Daniel P. Woods , et al.
Citation: Woods DP, Li W, Sibout R, Shao M, Laudencia-Chingcuanco D, Vogel JP, et al. (2023) PHYTOCHROME C regulation of photoperiodic flowering via PHOTOPERIOD1 is mediated by EARLY FLOWERING 3 in Brachypodium distachyon. PLoS Genet 19(5):
Editor: Claudia Köhler, Max Planck Institute of Molecular Plant Physiology: Max-Planck-Institut fur molekulare Pflanzenphysiologie, GERMANY
Received: October 13, 2022; Accepted: March 17, 2023; Published: May 10, 2023
This is an open access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication.
Data Availability: All mutant lines are available upon request to the following email address: email@example.com (Department of Biochemistry, University of Wisconsin) with no restrictions.
Funding: This work was funded in part by the US National Science Foundation (Award IOS-1755224 to RMA) and by the Great Lakes Bioenergy Research Center, US Department of Energy, Office of Science, Office of Biological and Environmental Research (Award No. DE-SC0018409). The work from JPV, RS, and MS (proposal: 10.46936/10.25585/60001041) conducted by the U.S. Department of Energy Joint Genome Institute (https://ror.org/04xm1d337), a DOE Office of Science User Facility, is supported by the Office of Science of the U.S. Department of Energy operated under Contract No. DE-AC02-05CH11231. DPW was a Howard Hughes Medical Institute (HHMI) Fellow of the Life Sciences Research Foundation which supported the work while in Jorge Dubcovsky’s lab and paid his salary. JD was funded by HHMI. DLC is funded by USDA-ARS CRIS Project 2030-21430-014D. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
The transition from vegetative growth to flowering is an important developmental decision for which the timing is often directly influenced by the environment (e.g. [1–4]). This critical life history trait has been shaped over evolutionary time to enable reproduction to coincide with the time of year that is most favorable for flower and seed development. Moreover, breeding to adjust the timing of flowering in crops has been critical for adapting various crop varieties to changing environments and to increase yield (e.g. ).
In many plant species, changes in day-length and/or temperature provide seasonal cues that result in flowering during a specific time of year [1,6]. Many temperate grasses such as Brachypodium distachyon (B. distachyon), wheat, and barley that flower in the spring or early summer months in response to increasing day-lengths are referred to as long-day (LD) plants . B. distachyon is closely related to the core pooid clade comprising wheat, oats, barley, and rye and has a number of attributes that make it an attractive grass model organism suitable for developmental genetics research [8,9].
Variation in the LD promotion of flowering in temperate grasses such as wheat and barley can be due to allelic variation at PHOTOPERIOD1 (PPD1), a member of the pseudo-response regulator (PRR) gene family (PPD1 is also known as PSEUDO RESPONSE REGULATOR 37;PRR37) [10,11]. Natural variation in PPD1 resulting in either hypomorphic alleles as found in barley or dominant PPD1 alleles as found in tetraploid or hexaploid wheat impacts flowering [10–15]. Specifically, natural recessive mutations in the conserved CONSTANS, CONSTANS-LIKE and TIMING OF CAB EXPRESSION 1 (CCT) putative DNA binding domain in the barley PPD1 protein cause photoperiod insensitivity and delayed flowering under LD [11, 12], whereas wheat photoperiod insensitivity is linked to overlapping large deletions in the promoter region of PPD1 in either the A  or D genome homeologs . These deletions result in elevated expression of PPD1, particularly during dawn, causing rapid flowering even under non-inductive SD conditions . It is worth noting that although these wheat lines are referred to as photoperiod insensitive (PI) varieties they still flower earlier under LD than under SD if the timing of flowering is measured as the emergence of the wheat spike (heading time) . It has been hypothesized that the large deletion within the PPD1 promoter might remove a binding site for one or more transcriptional repressors . To date, natural variation studies of flowering in B. distachyon have not pointed to allelic variation at PPD1 and thus its role in LD flowering in B. distachyon is not known [17–21].
Variation in EARLY FLOWERING 3 (ELF3; also known as mat and eam) impacts photoperiodic flowering in grasses, including wheat [22,23], barley [24,25], and rice . In these plants, natural variation in ELF3 allows growth at latitudes that otherwise would not be inductive for flowering, enabling these crops to be grown in regions with short growing seasons . For example, early maturity (eam) loci have been used by breeders to allow barley to grow at higher latitudes in regions of northern Europe with short growing seasons [24,27]. The eam8 mutant in the barley ortholog of ELF3, is a loss-of-function mutation that accelerates flowering under SD or LDs [24,25] similar to elf3 loss-of-function alleles described previously in the eudicot model Arabidopsis thaliana (A. thaliana) . Moreover, loss of function of ELF3 in B. distachyon also results in rapid flowering under SD and LD, and expression of the B. distachyon ELF3 protein is able to rescue the A. thaliana elf3 mutant, demonstrating a conserved role of ELF3 in flowering across angiosperm diversification [29–31].
