Alternative Splicing Substantially Diversifies the Transcriptome during Early Photomorphogenesis and Correlates with the Energy Availability in Arabidopsis

Dark-Grown Seedlings Display Numerous AS Changes upon Exposure to Blue, Red, or White Light

Previous studies have established massive transcriptomic reprogramming in the switch from skoto- to photomorphogenesis (Jiao et al., 2007). To investigate the potential impact of AS on gene expression in response to altered light conditions, we analyzed transcriptomes of Arabidopsis seedlings grown for 6 d in darkness and exposed for 1 or 6 h to blue, red, or white light by RNA-seq. Two replicate samples for each time point and light quality, as well as corresponding dark controls were generated. Mapping of the 100-bp reads to the Arabidopsis genome (TAIR10 annotation) resulted in 86.0 to 207.6 × 10⁶ reads per time point, based on the two replicate samples each (Supplemental Table 1). AS events were extracted from TAIR10 and complemented by unannotated events that were found in our data resulting in 56,270 AS events. To determine quantitative changes in AS and gene expression, a previously established and validated computational pipeline was applied (Rühl et al., 2012; Drechsel et al., 2013; Drewe et al., 2013; Supplemental Data Sets 1 and 2). Upon 1 h exposure to blue (∼6 µmol m⁻² s⁻¹) or red light (∼14 µmol m⁻² s⁻¹) at intensities that, based on previous publications (Laubinger et al., 2004; Shen et al. (2007)), are expected to result in overall saturating effects on hypocotyl elongation, 81 AS events derived from 51 genes were significantly altered (false discovery rate [FDR] < 0.1; this FDR value is generally used unless otherwise mentioned; Supplemental Figure 1A). The number of AS changes massively increased after 6 h blue or red light illumination (Figures 1A and 1B). Additional AS shifts were detected upon exposure to white light (∼130 µmol m⁻² s⁻¹), representing more natural light conditions. Considering all three light settings, 700 AS events associated with 311 genes were significantly altered upon 6 h light exposure. As the rate of change in transcript steady state levels depends on transcript stability, early AS shifts will only be detectable for relatively unstable transcripts. We found only few consistent AS changes when comparing the 1- and 6-h time points for blue and red light (Supplemental Data Set 1). This limited overlap can be explained by overall weak AS changes at the early time point and the activation of downstream signaling cascades at 6 h versus 1 h light exposure. To investigate the potential role of AS in all of these light-regulatory processes, we focused our further analysis on the 6 h time point. We noticed that the AS responses for the three light qualities showed only a partial overlap, independent of the FDR cutoff value (Supplemental Figure 1B). While blue and red light primarily elicit CRY and PHY photoreceptor signaling, respectively, both signaling pathways should become active in response to white light. Thus, most of the AS changes observed in response to monochromatic light were also expected to be present upon white light exposure. To reduce the effect of AS fluctuation between samples, which is expected to be most prevalent for low-abundant splicing variants with few supporting reads, and to select for AS events that are more likely biologically relevant, we included an additional filter for effect size (change in splicing index [SI] > 0.05; Supplemental Figure 1C and Supplemental Data Set 3). Addition of the SI filter reduced the number of detected events, while the overlap between the different light qualities was still limited. We assumed that this might be caused by a too stringent FDR filter. Therefore, we considered next all events with an FDR < 0.1 in at least one color and then filtered for SI > 0.05 (Figure 1C). Applying this filter strategy resulted in strongly overlapping patterns of AS changes for all three light qualities, with 87.5 to 98.0% of all events altered in one color being also affected in at least one more light condition (Figure 1C). The majority of AS events were changed under all three light regimes, indicating common AS responses. Red light caused fewer significantly altered events upon FDR filtering and a lower median SI change of 0.079 compared with 0.137 and 0.136 in blue and white light, respectively (Supplemental Data Set 3). Red light thus had an overall weaker effect on AS than blue and white light. Both the occurrence of mostly common AS changes in response to different light qualities and, for some AS events, weaker quantitative effects of red light were confirmed by the validation experiments (see below).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Changes in AS and Gene Expression in the Course of Photomorphogenesis Triggered by Blue, Red, and White Light.

(A) and (B) Venn diagrams showing the numbers of significantly altered AS events (A) and the corresponding numbers of genes affected (B) upon 6 h exposure to blue (∼6 µmol m⁻² s⁻¹), red (∼14 µmol m⁻² s⁻¹), and white (∼130 µmol m⁻² s⁻¹) light (FDR < 0.1). Events or genes showing changes in opposing directions under two light conditions were excluded. Total numbers of events/genes changing under each light condition are given in parentheses.

