Unanticipated regulatory roles for Arabidopsis phytochromes revealed by null mutant analysis

Abstract

In view of the extensive literature on phytochrome mutants in the Ler accession of Arabidopsis, we sought to secure a phytochrome-null line in the same genetic background for comparative studies. Here we report the isolation and phenotypic characterization of phyABCDE quintuple and phyABDE quadruple mutants in the Ler background. Unlike earlier studies, these lines possess a functional allele of FT permitting measurements of photoperiod-dependent flowering behavior. Comparative studies of both classes of mutants establish that phytochromes are dispensable for completion of the Arabidopsis life cycle under red light, despite the lack of a transcriptomic response, and also indicate that phyC is nonfunctional in the absence of other phytochromes. Phytochrome-less plants can produce chlorophyll for photosynthesis under continuous red light, yet require elevated fluence rates for survival. Unexpectedly, our analyses reveal both light-dependent and -independent roles for phytochromes to regulate the Arabidopsis circadian clock. The rapid transition of these mutants from vegetative to reproductive growth, as well as their insensitivity to photoperiod, establish a dual role for phytochromes to arrest and to promote progression of plant development in response to the prevailing light environment.

Plants rely on light as an energy source for photosynthesis and thus possess photosensor proteins to mediate responses to changes in light quantity, spectral quality, direction, and duration for optimal growth and development. Notable among these are the phytochromes, linear tetrapyrrole (bilin)-containing light sensors, which primarily detect the level of red (R) and far-red (FR) light in the environment (1). The long wavelength region of the visible light spectrum is critical for plant development, because both the production of chlorophyll and optimal function of the photosynthetic apparatus heavily rely on the absolute and relative flux of R and FR. It is for this reason that the phytochrome (phy) family has expanded and diversified among the extant seed plants (2). Molecular phylogenetic reconstructions provide evidence for three primary phy lineages, encoded by the PHYA, PHYB, and PHYC gene families, reflecting two rounds of duplications of an ancestral phy gene concomitant with the emergence of seed plants on land (3). Although nearly all angiosperms possess representatives of these three lineages, additional rounds of duplication of the PHYB locus have yielded new members, e.g., PHYD and PHYE, in some eudicot plant lineages such as Arabidopsis thaliana (4, 5).

Our present understanding of the regulatory roles of individual phys is best known for Arabidopsis thaliana and Oryza sativa (rice), owing to the extensive genetic and molecular resources for these species. The picture drawn from physiological analysis of phy mutants in these species indicates that the three phy lineages possess overlapping and distinct roles to entrain plant development with the prevailing light environment (6, 7). Moreover, such studies indicate that phyA performs a dominant role during seedling establishment in low light environments, whereas phyB is the major regulator of shade avoidance behavior in adult plants. The function of phyC has been more difficult to establish, although its role in photoperiod detection and modulation of phyB responses has been observed in both plant species. Based on these and other studies, it is also clear that the regulatory roles of phytochromes have continued to diverge within various plant lineages (5).

From studies on Arabidopsis, phyA appears to be the exclusive FR sensor, whereas phyB is the predominant R sensor, with phyC-E playing a less prominent role in R sensing (8–12). In rice, by contrast, phyB and phyA function as redundant R sensors, whereas phyA and phyC both perceive FR (13). This observation reflects a profound photosensory divergence of phyA and phyC lineages in eudicots and monocots. All rice and Arabidopsis phys are dimeric proteins, some of which, e.g., phyB-E in Arabidopsis and phyB-C in rice, can form heterodimers with each other (13, 14). The functional significance of heterodimer formation is unclear, although previous studies indicate that phyCs fail to homodimerize (15) and require other phys (i.e., phyB or phyD) to accumulate (9, 13, 15). The ability to homodimerize might have been lost multiple times in evolution because Arabidopsis phyE, like phyC, is also an obligate heterodimer (15).

