The Expression of ELF4 -Like Genes Is Inﬂuenced by Light Quality in Petunia

: The signals from photoreceptors modify plant morphogenesis and regulate the timing of ﬂowering. In the long-day plant petunia, ﬂowering is accelerated under blue (B) and white (W) light compared to red (R) light. In Arabidopsis thaliana L., ELF genes are involved in circadian clock-associated regulation of ﬂowering under di ﬀ erent light conditions. In this study, we aimed to assess the involvement of ELF genes in control of ﬂowering by light quality in petunia. Two ELF4 -like genes, PhELF4-1 and PhELF4-2 with 76% and 70% similarity to orthologues in pepper but low overall similarity to ELF genes in A. thaliana L., were characterized in petunia and their expression patterns studied under di ﬀ erent light qualities. Both genes showed a rhythmic expression pattern and higher expression under B light from light emitting diodes (LED) and W light from ﬂuorescent lamps than under R LED light from LED. For both genes, the expression peaked towards the end of the day, 12 h after start of a 14 h photoperiod. Compared with PhELF4-2 , PhELF4-1 expression showed higher amplitude with signiﬁcantly higher peak expression. As investigated for PhELF4-1 , such an expression rhythm was kept for two days after transfer of the plants to continuous lighting using B LED, indicating a circadian rhythm. PhELF4-1 also responded with a phase shift after transfer to short days of an 8 h photoperiod. These results indicate that PhELF4 -like genes in petunia are under photoperiodic control involving a circadian clock and play a role in signal transduction from one or more B light photoreceptors.


Introduction
In general, the transition to flowering in plants is well known to be regulated by the light conditions. In this respect, photoperiod is a crucial factor for floral induction in both long-day plants such as Arabidopsis thaliana L. and short-day plants like Chrysanthemum. Not only photoperiod but also the light quality affects induction of flowering in a wide range of plant species.
Photomorphogenesis, including floral induction, occurs in response to different wavelengths absorbed by different photoreceptors such as phytochromes and cryptochromes [1]. By this, the light receptors initiate signaling pathways which affect expression of genes and proteins controlling the flowering time [1,2]. PHYTOCHROME B (PHYB) is known to control the expression of the floral induction gene FLOWERING LOCUS T (FT) [3] and CRYPTOCHROME 2 (CRY2) stabilizes the CONSTANS (CO) protein, which in turn induces the expression of the FT gene [4,5]. On the other hand, transplanted individually to 10 cm pots filled with the same commercial growth medium used for germination. The seedlings were fertilized with a commercial fertilizer once a week (Hyponex, N:P:K = 6:10:5, Hyponex, Osaka, Japan).

Experimental Growth Conditions
When four to five true leaves had emerged at day 21 after sowing, the seedlings were transferred to growth cabinets (20-

Gene Identification by 3 EST High Throughput Sequencing
Three petunia shoots were sampled from each three plants at day 0, 7 and 14 of the treatments with R or B LED light described above. Each plant shoot was frozen in liquid nitrogen and stocked in the deep freezer (−80 • C). Total RNA extraction from the bulked shoot samples of each light quality treatment were performed using an RNeasy Plant Mini Kit (QIAGEN, Hilden, Germany) and then treated with DNase using an RNase-Free DNase set (QIAGEN, Hilden, Germany). The RNA sample qualities were checked by a Bioanalyzer (Agilent 2100 Bioanalyzer, Agilent Technologies Inc., Santa-Clara, CA, USA). Single strand cDNA was synthesized from 2 µg total RNA using a T7-Oligo (dt) primer with a T7-promoter. Furthermore, double strand cDNA was synthesized by DNA polymerase and RNase H. Subsequently, RNA amplification from this double strand cDNA was done by T7 RNA polymerase. Thereafter, single strand cDNA was synthesized again from 3 µg RNA template and random primers, and then double strand cDNA was synthesized using a biotin labeled oligo (dt) primer with a 454B adapter. A 454A adapter was attached into the synthesized double and single strand cDNA and was used as the sequence template DNA after alkalizing treatment. The sequence template DNA was bonded into capture beads and amplified by emulsion PCR. Thereafter, these beads were recovered and counted. An appropriate number of beads was added into a PicoTiterPlate and sequencing was done by a Genome Sequencer FLX system (454 Life Sciences, Branford, CT, USA). Trimming treatments were carried out to remove the adapter sequence and the polyA sequence from all sequence data. After data trimming, the assembly of the sequence data was done by MIRA3 (version 3.2.0) [17].

