purification using RNase-free DNase I according to the manufacturer's instructions. The cDNA of each sample was synthesized by adding 1 μg of total RNA in a 20 μl reaction and using the iScript cDNA Synthesis Kit (Bio-Rad, Hercules, CA). cDNAs were diluted to 120 μl with water and 3 μl was used in a 15 μl PCR reaction. The primers used in the reverse transcription (RT)-PCR-based expression analyses were as follows: EFO1, 5′-CGTTGTTTCCTCCGCATCCA-3′ and 5′-GTGATCGTCACGTGACTCTA-3′; EFO2, 5′-GAAGAATCTGTCGGTGTTGT-3′ and 5′-CGGGTCAACGAGTTTCCTTA-3′; CO, 5′-TGTCTTTGTTTCTGTCATTAGGCATAC-3′ and 5′-TTTGGGCGTTCTTGGGTGTG-3′; FT, 5′-AGTGGCTGCGGAGGAAGAAG-3′ and 5′-ATAGGCATCATCACCGTTCGTTAC-3′; and SOC1, 5′-TGTTCATTGGGTTACCTTGTTCTTC-3′ and 5′-TTGTTACTTGTCCTTATACACTCTCAG-3′. The constitutively expressed actin 2 gene (ACT2, At3g18780) was included as a control. The primers for amplifying ACT2 were: 5′-CGAGGGTTTCTCTCTTCCTC-3′ and 5′-TCTTACAATTTCCCGCTCTG-3′.
For the circadian experiment, seeds were sown on soil and plants were grown in a chamber under a 12/12 h light/dark photoperiod regime. After 18 d, the chamber was set to continuous light, and aerial tissues of the plants were harvested at 4 h intervals starting after 4 h in continuous light conditions. The transcript levels of EFO1 and EFO2 were determined by quantitative real-time RT-PCR (qRT-PCR) performed on an iCycler iQ Real Time PCR Detection System using the iQ SYBR Green Supermix (Bio-Rad). The transcript levels of each gene were calibrated to that of ACT2 within the sample. The relative expression level was normalized against that of ZT4 using the method as described (Livak and Schmittgen, 2001). The primers used were: EFO1, 5′- GTACCGAGGGCATGTGAATAGC-3′ and 5′-CCCACCTCTTATCGTACACAAACAC-3′; and EFO2, 5′-GAACAAGCACAACAACAAGAAGAAG-3′ and 5′-GGAGGAGGAGGAGGAAGATACG-3′. The primers for ACT2 were the same as described above.
Results
Over-expression of either EFO1 or EFO2 promotes flowering and hypocotyl elongation in Arabidopsis
From screening Arabidopsis lines, each overexpressing one of approximately 10 000 full-length and non-redundant genes under the control of the constitutive CaMV 35S promoter (Alexandrov et al., 2006, 2009), 17 genes were identified that conferred early flowering; a role for several of these genes in the control of flowering time was previously unknown. The transgene (At5g52250) contained in one line was designated as EFO1 (Fig. 1A, B). Subsequent analysis of the transgenic lines overexpressing another gene (At5g23730), the only Arabidopsis gene that encodes a protein with substantial sequence similarity to EFO1 (Fig. 2), showed that they were also early flowering (Fig. 1A, B). This gene was designated as EFO2.
Overexpression of EFO1 or EFO2 promotes flowering and hypocotyl elongation in Arabidopsis. (A) Flowering time in LDs. Two representative overexpressing lines of each gene are shown (n=12). The y-axis indicates the days to bolting (open bar) or the number of rosette leaves (grey bar). The error bars represent standard deviation. (B) Flowering time in SDs as shown in (A). (C) Hypocotyl length. Seedlings (n=20) were grown on agar plates for 9 d in LDs (open bar) or SDs (grey bar). (D) RT-PCR analysis of EFO1 and EFO2 expression in WT and the over-expression lines. Total RNA was isolated from 7-d-old seedlings grown on agar plates in SDs. The number of PCR cycles used was 30. The constitutively expressed ArabidopsisACT2 was used as a control.