Work in A. thaliana has shown that ELF3 is an important component of the circadian clock that acts as a bridge protein within a trimeric protein complex that also contains LUX ARRTHYHMO (LUX), and EARLY FLOWERING 4 (ELF4) and is referred to as the evening complex (EC) . Loss-of-function mutations in any of the proteins that make up the EC results in disrupted clock function and rapid flowering [33–36]. The peak expression of the EC at dusk is involved in the direct transcriptional repression of genes that make up the morning loop of the circadian clock including A. thaliana PRR7 and PRR9, which are paralogs of grass PPD1 and PRR73 [37–39]. Recently, it has been shown that the EC also directly represses PRR37, PRR95, and PRR73 in rice (PRR37 is the rice ortholog of PPD1), indicating conservation of the role of the EC across flowering plant diversification . Furthermore, elf3 mutants in barley, wheat, and B. distachyon have elevated PPD1 expression [23,24,30] indicating ELF3 may impact flowering in part via PPD1, but to what extent remains to be determined.
The photoperiod and circadian pathways converge in the transcriptional activation of florigen/FLOWERING LOCUS T1 (FT1) in leaves [6,41]. In temperate grasses, PPD1 is required for the LD induction of FT1, whereas in A. thaliana CONSTANS (CO) is the main photoperiodic gene required for FT1 activation in LD [16,42,43]. There are two CO-like genes in temperate grasses. Interestingly, in the presence of functional PPD1, co1co2 wheat plants have a modest earlier heading phenotype suggesting they are in fact mild floral repressors, but in the absence of PPD1, CO1 acts as a flowering promoter under LD . To date, no null co1co2 double mutants have been reported in B. distachyon. However, RNAi knock-down of co1 results in a 30-day delay in flowering under 16h LD  and overexpression of CO1 leads to earlier flowering in SD . These results indicate that in B. distachyon CO1 has a promoting role in flowering even in the presence of a functional PPD1 gene, and suggest potential differences in the role of CO1 in the regulation of flowering between B. distachyon and wheat.
Once FT1 is activated by LD it interacts with the bZIP transcription factor FD which triggers the expression of the MADS-box transcription factor VERNALIZATION1 (VRN1) [45,46]. VRN1 in turn upregulates the expression of FT1 forming a positive feedback loop that overcomes the repression from the zinc finger and CCT domain-containing transcription factor VERNALIZATION2 (VRN2) [17,47–50]. The FT1 protein is then thought to migrate from the leaves to the shoot apical meristem, as shown in A. thaliana and rice [51,52], to induce the expression of floral homeotic genes including VRN1, thus converting the vegetative meristem to a floral meristem under favorable LD photoperiods.
Light signals are perceived initially by photoreceptors that initiate a signal transduction cascade impacting a variety of developmental responses to light . The sensing of light is accomplished by complementary photoreceptors: phytochromes that perceive the ratio of red and far-red light, cryptochromes, phototropins, and Zeitlupe family proteins that detect blue light, and UV RESISTANCE LOCUS 8 that detects ultraviolet B light [54,55]. The phytochromes form homodimers that upon exposure of plants to red light undergo a confirmation shift to an active form causing the activation of a suite of downstream genes . Exposure of plants to far-red or dark conditions causes reversion of the phytochromes to an inactive state .
There are three phytochromes in temperate grasses referred to as PHYTOCHROME A (PHYA), PHYTOCHROME B (PHYB), and PHYTOCHROME C (PHYC) . Functional analyses of these phytochromes in temperate grasses revealed that PHYB and PHYC play a major role in the LD induction of flowering because loss-of-function mutations in either of these genes results in extremely delayed flowering [57–59] whereas loss-of-function mutations in PHYA in B. distachyon results in only a modest delay of flowering under inductive LD, indicating PHYB and PHYC are the main light receptors required for photoperiodic flowering in temperate grasses .The important role of PHYC in photoperiodic flowering is not universal as loss of phyC function in A. thaliana and rice only has small effects on flowering [60,61].
In temperate grasses, PHYB and PHYC are required for the transcriptional activation of a suite of genes involved in the photoperiod pathway, including PPD1, CO1, and FT1, and ectopic expression of FT1 in the B. distachyon phyC background results in rapid flowering—a reversal of the phyC single-mutant, delayed-flowering phenotype [57,58,62]. Moreover, consistent with PHYB/C acting at the beginning of the photoperiodic flowering signal cascade, expression of genes encoding components of the circadian clock are also severely dampened in the phyB and phyC mutant backgrounds [29,57–59]. An exception to this is that the expression of ELF3 is not altered in the temperate grass phytochrome mutants [29,58,59]. Recently, in B. distachyon it has been shown that PHYC can interact with ELF3, and this interaction destabilizes the ELF3 protein indicating that the regulation of ELF3 by PHY is at least in part at the protein level consistent with previous studies from A. thaliana, rice, and the companion study in wheat [23,29,40,63–66]. At present it is not clear to what extent the regulation of ELF3 by PHYs is critical for photoperiodic flowering.
