Casting Away the Shadows: Elucidating the Role of Light-mediated Posttranscriptional Control in Plants^†

Light-mediated regulation of AS

During transcription, pre-mRNAs undergo processing to give rise to mature mRNAs. First, the pre-mRNA receives a 7-methylguanosine cap at its 5′ end to protect the molecule against degradation, and then, it is subjected to splicing to remove the introns, sequences that will not appear in mature mRNA 17, 18. The splicing process is catalyzed by the “spliceosome,” a dynamic complex consisting of five small nuclear ribonucleoproteins particles (snRNPs; U1, U2, U4, U5 and U6) and a variety of auxiliary proteins 19. During this cleavage and rejoining of the RNA, the spliceosome assembles on exon–intron boundary sequences known as 5′ donor splice sites (5′SS) and 3′ donor splice sites (3′SS), while the auxiliary RNA-binding proteins allow or inhibit the recruitment of the spliceosome to neighboring splice sites by recognizing other sequence motifs present in the pre-mRNA 17, 18. Not all splicing sites are used constitutively; exons may be combined in many different ways through AS. The use of alternative 5′ and 3′ splicing sites causes that variable portions of introns to be removed and variable portions of exons to remain in the mRNA (Fig. 1). Exon skipping (ES) involves the removal of an exon together with their flanking introns (Fig. 1). Finally, intron retention (IR) gives place to the inclusion of an intron in the mature mRNA, being this the most frequent AS event reported in plants (Fig. 1) 20. Variability in the splicing pattern of a single pre-mRNA generates proteins harboring different domain combinations; thus, AS considerably increases the coding capacity of a genome 17. Likewise, AS can generate aberrant isoforms that are targets for RNA degradation. The retention of introns alters the open reading frame, which can result in a premature termination codon (PTC). Many PTCs are recognized by the nonsense-mediated decay (NMD) mechanism that leads to the degradation of the transcripts 21, 22, establishing a connection between transcript level regulation and AS 23.

All previously mentioned posttranscriptional processing steps are subject to regulation and have been extensively studied in plants 24, 25; nevertheless, our current understanding of the role of light-guided posttranscriptional mechanisms is limited to specific examples. The first light-regulated AS event was described for the HYDROXYPYRUVATE REDUCTASE gene in pumpkin in 1999 8, and it was not until 2014 that high-throughput sequencing technologies were employed to address this issue on a genomewide scale 14. Wu et al. analyzed transcriptome changes during a 1- or 4-h exposure to red or blue light in the moss Physcomitrella patens (Fig. 2a), investigating the AS response at these wavelengths and addressing the existence of putative light-dependent mechanisms controlling splicing. The authors reported thousands of IR events occurring specifically in response to either blue (6718 IR events) or red (6824 IR events) light. Of all differentially spliced IR events observed in the experiments, 2748 were in common to both wavelengths (Fig. 2b), representing ∼29% of all events observed in the red- or blue-light conditions. By performing a Gene Ontology (GO) term analysis using the top 1000 events regulated either by red or blue light, we found a significant enrichment of terms related to RNA translation. Many of the blue-light-regulated IR events involved genes encoding proteins related to photosynthesis, implying that blue light immediately regulates photosynthetic activity by splicing transcripts required for this process. Neither splicing-related nor light-signaling genes appeared to be enriched in our analysis; however, the authors validated several IR events for a hand-picked subset of transcripts belonging to these processes. To explore this issue further, we performed a new meta-analysis of these data. Based on gene lists provided by the authors, we found that splicing-related genes represent around 1% of the genes with altered IR under red or blue light exclusively, as well as the shared IR events (P < 8.075e-10). On the other hand, light-signaling-related genes comprised less than 0.7% of the statistically significant IR events in all of these data sets (P < 5.204e-09) (Fig. 2d). This evidence supports the conclusions made by Wu et al. that light regulates the AS of several splicing factors, suggesting that AS could be a mechanism to regulate downstream transcripts such as those involved in photomorphogenesis. Additionally, the authors analyzed previous RNA-seq data from a mutant defective in phytochrome chromophore biosynthesis 26 together with data from a set of phytochrome knockout mutants to identify whether these photoreceptors mediate red-light-regulated AS. The red-light-induced AS observed in wild-type plants was reduced in this series of mutants, indicating that phytochromes participate in splicing regulation. Remarkably, Wu et al. showed that more than 70% of the top 1000 IR events upregulated by 1 h of either red- or blue-light illumination displayed no change in gene expression, revealing that light-activated IR is not associated with transcriptional activity. Interestingly, a repetitive GAA cis-element was enriched at both the 5′SS and 3′SS of retained introns in red- and blue-light-responsive AS events, suggesting that this GAA motif is an exonic splicing regulator that functions in light-regulated IR.