(C) Venn diagram of significantly altered AS events after 6 h light exposure upon additional filtering based on the effect size (change in SI > 0.05). Only events with an FDR < 0.1 under at least one light condition were considered for SI analysis. Events with SI changes in opposite directions under two conditions were excluded. Events are grouped according to their SI only.

(D) Venn diagram of genes changing in total expression (TX) upon 6 h exposure to blue, red, and white light (FDR ≤ 0.1). Total numbers of genes changing under each light condition are given in parentheses.

(E) and (F) Genes exhibiting changes in AS, TX, or both upon 6 h white (E) and blue (F) light exposure.

(G) Gene Ontology term analysis of genes undergoing AS or TX changes, upon 6 h white or blue light exposure compared with all AS events detected and all genes in MapMan, respectively. met., metabolism; misc., miscellaneous; cell div. and dev, cell division and development; asterisks indicate terms overrepresented compared with all AS events and all genes in MapMan, respectively, with Bonferroni-corrected P < 0.05.

Of all 56,270 AS events detected in this analysis, 46.9, 22.4, 21.2, and 9.5% corresponded to alternative 3′ splice sites, regulated introns (varying rate of intron retention/splicing), alternative 5′ splice sites, and regulated exons (cassette exons), respectively (Supplemental Figure 2A). The AS analysis in this work is based on a heuristic method, and it was demonstrated to compare favorably to related approaches (Kahles et al., 2016). Similar frequencies of the different AS types have been observed in previous studies using the same, but also with different AS analysis pipelines, and are also found for the TAIR10 annotation (Supplemental Table 2). According to these data sets, alternative 3′ splice sites are most abundant, whereas a previous survey of AS in Arabidopsis identified intron retention as the prevalent AS type (Marquez et al., 2012). These discrepancies most likely result from using different computational approaches for defining splicing variants, including many low-abundant isoforms that might not be biologically relevant. Indeed, a very different distribution was found for the 700 light-regulated AS events: cassette exons and regulated introns were enriched, representing 18.0 and 37.1%, respectively, of all AS events altered in response to at least one light quality, while lower fractions of alternative 3′ (27.3%) and 5′ (17.6%) splice sites were observed. Among the light-regulated AS events, we also identified several exitrons (Supplemental Table 3), a class of cryptic introns that reside within the coding region of transcripts (Marquez et al., 2015). Analyzing the direction of the shift for the AS events that were significantly altered in response to at least one light quality, we observed strong biases for the intron retention and cassette exon events (Supplemental Figure 2B). In 74% of the significantly changed intron retention events, light triggered a shift toward the spliced, i.e., shorter transcript variant, while in 67% of the cassette exons a relative increase of the skipping variant was detected. Proportions of alternative up- and downstream 5′ and 3′ splice site usage, respectively, were also elevated in response to light, but this effect was less pronounced than for the other two AS types. To assess the potential consequences of light-regulated AS on the expression of the corresponding genes, we compared the positions of all events to those displaying significant changes (Supplemental Figure 2C and Supplemental Data Sets 4A to 4D). For the light-regulated AS events, the fractions of events associated either with the coding sequence or the 3′ untranslated region (UTR) were decreased and increased, respectively, in comparison to all events. For all subsets, however, most of the AS events overlapped with the coding sequence. Based on their positions within the pre-mRNAs, light-triggered AS events can affect the coding and regulation potential of the resulting mRNAs.

Previous studies have revealed widespread coupling of AS and nonsense-mediated decay (NMD) in Arabidopsis (Kalyna et al., 2012; Drechsel et al., 2013). To assess the prevalence of coupled AS-NMD in the context of light regulation, the occurrence of NMD-triggering features in the corresponding splicing variants was analyzed. To this end, the AS events were integrated into the representative transcript isoform from TAIR10, followed by the detection of upstream open reading frames, premature termination codons (PTCs), and long 3′ UTRs (Supplemental Data Sets 4A to 4D). Remarkably, 77.2% of all light-regulated AS events exhibit NMD features within the splicing isoform that is relatively more abundant in the dark samples. Furthermore, 61.1% of all events showed a relative switch from a putative NMD target to a non-NMD-regulated transcript variant upon light exposure. The corresponding fractions were even larger when only considering events within the coding sequence, which accounted for most NMD-triggering features in those transcripts. Further evidence for coupling of light-regulated AS and NMD was provided by comparing the sets of significant events from this study and from a previous analysis upon NMD impairment (Drechsel et al., 2013): ∼10% of all light-regulated events have previously been established to involve NMD control (Supplemental Data Sets 4E to 4G). Notably, the seedlings analyzed in this and the previous work substantially differed in their developmental stage and growth conditions. The frequency of coupled AS-NMD was analyzed in light-grown seedlings, while this study revealed that most of the light-regulated AS-NMD events showed downregulation of the putative NMD form in light. Thus, the overlap might be even higher when analyzing seedlings cultivated under identical conditions. In conclusion, light-triggered AS typically mediates a switch from a presumably NMD-regulated transcript to a protein-coding alternative variant, enabling the activation of gene expression in the transition from skoto- to photomorphogenesis.