Sequence alignment of EFO1 and EFO2. The numbers on the right refer to the positions of amino acid residues. Identical residues are shaded in black and conserved residues in grey. The roman numerals indicate the WD repeats highlighted with solid lines in EFO1 (top) and EFO2 (bottom). The asterisks indicate the WDxR motif. Arrowheads indicate the positions where T-DNA is inserted in efo1-1 (top) and efo2-1 (bottom).
In the segregating progeny of multiple independent transgenic lines of EFO1 or EFO2, both homozygous and heterozygous transgenic plants flowered earlier than null segregants, suggesting that the early-flowering phenotype is dominant. Two independent lines of each gene that contain a single copy of the transgene were selected and homozygous transgenic lines were identified for phenotypic analysis (Fig. 1D). Among all lines tested, the line 35S:EFO1-02 showed the strongest early-flowering phenotype, the other EFO1- and EFO2-overexpressing lines flowered slightly, but reproducibly, earlier in LDs (Fig. 1A). In SDs, the early-flowering phenotype became more pronounced in all overexpressing lines (Fig. 1B).
In addition to early flowering, the EFO1- and EFO2-overexpressing lines, compared to WT, share another attribute: elongated hypocotyls (Fig. 1C). These results indicate that EFO1 and EFO2 may play a role in the regulation of flowering time and hypocotyl elongation.
EFO1 and EFO2 encode putative WD-domain proteins
EFO1 and EFO2 share 56% sequence identity over their entire length (Fig. 2), and both contain multiple WD motifs (also known as the WD40 motif or repeat). The sequences in the domains are highly conserved (63% identity). WD domains have a conserved β-propeller structure, but accurate prediction of the domain based on a sequence is difficult (Smith et al., 1999). Although the C-terminal region of EFO1 was not predicted to contain a seventh WD repeat by the SMART program (Schultz et al., 1998), it may constitute one, considering the high sequence similarity to the seventh WD repeat in EFO2 (Fig. 2).
There are more than 200 proteins predicted to contain a WD domain in the Arabidopsis genome (van Nocker and Ludwig, 2003). Sequence analysis showed that the WD domains of EFO1 and EFO2 have little sequence similarity (<20% identity) to other members in the superfamily except CONSTITUTIVE PHOTOMORPHOGENIC 1 (COP1), a photomorphogenesis repressor functioning as an E3 ubiquitin ligase (Deng et al., 1992), and SUPPRESSOR OF PHYA-105 (SPA1), another photomorphogenesis repressor mediating PHYA signalling, and other members (SPA2, SPA3, SPA4) in the SPA family (Laubinger et al., 2006). All five proteins contain a domain of seven WD repeats, with the sequence identity to those of EFO1 and EFO2 ranging from 31% to 35% (see Supplementary Table S1 at JXB online). In a reciprocal BLAST search (Rivera et al., 1998), an open reading frame encoding a polypeptide homologous to EFOs was identified in a single-celled green alga (Chlamydomonas reinhardtii) and a multicellular moss (Physcomitrella patens subsp. patens), respectively (see Supplementary Table S1 at JXB online). Neither open reading frame is fully annotated in that both lack the start codon. Although only five WD repeats were predicted in the two partial polypeptides, they shared the highest similarity to EFO1 and EFO2 among all the WD-domain proteins in Arabidopsis. Interestingly, the most distantly related sequence (Chlamydomonas) has similar sequence identities to the WD domains of EFO1, EFO2, COP1 and SPA proteins, suggesting that these WD domains may have descended from a common ancestor during evolution. In a recent study (Lee et al., 2008), EFOs, COP1, and SPAs were classified into a subgroup in the WD-domain protein superfamily because each contains a WD repeat with a conserved WDxR motif (Fig. 2).