Owing to the regulatory complexity introduced by phy heterodimerization, understanding the specific role of individual phys requires removal of all other phy species. As a baseline for such analyses, it is important to establish the phenotype of a given plant species that lacks all of its phys. A rice phyABC triple null mutant in the Nipponbare cultivar was the first reported phy-less plant species (16). Blind to both R and FR as evaluated by seedling photomorphogenesis, rice phyABC seedlings failed to accumulate detectable chlorophyll under continuous R (Rc) and lacked a transcriptomic response to an R pulse. However, this mutant was able to complete its life cycle under white light, albeit with greatly altered morphology (e.g., increased elongation of internodes even during vegetative stages) and reduced fertility due to an anther dehiscence defect (16). By contrast, an Arabidopsis phyABCDE quintuple null mutant in the Col accession necessitated the presence of a flowering locus T (ft-1) mutation to ensure germination (17). Unlike rice null mutants, the Arabidopsis phyABCDE mutants retained the ability to synthesize some chlorophyll under R yet failed to develop beyond the cotyledon stage. The retention of rhythmic leaf movement in this mutant also indicated that phys are dispensable for clock maintenance (17).

In view of the extensive literature on phy mutants in the Ler accession of Arabidopsis, we sought to secure a phy null line in the same genetic background. The present work describes the isolation and phenotypic characterization of phyABCDE quintuple and phyABDE quadruple mutants in the Ler background. Because both possess a functional allele of FT, these mutants permit measurements of photoperiod-dependent flowering behavior in the absence of phys and in the presence of stand-alone phyC. Our studies show that phys are not required for the completion of the Arabidopsis life cycle under high fluence rate R despite an almost complete lack of transcriptomic response to R, establish that Arabidopsis phyC is nonfunctional in the absence of other phys, and provide unanticipated insight into the regulatory role of phys in the circadian clock function.

Results

Isolation of phyABCDE Null Mutants in the Ler Accession.

A phyA-201,B-1,C-1,D-1,E-1 null mutant (abbreviated as phyABCDE hereafter) was obtained from a cross between a transgenic line YHB^g/phyAB that expresses a constitutively active allele of PHYB (18) and phyBCDE (19) both in the Ler background. After confirming the viability of phyABCDE, additional mutant lines were obtained from a direct cross between phyABDE (20) and phyBCDE. Besides genotyping at the DNA level (see Table S1 for primers used), immunoblot analyses were performed to validate the identities of newly isolated phyABDE and phyABCDE mutants (Fig. 1A). As expected, phyA, phyB, phyD, and phyE proteins were undetectable in both mutant lines, whereas phyC was not present in phyABCDE and could be detected in phyABDE only after long exposure. The phyC level in phyABDE was less than that in other phy mutants examined (15).

Fig. 1.

phyABCDE and phyABDE mutants are photomorphogenically similar. (A) Immunoblot analysis confirms the identities of phyAB(C)DE mutants; the weak phyC band of phyABDE is detected after long exposure (bottom blot). (B) White light-grown adult plants on soil under short-day conditions for 6 wk, from left to right: WT (Ler), phyB, phyAB, phyBCDE, phyABDE, and phyABCDE. (Scale bar, 2 cm.) (C) Rc50-grown, 5-wk-old adult plants on soil, the plant order is same as B (Scale bar, 1 cm.) (D) Hypocotyl lengths of 4-d-old seedlings grown in darkness or under 50 µmol⋅m⁻²⋅s⁻¹ fluence rate of continuous red (Rc), white (Wc), or blue (Bc) light (mean ± SEM, n = 30∼50). (E) Germination of phyAB(C)DE mutants vary and are promoted more effectively by GA₄₊₇ than by GA₃. Seeds were sown on phytagar plates with (MS+) or without MS salts (MS−) and supplied with or without 100 µM GA, stratified for 4 d, and then grown under Rc50 for 4 d before germination scoring. All mutant lines tested were independently grown and harvested; the two phyABDE lines are plotted in blue and the three phyABCDE lines in red.

Seedling Photobiology and Seed Germination of phyAB(C)DE Mutants.

We compared the phyABDE and phyABCDE mutants to define any possible physiological activities regulated by the low level of phyC in phyABDE. As shown in Fig. 1B, adult white light-grown phyABDE and phyABCDE [collectively called phyAB(C)DE hereafter] plants were both slender and capable of reproductive development. Grown under Rc at a moderate fluence rate (50 µmol⋅m⁻²⋅s⁻¹) on soil, most phyAB(C)DE plants could not survive, but some were able to produce three to four tiny rudimentary leaves (Fig. 1C). Under a higher fluence rate of Rc (150 µmol⋅m⁻²⋅s⁻¹), phyAB(C)DE mutants produced flowers and set seeds (Fig. S1A), suggesting that phy-less Arabidopsis plants can fulfill their life cycle when provided sufficient R illumination. When grown on Murashige and Skoog (MS) salt medium, however, phyAB(C)DE mutants performed considerably worse than on soil, exhibiting similar phenotypes to the phy null mutant in the Col background (Fig. S1 B and C) (17).

