Phytochrome controls alternative splicing to mediate light responses in Arabidopsis

Significance

Plants adapt to their fluctuating environment by monitoring surrounding light conditions through several photoreceptors, such as phytochrome. It is widely believed that upon absorbing red light, phytochrome induces plant light responses by regulating the transcription of numerous target genes. In this study, we provide clear evidence that phytochrome controls not only transcription, but also alternative splicing in Arabidopsis. We reveal that 6.9% of the annotated genes in the Arabidopsis genome undergo rapid changes in their alternative splicing patterns in a red light- and phytochrome-dependent manner. Our results demonstrate that phytochrome simultaneously regulates two different aspects of gene expression, namely transcription and alternative splicing to mediate light responses in plants.

Abstract

Plants monitor the ambient light conditions using several informational photoreceptors, including red/far-red light absorbing phytochrome. Phytochrome is widely believed to regulate the transcription of light-responsive genes by modulating the activity of several transcription factors. Here we provide evidence that phytochrome significantly changes alternative splicing (AS) profiles at the genomic level in Arabidopsis, to approximately the same degree as it affects steady-state transcript levels. mRNA sequencing analysis revealed that 1,505 and 1,678 genes underwent changes in their AS and steady-state transcript level profiles, respectively, within 1 h of red light exposure in a phytochrome-dependent manner. Furthermore, we show that splicing factor genes were the main early targets of AS control by phytochrome, whereas transcription factor genes were the primary direct targets of phytochrome-mediated transcriptional regulation. We experimentally validated phytochrome-induced changes in the AS of genes that are involved in RNA splicing, phytochrome signaling, the circadian clock, and photosynthesis. Moreover, we show that phytochrome-induced AS changes of SPA1-RELATED 3, the negative regulator of light signaling, physiologically contributed to promoting photomorphogenesis. Finally, photophysiological experiments demonstrated that phytochrome transduces the signal from its photosensory domain to induce light-dependent AS alterations in the nucleus. Taking these data together, we show that phytochrome directly induces AS cascades in parallel with transcriptional cascades to mediate light responses in Arabidopsis.

Phytochromes are the red/far-red light (R/FR) receptors by which plants monitor their surrounding light environment and modulate their growth, development, and metabolism accordingly. Upon absorption of R, phytochromes are converted from the biologically inactive Pr form into the active Pfr form, whereas FR irradiation converts Pfr back to Pr. Arabidopsis has five molecular species of phytochrome, phyA to phyE, among which phyA and phyB play predominant roles in seedling de-etiolation, a critical process during which the plant switches from heterotrophic to autotrophic growth (1). PhyA and phyB display distinct light responsiveness; whereas light-stable phyB mediates R/FR reversible responses, light-labile phyA is responsible for sensing continuous FR (high irradiance response) and very weak light of various wavelengths (very low fluence response) (1).

A phytochrome molecule consists of an N-terminal chromophoric domain and a C-terminal dimerization domain (2). The N-terminal domain is sufficient for transducing light signals in the nucleus to induce photomorphogenesis (3, 4). Pfr is translocated from the cytoplasm to the nucleus (5, 6), where it inhibits two major negative regulators of photomorphogenesis—that is, phytochrome-interacting factors (PIFs) and CONSTITUTIVE PHOTOMORPHOGENIC 1 (COP1)—to elicit light responses in plants (7). PIFs are basic helix–loop–helix transcription factors that directly interact with Pfr (8). This R-dependent interaction results in the phosphorylation and proteasome-mediated degradation of PIFs (9, 10). In contrast, COP1 is an E3 ubiquitin ligase that mediates the ubiquitination and proteasomal degradation of photomorphogenesis-promoting transcription factors, including LONG HYPOCOTYL 5 (HY5) (11). Therefore, phytochrome regulates the transcription of light-responsive genes by modulating the protein stability of negatively acting (such as PIFs) and positively acting (such as HY5) transcription factors to mediate photomorphogenesis.