Here, we show in B. distachyon by analyzing phyC/efl3 double mutant plants that indeed the light signal perceived by phytochromes is mediated through ELF3 for photoperiodic flowering. The extreme delayed flowering of the phyC mutant disappears in the phyC/elf3 double mutant which flower as rapidly as the elf3 single mutant. Moreover, the expression profiles of genes in the photoperiod pathway are similar between elf3 and phyC/elf3 mutants compared to phyC mutants. Thus, elf3 is completely epistatic to phyC at the phenotypic and molecular levels. Furthermore, we show strong, environment-dependent genetic interactions between ELF3 and PPD1, which indicates that PPD1 is a main target of ELF3-mediated repression of flowering. These results provide a genetic and molecular framework to understand photoperiodic flowering in the temperate grasses.
Rapid flowering of elf3 is epistatic to the delayed flowering of phyC
Previous studies in B. distachyon showed that PHYC can affect the stability of the ELF3 protein, and that the transcriptome of a phyC mutant resembles that of a plant with elevated ELF3 signaling . Thus, it has been suggested that the extreme delayed flowering phenotype of the phyC mutant  could be mediated by ELF3 . To test the extent to which the translation of the light signal perceived by PHYC to control flowering is mediated by ELF3, we generated elf3/phyC double mutant plants and evaluated the flowering of the double mutant relative to that of elf3 and phyC single mutants as well as Bd21-3 wild type under 16h-LD and 8h-SD (Fig 1).
The rapid flowering of the elf3 mutant is epistatic to the delayed flowering of the phyC mutant (A) Representative images of Bd21-3 wild-type, elf3, phyC and elf3/phyC double mutant plants grown in a 16h photoperiod at 90d after germination. Bar = 17cm. (B, D) Flowering times under 16h (B) or 8h daylengths (D) measured as days to heading of Bd21-3, elf3, phyC, and elf3/phyC. (C) Flowering phenotypes under 16h (C) or 8h daylengths (E) measured as the number of leaves on the parent culm at time of heading. Bars represent the average of 8 plants ± SD. Arrows above bars indicate that none of the plants flowered at the end of the experiment (150d). Letters (a, b) indicate statistical differences (p < 0.05) according to a Tukey’s HSD test used to perform multiple comparisons.
Under both 16h LD and 8 SD photoperiods, we found that elf3 is epistatic to phyC. Specifically, in LD elf3/phyC double mutants flowered rapidly by 38 days with 6.9 leaves similar to elf3 mutants that flowered by 34 days with 6.6 leaves (Fig 1A and 1B). In contrast, phyC mutants had not flowered after 150 days with greater than 20 leaves when the experiment was terminated, and Bd21-3 wild-type flowered by 72 days with 12 leaves consistent with previous studies [49,58]. In 8h SD, elf3/phyC double mutants also flowered rapidly by 54 days with 8.8 leaves similar to elf3 mutants that flowered by 48 days with 8.5 leaves (Fig 1D and 1E). In contrast, both Bd21-3 wild-type and phyC mutants had not flowered by 150 days with >18 leaves when the experiment was terminated (Fig 1D and 1E). These results indicate that the extreme delayed flowering mutant phenotype of phyC in B. distachyon is mediated by ELF3.
To determine if PHYC affects the expression of ELF3, we analyzed ELF3 mRNA levels across a diurnal light cycle (16h light and 8h dark). There were no significant differences in ELF3 expression at any time point in the phyC mutant relative to wildtype (S1 Fig) indicating that PHYC does not affect the transcriptional profile of ELF3 in B. distachyon consistent with results from A. thaliana and wheat (67, 69).