Examined at the seedling stage, phyABDE and phyABCDE mutants were indistinguishable under all light conditions (Fig. 1D; Fig. S1D). Both mutants were etiolated under Rc and had longer hypocotyls than phyAB and phyBCDE under continuous white (Wc). Under continuous blue (Bc), phyB and phyBCDE were similar to WT, whereas phyAB was longer than WT, indicating that phyA modulates blue light–induced photomorphogenesis, consistent with a previous finding (21). phyAB(C)DE seedlings were longer than phyAB, but still much shorter than cry1cry2, showing that blue light signaling is moderately impaired in phyAB(C)DE mutants. Photomorphogenesis under FRc was as deficient in phyAB(C)DE as in phyA (Fig. S1D), consistent with previous conclusions that phyA is the sole FR photoreceptor in Arabidopsis.

The phyABCDE mutant in the Col accession was reported to require the ft mutation and gibberellic acid (GA₄) treatment for efficient seed germination (17), which was not the case for the Ler phyAB(C)DE mutants. Independently grown and harvested phyAB(C)DE seeds exhibited variable germination capacity, with some lines exhibiting >80% germination rate on the MS salt plates (Fig. 1E). Comparing the germination of the same mutant line on phytagar plates with and without the MS salts, it is evident that some nutrient elements of the MS salts greatly promote phyAB(C)DE germination. Most of the time, phyAB(C)DE seeds exhibited >40% germination, which is sufficient for analysis work using seedlings as the materials. Consistent with the previous report (17), we demonstrated that GA₄ promotes germination more effectively than GA₃ of the mutant lines with low germination capacity (Fig. S2A). We also found that 25 µM of GA₄ was as effective as 100 µM for promoting good germination (Fig. S2B).

phyAB(C)DE Mutants Can Synthesize Chlorophyll Under Red Light.

Rc-grown phyAB(C)DE seedlings were etiolated and pale green, and some had cotyledons that were completely enclosed by testa and never expanded (Fig. 2A). Chlorophyll fluorescence assays showed that the Rc-grown mutants can synthesize chlorophyll (indicated by their peak fluorescence at 670 nm) at a level ∼30- to 50-fold lower than WT (Fig. 2B). The newly isolated phyABDE lines from this study had the same chlorophyll level as phyABCDE. The original/parental phyABDE line (20) repeatedly had a chlorophyll level twofold higher than the newly isolated phyAB(C)DE mutants under various Rc irradiation levels and seedling ages tested (Fig. S3). In addition, the original phyABDE line exhibited unusually long hypocotyls even in darkness and narrower and longer leaves under Wc-phenotypes not observed in the new phyABDE lines. When dark-grown seedlings were exposed to R for 15 min, phyAB(C)DE converted protochlorophyllide into chlorophyll(ide) to a similar extent as WT and phyAB (Fig. 2C). When 4-d-old, dark-grown seedlings were exposed to Rc over a 24-h period, phyAB(C)DE mutants accumulated chlorophyll 10- and 3-fold lower than that of WT and phyAB, respectively (Fig. S4). During the first 3 h Rc, there was no difference in chlorophyll accumulation between phyAB and phyAB(C)DE, implying that phyC-E contribute to prolonged light-dependent chlorophyll accumulation in Arabidopsis. When exposed to Wc, phyAB(C)DE accumulated chlorophyll at a much higher level than under Rc, confirming that these mutants are more robust under wide spectrum light. Collectively, the phy null mutants retained a basal capability of chlorophyll synthesis under R, and there was no significant difference between phyABDE and phyABCDE.