Accumulating evidence suggests that light regulates not only transcription but also other aspects of gene expression, such as chromatin reorganization (12), translation (13, 14), posttranslational modification (11, 15), and pre-mRNA splicing (16–18). Pre-mRNA splicing is a posttranscriptional event that is carried out by a macromolecular complex called the spliceosome. Alternative splicing (AS) produces multiple transcripts from a single gene by using different splice sites. AS qualitatively expands transcriptome diversity, whereas transcriptional regulation quantitatively controls the transcriptome. Because conventional microarray experiments do not detect qualitative alterations in the transcriptome, AS has been investigated to a lesser extent than has transcriptional regulation. Selection of alternative splice sites is mediated by trans-acting splicing factors, such as serine/arginine-rich (SR) proteins. These factors bind to cis elements on the pre-mRNA to promote or inhibit the recruitment of spliceosome components to the adjacent alternative splice sites (19). Therefore, the regulation of AS depends on the expression level and posttranslational modification of SR proteins and other splicing factors (20). It has been reported that light affects the AS of several genes in plants (16–18). However, the physiological significance of this finding and the identity of the photoreceptors that mediate light-regulated AS remain unclear. Recently, we identified rrc1 (reduced red-light responses in cry1cry2 background 1), an Arabidopsis mutant that is impaired in phyB-mediated light responses (21). Because RRC1 encodes a novel SR-like protein and rrc1 mutants show aberrant AS profiles, we hypothesized that RRC1 controls AS to mediate phyB signaling. Furthermore, we proposed that phyB is involved in the R-dependent AS changes of several genes (21).

Here, we present clear evidence that phytochrome induces genome-wide AS alterations in Arabidopsis. mRNA sequencing (mRNA-seq) analysis revealed that over 1,500 genes underwent changes in their AS patterns within 1 h of R exposure, in a phytochrome-dependent manner. Gene Ontology (GO) analysis suggested that RNA splicing-related genes were the most enriched among those genes that showed rapid phytochrome-dependent changes in AS in response to continuous R (cR), whereas transcription factor genes were the main early targets of transcriptional regulation by phytochrome, as previously reported (22). Our results demonstrate that phytochrome controls not only transcription but also AS to mediate R responses in Arabidopsis.

Results and Discussion

Genome-Wide Analysis of AS Control by Phytochrome During De-Etiolation.

In Arabidopsis, phyA and phyB play predominant and partially redundant roles in every step of seedling de-etiolation, from changes in gene expression to changes in morphology (1, 22). To identify the genes that show phytochrome-dependent AS changes in the Arabidopsis genome, we performed mRNA-seq analysis and determined a series of time-course transcriptome profiles of 4-d-old, dark-grown WT and phyAphyB double-mutant (phyAB) etiolated seedlings that were exposed to cR (8.3 µmol⋅m⁻²⋅s⁻¹) for 1 h (R1h) or 3 h (R3h) (Fig. 1A). For the dark control, transcriptome profiles were also determined for 4-d-old WT etiolated seedlings with or without additional growth in darkness for 3 h (D3h and D0h) (Fig. 1A).

Fig. 1.

Red light triggers global transcriptomic changes that are mediated by phytochrome in Arabidopsis. (A) Flowchart for the analysis of red light- and phytochrome-dependent changes in AS or TX. For transcriptomic profiling with mRNA-seq, total RNA was extracted from 4-d-old WT and phyAphyB double-mutant (phyAB) etiolated seedlings that were exposed to continuous red light (8.3 µmol⋅m⁻²⋅s⁻¹) (pink shading) for 1 h (R1h) or 3 h (R3h). For the dark control, total RNA was also extracted from 4-d-old WT etiolated seedlings with or without additional growth in darkness (gray shading) for 3 h (D3h and D0h). (B) Venn diagrams showing the number of genes that displayed phytochrome-dependent changes in either AS (AS_only) or TX (TX_only), or in both AS and TX (AS&TX), at R1h and R3h. AS, P < 0.05 (Pearson’s χ² test); TX, FDR < 0.01 (FDR by LIMMA). (C) GO analysis of genes that displayed phytochrome-induced changes in AS or TX. The bars represent the proportion of phytochrome-regulated genes in each category at R1h or R3h, or that of all of the genes in each category to all of the annotated genes in the genome (Genome). Asterisks indicate the statistical significance (FDR < 0.05 by LIMMA) that was calculated based on relative abundance in the genome.