Effect of mutations in PHYC and ELF3 on the transcriptional profiles of flowering time genes
To further understand how PHYC and ELF3 affect flowering at a molecular level, we compared the mRNA levels of B. distachyon orthologs of the photoperiod and vernalization pathway genes FT1, VRN1, PPD1, VRN2, CO1, and CO2 across a diurnal cycle in 16h LD in the phyC and elf3 single mutants versus the elf3/phyC double mutant (Fig 2). We were particularly interested in determining how the expression profiles of “flowering-time genes” in the elf3/phyC double mutant compared to the elf3 and phyC single mutant. The newly expanded fourth leaf was harvested for gene expression analyses because at this developmental stage in 16h daylengths the meristems of all of the plant genotypes are at a vegetative stage and thus are developmentally equivalent. Consistent with the rapid flowering of the elf3 and elf3/phyC mutants, the mRNA expression levels of FT1 and VRN1 in these lines are significantly higher than the levels in wild-type and phyC mutants across all the time points tested (Fig 2A and 2D). Moreover, the overall expression profiles of FT1 and VRN1 in the elf3 and elf3/phyC mutants were similar throughout the day. This is in contrast to the phyC mutant in which FT1 and VRN1 mRNA levels were lower than wild type throughout the day consistent with the delayed flowering phenotype of phyC. Although the elevated levels of FT1 and VRN1 in both elf3 and elf3/phyC are consistent with their rapid flowering, the expression of the floral repressor, VRN2, exhibits a similar elevated expression profile throughout the day in both elf3 and elf3/phyC relative to wild-type or phyC single-mutants (Fig 2E). The elevated VRN2 expression levels in elf3 mutant plants are consistent with previous results in B. distachyon and other grasses [23,29,30,40,64]. The transcriptional profile of CO1 was similar in both the elf3 and elf3/phyC mutants with elevated expression compared to wild-type between zt4-8 and then lower than wild-type between zt12-20 (Fig 2C). A similar expression pattern was found for Hd1 (the rice CO homolog) in the elf3-1/elf3-2 double mutant in rice . Consistent with previous reports, CO1 expression levels remained low in Brachypodium phyC mutants throughout a diurnal cycle . By contrast CO1 expression is increased in the phyC mutants in wheat  indicating another difference in the regulation of CO1 between these two species. Lastly, the CO2 expression profiles were similar between wild-type, elf3, and elf3/phyC, whereas CO2 mRNA levels were lower in phyC throughout a diurnal cycle (Fig 2F). In summary, the transcriptional profiles of FT1, VRN1, VRN2, CO1, and CO2 are similar between elf3 and elf3/phyC mutants consistent with ELF3 acting downstream from PHYC in the photoperiod flowering pathway.
Fig 2. Effect of loss-of-function mutations in ELF3 and PHYC on the transcriptional profiles of six flowering time genes in 16h LD.
Normalized expression of (A) FT1, (B) PPD1, (C) CO1, (D) VRN1, (E) VRN2, and (F) CO2 during a 24h diurnal cycle in Bd21-3 (black line), elf3 (blue line), phyC (gray line) and elf3/phyC double mutant (orange line). Plants were grown in LDs until the fourth-leaf stage was reached (Zadoks = 14) at which point the newly expanded fourth leaf was harvested at zt0, zt4, zt8, zt12, zt16, and zt20. Note the zt0 value and zt24 value are the same. The average of four biological replicates is shown (three leaves per replicate). Error bars represent standard deviation of the mean. Data were normalized using UBC18 as done in .
The transcriptional profile of PPD1 indicates a more complex interaction between PHYC and ELF3. In wild type, the expression levels of PPD1 peak at zt12 with the lowest expression level at dawn and during the evening consistent with previous reports of PPD1 expression patterns in B. distachyon [29,30] (Fig 2B). In both the elf3 and elf3/phyC mutants, we observed increased PPD1 expression relative to wild-type at dawn and during the evening with expression levels similar to wild-type at zt12. Interestingly, the increased expression of PPD1 observed at dawn and during the night in the elf3/phyC background was significantly lower than that of the elf3 single mutant suggesting PHYC may impact PPD1 expression via additional genes beyond ELF3. In contrast, PPD1 expression levels were reduced in the phyC mutant relative to wild type, elf3, and elf3/phyC mutants throughout a diurnal cycle, consistent with the reduced FT1 expression and delayed flowering phenotype of the phyC mutant.
Identification and mapping of a ppd1 mutant in B. distachyon
To determine the role of PPD1 in flowering in B. distachyon, the genome-sequenced, sodium-azide mutant line NaN610 with a predicted high-effect mutation impacting a splice acceptor donor site in PPD1 (BdiBd21-3.1G0218200) was obtained from the Joint Genome Institute (JGI) (; https://phytozome-next.jgi.doe.gov/jbrowse/). A quarter of the NaN610 M3 seeds received were segregating for an extremely delayed flowering phenotype (Fig 3B–3D).
Due to the high mutant load of these NaN mutant lines, we validated through mapping that the delayed flowering phenotype is associated with PPD1 (Fig 3E and 3F). We backcrossed NaN610 with Bd21-3 and confirmed a quarter of the plants in the BC1F2 population (n = 380) were delayed flowering, demonstrating the recessive nature of the mutant. Three Derived Cleaved Amplified Polymorphic Sequences (dCAPs) markers closely linked with PPD1 were developed based on the variant’s information for the NAN610 line, with one of the dCAPs primers located within the PPD1 locus itself (Fig 3E and S1 Table). This approach allowed us to map the causative lesion to within a 1Mb interval (13.1Mb-14.2Mb) on the top arm of chromosome 1, demonstrating the delayed flowering phenotype is tightly linked with PPD1 (Fig 3E).