The RNA-seq data were also analyzed for differential gene expression (Anders and Huber, 2010; FDR ≤ 0.1; Supplemental Data Set 2). In line with previous findings (Jiao et al., 2007), a substantial fraction of all genes were significantly up- or downregulated in response to light (Figure 1D; Supplemental Figure 3). Out of 33,602 genes in the TAIR10 annotation, 23,432 genes were expressed in our data set when considering all samples (FDR ≤ 0.1; method based on Gan et al., 2011). Furthermore, 10,271 (43.8%) of the expressed genes showed altered transcript levels in response to at least one light quality for the 6 h time point. White, blue, and red light changed the expression of 9336, 4381, and 4251 genes, respectively. When setting a threshold of an at least 2-fold change in transcript levels, 3439, 2406, and 2020 were differentially expressed upon seedling exposure to white, blue, and red light, respectively. This adds up to a total number of 4310 genes, corresponding to 18.4% of all expressed genes. Patterns of differential gene expression in response to blue and red light showed a huge overlap and most of the changes in transcript levels upon illumination with monochromatic light were also detected under white light. Furthermore, blue and red light affected the expression of a comparable number of genes, while on the level of AS red light was less effective than blue light. Moreover, many transcriptional changes were only found in response to white light, possibly as part of an adaptive program that is not activated by weak, monochromatic light. When considering differential expression separately for up- and downregulated genes, slightly more genes were induced than repressed at the 6 h time point. We also analyzed differential gene expression for the samples exposed to light for 1 h (Supplemental Figure 3). In line with the observations on the level of AS, fewer changes were detected for the 1 h compared with the 6 h time point. For this earlier time point, the number of induced genes was approximately twice the number of downregulated ones. Fewer down- than upregulated genes cannot only be explained by a lower number of repressed than induced genes, but also by the stability of the transcripts: A significant decrease in steady state transcript levels as a result of diminished transcription within 1 h is expected to be detectable only for highly unstable transcripts.

To test if light affects both expression levels and AS of genes, the corresponding gene lists were compared separately for white, blue, and red light (Figures 1E and 1F; Supplemental Figure 1D). For all light qualities, a substantial fraction of the genes showing changes in AS had unchanged total transcript levels. Given that AS can contribute to quantitative gene control, for example by generating destabilized NMD targets (Drechsel et al., 2013), the number of light-regulated genes displaying both altered AS and differential gene expression in a splicing-independent manner might be even smaller. In summary, the altered light status triggers complex transcriptome reprogramming, involving changes in both gene expression and, for a smaller, mostly distinct set of genes, AS.

Analyzing the functional categories of genes associated with light-regulated AS revealed an overrepresentation of the terms “RNA” and “metabolism” for blue light and “RNA” for white light (Figure 1G; Supplemental Data Set 5). The overrepresentation of the “RNA” category is in line with previous publications (Filichkin et al., 2010; Rühl et al., 2012; Drechsel et al., 2013), showing extensive regulated AS for genes involved in RNA metabolism. Since numerous intergenic regions are expressed in an NMD-regulated manner (Drechsel et al., 2013), we compared read accumulation in intergenic regions for the dark- and light-exposed samples (Supplemental Data Set 6). Several of these transcriptional units were found to overlap with previously identified long intergenic RNAs (Liu et al., 2012a). Read coverage for some of these regions differed substantially between the light conditions tested here. However, total expression levels were low in most cases and further studies are required to test the functional relevance of these transcripts.

Finally, to rule out the occurrence of rhythmic expression in the absence of light, transcript levels of circadian genes were analyzed in the dark and light samples (Supplemental Table 4). When comparing the 0 and 6 h dark samples, no significant change was detectable for any of the genes. By contrast, light altered the expression of several of these genes, in line with the known role of light in influencing circadian expression patterns (Jiao et al., 2007).

Validation of Light-Regulated AS Events

We next selected candidates from the list of light-regulated AS events for an independent experimental validation (Figure 2). This selection covered different functional categories of genes, including splicing factors (Figures 2A to 2D), putative transcription factors (Figures 2E to 2G), and a photosynthetic component (Figure 2H). All of the candidate events were confirmed, in line with the high validation rate observed in previous studies that applied the same pipeline for AS analysis (Rühl et al., 2012; Drechsel et al., 2013). For some genes, several AS events were detected and we focused our analysis on the major splicing variants, which were also sequenced (Supplemental Figure 4). For all events, the AS ratios were changed in response to blue, red, and white light. However, the extent of splicing change differed for some candidates (Figure 2). For those candidates, white light generally caused the strongest AS shifts. Furthermore, for three out of nine candidates, a weaker change in response to red compared with blue and white light was observed. These findings are in agreement with differences in the SI changes for the three light qualities from the RNA-seq data.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Validation of Light-Dependent AS in Genes from Different Functional Categories.