AS patterns can be determined based on the number of splice-junction reads within a gene. Although 45,372 splice junctions have been annotated in TAIR10 (The Arabidopsis Information Resource), it is likely that many other splice junctions exist. To characterize AS events at the genomic scale, we additionally identified 37,545 novel splice junctions in 32,398 annotated genes (SI Text). For each of these 82,917 previously known and novel splice junctions, the number of junction reads was counted and normalized by reads per kilobase of exon model per million mapped reads (RPKM) for the corresponding genes under the following conditions (genotype_treatment): WT_D0h, WT_D3h, WT_R1h, WT_R3h, phyAB_D0h, phyAB_R1h, and phyAB_R3h (Fig. 1A). Then, AS was defined as being regulated by phytochrome at R1h if the relative number of junction reads in WT_R1h significantly differed, in the same direction, from that in WT_D0h and from that in phyAB_R1h (Fig. 1A). Using the same criteria, but with R1h being replaced with R3h, we identified phytochrome-regulated AS at R3h; however, in this case, we added another criterion, namely that the number of junction reads should not differ between WT_D0h and WT_D3h, because such a difference might indicate that AS is controlled by the circadian clock and not by phytochrome (Fig. 1A).

The mRNA-seq coverage was high enough (×236 to ×546) (Table S1) to allow robust statistical analyses that are likely to cover genome-wide light-responsive AS changes specifically controlled by phyA and/or phyB. Using this method, we found that 2,230 genes (∼6.9% of the 32,398 annotated genes in the Arabidopsis genome) displayed phytochrome-regulated AS patterns at R1h and/or R3h (Pearson’s χ² test, P < 0.05) (Fig. 1B and Dataset S1 A–C). To determine the steady-state transcript level (TX), we determined the RPKM value for each of the 32,398 annotated genes in TAIR10. We next established the phytochrome-regulated TX using the same criteria as for the AS analysis (Fig. 1A), and found that TX was regulated by phytochrome in 5,096 genes [false discovery rate (FDR) < 0.01] (Fig. 1B and Dataset S1 A, D, and E). Interestingly, 1,505 genes rapidly altered AS patterns within 1 h of transfer to cR in a phytochrome-dependent manner, and 88% of these genes did not show any significant phytochrome-mediated changes in TX at this time point (Fig. 1B). This result suggests that phytochrome is directly involved in the genome-wide regulation of AS.

Functional Categorization of Genes Under Phytochrome-Mediated AS Regulation.

We next performed GO analysis of the genes that displayed phytochrome-dependent changes only in AS (AS_only), only in TX (TX_only), or in both AS and TX (AS&TX). At R1h, GO terms related to RNA splicing were significantly enriched among AS_only genes (Fig. 1C and Dataset S1F), whereas those related to transcription were overrepresented in TX_only genes at R1h, as previously reported (Fig. 1C and Dataset S1G) (22). Consistent with this result, phytochrome-mediated AS was found in 15% (58 of 395) of all splicing-related genes known in Arabidopsis at R1h (Table S2 and Dataset S1I) (23). These 58 genes include splicing regulators involved in splice site selection and spliceosome recruitment, such as 7 of the 18 SR proteins (RS31, RS40, RS41, RSZ33, RSZ34, SR34a, and SR34b), U1 small nuclear ribonucleoprotein 70 kDa, and U2 auxiliary factor 65a (U2AF65a) (Dataset S1I). Interestingly, 56 (81%) of the 69 splicing-related genes that were subject to phytochrome-mediated AS and/or TX control at R1h were AS_only genes (Table S2 and Dataset S1I). In contrast, the transcription factor genes that were under phytochrome-mediated gene expression control at R1h were mostly TX_only genes (210 of 358 = 59%), and this proportion is significantly higher than that of AS_only genes (124 of 358 = 35%) (Fisher’s exact test, P = 0.0001) (Table S2 and Dataset S1J). These results strongly suggest that RNA splicing-related genes and transcription factor genes are the main early targets of phytochrome-mediated AS and transcriptional regulation, respectively (Fig. 2). The finding that the phytochrome-mediated control of the expression of splicing-related genes at R3h was more prominent at the TX level than at the AS level (Table S2) suggests that phytochrome also controls AS through transcriptional regulation of splicing factors at a later time point after the onset of cR.