In the same year that Wu et al. employed RNA-seq technology to analyze light-regulated AS in Physcomitrella patens, Shikata et al. used similar resources to address light-mediated AS during photomorphogenesis in Arabidopsis 15. They identified genes with phytochrome-dependent AS changes by performing a genomewide transcriptome profiling on a series of time courses of four-day-old wild-type and phyAphyB double-mutant (phyAB) etiolated seedlings exposed to continuous red light for 1 or 3 h (Fig. 2a). After 1 h of light treatment, 1505 transcripts exhibited phytochrome-mediated AS, whereas 1678 genes were differentially expressed, with only 187 transcripts moderated by both processes. The 3-h treatment revealed 1116 and 4382 transcripts with phytochrome-dependent AS and differential expression, respectively, with 396 transcripts affected by both regulatory processes. These observations suggest that AS regulation is as important as transcriptional regulation in early phytochrome-mediated light signaling, while after 3 h of treatment, differential expression of transcripts increased considerably to become the dominant process. To further characterize the phytochrome-responsive transcripts, the authors performed a GO analysis on different subsets of genes. After 1 h of red-light treatment, GO terms related to RNA splicing were significantly enriched among transcripts that exhibited phytochrome-dependent AS alone, while genes showing phytochrome-dependent differential expression were enriched in categories associated with light signaling. Of the genes that presented only phytochrome-dependent AS changes, 58 were splicing-related genes, such as serine/arginine-rich (SR) proteins, the U1 small nuclear ribonucleoprotein and the U2 auxiliary factor 65a. The authors reported that the phytochrome-mediated regulation of the splicing-related genes after 3 h of red-light treatment was achieved more at the expression level than the AS level. Furthermore, the GO terms linked to photosynthesis and plastid/chloroplast function were enriched in the phytochrome-mediated AS gene subset after 3 h of treatment, suggesting that this process is also involved in light-induced chloroplast differentiation during seedling de-etiolation. Besides these observations, Shikata et al. mentioned a set of red-responsive but phytochrome-independent AS alterations in both treatments. To address the biological significance of these genes, we performed a new meta-analysis of these data and found that the number of differentially spliced AS events regulated via the phytochromes (2230) was similar to the number of events affected in a phytochrome-independent manner (2249). The latter involves several pathways such as blue-light and UV-light photoreceptors as well as photoreceptor-independent pathways, therefore attributing 50% of the response only to phyA and phyB (Fig. 2b). Similar to the authors’ analysis, we performed a GO-term assessment of the data and found a significant enrichment of RNA processing-related transcripts among genes exhibiting phytochrome-dependent and phytochrome-independent AS (Fig. 2c). Furthermore, we showed that light-stimulus-related transcripts (e.g. light signaling and photosynthesis) were enriched among the differentially expressed genes. The same observation was made for those transcripts that experienced both differential expression and light-regulated AS in a phytochrome-dependent or phytochrome-independent manner (Fig. 2c). Our findings reinforce the concept that a significant number of transcripts with light-modulated AS encode splicing-related proteins and that the genes that exhibit light-regulated AS in addition to differential expression are mainly related to light-signaling pathways and photosynthetic processes. Shikata et al. also addressed the physiological relevance of phytochrome-mediated AS by comparing the effect of overexpressing two alternatively spliced isoforms of the light-signaling gene SPA1-RELATED 3 (SPA3) with plants overexpressing the functional full-length isoform in wild-type Arabidopsis. Transgenic plants carrying the alternatively spliced isoforms of SPA3 observed under red light displayed a short hypocotyl phenotype and resembled spa3 knockout mutants, demonstrating that the proteins encoded by these isoforms have a dominant-negative effect on the endogenous SPA3. Finally, Shikata et al. investigated the role of phyA and phyB in the control of AS in the SR gene transcripts ARGININE/SERINE-RICH SPLICING FACTOR 31 (RS31) and SERINE-RICH PROTEIN SPLICING FACTOR 34B (SR34b), analyzing their splicing under different red/far-red treatments and in the phytochrome knockout mutants phyA, phyB and the double null phyAB. They showed that red-light-induced alterations in the splicing of RS31 were suppressed by a subsequent pulse of far-red light, indicating that defects on this AS event are also affected by the most distinguishing feature of typical phytochrome-mediated responses: red/far-red reversibility. By contrast, SR34b did not show clear red-/far-red-light reversibility. Despite this, both RS31 and SR34b displayed red-light-induced AS changes in the phyAB double mutant 15.