Fig. 2.

phyAB(C)DE mutants can synthesize a low level of chlorophyll under red light. (A) Rc50-grown, 5-d-old phyAB(C)DE seedlings have a nearly etiolated phenotype with marginal greening; some seedlings have cotyledons fully enclosed by a seed coat. (B) Five-day-old phyAB(C)DE seedlings accumulate very low levels of chlorophyll (n = 3, SD is given for the peak value). (C) Dark-grown phyAB(C)DE seedlings can efficiently photoconvert dark-accumulated protochlorophyllide into chlorophyll(ide) after exposure to Rc50 for 15 min, similar to WT and phyAB (n = 3).

phyAB(C)DE Mutants Are Nearly Transcriptionally Blind to Red Light.

To determine global gene expression changes in phyAB(C)DE mutants in response to R, we performed transcriptomic analysis using Affymetrix ATH1 microarray chips. Our previous work indicated that 2,112 genes had statistically significant, more than twofold (SSTF) expression change in WT grown for 4 d under Rc50 compared with WT grown in the dark (18). In the present studies, WT control microarray measurements revealed a similar number of Rc-regulated SSTF genes (i.e., 2,068 genes) after normalization of the WT dataset with the phyAB(C)DE mutant datasets. By contrast, only two and nine genes exhibited SSTF expression changes in Rc50-grown phyABCDE and phyABDE, respectively (Fig. 3A). Both genes from phyABCDE were among the nine genes from phyABDE. Four of the nine genes are stress responsive loci, suggesting that plants lacking phys are more sensitive to light stress (Fig. 3B). In addition, two genes showed an opposite expression pattern in the phyAB(C)DE mutants and WT, so the light regulation of these genes was masked by the presence of phys. We also measured transcriptomic changes in mutant seedlings in response to 2 h of R following 4 d of dark growth. Only four genes in phyABCDE and one gene in phyABDE exhibited SSTF expression changes to the short-time R treatment (Dataset S1). Notably, At5g53710 encoding an unknown stress-responsive protein was induced in phyABCDE by 2-h or 4-d R exposure, reinforcing the interpretation that phyABCDE perceives R as a stress (Fig. 3C). Overall, we conclude that phyAB(C)DE mutants are nearly blind to R at the transcriptomic level.

Fig. 3.

Transcriptomic analysis of phyAB(C)DE response to red light. (A) Venn diagram of red light responsive genes in 4-d-old WT, phyABDE, and phyABCDE (D, dark; Rc = 50 µmol⋅m⁻²⋅s⁻¹ red light). (B) Expression patterns (Rc vs. D) of the nine Rc responsive genes in phyABDE; white dots denote significantly differential expression, and asterisk indicates stress-responsive genes. (C) Expression levels of the two Rc-inducible genes in phyABCDE, At5g53710 and At5g15960/70(KIN1/2). Expression levels are normalized to WT-D of each gene; DR2 = 4-d darkness followed by 2 h of Rc50 exposure. *Statistical significance (adjusted P < 0.05) from the same genotype grown in the dark.

Flowering Behavior of phyAB(C)DE Mutants Is Insensitive to Photoperiod.

The initial phyABCDE lines isolated from crosses of YHB^g/phyAB × phyBCDE and of phyABDE × phyBCDE flowered consistently later than phyABDE (Fig. S5A). This observation suggested that phyC may promote flowering. When overexpression of Col or Ler alleles of PHYC in phyABCDE (independent line n = 16 and 40, respectively) failed to confer the early flowering phenotype of phyABDE, we transformed phyABDE mutants with a PHYC RNAi construct to knock down the already very low level of phyC. A delayed flowering phenotype was not observed in 84 independent transformants. To test whether the later flowering trait was due to a mutation linked to any of the phy alleles, phyABCDE was backcrossed to Ler. Although most of newly resultant phyABCDE lines were late flowering, a small number of phyABCDE lines flowered as early as phyABDE. We also isolated early (predominant) and late-flowering (rare) phyABDE lines from the backcrossed F₂ population. Thus, the later flowering behavior of the parental phyABCDE line was not due to the phyC mutation but reflected an unknown phyC-linked locus in the Ws background from which the phyC-1 allele was originally isolated (19). Alternatively, this result could be due to hybrid vigor between Ler and Ws on chromosome V. Fig. S5B shows morphological differences between early and late-flowering phyABCDE lines. The two types of phyABCDE mutants were indistinguishable under Rc.