Fig. 2.

A model depicting the early signaling cascades of phytochrome that regulate genome-wide gene expression in response to red light. Red arrows indicate pathways that are regulated by AS; blue arrows indicate pathways subjected to transcriptional regulation.

We further investigated the results of GO analysis to infer the physiological significance of phytochrome-mediated AS, and found that a number of photosynthesis- or plastid/chloroplast-related GO terms were overrepresented, not only in TX_only genes as previously reported (22), but also in AS_only genes and AS&TX genes (Fig. 1C and Dataset S1 F–H), suggesting that phytochrome-mediated AS is also involved in light-induced chloroplast differentiation during seedling de-etiolation.

Because GO analysis is not sufficient to fully infer the physiological effects of phytochrome-mediated AS control, we manually generated a list of Arabidopsis genes reported to be important for light signaling in vivo (Dataset S1K). Among the 184 genes identified, 38 (21%) and 88 (48%) showed phytochrome-regulated AS and TX alterations, respectively, at R1h and/or R3h (Table S3 and Dataset S1K). Phytochrome-regulated AS and TX genes were significantly enriched in this list relative to the whole Arabidopsis genome (Fisher’s exact test, P = 8.82 × 10⁻¹⁰ for AS and P = 1.10 × 10⁻¹⁵ for TX) (Table S3), suggesting that not only phytochrome-mediated TX, but also AS was enriched among the light signaling genes. Furthermore, we found that genes under phytochrome-mediated AS control were enriched in categories such as “COP/DET,” “clock,” “phy-, cry-, or phot-signaling,” “shade avoidance,” and “photoreceptor” (Table S3). These findings strongly suggest that both phytochrome-mediated TX and AS play important and specific roles in light signaling in Arabidopsis.

Experimental Validation of mRNA-seq Data.

To validate the mRNA-seq data, we manually selected 19 genes from the categories that were overrepresented among the AS_only and AS&TX genes, and confirmed phytochrome-dependent and cR-responsive AS alterations for 10 of the 19 genes (validation rate, 10 of 19 = 53%) using semiquantitative RT-PCR (sqRT-PCR) analysis (Fig. 3 and Fig. S1). Because sqRT-PCR is intrinsically less sensitive than mRNA-seq, this result suggests that phytochrome-dependent AS regulation occurs in at least 53% of the 2,230 genes that were detected in our mRNA-seq analysis. The validated genes were related to splicing (RS31, SR30, SR34a, SR34b, and U2AF65a), phytochrome signaling [SPA1-RELATED 3 (SPA3) and PSEUDO-RESPONSE REGULATOR 7 (PRR7)], the circadian clock [LATE ELONGATED HYPOCOTYL (LHY) and PRR7], and photosynthesis [PHOTOSYNTHETIC NDH SUBCOMPLEX B 4 (PnsB4)], strongly suggesting that phytochrome-induced AS alterations contribute to these functional categories.

Fig. 3.