EFO1 and EFO2 are regulated by the circadian clock
To understand the physiological function of EFO1 and EFO2, their expression was first analysed in different organs at various developmental stages based on the online AtGenExpress microarray data set (Schmid et al., 2005). The expression profiles of the two genes are clearly distinct (see Supplementary Fig. S1 at JXB online). The transcript level of EFO1 is very low and close to baseline in leaf, stem, and the shoot apex, whereas EFO2 is expressed in all tissues. Consistent with this observation, RT-PCR analysis suggested that EFO2 is more abundant than EFO1 (Fig. 1D).
Although little is known about the function of EFO1 and EFO2, genome-wide gene expression profiling analyses revealed that they are under the regulation of a wide spectrum of light. They may be involved in phytochrome-mediated signalling (Tepperman et al., 2006), and ELONGATION HYPOCOTYL 5 (HY-5)-mediated responses to high irradiance, blue and UV-B light (Ulm et al., 2004; Brown et al., 2005; Oravecz et al., 2006; Kleine et al., 2007). It is shown that both hypocotyl elongation and flowering time are disturbed in the EFO1- and EFO2-overexpressing lines. Changes in these two developmental processes are often associated with defects in circadian regulation (McClung, 2006). All these data prompted us to examine whether the two genes are regulated by the circadian clock.
Quantitative real-time RT-PCR analyses showed that EFO1 and EFO2 are indeed circadian-regulated with a similar rhythmic pattern (Fig. 3). They are highly expressed at night and peak at the beginning of the day, and the expression gradually decreases during the day and reaches a trough at the end of the day.
Circadian regulation of EFO1 and EFO2 expression. Plants were grown under a 12/12 h light/dark photoperiod regime for 18 d and then switched into continuous light. Aerial tissues were harvested and qRT-PCR analyses were performed as described in the Materials and methods. The x-axis indicates zeitgeber time. Open bars denote subjective day and grey bars subjective night. The y-axis represents the relative level of transcript normalized against that at ZT4.
EFO1 and EFO2 have overlapping roles in hypocotyl elongation and vegetative development
To gain insight into the function of EFO1 and EFO2, T-DNA insertion mutants of the two genes were obtained from the SALK collection (Alonso et al., 2003). The homozygous mutants were identified and subsequently named efo1-1 and efo2-1, respectively. Each mutant contains a T-DNA insertion that disrupts the WD domain as shown in Figs. 2 and 4A. RT-PCR analysis showed that no full-length transcript of EFO1 in efo1-1 or EFO2 in efo2-1 was detectable (Fig. 4B), suggesting that they are loss-of-function mutants.
Phenotype of efo1-1, efo2-1, and efo1 efo2. (A) Diagrams showing the positions of T-DNA in efo1-1 and efo2-1. The 5' and 3' untranslated regions (open bars), and the start and stop codons are marked. (B) RT-PCR analysis of EFO1 and EFO2 expression in efo1-1 and efo2-1. Total RNA was isolated from 7-d-old seedlings grown on agar plates in SDs. The number of PCR cycles used was 35 for amplifying EFO1 and EFO2, and 30 for ACT2. (C) Rosettes of WT and mutants grown in SDs. Bar=2 cm. (D) Early flowering of efo2-1. The plants were grown in SDs. (E) Hypocotyl length. Seedlings (n=20) were grown on agar plates for 9 d in LDs (open bar) or SDs (grey bar). (F) Rosette growth. Plants (n=12) were grown in soil for 16 d in LDs (open bar) or 30 d in SDs (grey bar). (G) Flowering time in LDs. The days to bolting (open bar) and the number of rosette leaves (grey bar) were measured for each genotype (n=12). The numbers on the top of bars indicate the average number of days required to produce one rosette leaf (days to bolting/number of rosette leaves). (H) Flowering time in SDs as shown in (G). (This figure is available in colour at JXB online.)
Although overexpression of EFO1 promoted flowering and hypocotyl elongation, in both LDs and SDs, no apparent difference in the phenotypes examined, including hypocotyl length, leaf and rosette size, flowering time, and plant height, was observed between efo1-1 and WT (Fig. 4; see Supplementary Fig. S3 at JXB online).