Another genomewide transcriptome profiling study assessing the light regulation of AS was performed in 2016 by Mancini et al. 16. Here, the authors sought to expand our understanding of AS in light-grown plants from a few reported examples to the whole transcriptome, and to identify additional possible posttranscriptional mechanisms involved in regulating circadian clock function and/or the floral transition. To achieve this, these workers used a 2-h pulse of light given in the middle of the night to Arabidopsis plants entrained to a 12-h light:12-h dark photoperiod (Fig. 2a). An interesting aspect of this work was the bioinformatics methodology employed to detect AS events from the RNA-seq data; multiexonic genes were partitioned into features defined as “bins,” corresponding to exonic regions (exon-bins), intronic regions (intron-bins) and AS regions (AS-bins). First, the differentially spliced AS-bins between treatments were determined with an approach similar to the differential expression analysis. The selection of differentially spliced AS-bins was then restricted to those bins for which the differential AS analysis was supported by changes in the number of splice junctions. This approach makes it possible to compute the splicing metrics PSI (percent spliced-in) and PIR (percent intron retention) 27, 28, a bioinformatics improvement that increases the statistical strength of AS event detection. The authors identified a total of 382 genes with AS events regulated by the light pulse treatment, as well as 4341 genes whose mRNA levels significantly changed. Interestingly, less than half of the genes that displayed an altered AS pattern under light showed alterations at the total mRNA level, further indicating that light affects AS through mechanisms distinct from those by which it affects gene expression. From the different type of AS events, IR were emerging as the most frequent in response to light 16 (Fig. 2d). Interestingly, light promoted intron retention of some genes, whereas it enhanced intron exclusion in others 16, indicating that light regulates AS rather than affecting splicing in general. Furthermore, a GO-term analysis showed that categories associated with RNA processing were significantly enriched among transcripts with light-regulated AS patterns, but not among the genes whose mRNA levels alone were affected. To fully appreciate the significance of these data and to contrast them with findings in Physcomitrella, we performed a similar meta-analysis of the RNA-seq data from Mancini et al. As reported by the authors, GO terms related to RNA processing were enriched in AS events, particularly in the categories of mRNA processing and RNA splicing. Also, the genes that were differentially expressed in the light treatment were enriched in light-stimulus-related GO categories (Fig. 2c). Interestingly, Mancini et al. reported that, among the genes showing light-regulated AS events, several encode SR proteins and heterogeneous nuclear ribonucleoproteins (hnRNPs) that are known to modulate the recruitment of spliceosomal snRNPs to donor and acceptor splicing sites 18. In addition, the authors evaluated the effect of white light on the AS of SERINE-RICH PROTEIN SPLICING FACTOR 30 (SR30) in light-grown wild-type, phyAB and cryptochrome 1 cryptochrome 2 (cry1cry2) double-mutant plants. They found that light had a similar effect on the AS of SR30 in the wild type and in the phyAB and cry1cry2 mutants, suggesting that a significant part of the light effect may be modulated by an intermediate molecule generated during photosynthesis, rather than by these photoreceptors. Furthermore, the authors reported that the strong effect of a 2-h red-light pulse on the AS of SR30 was not impaired in the quintuple mutant phyABCDE. This is in concordance with the observation of Shikata et al., who found that approximately 43.6% of all AS events regulated by red light were independent of phyA and phyB. In addition to studying the role of trans-acting factors, Mancini et al. also looked for cis-sequences that might play a role in light-induced AS. They screened the flanking regions of the exon–intron boundaries for potential cis-regulatory sequences involved in mediating the light effects on AS, particularly focusing on 232 introns that showed differential retention in response to the light pulse 14. In contrast to the findings of Wu et al., no enriched motifs were found near the 5′SS and 3′SS 16.