Fig 3. Identification of a ppd1 mutant.
(A) Gene structure of PPD1 showing the location of the nucleotide change of the sodium azide-induced mutation; orange bar indicates the region that encodes the CCT domain. Below the gene structure diagram is a gel image of the reverse transcription polymerase chain reaction (PCR) (30 cycles of amplification) showing PCR products of PPD1 cDNA in Bd21-3 and ppd1 mutant plants. The location of primers used in each reaction are shown in the diagram above the gel image. (B) Representative photo of Bd21-3, heterozygous, and homozygous ppd1 plants grown in a 20h LD. Picture was taken 60d after germination in 20h LD, bar = 5cm. (C and D) Flowering time was measured as days to heading (C) and the number of leaves on the parent culm at time of heading (D), ** indicates statistical differences (p < 0.01), *** indicates statistical differences (p < 0.001) by Student’s t-test. (E) Fine mapping of ppd1 in a population of 380 BC1F2 individuals. Individuals with seven different haplotypes were identified by three dCAPS markers and flowering times of each haplotype were determined in the F3 generation. Black, grey, and light grey rectangles represent NAN610, heterozygous, and Bd21-3 genotypes, respectively. Variants around the PPD1 locus from the NAN610 line are shown with black dots, and yellow arrows indicate the coding genes within the mapped interval with the specific effect on the coding region indicated.
To confirm that the predicted splice site mutation does in fact impact the splicing of PPD1, we sequenced the mRNA products of the ppd1 NaN610 mutant line and Bd21-3 (Fig 3A). We found that the splice site mutation resulted in the mis-splicing of the sixth intron, generating a reading frame shift resulting in a truncated protein lacking the conserved CCT domain (Fig 3A). The extremely delayed flowering of the B. distachyon ppd1 mutant is consistent with the ppd1 null mutants described in wheat, which take >120 days to head under inductive LD conditions [16,43], demonstrating PPD1 is required for LD flowering broadly within temperate grasses.
Genetic interactions between ELF3 and PPD1 under long and short days
We and others have shown that PPD1/PRR37 expression is increased in an elf3 mutant background in B. distachyon, rice, and wheat (Fig 2B; [29,30,40,66]). Moreover, a CHIPseq analysis of ELF3 demonstrated that PPD1 is directly bound by ELF3 in a time-of-day-responsive manner [29,40]. Thus, ELF3 acts as a direct transcriptional repressor of PPD1 but the extent to which this explains the rapid flowering in the elf3 mutant has not been tested. Therefore, we generated an elf3/ppd1 double mutant to explore the genetic interactions of these two genes under a highly inductive 20h LD, inductive 16h LD, and non-inductive 8h SD (Fig 4).
Fig 4. Genetic interactions between the delayed flowering ppd1 mutant and the rapid flowering elf3 mutant.
Representative image of Bd21-3 wild-type, rapid flowering elf3 mutant, delayed flowering ppd1 mutant, and delayed flowering elf3/ppd1 double mutant grown in a 20h photoperiod (A), 16h photoperiod (D), and 8h photoperiod (G). Picture was taken after 110d, for the 20h LD (A) and 140d after germination for the 16h LD and (D) 8h SD. Scale bar = 5cm. (B, E, H) Flowering times under 20h (B), 16h (E), 8h (G) measured as days to heading of Bd21-3, elf3, ppd1, and elf3/ppd1. Flowering times under 20h (C), 16h (F), and 8h (I) measured as the number of leaves on the parent culm at time of heading. The 8h experiment was repeated three times. The first experiment resulted in ppd1 plants that stopped producing new leaves before wild type. One possibility for the cessation of new leaf production in ppd1 plants in this experiment is that the meristem transitioned to flowering, but then did not proceed to heading. However, in two subsequent experiments ppd1 plants continually produced new leaves for the duration of the experiment similar to wild type and this data is shown in (I). Data for all three experiments are shown in S1 Data for Fig 4. When grown under non-inductive conditions for 120 days or more, a few B. distachyon plants flower; we consider this a stochastic flowering response because the majority of plants do not flower. Bars represent the average of 8 plants ± SD. Arrows above bars indicate that none of the plants flowered at the end of the experiment (150d, >20 leaves). Letters (a, b, c, d) indicate statistical differences (p < 0.05) according to a Tukey’s HSD test used to perform multiple comparisons.
Under all photoperiods, the elf3/ppd1 double mutant flowered significantly later than the elf3 single mutant (Fig 4). Interestingly, under 20h LD, the elf3/ppd1 double mutant flowered earlier than ppd1 by 16.2 days forming 3.0 fewer leaves whereas under 16h days elf3/ppd1 mutant flowered significantly later than ppd1 by 13.7 days with 2.5 more leaves. In 8h SD, only elf3 mutant plants were able to flower; Bd21-3, ppd1, and elf3/ppd1 all failed to flower by the end of the experiment. It is also worth noting that elf3/ppd1 double mutants are still able to respond to different photoperiods, with longer days resulting in significantly earlier flowering plants than under shorter days (Fig 4B and 4E and 4H). These results indicate that there are strong genetic interactions between ELF3 and PPD1 under different photoperiods, that PPD1 is a key flowering regulator downstream of ELF3, and that there is a residual photoperiodic response that is independent of these two genes.