RACE PCR for Isolation of ELF4 Like Genes
Shoots of petunia plants with five true leaves exposed to a 14 h photoperiod from W fluorescent lamps for 2 days were sampled and total RNA was extracted as described in Section 2.3. As a result of the 3 EST sequencing (2.3), 3 different clones of EARLY FLOWERING 4 (ELF4)-like genes were detected. Two EST fragments of ELF4-like genes (PhELF4-1 and PhELF4-2) were extended by 3 and 5 RACE PCR (3 -FULL RACE Core Set, TAKARA BIO INC., Kusatsu, Shiga, Japan). DNA fragments extended by RACE PCR were TA cloned (pGEM-Vector System 1, Promega Co., Madison, Wisconsin, USA) and transformed into competent cells (ECOS Competent Escherichia coli DH5α, TAKARA BIO INC., Kusatsu, Shiga, Japan) and incubated on LB plates. After inserted colony selection, transformed E. coli was incubated in liquid medium and then the plasmid was extracted by a plasmid DNA extraction kit (Plasmid DNA Purification, TAKARA BIO INC., Kusatsu, Shiga, Japan). Subsequently, extracted plasmids were sequenced by a DNA sequencer (PRISM 3130-Avant, Applied Biosystems Co., Foster, CA, USA).

Expression Analysis of ELF4-Like Genes and Recordings of Developmental Parameters
In each of four separate experiments (a-d below), 2-3 leaves from each of 3 petunia plants were collected for RNA extraction per sampling time point per light quality treatment (the W, B and/or R light treatments described above; experiment a-c) or photoperiod (experiment d; time courses in experiment b-d) were collected for RNA extraction ( Table 1). The sampled leaves from one plant served as a biological replicate, and there were accordingly 3 biological replicates per sampling time per light quality/photoperiod treatment. The sampling started at the 4-5 leaf stage and was performed according to the following schedule:

Statistical Analyses
After confirming normality using the Ryan-Joiner test and homoscedasticity by Levene's test, analysis of variance (ANOVA) in the general linear model (glm) mode was carried out using the Minitab 19 software (Minitab Inc, PA, USA) (p ≤ 0.05). To test for differences between means, Tukey's post hoc test (p ≤ 0.05) was used (Minitab 19). For the gene expression data, log-transformed data was used in the statistical analyses due to lack of normality or/and homoscedasticity of the nontransformed data, whereas for growth parameters nontransformed data was used. One-way ANOVA glm was used to evaluate the effect of light quality on ELF4-1 and ELF4-2 transcript levels in experiment a (one sampling time point) and effect of photoperiod (B light) on developmental parameters in experiment d. Two-way ANOVA was used to assess the effect of light quality on the ELF4-1 and ELF4-2 transcript levels during time courses, i.e., with light quality (experiment b, c) or photoperiod (experiment d) and time as factors.

Floral Induction and Light Signal Transduction Genes Expressed in Petunia under Different Light Qualities as Revealed by 3 EST High Throughput Sequencing
3'EST high throughput sequencing showed that there were EST clones of 167,634 reads in the R LED treatment and 147,129 reads in the B LED treatment. Totally, 288,779 assembled contig fragments and 6737 singlet fragments were detected after filtering and assembly steps. An NCBI-blastx search for these contig and singlet fragments revealed 3143 and 2048 specifically expressed EST clones in the R and B LED treatments, respectively.
Thirty-seven genes related to floral induction and light quality signal transduction were isolated ( Table 2). For APETALA (AP) and CO-like genes, there was no difference in read number between the R and B light conditions. CRY-like genes encoding B light receptors did also not show any clear difference between these light quality treatments, but PHYA-like genes encoding R-FR light receptors showed higher expression under the B LED treatment as compared to the R LED irradiation. For ELF4-like genes, three different clones were identified. One of the EST clones (PhELF4-1) had higher number of reads under the B LED exposure than under the R LED (R 1 vs. B 14), whereas another (PhELF4-2) showed higher number of reads under the R light than the B light (R 4 vs. B 0). Furthermore, as the results of preliminary qRT-PCR analysis on three clones of ELF4 like genes using the same bulk samples, we found higher gene expressions under blue light than red light in PhELF4-1 and PhELF4-2 (Data not shown). From those results, we focused on PhELF4-1 and PhELF4-2 and tried to get their full length cDNA clones.