So far, we have reviewed and analyzed the available data regarding light-regulated genomewide transcriptome changes in plants. All three studies showed that light induces rapid changes in the AS of splicing-related genes in light-grown plants or etiolated seedlings, suggesting that light regulates AS patterns mostly through its effect on splicing factors themselves. The data presented by Shikata et al. suggest that this phenomenon occurs during the early stages of signal transduction, and that possibly the large changes observed in differential gene expression are a result of early signal amplification. This indicates that light-regulated AS of splicing factors may also control the function of light-signaling-related genes and/or other transcripts. Although the reports we reviewed emphasize the fact that photoreceptors play an important role in controlling AS in plants, it is clear that a significant part of the light effect does not appear to operate through traditional photoreceptor pathways. This highlights the relevance of a retrograde signaling circuit connecting photosynthetic activity in the chloroplast to the regulation of AS in the nucleus 29; however, little is known about this circuit. Despite the progress made in this field, important questions remain unanswered, such as how light modulates AS and how light triggers the initial steps that result in AS.

MiRNA posttranscriptional regulation contributes to light responses

Small RNAs are involved in many stages of plant development and several small RNA pathways in plants contribute to plasticity and adaptation. MiRNAs are a predominant class of small RNAs involved in posttranscriptional gene regulation by reducing the levels of their target genes. Many targets for miRNAs encode transcription factors 30; thus, reducing the expression of their encoding genes affects their downstream targets as well, causing a major transcriptomic change in the cell.

Few studies at a genomewide level describe how miRNAs respond to and transmit light signals in plants 14-16, 31, and none have examined the biological relevance of light-mediated miRNA regulation. Transcriptome profiling experiments with microarrays or, most recently, RNA-seq revealed several miRNA genes that were differentially expressed following a variety of light treatments in Arabidopsis seedlings. For instance, a pulse of white light in the middle of the night induced the expression of miR398b, miR398c, miR163, miR157c and HUA ENHANCER 1 (HEN1), an enzyme participating in miRNA biogenesis, while the same treatment repressed miR408, miR834a and miR822a 16. Similarly, red light applied to four-day-old etiolated seedlings altered the expression of miR163, miR156c, miR157c, miR169 l and miR824a 15. By contrast, two-week-old plants exposed to UV light had diminished miR160, miR165, miR167 and miR939 levels 32 (Fig. 3). Based on these observations, it is clear that miRNA transcripts respond to specific wavelengths rather than to light more generally.

At least two papers have addressed the role of light-responsive miRNAs in early plant development in detail. Light induces the expression of MIR171, which targets transcripts encoding scarecrow-like proteins (SCL6/22/27) that negatively regulate chlorophyll biosynthesis in Arabidopsis 33 (Fig. 3). It was recently shown that SCL27 binds to the promoter of PROTOCHLOROPHYLLIDE OXIDOREDUCTASE C (PORC), which encodes a key enzyme in chlorophyll biosynthesis. Plants overexpressing MIR171c resemble scl6scl22scl27 triple mutants, with higher levels of PORC transcripts and, concomitantly, higher chlorophyll contents. Additionally, SCL27 binds to MIR171, activating its expression in a regulatory feedback loop 34. On the other hand, DELLA proteins upregulate the expression of POR genes 35 and promote chlorophyll biosynthesis 34 by acting via the SCLs; the interaction between the DELLA protein REPRESSOR OF GA (RGA) and SCL27 decreases the capacity of SCL27 to bind to DNA promoters. Light therefore triggers two parallel modes of releasing the inhibition on PORC to increase chlorophyll biosynthesis: promoting DELLAs to reduce the SCL's capacity to decrease PORC expression, and activating miR171 to clear SCL transcripts. In the second report, Chung et al. described the role of light-inducible miR163 in seed germination and primary root elongation 36. The expression pattern of miR163 and its target PARAXANTHINE METHYLTRANSFERASE 1 (PXMT1) are inversely correlated and are regulated by a variety of light wavelengths during seedling de-etiolation (Fig. 3). The authors showed that miR163 mediates seedling root growth through downregulating PXMT1 and found that seeds of the miR163 knockout mutant have delayed germination 36. Altogether, the results of this study suggest that miR163 functions in the early stages of growth, from seed germination to seedling root growth, by light-induced targeting of PXMT1.