Effect of mutations in ELF3 and PPD1 on the transcriptional profiles of flowering time genes
To understand how ELF3 and PPD1 affect flowering at a molecular level, we measured the mRNA levels of FT1, VRN1, PPD1, VRN2, CO1, and CO2 in the elf3 and ppd1 single mutants and the elf3/ppd1 double mutant across a diurnal cycle in 16h LD (Fig 5). As noted before, FT1 and VRN1 expression levels were elevated in the elf3 mutant background; however, in the elf3/ppd1 double mutant, expression of these genes remained low and resembled the expression profile of ppd1 single mutants (Fig 5A and 5D). The low expression levels of FT1 and VRN1 in ppd1 and ppd1/elf3 mutants is consistent with the delayed flowering phenotype of both of these mutants in 16h LD. The VRN2 expression profile was similar between wild type and ppd1 mutant plants with low expression levels at dawn and increased expression throughout the light cycle before expression levels dropped in the dark (Fig 5E). Interestingly, VRN2 expression levels are similarly elevated throughout the day in elf3 and elf3/ppd1 mutants compared to wild type (Fig 5E).
Fig 5. Effect of loss-of-function mutations in ELF3 and PPD1 on the transcriptional profiles of six flowering-time genes in 16h LD.
The fourth newly expanded leaves were harvested every 4h over a 24-hour period; three biological replicates (two leaves per replicate) were harvested at each time point for each genotype. Diurnal expression of FT1 (A), PPD1 (B), CO1 (C), VRN1 (D), VRN2 (E), and CO2 (F) were detected in Bd21-3 (black line), elf3 (blue line), ppd1 (grey line) and elf3/ppd1 double (orange line). Bars represent the average of three biological replicates ± SD. Letters (a, b, c, d) indicate statistical differences (p < 0.05) according to a Tukey’s HSD test used to perform multiple comparisons, letter color corresponds to the four different genotypes. The black, gray, and orange lines overlap given the scale used to show ELF3 expression in the same graph; the orange line is arbitrarily shown on top. Specific expression values are shown below the lettered statistical test. Raw data is included in S1 Data file.
Consistent with the expression patterns of PPD1 in wild type and elf3 shown in Fig 2, the expression levels of PPD1 peak at zt12 in wild-type and the elf3 mutant has increased PPD1 expression relative to wild type at dawn and during the evening (Fig 5B). PPD1 expression levels in the ppd1 mutant should be interpreted with caution because we do not know the effect of the splice site mutation on the mRNA stability. Significantly higher levels of PPD1 expression were observed in ppd1 relative to wild type at ZT8 and ZT16, and in elf3/ppd1 relative to elf3 at dawn. However, the expression patterns of PPD1 were most similar between ppd1 and wild type and between elf3/ppd1 and elf3 (Fig 5B).
CO1 and CO2 expression both exhibit peak expression in wild type at zt12 with expression dampening in the evening consistent with previous reports [29,58]. Interestingly, expression levels of CO1 and CO2 were elevated between zt4-8 in elf3 compared with wild-type. However, at zt16 and zt20, expression levels were similar in wild type, elf3, and elf3/ppd1 mutants, whereas at zt12 the expression of CO1 was reduced in elf3 compared to wild type. In contrast, the expression levels of CO1 and CO2 were lowest in ppd1 compared to the other lines at zt8. In the elf3/ppd1 mutants, CO1 and CO2 expression was most similar to ppd1 in the morning and most similar to elf3 in the evening (Fig 5C and 5F). These results indicate complex interactions between PPD1 and ELF3 in the regulation of CO1 and CO2.