Validation of phytochrome-regulated alternative splicing. Total RNA from 4-d-old WT and phyAphyB (phyAB) etiolated seedlings exposed to continuous red light (8.3 µmol⋅m⁻²⋅s⁻¹) for 0 h (D0h), 1 h (R1h), and 3 h (R3h) was analyzed by sqRT-PCR. The regions that were predicted to display differential AS patterns were amplified with gene-specific primers, and the DNA concentration of each PCR product of different sizes was quantified. AS events in each gene, which were designated as mRNA-x (where x represents 1–6), are shown as diagrams above the bar graphs. mRNA-1 represents splice variants that are considered to encode functional full-length proteins in each gene. Boxes and lines indicate exons (numbered) and spliced-out introns, respectively. The white and gray regions in the boxes indicate untranslated regions and coding regions, respectively. AS patterns were expressed as ratios of the indicated splice variants to the total transcripts, and the values are shown as the mean ± SE (n = 3). Asterisks indicate statistical significance relative to D0h samples, as determined using Student t test (*P < 0.05; **P < 0.01).

Physiological Contribution of Phytochrome-Mediated AS Control.

To gain insight into the roles of phytochrome-mediated AS regulation during de-etiolation, we determined the sequences of the splice variants of the 10 validated genes, including SPA3 (Fig. S1A). SPA3 is a member of the SPA family proteins, which interact with COP1 to form the COP1–SPA complex and function as negative regulators of photomorphogenesis in concert with COP1. The COP1–SPA complex is part of a multimeric E3 ubiquitin ligase containing CULLIN 4 and DAMAGED DNA-BINDING PROTEIN 1 (DDB1) (11). Both COP1 and SPA proteins possess a central coiled-coil domain, followed by C-terminal seven WD40 repeats. COP1 and SPA proteins interact with each other through their coiled-coil domains (24). SPA proteins are known to associate with DDB1 through the fourth repeat of the WD40 repeats (25). Interestingly, sqRT-PCR analysis revealed that phytochrome promoted the retention of intron 4 and selection of the alternative 5′ splice site within intron 4 of SPA3, both of which produce premature termination codons and result in the loss of most of the WD40 repeats, including the fourth repeat (Fig. S1A). These splice variants, named mRNA-3 and mRNA-2, respectively, likely encode truncated SPA3 proteins that have dominant-negative effects on the function of the endogenous COP1–SPA complex, because they are unable to bind to DDB1 but still retain the interaction with COP1. Thus, it was suggested that the phytochrome-induced AS of SPA3 promotes photomorphogenesis by increasing the amount of dominant-negative variants of SPA3.

To investigate this possibility, we overexpressed mRNA-2 and mRNA-3 of SPA3, as well as the full-spliced mRNA-1 that likely encodes a functional full-length SPA3 (Fig. S1A) in the WT background to obtain overexpression (OX) lines of each isoform. In the mRNA-3 OX construct, all of the splice sites within intron 4 were mutated to retain an unspliced intron 4 in the transcript (Fig. 4A). sqRT-PCR analysis showed that each isoform was specifically overaccumulated in each line, as expected (Fig. 4B). Then we examined the seedling de-etiolation phenotype of these lines and found that, although no clear phenotype was observed in darkness, obviously shorter- and longer-hypocotyl individuals than WT seedlings were segregated under cR (11 µmol⋅m⁻²⋅s⁻¹), at reasonable ratios, from T2 populations of the mRNA-2 and mRNA-3 OX lines and from those of the mRNA-1 OX lines, respectively (Fig. 4C). The R-dependent short-hypocotyl phenotypes of the mRNA-2 and mRNA-3 OX lines are reminiscent of those of spa3 mutants (26), demonstrating that the proteins encoded by mRNA-2 and mRNA-3 have dominant negative effects at least on the endogenous SPA3. Taken together, these data indicate that light-regulated, phytochrome-mediated AS control indeed contributes to promoting seedling de-etiolation.

Fig. 4.