Identification of Two Full Length ELF4-Like Genes in Petunia
Two full length cDNA sequences of ELF4 like genes, PhELF4-1 (604 bp) and PhELF4-2 (724 bp), were isolated by RACE PCR (Supplemental Figure S2). Both clones have the DUF1313 protein domain (domain of unknown function), known from ELF4 genes of other plant species such as A. thaliana, pepper (Capsicum annuum) and tomato (Solanum lycopersicum) (Figure 1). The predicted amino acid sequence of PhELF4-1 is 76% and 65% similar to the ELF4 proteins of pepper and tomato, respectively. For PhELF4-2, the predicted amino acid sequence has 70% and 58% similarity with the corresponding sequences of pepper and tomato, respectively. However, there is no overall significant similarity between PhELF4-1 and PhELF4-2 and ELF4 protein in A. thaliana.

Gene Expression of ELF4-Like Genes in Petunia under Different Light Conditions
Towards the end of the second day of the light quality treatments (12 h into the 14 h photoperiod), the expression level of PhELF4-1 differed significantly (p = 0.0001) in petunia plants exposed to different light quality treatments, with approximately 5 times higher transcript levels under the W and B light than the R LED light (Figure 2). The ELF4-2 transcript levels did not differ significantly under the different light qualities; there was a slight tendency only of higher expression levels under the W and B light than under the R light.
The diurnal expression levels of PhELF4-1 and PhELF4-2 were also measured every 4 h during the first and 2nd day of exposure to the different light quality treatments. For both genes, the transcript levels differed significantly between the light quality treatments (p = 0.015 for ELF4-1, p = 0.02 for ELF4-2). In addition, the expression levels of both genes differed between different time points (p = 0.0001), and there was a significant interaction between light treatment and time point for ELF4-1 (p = 0.048), but not ELF4-2 (p = 0.55). Under all light sources, both genes generally showed low expression levels in the morning and the middle of the day, followed by increased expression towards the end of the day with peak expression 12 h after the start of the 14 h photoperiod (Figure 3). This was followed by a rapid decline to low expression levels that persisted or decreased further during the remaining night period (Figure 3). For PhELF4-1, the amplitude of the peak expression was significantly higher (p < 0.05) about three times) under the W fluorescent lamps and the B LED panels than under the R LED panels ( Figure 3A). In addition, the expression under the B light became stronger at the 2nd day of the experiment with 6 times higher expression levels under the B as compared to the R light, whereas the expression under the W light was similar the two consecutive days. The PhELF4-2 transcript level was also higher (p < 0.05) under the B and W light than under the R light but the difference (about 3-fold) was similar both days ( Figure 3B).  The expression levels of PhELF4-1 and PhELF4-2 genes were also measured at different developmental stages of petunia plants grown under the light quality treatments for 35 days (sampling 12 h into the 14 h photoperiod). In this study, the ELF4-1 expression level differed significantly between the treatments (p = 0.023) but this was not the case for ELF4-2 (p = 0.888). For both genes there were significant differences in transcript levels between different time points (p = 0.0001). Under all light qualities, the expression levels of PhELF4-1 and PhELF4-2 genes increased up to the 7th day of the treatments and then decreased until the end of experiment (Figure 4). At day 7, 14 and 21 the PhELF4-1 expression levels under the W fluorescent lamps and the B LED panels were significantly higher (p < 0.05) as compared to under the R LED panels, i.e., with the largest difference (about twice) in expression level at day 7. Thereafter the expression levels under the W and B light kept being high until the 21th and 28th day of the experiment, respectively, before declining to levels similar to under the R light exposure ( Figure 4A). For PhELF4-2 there was no such significant difference in expression levels among the light qualities and the expression levels were decreased already at the 14th day of the treatments ( Figure 4B). The diurnal expression pattern of the PhELF4-1 gene under the B LED panels was also compared during 3 consecutive days in petunia plants grown under a 14 h photoperiod and plants transferred to a 8 or 24 h photoperiod (i.e., 14 h/10 h, 8 h/16 h or 24 h/0 h day/night, respectively). The transcript level differed significantly between the photoperiods and the different time points, and a significant interaction between these two factors was observed (p = 0.0001 in all cases). Under the 14 h photoperiod treatment, the expression peak was observed 12 h into the irradiation period during all 3 days ( Figure 5A). However, when the seedlings were moved to a 8 h photoperiod, the timing of the peak expression was shifted. At the 1st day of the experiment, the expression peak was evident after 12 h (2 h into the night), but at the 2nd and 3rd day, the expression peaks were observed after 8 h of irradiation. Furthermore, when petunia plants were moved into continuous lighting, the expression peak was present at 12 and 36 h after start of the treatment. However, the peak levels gradually decreased, and the peak became less distinct at the 3rd day of the experiment. The plant growth and the percentage of blooming plants were also recorded after 4 weeks under the different photoperiods provided by B LED panels (Table 3)  * Each value shows averaged data (n = 6), and different letters within the same column show significant differences among treatments by Tukey's test (p < 0.05).