Based on genotyping (see below), we determined that the early flowering behavior is the authentic phenotype of the phyABCDE mutant. Evaluated by rosette leaf number, authentic phyABDE and phyABCDE lines flowered very early under both long-day (LD) and short-day (SD) conditions (Fig. 4). Both exhibited a delay in days to flowering under SD; however, this delay was probably due to insufficient photosynthesis that limited growth and development (Fig. 4; Fig. S6). The flowering behavior of phyAB(C)DE illustrates their insensitivity to photoperiod, as neither mutant displayed flowering delay under SD. By comparison, phyBDE mutants flowered as early as phyAB(C)DE under LD, but later under SD, indicating that phyA can delay flowering under SD in the absence of type II phys (Fig. 4). Indeed, phyBCDE mutants flower later than phyABCDE lines under SD (Fig. 4). The flowering phenotypes of phyAB(C)DE lines support the conclusion that phyC does not regulate flowering in the absence of other phys.

Fig. 4.

Flowering of phyAB(C)DE mutants is insensitive to photoperiod. The data are presented as mean with SEM (n = 20). LD, long-day conditions (16-h L/8-h D); SD, short-day conditions (8-h L/16-h D).

Genotyping Distinguishes Between Early and Late-Flowering phyAB(C)DE Lines.

Seedling microarray data revealed that the parental late-flowering phyABCDE line had unusually high expression of FLC, a flowering repressive gene that integrates signals from both vernalization and autonomous pathways (22). Indeed, the FLC expression in the early flowering phyABCDE line was reduced to a level similar to WT (Fig. S5C). Although both FLC and PHYC are located on chromosome V (ChrV), the long distance between FLC (at 3.2 Mb) and PHYC (at 14.0 Mb) is inconsistent with the close linkage between phyC and the late-flowering locus inferred by genetic analyses. Association mapping excluded linkage of loci on the bottom arm of ChrV with the flowering behavior. The parental phyBCDE line used for constructing phyABCDE was found to contain Ws alleles in the entire top arm of ChrV, presumably from the original Ws phyC-1 mutant (Fig. S5D). The Ws NGA76 marker allele at 10.4 Mb always cosegregated with phyC-1 and was not linked with the flowering phenotype. That the pericentric Ws NGA76 marker cosegregated with the pericentric phyC-1 allele is consistent with the rare recombination frequency of loci near the centromere (23). By contrast, three markers at the top arm of ChrV were linked with the flowering phenotype to varying degrees. The early flowering phyABCDE lines all had Ler alleles for these markers, whereas the late-flowering lines contained Ws alleles. We thus conclude that a variant Ws locus in the top arm of ChrV that activates FLC expression is responsible for the delayed flowering of the parental phyABCDE lines (Fig. S5D). Fine mapping of this locus is beyond the scope of this work. An early flowering phyABCDE line with all Ler alleles in this region was further backcrossed with Ler WT. All progeny phyABCDE and phyABDE mutant lines from this second backcross flowered early. These data support that the early flowering phenotype of Ler phyAB(C)DE mutants is authentic.

ATHB2 Retains Response to Changes in R/FR Ratio in phyAB(C)DE Mutants.

A previous study of the phyABDE mutant showed that the shade-inducible gene ATHB2 was still responsive to the change in R/FR ratio: a result attributed to the residual phyC function (20). We therefore reexamined this response in newly isolated phyAB(C)DE mutants under the same growth conditions and treatment (20). As expected from previous studies (24), transfer of WT plants to simulated shade (R:FR = 0.2) resulted in dramatic increase in ATHB2 transcript abundance compared with R treatment alone (Fig. 5A). Although the ATHB2 transcript levels in light-grown phyAB(C)DE were already elevated compared with WT, they further increased in response to transfer to darkness (Fig. 5A). A similar but weaker increase was also seen in WT, implying that other processes can suppress ATHB2 expression in the light, e.g., photosynthesis. The transcript increase was more pronounced in phyAB(C)DE when the dark period was extended from 4 to 6 h, whereas 2-h R treatment following 4-h dark prevented this enhancement. By contrast, when the 2-h R treatment was replaced with simulated shade (R:FR = 0.2) with the same R fluence rate, ATHB2 expression increased (the P values of statistical significance were slightly higher than 0.05 due to great variation among biological replicate sets). As both mutants behaved similarly to simulated shade, the residual phyC does not contribute to the expression alternation of ATHB2. PIL1, another shade-inducible gene, maintained a very high expression level in light-grown phyAB(C)DE and did not respond significantly to the dark treatment (Fig. 5B). Intriguingly, phyABCDE, but not phyABDE, displayed a marked, but variable, increase in PIL1 transcript abundance following simulated shade treatment. This increase cannot be attributed to phy function but may represent a stress response in these plants.