The EFO2-overexpressing lines had enhanced hypocotyl elongation (Fig. 1C). Conversely, hypocotyl length of light-grown efo2-1 plants, compared with the WT, was significantly reduced (Fig. 4E). Etiolated seedlings of efo2-1 also exhibited a reduction in hypocotyl elongation, but to a much lesser extent relative to the light-grown seedlings (see Supplementary Fig. S2 at JXB online). Thus, EFO2 is a key regulator of hypocotyl elongation, and its function is largely light-dependent. In addition, the efo2-1 mutant was notably slower in growth as shown by leaf and rosette size in both LDs and SDs (Fig. 4C, F). The rate of leaf formation in efo2-1 was reduced relative to WT in LDs, and the reduction was even more severe in SDs (Fig. 4G, H), indicating that EFO2 may play a role in controlling plastochron length. Inflorescence stems of efo2-1 were shorter than those of WT (see Supplementary Fig. S3 at JXB online). Therefore, EFO2 is involved in the regulation of not only hypocotyl elongation at the seedling stage but also leaf development and stem elongation in adult plants.
Given that the sequences of EFO1 and EFO2 are highly conserved, and their overexpression exerts similar effects on hypocotyl elongation and flowering, the lack of any apparent defects in efo1-1 may be due to low expression of EFO1 as described above and/or functional redundancy of EFO1 and EFO2. Therefore, a double mutant was created. Compared with efo2-1, the growth of efo1 efo2 was further stunted (Fig. 4C). The leaves of efo1 efo2 were rolled under and much smaller and narrower at the early developmental stage; however, they continued to expand when the leaves of efo2-1 reached a maximum size. The difference in the leaf size of efo1 efo2 and efo2-1 largely diminished eventually, although the leaves of both mutants were significantly smaller than those of WT and efo1-1 (Fig. 4D). The leaves of efo1 efo2 appeared to expand at a slower rate than those of efo2 but with an extended duration. Hypocotyls and inflorescence stems of efo1 efo2 were slightly shorter than those of efo2-1 (Fig. 4E; see Supplementary Fig. S3 at JXB online). These findings revealed that there is an additive effect of EFO1 and EFO2 in multiple developmental processes.
EFO2 is a repressor of flowering
In addition to the defects observed during vegetative development, the efo2-1 mutant bolted slightly earlier than WT in LDs, and produced 8.3 rosette leaves in total compared with 11.1 in WT (Fig. 4G). In SDs, the early flowering of efo2-1 became more striking (Fig. 4H). WT plants produced around 42 more rosette leaves in SDs than LDs, whereas the difference of rosette leaf numbers of efo2-1 under the two photoperiods was around 12. Therefore, the flowering time of efo2-1 is less sensitive to the change in day length, and the mutant has a much reduced photoperiod response in flowering. EFO2 negatively regulates flowering in response to photoperiod.
Overexpression of EFO2 resulted in an early-flowering phenotype (Fig. 1A, B); interestingly, its loss-of-function mutant efo2-1 was also early flowering (Fig. 4D, G, H). To address whether the phenotypes in efo2-1 were caused by loss of function of EFO2, the CaMV 35S promoter-driven EFO2 was introduced into efo2-1. Twelve independent transformants (T1) were selected, and the phenotype of these lines was further examined in the T2 generation. For 11 of the 12 lines analysed, the EFO2 transgene partially or completely rescued the defects of growth and flowering time in efo2-1 (see Supplementary Fig. S4 at JXB online), suggesting that the efo2-1 mutation was responsible for the mutant phenotypes. The fact that either overexpression or loss of function of EFO2 confers early flowering indicates a complex mechanism underlying the regulation of flowering time by EFO2 (see Discussion). Such a phenomenon was previously reported for PHYB (Bagnall et al., 1995). PHYB is also a negative regulator of flowering time and results in early flowering when overexpressed or knocked-out.