Splicing variants of genes encoding splicing factors ([A] to [D]), putative transcription factors ([E] to [G]), a photosynthetic component (H), and a hypothetical chloroplast protein (I) were coamplified from samples grown in darkness and collected at 0 h or after 6 h exposure to white (∼130 µmol m⁻² s⁻¹), blue (∼6 µmol m⁻² s⁻¹), or red (∼18 µmol m⁻² s⁻¹) light (top, middle, and bottom bars) and quantified using a Bioanalyzer. Shown are representative agarose gels with double arrowheads pointing at 300 bp of a DNA size ladder with 100-bp increments, and PCR products from 0 h (left) and 6 h white light (right) samples. The variants quantified are labeled (.1 or .2), and partial ([A] to [C], [E], [G], and [I]) or full ([D], [F], and [H]) gene models are shown with introns represented by lines and exons by boxes. Regions colored in dark gray are UTRs, and asterisks mark the introduction of a premature termination codon. Solid arrowheads show the positions of the primers used, and the arrow in (C) indicates a splice-junction-spanning primer. Bars give average relative splice form ratios with the ratio in darkness set to 1, as indicated by the light-gray background bar for each color. Error bars are sd, n = 3. Scale bars beneath the models represent 500 bp.

Light-Regulated AS of RRC1 Results in a Self-Enforcing Circuit

Previous work had identified the putative splicing factor RRC1 as a novel component of PHY-dependent light signaling (Shikata et al., 2012b). Interestingly, our RNA-seq data suggested that the inclusion of the third exon of RRC1 is regulated in a light-dependent manner (Figure 3A). Analyzing the AS pattern of this region via RT-PCR supported the notion that blue, red, and white light caused a shift toward the inclusion variant compared with dark samples (Figure 3B). Separate quantitation of the two splicing variants revealed that light exposure resulted in slightly elevated levels of the representative RRC1.1 variant and diminished amounts of RRC1.2 (Supplemental Figure 5A). Opposite changes in the levels of the two splicing variants were also observed for the light-regulated event in SR30 (Supplemental Figure 5B), indicating that those shifts in splicing variant ratios are caused by AS and not by an altered transcript turnover rate. The exon skipping variant RRC1.2 gives rise to a frame shift, resulting in a PTC two exons further downstream. We therefore assumed that this AS variant is targeted by NMD, which was corroborated by its accumulation in two mutants impaired in NMD activity (Figure 3C).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Light Promotes AS of RRC1 to the Variant Required for Functioning in Phytochrome Signaling.

(A) Models of RRC1 major splicing variants showing exons as boxes and introns as lines. UTRs are dark gray. The positions of the coamplification primers are given by arrowheads, and the insertion site of the T-DNA in the rrc1-2 mutant is indicated. The asterisk marks the introduction of a premature termination codon in the RRC1.2 variant. The coamplified PCR products in 0 h (left) and 6 h white light (right) samples separated on a gel are shown with the double arrowhead pointing at 100 bp of a 50-bp ladder. Transcript models are aligned to corresponding amplification products. Below the transcript models, coverage plots show representative RNA-seq results for a 0 h and 6 h light sample. The alternatively spliced region is shown in black.

(B) Confirmation of light-dependent AS under ∼130 µmol m⁻² s⁻¹ white (left), ∼6 µmol m⁻² s⁻¹ blue (middle), and ∼18 µmol m⁻² s⁻¹ red (right) light. Splicing variants were coamplified from samples grown in darkness for 6 d and collected at 0 h or after 1 h or 6 h exposure to light, and quantified using a Bioanalyzer. Bars give average splice form ratios with the ratio in darkness set to 1. Error bars are sd, n = 3.

(C) Splicing variants were coamplified from etiolated wild-type plants, or indicated NMD-deficient mutants, and quantified as in (B). Ratio in the wild type is set to 1; error bars are sd, n = 3.

(D) Violin plots showing the distribution of the relative hypocotyl lengths measured in red light (∼10 µmol m⁻² s⁻¹) for rrc1-2, wild-type, and complementation lines (top). In each violin, the dashed line represents the median, and the dotted lines the quartiles. All hypocotyls were normalized to the average length in darkness of each line. Complementation constructs express tagged splicing variants (.1 or .2) or the genomic sequence (g) under control of the CaMV 35S promoter. Asterisks indicate P values from Mood’s median test compared with the wild type: *P < 10⁻², **P < 10⁻¹¹, and ***P < 10⁻¹⁷. Exact P values for all comparisons are provided in Supplemental Figure 6E; n is indicated above each genotype. For each complementation construct, three independent F1 lines were each analyzed once in a total of three independent experiments. Bottom panel shows representative seedlings from hypocotyl assays in darkness and under red light (cR).