Fig. 5.

Expression response of ATHB2 (A) and PIL1 (B) to various light treatments. Three-week-old plants grown on soil under SD conditions (8-h L/16-h D) at 16 °C were transferred to darkness for 4 or 6 h, 4 h followed by 2 h of red light (30 µmol⋅m⁻²⋅s⁻¹), or 2 h of red plus far-red light (R:FR = 0.2) treatments. Expression levels are the means from three biological replicates ± SD.

phyAB(C)DE Mutants Maintain Circadian Rhythms Under Rc, with Reduced Responsiveness of Period to Fluence Rate.

Both temperature and light cues ensure correct synchronization between the endogenous clock and the environment (25, 26), with phys affecting circadian phase, period, and output amplitude of gene expression (27, 28). To test whether circadian rhythms of gene expression are maintained in phyAB(C)DE seedlings under Rc, we introduced the clock-regulated, enhanced luciferase reporter pCCA1::LUC2 into both mutants. Both phyABDE and phyABCDE seedlings retained robust rhythms of bioluminescence following transfer from 12 L:12 D light cycles to Rc, although the initial phase of peak bioluminescence for the two mutants were earlier than that of the WT (Fig. 6A). The periods were similar for both mutants; however the amplitude of rhythmic bioluminescence in both mutants was greatly reduced in comparison with WT. Because circadian periods of many diurnal species, including plants, are shortened in response to higher fluence rates of constant light, a phenomenon formalized by Aschoff (29), we undertook comparative period measurements under a range of Rc fluence rates. For the WT as expected, we observed a fluence rate-dependent shortening of circadian period, with the seedlings most responsive between ∼5 and ∼30 µmol⋅m⁻²⋅s⁻¹ Rc (Fig. 6B). Intriguingly, phyAB(C)DE seedlings did not simply display longer period phenotypes as might be expected from data reported for single phyA and phyB mutants (27). Instead, the period of phyAB(C)DE mutants was much less dependent on the fluence rate of Rc, exhibiting a shorter period than WT under lower fluence rates and a longer period under higher fluence rates (Fig. 6B). No measurable difference was observed between phyABDE and phyABCDE mutants. These data show that phys are not required for clock maintenance under Rc and implicate that phys can both increase and decrease the rate of the clock.

Fig. 6.

Circadian rhythms in the phyAB(C)DE mutants. (A) Normalized bioluminescence of seedlings containing a pCCA1:LUC2 reporter construct. Plants were entrained to 12 L:12 D cycles for 6 d before being moved to 27 µmol⋅m⁻²⋅s⁻¹ Rc. Data presented for each line were normalized to the average bioluminescence over 72 h following background subtraction. (B) phyAB(C)DE mutants have a shorter period in comparison with WT at low fluence rates, but a longer period at higher red light fluence rates. Seedlings were entrained as in A before being moved to Rc at the indicated fluence rate. Error bars indicate SEM (n ≥ 6).

Discussion

Although Arabidopsis phy null mutants in the Col accession have been described previously (17), null mutants have not been secured in the Ler accession for which an extensive literature on phy function is available. In contrast to the earlier report, we show that the Ler phy-less mutant is robust, and as such, represents a valuable tool for studying the photoregulatory functions of individual phys and their interaction with other family members in an otherwise isogenic background. The Ler phy-less mutant phenotype is quite stable, and resegregated mutants from two backcrosses with the Ler WT continue to produce viable seeds for propagation for multiple generations. This observation indicates that residual phy transmitted to the progeny is dispensable for continued viability. Our studies also reinforce that phyC requires other phys for activity, because all phenotypes examined for phyABDE are indistinguishable from those of phyABCDE. Although this loss of function is in part owed to greatly reduced phyC protein accumulation, the residual phyC in the phyABDE mutant lacks any photoregulated activity. These findings are consistent with the observation that, in Arabidopsis, phyC is an obligate heterodimer with either phyB or phyD (9, 14, 15), implicating monomeric phyC to be nonfunctional and/or degraded. This interpretation agrees with the observation that the rice phyAB mutant is essentially the same as the rice phyABC mutant phenotypically (16), yet contrasts with earlier observations that implicate regulatory function of phyC in the absence of other phys (8, 17). The reason for this difference is unclear but may reflect cryptic mutations at other loci that were not removed in the genotypes previously examined.

phy-Less Plants Are Viable but Developmentally Challenged.