Constitutive expression of ELF3 results in delayed flowering and lower PPD1, FT1, and VRN1 expression levels
In our previous study, we showed that overexpression of ELF3 in the elf3 mutant background results in strongly delayed flowering (, Fig 6A). However, this was done in the T0 generation, so we evaluated the flowering time and expression of downstream flowering-time genes in the T1 generation. We grew four UBI::ELF3/elf3 transgenic lines alongside Bd21-3 and elf3 in a 16h photoperiod, and harvested the newly expanded fourth leaf at zt4. This time point was chosen because expression of several critical genes such as CCA1, TOC1, LUX, PPD1, VRN2, CO1, and CO2 were significantly different in the morning in the elf3 single mutant compared with wild-type [29,30]; Figs 2 and 5). We first confirmed that all of the UBI::ELF3/elf3 transgenic lines had elevated ELF3 mRNA levels, and found indeed there is a significant increase of ELF3 expression in the transgenic lines (Fig 6C). To understand how UBI::ELF3/elf3 affects flowering, we evaluated expression levels of FT1, VRN1, PPD1, VRN2, CO1, and CO2 in wild type, elf3, and UBI::ELF3/elf3. Consistent with the delayed flowering, FT1 and VRN1 expression levels of UBI::ELF3/elf3 were reduced relative to wildtype compared to elevated levels elf3 relative to wildtype (Fig 6D and 6G). Also, the expression of PPD1, VRN2, CO1, and CO2 were decreased in UBI::ELF3/elf3, indicating ELF3 is playing a broad repressive role in regulating CCT domain containing genes responding to photoperiodic flowering.
Fig 6. Overexpression of ELF3 in the elf3 mutant delays flowering.
(A) Representative image of Bd21-3 wild type, elf3, and three independent transgenic lines of UBI::ELF3 in the elf3 background grown in a 16h photoperiod. Images were taken 120d after germination. Bar = 5 cm. The fourth newly expanded leaves were harvested at zt4 in 16h. (B–I), Normalized expression of ELF3 (C), FT1 (D),PPD1 (E), CO1 (F), VRN1 (G), VRN2 (H) in Bd21-3 wild type, elf3, and three UBI::ELF3/elf3 transgenic lines. Expression of CO2 is shown in S2 Fig. Bars represent the average of four biological replicates ± SD.
The phytochromes PHYC/PHYB and ELF3 connection
B. distachyon has an obligate requirement for LD to flower [8,49,67]. Previous studies have shown the important roles that both PHYC and ELF3 play in photoperiodic flowering in B. distachyon [29,30,58]. Specifically, mutations in phyC result in extremely delayed flowering whereas loss-of-function mutations in elf3 result in rapid flowering in either LD or SD [30,58]. Furthermore, phyC mutants resemble plants grown in SD both morphologically and at the transcriptomic level regardless of day-length whereas elf3 mutants resemble plants grown in LD both morphologically and at the transcriptomic level regardless of day-length [29,30,58]. Thus, we were interested in exploring the genetic relationships between PHYC and ELF3. The extreme delayed flowering phenotype observed in phyC mutant plants is mediated by ELF3 because phyC/elf3 double mutants flower rapidly in LD and SD similar to elf3 mutants. Similar genetic interactions between phyB and elf3 were also found in wheat in the companion study , suggesting these interactions are likely to be conserved broadly in temperate grasses. Loss-of-function mutations in phyB in wheat also result in delayed flowering similar to phyC . At present, no null phyB alleles have been reported in B. distachyon; however, PHYB is able to heterodimerize with PHYC in B. distachyon and wheat [29, 57], and both phyB  and phyC  mutants are extremely late flowering in wheat suggesting that both PHYs are likely required for photoperiodic flowering in the temperate grasses, perhaps because PHYB/PHYC heterodimers are required for flowering regulation.
Phytochrome regulation of ELF3 at the post-translational level rather than at the transcriptional level is likely to be the critical interaction impacting flowering. In A. thaliana, B. distachyon, and wheat, phyB/phyC mutants do not impact the circadian oscillation of ELF3 mRNA levels [29,59,68]. However, in all three species PHYB and PHYC have been shown to interact with the ELF3 protein, but the stability of the ELF3 protein upon exposure to light differs between A. thaliana and temperate grasses [29,63,66,69]. Specifically, in A. thaliana, PHYB contributes to the stability of the ELF3 protein during light exposure leading to ELF3 accumulation at the end of the day [63,68], whereas in rice ELF3 is degraded and or modified during light exposure in a PHY-mediated process . In temperate grasses, ELF3 protein accumulates during the night and is rapidly degraded or modified upon light exposure [29,67], and this is likely to be a PHY mediated response as well.
The differences in how phytochromes impact the stability of the ELF3 protein might explain the contrasting flowering phenotypes of the phyB/phyC mutants between A. thaliana and temperate grasses. In A. thaliana, phyB mutants flower more rapidly than wild type in either LD or SD and phyC mutants flower earlier under SD , whereas in temperate grasses phyB or phyC mutants are extremely delayed in flowering [57–59]. However, ELF3 acts as a flowering repressor in both A. thaliana and grasses [30,32]. In A. thaliana PHYB stabilizes the ELF3 protein; therefore, in phyB mutants, ELF3 is no longer stable leading to rapid flowering. In contrast, in temperate grasses and rice, in the absence of phyB or phyC the ELF3 protein is more stable leading to delayed flowering.