The efo2-1 mutation results in the up-regulation of FT
The circadian-regulated expression and the flowering phenotypes observed in the overexpressing lines and mutants pointed to a role for EFO2 in photoperiodic flowering. To test this hypothesis, the expression of key genes involved in the pathway in the mutants was examined. No dramatic difference was found in the transcript levels of CO or SOC1 at the same developmental stage of WT, efo1-1, efo2-1, and efo1 efo2 (Fig. 5A). The diurnal expression pattern of CO was also almost identical in the WT and in the mutants (Fig. 5B). However, compared with efo1-1 and the WT, the FT transcript in efo2-1 and efo1 efo2 was up-regulated (Fig. 5A, B). The elevation of the FT transcript was specifically associated with the efo2-1 mutation, whereas the transcript levels of CO are similar in WT and the mutants. Thus, efo2-1 derepresses FT expression, which appears to be independent of the CO transcript level.
Derepression of FT by efo2-1. (A) Transcript abundance of CO, FT, and SOC1 determined by RT-PCR analysis. Total RNA was isolated from 7-d-old and 14-d-old plants grown on agar plates in SDs. The number of PCR cycles used was 35 for amplifying CO and FT, and 30 for SOC1 and ACT2. (B) Diurnal expression of CO and FT. Total RNA was isolated from 7-d-old plants grown on agar plants in SDs. The number of PCR cycles used was 30.
EFO2 acts upstream of CO/FT
To define the role of EFO2 in photoperiodic flowering further, the efo2 co, efo2 ft, and efo2 soc1 double mutants were created. In LDs, as described previously (Yoo et al., 2005; Balasubramanian et al., 2006), co-9, ft-10, and soc1-2 all flowered significantly later than WT (Fig. 6A). The efo2 co double mutant bolted at the same time as co-9; efo2 ft bolted slightly earlier than ft-10 (Fig. 6A). The rosette leaf number of efo2 co and efo2 ft was, to different extents, reduced compared with that of co-9 and ft-10, respectively. In SDs, co-9 and ft-10 have no effect on flowering time in the single mutants, but they completely abolished the early flowering of efo2-1 (Fig. 6B). Therefore, although the efo2-1 mutation does not appear to affect CO transcriptionally (Fig. 5), both CO and FT are required for the early flowering of efo2-1. In both LDs and SDs, efo2 soc1 bolted earlier than soc1-2 and produced many fewer rosette leaves (Fig. 6A, B). These results indicate that the activity of EFO2 is largely dependent upon CO and FT, and to a lesser degree upon SOC1, especially in SDs.
Flowering time of efo2 co, efo2 ft, and efo soc1. The days to bolting (open bars) and the number of rosette leaves (grey bars) were recorded for each genotype (n=12) in both LDs (A) and SDs (B).
Distinct roles of EFO1 and EFO2 in flowering
In contrast to functional redundancy in vegetative development, the loss-of-function of EFO1 did not promote the early flowering of efo2-1 further, and no additive effect in flowering was observed in efo1 efo2. In fact, efo1 efo2 bolted later and produced a few more rosette leaves than efo2-1 in both LDs and SDs (Fig. 4G, H), which seems to suggest that the early flowering of efo2-1 was partially suppressed by efo1-1 in efo1 efo2, and EFO1 promotes flowering in the efo2-1 mutant background. It should be noted, however, that although efo1 efo2 bolted significantly later than efo2-1, the FT transcript level of efo1 efo2 was at least as high as that of efo2 in 7-d-old plants, and even higher in 14-d-old plants (Fig. 5A, B). This result implies that efo1-1 does not suppress the expression of the major floral integrator FT induced by efo2-1. Since EFO1 and EFO2 have a redundant function in controlling vegetative development, the late bolting of efo1 efo2 might be caused by compound inhibition of growth rather than by a role of efo1-1 in flowering (Fig. 4C).