Functional analysis of SPA3 splice variants in transgenic Arabidopsis. (A) Schematic illustration of constructs for generating OX lines of each SPA3 splice variant in the WT background. In the mRNA-3 OX construct, all of the splice sites within intron 4 were mutated, as indicated by the arrows. 35S, Cauliflower mosaic virus 35S promoter; boxes, spliced exons; horizontal lines, unspliced intron 4; asterisks, premature termination codons. (B) Ratios of each SPA3 isoform to the total transcripts in the OX lines. For each construct shown in A, two independent representative lines that segregated about 3:1 for kanamycin-resistance in the T2 generation were chosen for analysis. These lines, shown on the top, were used in B and C. Total RNA from WT plants and kanamycin-resistant T2 plants for each line, grown under continuous white light (35 µmol⋅m⁻²⋅s⁻¹) for 10 d, was analyzed by sqRT-PCR. (C) Box plots of hypocotyl lengths of the SPA3 isoform OX lines. Seedlings of WT and T2 segregating populations of mRNA-1 OX, mRNA-2 OX, and mRNA-3 OX lines were grown under cR (11 µmol⋅m⁻²⋅s⁻¹) or in darkness (Dark) for 5 d, and hypocotyl lengths were determined. The top, middle, and bottom of the box indicate the 25th, 50th, and 75th percentiles, respectively. Dots and whiskers represent individual values and the spread of the data, respectively.

PhyA and PhyB Are Responsible for R-Dependent AS Control.

We previously showed that two SR protein genes, RS31 and SR34b, display cR-induced changes in AS, and that this response is partially reduced in phyB mutants (21). The mRNA-seq analysis conducted in this study revealed that RS31 and SR34b undergo phytochrome-mediated AS control (Dataset S1). These results strongly support the notion that at least phyB is involved in the cR-induced AS changes. To further examine the photophysiological properties and involvement of specific isoforms of phytochrome in the light-regulated AS events identified in this study, we subjected RS31 and SR34b to sqRT-PCR analysis using samples obtained from seedlings grown under various light conditions. These two SR protein genes were chosen as representatives, because the changes in the AS profiles of these genes are thought to be reflected in the AS of their numerous downstream target genes. We first changed the exposure time of etiolated seedlings to cR (8.3 µmol⋅m⁻²⋅s⁻¹), and found that the alterations in the AS patterns were dependent on the length of cR irradiation (Fig. S2A). We found that even a 2-min pulse of R (pR) could induce the response (Fig. S2A). Time-course analysis revealed that the AS changes were detected within 1 h and peaked 3 h after pR irradiation (Fig. S2E), as was the case for cR irradiation (21). Next, we exposed etiolated seedlings to various intensities of 2-min pR or 3-h cR to confirm that the AS responses were dependent on the fluence rate of R (Fig. S2 B and C). We then converted the unit on the x axes in Fig. S2 A–C into fluence, which is the product of fluence rate and exposure time, and these plots fit closely to one another (Fig. S2D), indicating that the AS responses to R obey the Bunsen–Roscoe reciprocity law, where the extent of a certain response is proportional to the total number of photons, irrespective of the fluence rate or irradiation time (27). In this case, the response was dependent on R fluence of 1–10⁶ µmol⋅m⁻², a range that corresponds to the fluence required for phyB-dependent seed germination in lettuce (Lactuca sativa) and Arabidopsis (28, 29), further supporting the notion that phyB is the major photoreceptor involved in this AS response.

The most distinguishing feature of phytochrome-mediated responses is R/FR reversibility, in which R-induced responses can be cancelled by subsequent FR irradiation. To examine whether the light-induced AS changes exhibit R/FR reversibility, we exposed WT etiolated seedlings to a 2-min pR, immediately followed by a 2-min pulse of FR (pFR) (Fig. 5 A and B). We found that the pR-induced AS alterations of RS31 were substantially suppressed by the subsequent pFR, indicating that this AS response is indeed controlled by a light-stable phytochrome, such as phyB. However, SR34b did not show a clear R/FR reversible response (Fig. 5B), probably because of the very low fluence response of phyA where even pFR can induce the response (30).

Fig. 5.