Mechanisms of miRNA regulation by light

MiRNA genes are generated from endogenous loci, and their resemblance to protein-coding genes means their transcript abundance is in part guided by the same rules 37, for example, the existence of cis-elements in their promoters that respond to environmental and endogenous stimuli. In addition to transcriptional control, miRNA production involves their excision from the stem loop of a fold-back precursor; hence, mature miRNA levels could also be adjusted by modulating their biogenesis. During processing, an imperfect double-stranded RNA (dsRNA) structure is recognized by the miRNA processing machinery and processed to release a ∼21-nucleotide small RNA. In plants, all miRNA-generating events take place in the nucleus 38, while mature miRNA acts in the cytoplasm of the cell. The ribonuclease III DICER-LIKE 1 (DCL1) is assisted by accessory proteins to produce two staggered cuts in the dsRNA precursor that release the miRNA/miRNA* duplex 39, 40. The HYPONASTIC LEAVES 1 (HYL1) 41, 42 and SERRATE (SE) proteins 43, 44 assist miRNA processing 45, 46 by improving the efficiency and precision of DCL1 cleavages 47. Other regulators have been lately identified such as C-TERMINAL DOMAIN PHOSPHATASE-LIKE 1 (CPL1)/FIERY2 (FRY2), which participates in the dephosphorylation of HYL1 needed for optimal HYL1 activity 48. Later, the miRNA/miRNA* molecule is methylated by the action of HEN1 49 to stabilize the duplex and prevent its degradation 50-52. To complete miRNA maturation, the duplex is exported to the cytoplasm 53 and loaded into the RNA-induced silencing complex, and the miRNA* is degraded. Once in the silencing complex, the single-stranded mature miRNA binds to ARGONAUTE 1 (AGO1) and is ready to scan for its messenger targets to regulate gene expression.

Light may regulate the expression or maturation of miRNAs. Motifs known to confer light responsiveness have been described in upstream regions of a subset of miRNA genes 32, 54, suggesting that miRNA levels react to different light signals and help to attenuate the expression of their light-responsive target genes. Moreover, LONG HYPOCOTYL 5 (HY5), a bZIP transcription factor and positive regulator of plant photomorphogenesis, mediates gene transcription indirectly, partially by targeting miRNAs. Genomewide mapping of HY5-binding sites showed that miR156d, miR172b, miR402, miR408, miR775, miR858, miR869 and miR1888 bind directly to this protein 13, and that 21 targets of those miRNAs were upregulated in the hy5 knockout mutant 13, 55, implying that those target genes could be repressed by HY5 via the activation of specific miRNAs. HY5 also binds to the promoters of clock genes such as CIRCADIAN CLOCK-ASSOCIATED 1, LATE ELONGATED HYPOCOTYL, TIME OF CAB EXPRESSION 1 and EARLY FLOWERING 4, but the binding of HY5 alone is not sufficient to maintain proper circadian rhythms 55. Thus, either through cis-elements on promoters or through HY5 induction, light controls miRNA abundance and therefore this posttranscriptional process could participate in repressing or at least fine-tuning the expression amplitude of genes in the plant's response to light.