Regulation of hypocotyl elongation by EFO2 is independent of its role in flowering-time control
The efo2-1 mutation has pleiotropic effects on developmental processes, such as hypocotyl elongation, leaf expansion, and flowering time. CO, FT, and SOC1 all appear to be specific in regulating flowering time and have no reported role in the vegetative phase of development (Koornneef et al., 1998; Lee et al., 2000; Onouchi et al., 2000). It was noted that leaves and rosettes of efo2 co, efo2 ft, and efo2 soc1 were smaller than those of co-9, ft-10 and soc1-2 at the same developmental stage, respectively (see Supplementary Fig. S5 at JXB online). It was also found that the hypocotyl length of efo2 ft and efo2 soc1 was comparable to that of efo2-1. Hypocotyls of efo2 co were longer than those of efo2-1 but still shorter than those of WT and co-9 (Fig. 7). Therefore, although the early flowering phenotype of efo2-1 is completely or partially repressed in the double mutants, the inhibition of hypocotyl elongation resulting from efo2-1 is not, or to a lesser extent, suppressed, indicating that the involvement of EFO2 in hypocotyl elongation is largely independent of its role in flowering.
Effect of efo2-1 on hypocotyl elongation. Seedlings (n=35) were grown on agar plates for 9 d in SDs.
Discussion
From screening a population of Arabidopsis overexpression lines for early flowering, EFO1 and EFO2, which encode putative WD-domain proteins with substantial sequence identity, were identified. Overexpression of either EFO1 or EFO2 promotes flowering. Subsequent analysis of loss-of-function mutants demonstrated that EFO2 is a repressor of photoperiodic flowering in Arabidopsis, whereas the involvement of EFO1 in flowering is minimal or indirect. In addition to flowering time, EFO2 influences hypocotyl elongation and other developmental processes, such as leaf expansion and formation, and stem elongation. Double mutant analysis revealed that EFO1 has overlapping roles with EFO2 in promoting vegetative growth. EFO1 and EFO2 may act at one of the converging points in multiple independent, but yet interconnected, developmental processes (Fig. 8).
A schematic diagram depicting regulatory roles of EFO1 and EFO2 in flowering and vegetative development. Lines with an arrowhead indicate promotion and lines with a bar inhibition. Thin lines indicate speculative relationships. The question mark denotes unknown factor(s) which are under the regulation of EFO1 and/or EFO2 when they are over-expressed to promote flowering.
EFO1 and EFO2 are closely related putative WD-domain proteins
In Arabidopsis, several WD-domain proteins have been shown to regulate flowering. FY, VIP3, and FVE are involved in controlling the expression of FLOWERING LOCUS C (FLC) (Simpson et al., 2003; Zhang et al., 2003; Ausin et al., 2004; Kim et al., 2004), a MADS-box gene and potent flowering repressor (Michaels and Amasino, 1999; Sheldon et al., 1999), which is a key component in the vernalization and autonomous pathways. A recent study indicates that LIGHT-REGULATED WD1 (LWD1) and LWD2, two WD-domain proteins which share high sequence identity, are functionally redundant and modulate the circadian clock to suppress photoperiodic flowering (Wu et al., 2008). EFO1 and EFO2 have little sequence similarity with any of these WD-domain proteins. High sequence conservation and partial functional redundancy of EFO1 and EFO2 suggested that they form a small WD-domain protein subfamily.