PhyA and phyB control the red light-dependent alternative splicing of RS31 and SR34b. (A) Illustration of the experimental procedure. Four-day-old etiolated seedlings were exposed to a 2-min pR (30 µmol⋅m⁻²⋅s⁻¹), 2-min pFR (60 µmol⋅m⁻²⋅s⁻¹), pR followed by subsequent pFR (pR/pFR), or were kept in darkness (D). After the exposure, the seedlings were placed in the dark for 3 h and then harvested for extraction of total RNA. (B) AS alterations in response to pR, pR/pFR, or pFR. sqRT-PCR was performed for RS31 and SR34b, and each PCR product was quantified. The ratios of mRNA-1 (for RS31) or mRNA-1 and mRNA-2 (for SR34b) to the total transcripts were compared with those of the D control, and the relative values are shown as the mean ± SE (RS31, n = 4; SR34b, n = 3). (C) pR-induced AS alterations depend on both phyA and phyB. For sqRT-PCR analysis, total RNA was extracted from pR, pFR, and D samples of WT, phyA, phyB, and phyAphyB (phyAB). Details are as in B. Data are shown as the mean ± SE (RS31, n = 5; SR34b, n = 4). (D) The N-terminal domain of phyB is sufficient to bring about the phyB-dependent AS changes. Total RNA for sqRT-PCR was extracted from pR and D samples of phyA, phyAB, and transgenic lines expressing PBG or NG-GUS-NLS in the phyAB mutant background. Details are as in B. Data are shown as the mean ± SE (n = 3). Asterisks show statistical significance compared with pR samples (B), WT (C), and phyAB (D), as determined using Student t test (*P < 0.05; **P < 0.01).

To further confirm that phytochrome is required for the light-responsive AS changes, we examined the pR-induced AS alterations in the mutants that lacked phyA and/or phyB, which are the major molecular species of phytochrome in etiolated seedlings. Although the pR-induced AS changes were marginally reduced in both phyA and phyB single-mutants, the response was almost completely abolished in phyAB double-mutants (Fig. 5C). These results clearly demonstrate that phyA and phyB are photoreceptors that play dominant and partially redundant roles in mediating R-responsive AS changes during seedling de-etiolation.

Furthermore, we observed that pFR induced smaller AS changes than did pR (Fig. 5 B and C). Because phyA is thought to be the sole receptor for FR in Arabidopsis etiolated seedlings (1), we investigated whether the pFR-induced AS response is dependent on phyA and on the fluence of FR. As expected, although the AS patterns of RS31 and SR34b in WT plants were altered in a FR fluence-dependent manner, the phyA mutant did not show any significant alterations in AS even under various fluences of FR (Fig. S2 F and G). These results demonstrate that phyA mediates AS alterations in response to FR.

Finally, we investigated which domain in the phytochrome molecule underlies the signaling that induces the light-responsive AS alterations. The N-terminal domain of phyB fused to GFP (NG), when dimerized and targeted to the nucleus by the activities of β-glucuronidase (GUS) and a nuclear localization signal (NLS), respectively (NG-GUS-NLS), is biologically fully functional in all of the phyB-mediated responses examined (3). When NG-GUS-NLS and full-length phyB fused to GFP (PBG) were overexpressed in the phyAB double-mutant background, we found that NG-GUS-NLS was as active as PBG and the endogenous phyB present in phyA single-mutants in mediating pR-responsive AS changes and seedling de-etiolation under cR (Fig. 5D and Fig. S3). These results indicate that phyB transduced the signal from its N-terminal domain to induce the light-dependent AS alterations in the nucleus. Taken together, through detailed photophysiological experiments, we have clearly demonstrated that phyA and phyB are photoreceptors that mediate light-induced AS alterations in Arabidopsis.

Concluding Remarks.