Discussion
In our previous studies [15,16], flowering of petunia (Petunia × hybrida) was shown to be delayed in plants grown under R LEDs and accelerated in plants grown under B LEDs. In addition, the presence of PehFT genes was reported, but their function in promotion of flowering is unclear since the expression of PehFT was not increased under the B light conditions [16]. In this study, we aimed to identify genes associated with the regulation of flowering time by light quality in petunia.
FR and B light are known to induce flowering in the quantitative long-day plant A. thaliana [1,12]. This is associated with absorption of light of different wavelengths by the photoreceptors PHYA, PHYB and CRYs, and their signaling pathways affecting the floral induction genes and proteins controlling flowering time [1,2]. In A. thaliana it is known that PHYB controls the expression of the floral induction gene FT by sending signals to PHYTOCHROME AND FLOWERING TIME1 (PFT1) [3]. Signals from CRY2 induces flowering by stabilizing the CO protein through CONSTITUTIVE PHOTOMORPHOGENIC1 (COP1) [5]. Furthermore, CRY2 also acts by binding to the CRYPTOCHROME-INTERACTING BASIC-HELIX-LOOP-HELIX1 (CIB1) protein and by directly inducing the expression of FT gene. In addition, not only CRY2, but another light-oxygen-voltage (LOV) domain photoreceptor, ZEITLUPE (ZTL), also mediates a signal of B light to prevent the CIB1 protein degradation in A. thaliana [18]. In this study, as a result of 3 EST high throughput sequencing, a number of photoreceptor and signal transduction genes were identified including PHYA, CRY, CO, ZTL, SOC1, FBP and AP. However, clear effects of different light qualities on expression levels (read numbers) could not be detected for those genes (Table 2). Furthermore, no light quality effect on PhFT was detected in this experiment. It may be possible that the bulked sample from different growth stages and young leaves and shoot apices used in this analysis could have resulted in dilution effects on gene expression in a specific organ such as the shoot apex. Furthermore, in this study, ELF-like genes were isolated from petunia. It was previously reported that ELF genes regulate the circadian rhythm [9,19]. In the circadian clock system in A. thaliana, different gene groups show different timing of their expression during the day and nighttime. CCA1 and LHY are expressed in the morning (known as the "morning loop" group I) [20,21], and LUX ARRHYTHMO (LUX), ELF3 and ELF4 are activated during the evening (known as the "evening loop") [8,22]. These genes are accordingly related to the circadian clock system in plants [23][24][25]. Furthermore, some proteins, for instance ELF4, ELF3 and LUX, form a complex body that can receive a signal from photoreceptors such as PHYB for regulation of GI expression [7]. Moreover, FLAVIN-BINDING, KELCH REPEAT, F-BOX1(FKF1) and GI are key genes for regulation of CO expression [26,27].
In A. thaliana, the ELF4 gene is one of the key genes involved in the core circadian clock for regulating various biological responses such as floral induction [7,28]. In this study, two types of ELF4-like genes, PhELF4-1 and PhELF4-2 were isolated. Although there is no overall significant similarity between the predicted PhELF4-1 and PhELF4-2 proteins in petunia and the ELF4 protein in A. thaliana, the two ELF4-like genes include a common domain DUF1313 that is also included in the ELF4 gene of A. thaliana [28]. Our analysis indicates that PhELF4 genes belong to the I(S/T/F)(K/R)V-type proteins of the DUF1313 family ( Figure 1). Furthermore, the size of the predicted PhELF4-1 and PhELF4-2 proteins are 4 to 5 times larger than corresponding genes in other plant species. Thus, it may be possible that PhELF4-1 and PhELF4-2 have other functions than other ELF4-like genes.