Interestingly, overexpression of ELF3 results in extremely delayed flowering in B. distachyon [29,30] (Fig 6A and 6B). Given that the regulation of ELF3 appears to occur at the protein level, one might not expect that overexpression would cause such a strong flowering delay. However, if the ELF3 protein is expressed at a high level such that the degradation machinery is unable to degrade much of the ELF3 protein during LD, then a strong flowering delay might occur. In support of this idea, the delayed flowering of overexpression of ELF3 is mitigated when plants are grown under constant light versus 16h LD . It is worth noting that although overexpression of ELF3 generally leads to delayed flowering in different plant species, there is considerable variation in the magnitude of this delayed flowering [29,63,66] (Fig 6A and 6B).
Similar genetic interactions between ELF3 and PHYB have also been observed in rice which is a SD-flowering plant that has two rice-specific ELF3 paralogs . Mutations in either paralog results in delayed flowering in SD or LD in contrast to the rapid flowering observed in temperate grasses containing elf3 mutations [40,71,72]. Also in contrast to the situation in temperate grasses, phyB mutants flower more rapidly than wild type in rice . Despite the flowering differences of the elf3 and phyB mutants between rice and temperate grasses, the flowering phenotype of phy mutants is ELF3 mediated because in both rice and temperate grasses elf3 is epistatic to phyB or phyC [Fig 1; 40, 67]. Moreover, PHYB and ELF3 proteins interact impacting the modification of ELF3 by light . The opposite roles that phytochromes and elf3 have on flowering in rice and temperate grasses is likely due, at least in part, to the reverse role that the downstream PPD1/PRR37 gene has on flowering. PPD1 is a promoter of flowering in LD temperate grasses but is a repressor of flowering in SD grasses such as rice [11,16,74,75] (Fig 3).
The ELF3 and PPD1 connection
The extremely delayed flowering of B. distachyon ppd1 mutant plants under LD is similar to the extremely delayed heading of ppd1 mutants in wheat . However, a previous study in B. distachyon using a CRISPR induced ppd1 mutant allele which has a 1bp deletion in the sixth exon of PPD1 has a milder delayed flowering phenotype with plants taking around 40 days to flower under 20h LD, whereas the mutant ppd1 plants presented here flower around 120 days in 20h LD ; Figs 3 and 4). In both studies, wild-type Bd21-3 plants flower on average between 25–30 days in 20h LD consistent with previous reports in B. distachyon [19,49,76]. The differences in flowering time between the two B. distachyon ppd1 mutant alleles suggests that the CRISPR induced ppd1 allele is a weaker hypomorphic allele than the ppd1 mutant allele characterized in this study. This is further supported by the fact that the ppd1 allele described here has an extremely delayed flowering phenotype similar to the null ppd1 wheat allele .
The ppd1/elf3 double mutant is delayed in flowering relative to elf3 mutant plants indicating that PPD1 is downstream of ELF3 in photoperiodic flowering. This is also consistent with the elevated PPD1 expression levels observed at dawn and dusk in the elf3 mutant relative to wild-type in temperate grasses [30,70] (Fig 5). Indeed, ELF3 binds to the PPD1/PRR37 promoter in B. distachyon, wheat, and rice indicating ELF3 is a direct transcriptional repressor of PPD1/PRR37 in grasses [29,40,66]. ELF3 does not have any known DNA binding activity and thus, the direct repression is likely to be due to ELF3’s interaction with a LUX transcription factor which, from studies in A. thaliana, recognizes GATWCG motifs that are also found in the PPD1 promoter in grasses [37,38,67]. Interestingly, photoperiod insensitivity in wheat is associated with deletions in the PPD1 promoter, which remove the LUX binding site and results in elevated PPD1 expression at dawn similar to the PPD1 expression dynamics observed in elf3 and lux mutant plants [10–13,39,77,78]. In the companion wheat paper, ChIP-PCR experiments show ELF3 enrichment of the DNA region around the LUX binding site in the PPD1 promoter, which is present within the region deleted in photoperiod-insensitive wheats. These results demonstrate that removal of the evening complex binding site leads to elevated expression in PPD1 and accelerated heading under SD in many photoperiod-insensitive wheats .
The characterization of elf3/ppd1 mutant plants under different photoperiods reveal complex interactions between the two genes and their downstream targets depending on the environment. For example, elf3/ppd1 mutant plants are earlier flowering than ppd1 mutant plants under 20h day-lengths, but are later flowering under 16h and 8h day-lengths. This is in contrast to elf3/ppd1 mutants in wheat, which head earlier than ppd1 under a 16h day-lengths indicating that ELF3 can delay heading independently of PPD1 in this condition . Thus, there are differences between B. distachyon and wheat in the effects of ELF3 on heading in the absence of PPD1. We speculate that these differences may be related to the different interactions observed between CO1 and other flowering genes (e.g. PHY) in these two species. For example, in wheat phyC and ppd1 mutants, CO1 expression levels are elevated compared to wild type, whereas in B. distachyon CO1 expression is reduced in both mutants [57,58,66].