In this study, using a combination of high-throughput mRNA-seq analysis and detailed photophysiological sqRT-PCR experiments, we have provided clear evidence that phytochrome controls genome-wide AS alterations in Arabidopsis. We extracted junction reads from high-coverage mRNA-seq data and systematically compared the relative frequency of use of each splice site among WT and phyAB double-mutant etiolated seedlings with or without cR irradiation, which enabled us to conduct robust statistical analyses of R-responsive AS changes that were specifically controlled by phytochrome. As a result, we were able to obtain a highly reliable list of 2,230 Arabidopsis genes that are under phytochrome-mediated AS regulation, which had a validation rate of more than 50% (Fig. 3). We also found that 1,722 genes displayed R-responsive but phytochrome-independent AS alterations at R1h or R3h, indicating that 56.4% of the genes that showed R-induced AS changes were under phytochrome-mediated AS control (Table S4). It is well known that the relative amount of splice variants could be affected by differential mRNA degradation such as nonsense-mediated decay (NMD), where alternatively spliced transcripts containing premature termination codons are recognized for degradation (31). However, the R-dependent changes in the ratio among splice variants that we observed in this study are considered to be because of AS but not to differential mRNA degradation, because the responses to R were still observed in the NMD-impaired mutant upf1-5 (32) (Fig. S4A), and because inhibiting transcription by the treatment with α-amanitin completely abolished the response to R (Fig. S4B).

Light-regulated AS has been reported not only in different species in higher plants (16–18) but also in a chlorophyte Chlamydomonas (33). Moreover, recently, phytochrome was suggested to be involved in R-dependent AS alterations in the moss Physcomitrella (34). Therefore, phytochrome-mediated AS regulation that we demonstrated here seems to be widely conserved in plants. It has been reported very recently that light affects AS of a subset of Arabidopsis genes through a chloroplast retrograde signal, but not through photoreceptor signaling (35). We thus analyzed the overlapping target genes that are regulated by both phytochrome and chloroplast retrograde signaling pathways and found that there is basically no correlation between these two pathways (Table S5), although they are not mutually exclusive and may coexist and act coordinately in certain conditions, sharing common target genes, such as RS31 (35).

GO and other statistical analyses suggested that phytochrome-mediated control of not only TX, but also AS, have significant roles in light signaling (Fig. 1C and Table S3). We experimentally validated the phytochrome-dependent and cR-responsive AS changes of several genes, including SPA3, which has a pivotal role in light signaling, and demonstrated that phytochrome-mediated AS control of SPA3 actually contributed to promoting photomorphogenesis (Figs. 3 and 4 and Fig. S1). Taken together, these data indicate that phytochrome induces genome-wide AS alterations to mediate light responses in Arabidopsis.

Intriguingly, our GO analysis revealed that splicing factor genes are the main early targets of phytochrome-mediated AS regulation, whereas transcription factor genes are the primary direct targets of transcriptional regulation mediated by phytochrome, as previously reported (22) (Fig. 2). This finding suggests that phytochrome initiates both AS and transcription cascades, by regulating the AS of splicing factor genes and the transcription of transcription factor genes, respectively, in response to R. It remains to be elucidated how phytochrome regulates the AS of splicing factor genes within 1 h of cR irradiation. Phytochrome probably regulates the expression level (mRNA and protein level) or posttranslational modification of more upstream splicing regulators. RRC1, an SR-like splicing factor that is required for phyB signaling, is among these candidate upstream factors, because the R-responsive AS alterations of RS31 an SR34b have been shown to be partially dependent on RRC1 (21). Moreover, we found that some of the early R-responsive and phytochrome-dependent AS changes were substantially reduced in rrc1 mutants, whereas others were totally independent of RRC1 (Fig. S4C). Therefore, RRC1 is responsible for mediating phytochrome-induced AS alterations only in a subset of the target genes, which may account for the fact that phyB-mediated light responses are diminished only partially in rrc1 mutants (21).

Recently, it has been shown that phytochrome also regulates the translation of protochlorophyllide reductase mRNA by directly interacting with the RNA-binding protein PENTA1 in the cytosol (13). Therefore, phytochrome has emerged as a direct regulator of various aspects of gene expression that interacts with different partners in different subcellular locations. It would be of particular interest to establish if a common initial mechanism underlies these apparently distinct modes of phytochrome signal transduction.

Materials and Methods

Plant materials and growth conditions, α-amanitin treatment, RNA preparations and RNA-sequencing analysis, read mapping to Arabidopsis genes, comparison of splicing events, statistical tests for determining overrepresented GO categories, sqRT-PCR analysis of alternative splicing patterns, quantitative RT-PCR analysis, and generation and analysis of transgenic plants are described in SI Text.