Among the numerous WD-domain proteins in the Arabidopsis genome, only the WD domains of COP1 and the SPA proteins share notable sequence identity with EFO1 and EFO2 (see Supplementary Table S1 at JXB online). A recent bioinformatics analysis classified them into a subgroup in the WD-domain protein superfamily because they all contain a signature motif WDxR within the WD domains, which is essential for binding the Damaged DNA Binding 1 (DDB1) protein, a component of the Cullin 4 (CUL4)-based E3 ubiquitin ligase complex (Lee et al., 2008). It was proposed that members in this group may function as the substrate receptor to recruit proteins selectively for ubiquitination to modify the proteins post-translationally for proteasomal degradation. Both COP1 and SPAs contain additional functional domains outside their WD domain (Deng et al., 1992; Laubinger et al., 2004, 2006), whereas EFOs contain almost solely WD repeats in their entire sequences with only a very short extension at the N-terminus (Fig. 2). The mutations in these genes have clearly distinct effects. Constitutive photomorphogenesis, a phenotype of the cop1 and spa mutants, was not observed in the efo mutants (see Supplementary Fig. S2 at JXB online). Despite the difference, the efo, cop1, and spa mutants do share some attributes, such as reduced hypocotyl elongation and retarded growth. The photoperiod response in flowering is completely abolished in cop1 and the spa1 spa3 spa4 triple mutant (McNellis et al., 1994; Laubinger et al., 2006), while the sensitivity to photoperiod is much reduced in efo2-1.
Overlapping and distinct functions of EFO1 and EFO2
Overexpression of either EFO1 or EFO2 promoted hypocotyl elongation (Fig. 1C). Consistent with the phenotype in the overexpressing lines, the loss-of-function of EFO2 inhibited hypocotyl elongation (Fig. 4E). In addition, efo2-1 exhibited slow leaf expansion, decreased rate of leaf formation, and reduced stem elongation (Fig. 4, see Supplementary Fig. S3 at JXB online). Although efo1-1 did not show such visible morphological defects, the phenotypes caused by efo2-1, except the rate of leaf formation, are enhanced in efo1 efo2. Thus, EFO1 and EFO2 have overlapping roles in promoting hypocotyl elongation, leaf expansion, and stem elongation. Despite their partially redundant functions, EFO2 clearly plays the dominant role in these developmental aspects.
The early-flowering phenotype of EFO1- and EFO2-overexpressing lines suggests a promotive role for both genes in floral transition (Fig. 1A, B). However, analysis of the single mutants revealed that EFO2 is a floral repressor, while EFO1 is normally not involved in the control of flowering time. The efo1 efo2 double mutant bolted later and produced a few more rosette leaves than efo2-1 in both LDs and SDs. This result raises the possibility that EFO1 promotes flowering in the absence of EFO2. Expression analysis, however, showed that efo2-1, but not efo1-1, induced the up-regulation of FT, and the transcript level of FT in efo1 efo2 was comparable to or even higher than that in efo2-1 (Fig. 5). Thus, the signal for floral transition was not suppressed in efo1 efo2. A more plausible explanation for the delay in the bolting of efo1 efo2 relative to efo2-1 is that it might reflect physiological effects resulting from the severely stunted growth of the double mutant rather than a direct role of EFO1 in flowering. In other words, EFO1 is required for the early-flowering of efo2-1 because of its role in growth promotion. Regardless, EFO1 appears to play a minimal role in flowering, whereas EFO2 is a floral repressor.
Analyses of overexpressing lines and loss-of-function mutants revealed that EFO1 and EFO2 have overlapping but distinct functions. The functional distinction between the two genes may be due to the differences in expression levels, and/or temporal and spatial expression patterns. As described above, EFO1 and EFO2 are differentially expressed and their transcripts differ in abundance. It is also possible that the two proteins have different biochemical activities, such as substrate specificity and binding affinity. Further molecular and biochemical characterization is needed to test these hypotheses.
EFO2 acts through CO/FT to regulate photoperiodic flowering
The expression of EFO2 is circadian-regulated (Fig. 3). The efo2-1 mutant flowers early in both LDs and SDs and has a reduced response to the change of day length in flowering (Fig. 4G, H). EFO2 suppresses FT transcript levels (Fig. 5). Genetic analysis demonstrated that both CO and FT are required for the early flowering of efo2-1 in both LDs and SDs (Fig. 6). All these results implicate that EFO2 is a negative regulator of photoperiodic flowering and acts upstream of CO/FT.