To test the functional significance of light-regulated AS of RRC1, complementation of the rrc1-2 mutant, carrying a T-DNA insertion (Figure 3A) and previously described as a knockdown allele (Shikata et al., 2012b), was performed. Subsequently, we determined hypocotyl lengths of seedlings grown in red light or darkness (Figure 3D). Median hypocotyl lengths in darkness were similar for all tested lines (Supplemental Figure 6A) and lengths measured in red light were normalized to the average dark value for each line to correct for a potential light-independent growth phenotype. In line with the previous report by Shikata et al. (2012b) and the role of RRC1 as a positive regulator of light signaling, rrc1-2 had longer hypocotyls than the wild type (Figure 3D). This phenotype could be rescued upon complementation with the splicing variant RRC1.1, but not RRC1.2, under control of the constitutive 35S promoter. Complementation with a corresponding genomic construct also resulted in significantly shorter hypocotyls compared with the rrc1-2 mutant, even though the median length was still slightly elevated compared with the wild type. The differences in hypocotyl lengths when comparing complementation with RRC1.1 and the genomic construct might be caused by varying levels of overexpression (Supplemental Figure 6B). Analysis of the RRC1 levels for the two splicing variants and total transcripts confirmed robust and specific expression of the constructs in all transgenic lines. However, the genomic construct resulted in massive overaccumulation of RRC1 transcripts compared with a more moderate increase in the cDNA lines. To exclude an effect of using a strong constitutive promoter for the complementation, the constructs based on the two AS variants were also expressed in the wild-type background. None of these lines displayed altered hypocotyl lengths compared with the wild type (Supplemental Figures 6C to 6E). Moreover, immunoblot analysis allowed the detection of a protein corresponding to the splicing variant RRC1.1, but not RRC1.2 (Supplemental Figure 6F), further suggesting that the exon skipping variant is subject to NMD and does not lead to RRC1 protein. In line with the transcript data, protein levels were much higher in the plants expressing the genomic construct compared with complementation with the RRC1.1 construct. The strong overexpression of RRC1 upon complementation with the genomic construct might result in perturbed downstream signaling, which would explain the only partial rescue of the hypocotyl elongation phenotype in case of this construct. To exclude such effects, the rrc1-2 mutant was also complemented with constructs under control of the RRC1 promoter. Indeed, the genomic RRC1 sequence under control of the endogenous promoter fully rescued the mutant phenotype (Supplemental Figures 7A and 7B). Expressing the two RRC1 splicing variants under control of the endogenous promoter resulted in transcript levels that were substantially lower than in the wild type, possibly due to the absence of introns (Supplemental Figure 7C). Accordingly, functional complementation of the hypocotyl phenotype was found only for the RRC1.1 line with the highest expression (Supplemental Figure 7D). As expected, none of the RRC1.2 lines showed a rescue of the mutant phenotype. Taken together, light-stimulated inclusion of the cassette exon represents a mechanism to induce functional RRC1 expression, thereby increasing the levels of a positive regulator of light signaling.

Contribution of Photoreceptors to AS Changes during Photomorphogenesis

Recent studies in P. patens (Wu et al., 2014) and etiolated Arabidopsis seedlings (Shikata et al., 2014) suggested a major role of PHY photoreceptors in red light-dependent AS, whereas altered AS patterns in leaves subjected to varying light conditions were attributed to retrograde signaling (Petrillo et al., 2014b). These seemingly controversial findings may result from different experimental settings and suggest that various factors can influence AS patterns under changing light conditions. To further address this intriguing aspect, we first compared the AS patterns of five confirmed candidates in etiolated wild-type and phyA phyB double mutant seedlings upon illumination with white light (Figure 4A). For all events, very similar patterns of light-induced AS changes in the comparison of wild-type and phyA phyB mutant seedlings were observed; significant differences between the wild type and the mutant were only found for single events and time points and did not correlate with the overall light response. Interestingly, seedling growth on sugar-containing medium, as in the RNA-seq experiment, shifted the AS ratio into the same direction as light (Figure 4A, right panels). The relative change upon 6 h of light exposure, however, was identical for seedlings grown without and with external sugar supply and also did not differ between the wild type and phyA phyB (Supplemental Figure 8).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Contribution of Phytochrome A/B Signaling to AS Control Becomes Visible in Red and Far-Red, but Not in White Light.