Our studies show that phyABCDE plants are viable, although their survival is conditional on the growth environment as reported previously (16, 17). We believe that survival reflects retention of minimal photosynthetic development, as phyABCDE null plants can synthesize chlorophyll and develop functional chloroplasts even under Rc. However, only under elevated fluence rates of Rc can the quintuple mutant complete the life cycle, arguably due to enhanced chlorophyll synthesis and light harvesting. The poor cotyledon expansion of phyABCDE seedlings frequently prevented shedding of their seed coats, which contributed to arrested seedling development and death. To a lesser extent, this also occurred in phyABCDE seedlings grown in white light (Fig. S7). The proportion of arrested development in the mutant population was much higher under SD than under LD conditions, suggesting that phy-less plants rely on high irradiation levels for survival.

Only two genes were SSTF induced in phyABCDE seedlings under Rc, both of which are stress related. The phyABDE used in the microarray studies was the original line that had more chlorophyll than phyABCDE. It is not surprising that this mutant had seven more SSTF-regulated genes, two of which are involved in starch metabolism. Compared with the >2,000 SSTF Rc-regulated genes in WT, the few SSTF genes in the two mutants reinforce the conclusion that the phyAB(C)DE mutants are nearly blind to R. Growth on MS agar plates was also stressful for phyAB(C)DE mutants. Even with sucrose supplementation, most quintuple plants failed to develop beyond the seedling stage—a problem observed in the previous study (17). In contrast to the Arabidopsis phyAB(C)DE mutants, rice phyABC mutants do not synthesize sufficient chlorophyll under Rc for development beyond the seedling stage (16). Under broad-spectrum white light, phyABCDE null mutants fared much better, presumably due to the activities of the cryptochrome, phototropin, or other blue/UVA light sensors or due to enhanced photosynthetic light conversion. The ft mutation was previously found necessary for germination of Col phyABCDE seeds (17). We too found that germination was reduced in some Ler phyAB(C)DE mutants, which could be mostly rescued by GA₄ treatment. However, some seed lots of the phyAB(C)DE mutants showed robust germination, suggesting that the physiological state of adult plants at the time of seed set plays a significant role in seed germination.

Flowering Is Insensitive to Photoperiod in the Absence of phys.

The ft-1 mutation present in the Col phyABCDE mutant makes flowering measurements problematic; thus, the photoperiod response of flowering was not addressed previously (17). Moreover, the ft-1 allele was originally derived from the Ler background, so genetic background effects could also complicate the interpretation of the flowering phenotypes of the mutant. Indeed, we encountered a similar problem when we examined the flowering of the originally isolated phyABCDE mutant that flowered later than the phyABDE mutant. After genetic background cleanup by backcrossing, phyABCDE flowered as early as phyABDE under both LD and SD conditions. Thus, the phy-less mutants appeared to be insensitive to photoperiod. In this regard, the rice chromophore-deficient se (30) and phyABC mutants (16) are both insensitive to photoperiod. Measured as days to flowering, both rice se and phyABC mutants flowered slightly later under SD than under LD conditions, similar to Arabidopsis phyAB(C)DE mutants. This result has been rationalized by the slower growth of the rice phyABC mutants under SD conditions (30); thus, the same appears true for the Arabidopsis phyAB(C)DE mutants. These data indicate that the elimination of phys confers photoperiod insensitivity. Given that phys alter CO stability in the photoperiodic pathway (31), we hypothesize that phy-less mutants have increased CO stability and therefore high FT expression, rendering them insensitive to photoperiod and early flowering. Alternatively, the photosynthetic deficiency of these mutants may contribute to a general stress that induces early flowering regardless of the light environment. Finally, previous genetic studies implicate phyC in the delay of flowering under SD photoperiods while also supporting the conclusion that the Ler PHYC allele is poorly active (9, 32). It is thus conceivable that the reduced regulatory activity of Ler phyC is responsible for the observed photoperiod insensitivity of our phyABDE mutants. Although experiments to assess this possibility are beyond the scope of this investigation, a potential polymorphism in a flowering locus linked to the pericentric PHYC allele on chromosome V (as was observed here) cannot be dismissed as an explanation for the previous observations.