Regulating miRNA biogenesis could hypothetically be another way of controlling their levels. Interestingly, several mutants of genes involved in almost each step of miRNA biogenesis mentioned above show hypersensitivity to light in Arabidopsis 10, 56, suggesting that light regulates miRNA biogenesis. Several alleles of the ago1 mutant have an exaggerated response to light 10, and furthermore, light regulatory pathways mediated by phyA are upregulated in ago1, demonstrating that deregulating the phyA pathway might account for part of the altered light response. The exact mechanism and the specific miRNAs involved remain to be clarified, but the phenotypic analysis of the ago1 mutant suggests that miRNAs might act as negative regulators of light-signaling pathways 10. Mutations in HYL1, HASTY and HEN1 cause Arabidopsis to have shorter hypocotyls than the wild type under both dark and light conditions, a phenotype that is dependent on the fluence of white light 56. This hypersensitivity to light implicates HYL1, HASTY and HEN1 as negative regulators in the de-etiolation process, although only the production of HEN1 was induced by light 56. Light-mediated induction of HEN1 is dependent on the photoreceptors phyA, phyB, cry1 and cry2, and on HY5 and HY5 HOMOLOG (HYH), all of which are involved in photomorphogenic behaviors. The increased levels of HEN1 in seedlings lead to an increase of miR157d and miR319. Tsai et al. showed that miR157d targets HY5 transcript 56, forming a negative feedback loop between HY5 and HEN1 during photomorphogenesis. This suggests an interesting example of a mechanism for light-mediated RNA biogenesis in photomorphogenesis.

CONSTITUTIVE PHOTOMORPHOGENIC 1 (COP1) is an E3 ligase involved in photomorphogenesis that promotes the degradation of light-regulated transcription factors and photoreceptors. COP1 represents another mechanistic link between light signaling and miRNA biogenesis, because the absence of COP1 impairs the accumulation of miRNAs. COP1 is believed to stabilize HYL1 by protecting it from degradation by an unknown protease in a light-dependent manner 57. This observation is surprising, as COP1 is associated with the proteolytic decay of proteins rather than their stabilization; therefore, the mechanism by which COP1 contributes to protein stabilization remains to be described.

These observations highlight the role of light in modulating the posttranscriptional mechanisms of gene expression, which allow plants to adapt to changing light environments.

Integrating the AS and miRNA pathway in response to light

So far, we have reviewed that two of the most important gene posttranscriptional mechanisms, namely AS and the miRNA regulatory pathway, respond to light, which is considered to be the most important environmental cue for plants. It would be intriguing to establish whether there is cross talk between these processes. Using data from the transcriptome studies described in this review, we searched for miRNAs or genes involved in their biogenesis that undergo AS in response to light. In Physcomitrella patens, most of the factors involved in miRNA processing undergo AS under light 14. Upon treatment with red or blue-light pulses, IR events are induced in SE, HEN1, CPL1 and DCL1 in Physcomitrella (Fig. 4a). Intriguingly, a certain light wavelength specificity was observed; only red light affects the AS of DAWDLE (DDL) thought to stabilize miRNA primary transcripts 58, 59, while blue light affects HYL1 splicing, but red light has no effect on HYL1 (Fig. 4a). Also, blue and red light can generate different isoforms of the same gene; for example, we observed that in HEN1, the second intron is retained after a red pulse, while blue light caused the retention of intron 7. Both treatments result in a nonfunctional HEN1 protein, as one disrupts the methyltransferase domain and the other generates a truncated protein containing only the RNA-binding domain. Furthermore, red light induces an alternative donor splice site event in DCL1, whereas blue light results in an alternative acceptor splice site in the transcript.

Developing a systematic bioinformatics search of the Arabidopsis transcriptome, Yang et al. revealed that at least 12% of high-confidence miRNA binding sites were affected by AS, as different isoforms of a gene diverge at the miRNA recognition sequence 60. In particular, AS observed in DCL1, TAS1 and SPL4 abolished the capacity of miRNAs to bind to them, as the binding sites disappear in their alternative isoforms. The biological relevance of AS in SPL4 was underscored using transgenic plants bearing isoforms with or without the miRNA binding site, which resulted in two distinguishable phenotypes 60.

Perhaps the strongest association between AS and the miRNA pathway in response to light is exemplified by COP1 protein. COP1 is an interesting protein linked not only to light-induced miRNA biogenesis 57 but also to AS. The AS form of HYH lacks a COP1-interacting motif and is therefore more resistant to selective protein degradation mediated by COP1 61.