Etiolated seedlings grown on plates with or without 2% sucrose for 6 d were exposed to ∼130 µmol m⁻² s⁻¹ white (A), ∼28 µmol m⁻² s⁻¹ red (B), or ∼15 µmol m⁻² s⁻¹ far-red (C) light for the indicated periods. Splice variants were coamplified and quantified using a Bioanalyzer. AS ratios in (A) and (C) were normalized to the one measured for the corresponding wild type at 0 h on plates without sucrose, separately for each replicate of wild-type and phyA phyB sample sets. In (B), ratios were normalized to the mean value of the wild-type replicates at 0 h (− Suc). Displayed are mean values + sd ([A], n = 5 to 7 for SR30 and RRC1; other candidates: n = 3; [B], n = 4 for SR30; other candidates n = 3; [C], n = 3). P values: *P < 0.05, **P < 0.01, ***P < 0.001, comparing the wild type and phyA phyB in an independent t test, or, if the wild type is set to 1, in a one-sample t test.

Our data did not provide evidence for a critical role of the two major red light photoreceptors PHYA and PHYB in triggering AS changes upon exposure to white light. Given that white light could trigger AS via red and blue light signaling, we next analyzed changes in AS upon exposure to red light (Figure 4B; Supplemental Figure 8). For four out of five events, red light resulted in an AS change in both wild-type and phyA phyB seedlings. Interestingly, the mutant showed a significantly weaker red light response than the wild type for MYBD. Furthermore, the AS ratio of PPL1 was not significantly changed in comparison of darkness and red light in the mutant. These data suggested the existence of alternative pathways controlling light-triggered AS and that a contribution of PHYA/PHYB only becomes detectable for some events under red light. We therefore assumed that exposure of seedlings to far-red light, which triggers PHYA signaling but does not support photosynthesis and the associated signaling, should result in more distinct AS responses in wild-type and phyA phyB seedlings. Far-red light caused an AS shift of all tested events in the wild type, albeit quantitative changes for SR30 were substantially lower than under white light (Figure 4C; Supplemental Figure 8). Remarkably, no or only very weak effects of far-red on the AS pattern in the mutant was detected, revealing the dependency on PHY under this particular light condition.

The differences in AS responses under white, red, and far-red light highlight the occurrence of multiple signaling pathways and that the major PHYs A and B are not essential in this process under white light. Besides PHY signaling, white light also activates the blue light-responsive cryptochrome photoreceptors. To test for a potential role of CRYs, AS changes upon exposure to white light were compared between wild-type and cry1 cry2 mutant seedlings. All tested events showed the same white light response in wild-type and cry1 cry2 seedlings (Figure 5A; Supplemental Figure 8). Overall similar AS changes in wild-type and cry1 cry2 seedlings were also detected upon blue light exposure (Figure 5B). The relative AS changes between darkness and 6 h blue light were slightly more pronounced in the wild type than in the mutant (Supplemental Figure 8); however, this quantitative difference was statistically significant only for SR30. Taken together, neither PHYA/B-dependent red light signaling nor CRY-mediated blue light signaling are essential in causing the AS changes in etiolated seedlings exposed to white light. This observation could be explained by alternate signaling through either PHYs or CRYs; in this case, however, no AS changes would be expected upon red and blue light exposure of the phyA phyB and cry1 cry2 mutant, respectively. While a role of other photoreceptor types cannot be fully excluded, it seems more likely that another, photoreceptor-independent signaling pathway is involved in light-responsive AS during photomorphogenesis.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

AS Shifts in Response to White Light Are Comparable in Wild-Type and cry1 cry2 Seedlings.

Etiolated seedlings grown on plates with or without 2% sucrose for 6 d were exposed to ∼130 µmol m⁻² s⁻¹ white (A) or ∼4 µmol m⁻² s⁻¹ blue (B) light for the indicated periods. Splice variants of the indicated genes were coamplified and quantified using a Bioanalyzer. AS ratios were normalized to the one measured for wild-type at 0 h on plates without sucrose. Displayed are mean values + sd (n = 3 to 4). Statistical comparison of the wild type and cry1 cry2 using independent t test, or, if the wild type is set to 1, in a one-sample t test (*P < 0.05).

Illumination of etiolated seedlings will not only activate photoreceptor signaling and photosynthesis, but also entrain the expression of circadian regulators. Previous reports revealed intricate links between the circadian clock and AS in plants (Sanchez et al., 2010; James et al., 2012). To address a potential impact of circadian regulators, we tested light-triggered AS in mutants defective in different components of the circadian clock (Supplemental Figure 9). Only for the prr7-3 prr9-1 double mutant a slightly weaker AS change was seen in case of RRC1, suggesting that overall circadian regulators do not play a major role in the control of these AS events in the early phase of photomorphogenesis.