Circadian Clock Period Length Is Nearly Insensitive to Rc Fluence Rate in phy-Less Plants.

Col phyABCDE plants were previously shown to maintain circadian rhythms of leaf movement under Wc, but not under Rc (17). Using the pCCA1::LUC2 reporter, we show that Ler phyAB(C)DE mutants maintain rhythmicity of CCA1 expression under Rc. However, a dramatic reduction in the amplitude of the bioluminescence signals in phyAB(C)DE seedlings was observed compared with WT. It was previously shown that Rc induces CCA1 expression (33) and that the amplitude of CCA1 rhythmic expression is reduced in the phyB-9 mutant (34). Similarly, the amplitude of CCA1 promoter-driven luciferase expression is dampened in darkness (35). The reduced amplitude we observed is thus likely a consequence of impaired Rc perception in phyAB(C)DE. Because phyAB(C)DE seedlings are smaller with delayed cotyledon expansion and true leaf emergence when grown under 12 L/12 D cycles, however, we cannot fully distinguish between this hypothesis and the possibility that the reduced bioluminescence is a consequence of delayed development. These data indicate that the previously reported arrhythmicity in leaf movement under Rc (17) is not caused by complete loss of oscillator function, but instead might reflect an overall low amplitude of clock-regulated processes and/or the extremely small leaves of the phyAB(C)DE seedlings.

In addition to reduced bioluminescence, we observed an early phase of pCCA1::LUC2 peak activity in the phyAB(C)DE mutants immediately following transfer to Rc. An early phase phenotype has previously been reported for phyB mutants harboring a pLHCB::LUC transgene in the Col accession under Wc, i.e., phyB-9 and oop1 (28). The early phase phenotypes of phyB-9 and oop1 were not evident in seedlings entrained to temperature cycles, suggesting that light signaling defects contribute to this phenotype (28). Intriguingly, phyB-9 has also been reported to differentially affect the phase of several clock components under Wc (34). In this latter study, the phase of pCCA1::LUC+ and pTOC1::LUC+ was comparatively unaffected, whereas GI and PRR9 promoters had early phases compared with WT (34). Impaired phy signaling to multiple points of the circadian system likely underlies the early phase phenotype of phyAB(C)DE seedlings.

Increased R fluence rates lead to a shortening of the circadian clock in Arabidopsis (27, 29). This response was impaired in phyAB(C)DE seedlings, consistent with the expectation that phys contribute to this fluence rate-dependent period shortening. However, we observed a modest shortening of circadian period in phyAB(C)DE as fluence rate increased, suggesting that phy-less seedlings maintained some sensitivity to R. Such sensitivity may derive from a metabolic signal induced by enhanced photosynthesis under increasing fluence rates or by increased oxidative stress. Interestingly, at fluence rates less than 10 μmol⋅m⁻²⋅s⁻¹ Rc, we observed an increased pace in the circadian oscillator in phyAB(C)DE compared with WT. This nonintuitive shortening of circadian period in phyB-9 and higher-order phy mutants under Wc has previously been reported (17, 34). Such data suggest that phys do not simply act as a light-induced accelerant of the clock mechanism. Instead, we hypothesize that P_r forms of phys act to delay the circadian system under low fluence rates, whereas light-activated P_fr forms act to increase the pace of the oscillator under higher light intensities.

Materials and Methods

Methodologies used for isolation and genetic, immunochemical, and phenotypic characterization of phyAB(C)DE mutants in the Ler genetic background can be found in SI Materials and Methods. Adapted from methods previously developed (18), transcriptome and RT PCR analyses are detailed in SI Materials and Methods. Construction and luciferase imaging of phyAB(C)DE lines harboring a pCCA1::LUC2 reporter are detailed in SI Materials and Methods.