CO is a core component in the control of photoperiodic flowering. CO promotes FT expression (Samach et al., 2000; Wigge et al., 2005), and the FT transcript level is highly and positively correlated with the activation of floral transition (Suarez-Lopez et al., 2001; Yanovsky and Kay, 2002). The activity of CO is regulated at both the transcriptional and post-transcriptional levels. CO transcription is regulated by the circadian clock, while CO is rapidly degraded in the dark and light is required for its accumulation (Valverde et al., 2004). Expression analysis showed that efo2-1 derepressed FT transcription, but this effect appears to be independent of CO transcript levels because the transcript abundance of CO at different developmental stages and the diurnal expression pattern were very similar in WT, efo2-1, efo1-1, and efo1 efo2 (Fig. 5). Thus, EFO2 may regulate CO post-transcriptionally. In this regard, it is intriguing to note that COP1 and SPAs physically interact with CO and modulate the abundance of CO protein (Laubinger et al., 2006; Jang et al., 2008; Liu et al., 2008). SPA1 also interacts with COP1 and modulates its activity in vitro (Hoecker and Quail, 2001; Saijo et al., 2003; Seo et al., 2003). The WD domain of COP1 is required for its interaction with CO (Jang et al., 2008; Liu et al., 2008). COP1 may target CO for degradation by ubiquitination (Liu et al., 2008). Given the sequence similarity as discussed above, it is possible that EFO2 may function as a substrate receptor and interact with COP1 or other E3 ubiquitin ligase to target CO for ubiquitination and proteasome-mediated degradation. Identifying proteins that interact with EFO2 will help gain more insight into the underlying mechanism of the regulation of CO by EFO2.
Under inductive LDs, efo2 co bolted at the same time as co-9 and efo2 ft slightly later than ft-10, whereas efo2 co and efo2 ft produced fewer rosette leaves than co-9 and ft-10, respectively (Fig. 7A). The reduction in the rosette leaf number of the double mutants relative to the single mutants may be the consequence of the slow rate of leaf formation caused by efo2-1 (Fig. 4G, H). Alternatively, EFO2 may have activities in repressing flowering under inductive LDs that are independent of CO and FT (Fig. 8). The CO- or FT-like genes could act in photoperiodic flowering mediated by EFO2 (Robson et al., 2001; Griffiths et al., 2003; Michaels et al., 2005). Nevertheless, our results suggest that the role of EFO2 in flowering is largely dependent upon the activities of CO and FT.
The mechanism is yet to be determined on how overexpression of EFO2, a floral repressor, promotes flowering. Since EFO2 has pleiotropic effects in multiple developmental processes, physiological changes may occur when constitutively and/or ectopically overexpressed, which could create a gain-of-function early-flowering phenotype. Such a scenario might account for the complex regulatory roles of PHYB in flowering (Bagnall et al., 1995). Like EFO2, PHYB does not appear to affect CO at the transcript level (Blazquez and Weigel, 1999), and it is known to promote the degradation of CO (Valverde et al., 2004). It would be interesting to investigate whether EFO2 acts in the PHYB-mediated pathway to regulate flowering. Another possibility is that constitutive and/or ectopic overexpression may enable EFO2 to bypass CO to activate downstream components in the pathways, such as FT and floral meristem identity genes (Fig. 8). Recent studies have revealed regulators that mediate CO-independent photoperiodic flowering (Jung et al., 2007; Wang et al., 2009; Yamaguchi et al., 2009). Further studies are needed to differentiate these possibilities.
We are grateful to Dr Richard Amasino and the Arabidopsis Biological Resource Center for providing mutant seed. We thank Dr Roger Pennell for advice on the manuscript. We acknowledge our many Ceres colleagues for excellent technical assistance. Part of the data in this work resulted from the collaboration between Monsanto and Ceres.
References
Author notes
Present address: Monsanto, 800 North Lindbergh Boulevard, St Louis, MO 63167, USA.
Present address: School of Plant Sciences, University of Arizona, Tucson, AZ 85721, USA.
Comments