AS Output Correlates with the Plant’s Energy Supply

Analysis of the light-responsive AS in etiolated seedlings revealed that not only light, but also the growth conditions had a major impact on the splicing outcome. Etiolated seedlings grown on sucrose-containing medium had AS ratios shifted into the same direction as observed for seedlings grown without sucrose and exposed for 6 h to light (Figures 4 and 5). To further dissect how light and sugar can alter AS patterns, wild-type seedlings grown in darkness and on sugar-free medium were transferred to liquid medium with or without sugar and kept in darkness or exposed to white light. To account for the osmotic effect of sugar supplementation, an additional control with an equimolar concentration of mannitol was included. No significant change in AS was detected upon 1 h incubation in darkness when comparing medium without supplement, with mannitol, and with sucrose (Figure 6A). In line with the previous findings, light exposure resulted in a pronounced AS shift already after 1 h of illumination. Interestingly, at the 6 h time point, the sugar-treated and dark-kept seedlings showed a pronounced AS shift in the same direction and of a similar extent as observed in light without sugar supply. For several events, the presence of both sugar and light caused an even stronger AS shift than the single treatments (Supplemental Table 5).

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

Sucrose and Light Cause Comparable AS Shifts.

(A) Seedlings were grown in darkness for 6 d and afterwards incubated in control medium (MS), mannitol (Man), or sucrose (Suc) solutions for 1 or 6 h in darkness or ∼130 µmol m⁻² s⁻¹ white light. Alternative splice forms were coamplified and quantified using a Bioanalyzer. Ratios were normalized to the corresponding control samples in darkness. Color scheme for dark (gray) and light (yellow) samples applies to all panels. Displayed are mean values (n = 3). Error bars are sd; P values: *P < 0.05, **P < 0.01, and ***P < 0.001 comparing mannitol and sucrose to the MS light and dark samples, respectively. Tests are independent t tests when not tested against 1, and one-sample t test when tested against 1. Detailed results from the statistical analysis are provided in Supplemental Table 5.

(B) AS analysis of SR30 in seedlings grown in darkness and incubated in control medium (MS), sucrose (Suc), mannitol (Man), glucose (Glc), or trehalose (Tre) solutions for 6 h in darkness or ∼130 µmol m⁻² s⁻¹ white light. Data are normalized to MS sample in darkness. Displayed are mean values (n = 3 to 5) + sd. Statistical comparison between MS and different sugars as detailed in (A); dark and light samples were analyzed separately.

(C) RT-qPCR analysis of transcript levels for SnRK1.1 targets. Sample description and data normalization, depiction, and statistical analysis as described in (A). Data displayed on a log scale.

Many of the light-induced AS events are expected to be coupled to NMD. To test if the AS shift under these conditions might be, at least to some extent, a consequence of altered NMD activity, we compared the AS response to light and sucrose in etiolated wild-type and NMD mutant seedlings. Analysis of three predicted AS-NMD events revealed identical AS shifts in wild-type and lba1 seedlings upon exposure to light and sucrose (Supplemental Figure 10), irrespective of the accumulation of the predicted NMD variant in the lba1 mutant. These data and the separate quantitation of splicing variants (Supplemental Figure 5) indicate that the changes occur on the level of AS and not downstream of it.

The strong effect of sucrose feeding on the AS output is in line with a previous study, suggesting that retrograde signaling contributes to AS control in light-grown plants (Petrillo et al., 2014b). Accordingly, light-mediated AS is suppressed in green plants upon chemical inhibition of photosynthesis by DCMU (Petrillo et al., 2014b; Supplemental Figure 11). Treatment of etiolated seedlings with the same inhibitor also slightly weakened, yet did not completely abolish the AS shift in our study (Supplemental Figure 11). The weaker suppression of light-mediated AS by DCMU in etiolated seedlings might be explained by the absence of an active photosynthesis apparatus. Indeed, different effects of DCMU on light- and dark-grown plants have been described before (Mancinelli, 1994). Alternatively, upon disruption of photosynthesis, photoreceptor-mediated AS control might become detectable in etiolated seedlings, but not in light-grown plants.

We next tested the effect of different sugars on the AS output. Treatment of etiolated seedlings with sucrose, glucose, or trehalose in the absence or presence of light revealed that sucrose was most effective (Figure 6B). Exposure to 0.2% sucrose caused a strong AS shift, which was further enhanced in the presence of 2% sucrose (Figure 6B). Glucose feeding caused a slightly weaker AS shift than sucrose. We also treated seedlings with trehalose to trigger accumulation of trehalose 6-phosphate (Schluepmann et al., 2004), which has previously been described as a signal for carbon availability (Schluepmann et al., 2012); however, trehalose exposure had only a minor effect on the splicing outcome. Based on these findings, we postulate that the AS output might be regulated in response to the plant’s energy supply, possibly mediated by the level of sucrose, the major transport form of photoassimilates.