The expression patterns of PhELF4-1 and PhELF4-2 increased sharply and showed peak expression towards the end of the day. Such an expression pattern coincided with previously reported expression of ELF4-like gene in petunia [29]. The expression patterns of both PhELF4-1 and PhELF4-2 resemble the ELF4 gene expression in A. thaliana [30]. The expression of PhELF4-1 also responded to a change in photoperiod and showed a robust circadian rhythm both under short and long day conditions. The model of the circadian clock in A. thaliana is composed of interlocked feedback loops [28]. Li et al. [30] reported that the ELF4 gene expression is controlled by the coordinated action and interaction between constant positive transcription factors and periodic negative transcription factors. Particularly, CCA1 and LHY are essential for establishing and keeping the circadian change of the ELF4 gene expression. In addition, the FAR-RED ELONGATED HYPOCOTYL 3 (FHY3) protein associates with the PHYA protein in the signal transduction related to the regulation of the circadian clock. Our results indicate that the PhELF4-1 and PhELF4-2 genes probably have ELF4 function as part of the core system of the circadian clock in petunia. Under short-day conditions using B LED panels as the light source, the petunia plants did not show any flowering during the experimental period of four weeks (Table 3). On the other hand, the signal output from the core circadian clock system including PhELF4 could regulate the floral induction under long days. As described above, photoreceptors such as cryptochrome and phytochrome would then contribute to the induction of the ELF4 gene by stabilizing the ELONGATED HYPOCOTYL 5 (HY5) protein [7]. The signals from the photoreceptors contribute to sustain robust rhythms of the circadian clock system. Although the different light qualities did not affect the circadian rhythm in expression of PhELF4-1 and PhELF4-2 as such, their amplitude of expression was affected. The expression peaks towards the end of day was higher under the W fluorescent lamps and B LED panels than under the R LED panels ( Figure 3) and for PhELF4-1 high expression levels under W and B light were kept during 3-4 weeks ( Figure 4A). Thus, it is possible that B light is required to induce increased expression of ELF4-like genes in petunia.
Doyle et al. [9] have reported that mutation of AtELF4 changed the expression patterns of circadian clock genes and caused early flowering under short-day conditions in A. thaliana. In addition to the role of ELF4 genes in A. thaliana in photoperiod perception and circadian regulation involved in flowering time control, a role in phytochrome B-mediated de-etiolation has been demonstrated [31], but in other plant species the function of ELF4 genes is still unknown. Liew et al. [32] reported that the DIE NEUTRALIS (DNE) locus, an orthologue of AtELF4, in garden pea (Pisum sativum) inhibited flowering under short-day conditions. However, PsDNE may be involved in the circadian clock system and be related to PTOC1, PsLATE BLOOMER 1 (PsLATE1) and other components, but its function may be less central than AtELF4 [33]. Previously, we found that the flowering of petunia is delayed in plants grown under R LEDs and accelerated in plants grown under B LEDs [15,34]. In this experiment, early flowering under long photoperiods of W and B light is associated with elevated expression of PhELF4-1 and PhELF4-2, suggesting their involvement in acceleration of flowering. Such a role of these genes is antagonistic to the early flowering observed under noninductive short days in the elf4 mutant of A. thaliana. Thus, although there is no direct evidence revealing the relationship between flowering time and PhELF4-1 and PhELF4-2 genes, possibly these genes have crucial functions in the circadian clock system-related positive output signal for floral induction in petunia.

Conclusions
In summary, the gene isolated in this study, PhELF4-1 and PhELF4-2, have robust circadian rhythms and their expression is influenced by photoperiod and light quality with the highest expression under light spectra containing B light. Since these genes include the DUF1313 domain characteristic of circadian-clock-related ELF4-like genes in other plant species, it appears that the PhELF4-1 and PhELF4-2 are orthologues of such genes in spite of being considerably longer than and having low overall similarity with ELF4 in A. thaliana.