Upstream Signaling Involved in Light- and Sugar-Mediated AS Changes

Both sugar feeding and light-driven photosynthesis alter the plant’s energy signaling, which might be an integration point resulting in the AS changes observed here. Independently acting systems for sensing the plant’s energy status have been described, including HEXOKINASE1 (HXK1) and SUCROSE-NON-FERMENTATION1-RELATED KINASE1 (SnRK1; Sheen, 2014). To test their potential relevance under the conditions of our experiments, transcript levels of HXK1, CHLOROPHYLL A/B BINDING PROTEIN1 (CAB1), which is known to be induced by light (Brusslan and Tobin, 1992), and the SnRK1 targets DARK INDUCED1 (DIN1) and DIN6 were measured in seedlings transferred to control or supplemented media and incubated for 6 h in light or darkness (Figure 6C; Supplemental Figure 12). Expression levels of HXK1 were unaffected by both sugar and light, while CAB1 transcript levels were elevated only in response to light. However, DIN1 and DIN6 levels correlated with the AS pattern shifts, altering in response to sugar and light. DIN1 and DIN6 transcript levels were reduced in response to light and, to an even greater extent, upon sugar exposure. As for the AS shifts, the maximum effect was visible when both sugar and light were supplied. The light- and sugar-responsive expression patterns of DIN1 and DIN6 are in agreement with previous reports (Thum et al., 2003; Baena-González et al., 2007).

The resemblance of the AS and DIN expression changes in response to light and sugar application would be in line with a coupling of these processes, which we first tested using a snrk1.1 mutant. Molecular characterization of the snrk1.1-3 mutant revealed that the T-DNA insertion resulted in an altered transcript and no detectable SnRK1.1 protein (Supplemental Figures 13 and 14). Comparing the AS patterns in etiolated wild-type and snrk1.1-3 seedlings showed a slightly different sugar response in darkness (Supplemental Figure 15A). However, both the AS and DIN expression responses (Supplemental Figure 15B) were overall similar in wild-type and mutant seedlings, suggesting remaining activity of the mutant allele or functional redundancy of the two close homologs SnRK1.1 and SnRK1.2. In line with this notion, previous studies have shown that the snrk1.1-3 mutant does not have an obvious growth phenotype (Mair et al., 2015) and that plants are impaired in development and stress responses only upon transient knockdown of both SnRK1.1 and SnRK1.2 (Baena-González et al., 2007).

Signaling through SnRK1 is dependent on its kinase activity and can be disrupted by treatment with the kinase inhibitor K252a (Baena-González et al., 2007). Thus, chemical inhibition of SnRK1 is expected to mimic its inactivation under conditions of increased energy availability. However, it should be noted that K252a has a broad target spectrum, resulting in the inhibition of various kinases, and not exclusively SnRK1. Treatment of etiolated seedlings with K252a in the dark changed the AS ratio for SR30, PPD2, and MYBD as sucrose supply or light exposure did (Figure 7). Furthermore, in the presence of K252a, the effect of sucrose supply in darkness on the AS ratio of SR30 and PPD2 was significantly enhanced compared with the corresponding controls without inhibitor. An additional effect of K252a on the AS ratio was also observed for some of the light-treated samples. In the case of RRC1 and PPL1, K252a treatment changed AS into the opposite direction compared with sucrose and light treatment. These different AS responses could be explained by the involvement of distinct sets of splicing factors and their regulation upon inhibitor treatment in a SnRK1-dependent and -independent manner, in line with the broad target spectrum of K252a. Thus, generation of a mutant specifically impaired in both SnRK1.1 and SnRK1.2 activities will be needed to test for a direct role of SnRK1 signaling in these splicing pattern changes. Taken together, our data suggest that in plants AS represents an integration point of multiple signaling pathways that are responsive to altered energy availability.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

Light- and Sucrose-Regulated AS Involves Kinase Signaling.

Wild-type seedlings were grown in darkness and then incubated in control medium (mannitol [Man]) or sucrose (Suc) solution for 3 or 6 h in darkness or ∼130 µmol m⁻² s⁻¹ white light. Incubation was performed in the absence (control) or presence of the kinase inhibitor K252a. AS ratios were determined as described before and are normalized to corresponding Man dark samples. Displayed are mean values + sd (n = 3); P values: *P < 0.05, **P < 0.01, and ***P < 0.001 comparing control and K252a treatment. Significant differences determined using independent t tests when not tested against 1, and one-sample t test when tested against 1.