Light in the transcription landscape: chromatin, RNA polymerase II and splicing throughout Arabidopsis thaliana’s life cycle

ABSTRACT

Plants have a high level of developmental plasticity that allows them to respond and adapt to changes in the environment. Among the environmental cues, light controls almost every aspect of A. thaliana’s life cycle, including seed maturation, seed germination, seedling de-etiolation and flowering time. Light signals induce massive reprogramming of gene expression, producing changes in RNA polymerase II transcription, alternative splicing, and chromatin state. Since splicing reactions occur mainly while transcription takes place, the regulation of RNAPII transcription has repercussions in the splicing outcomes. This cotranscriptional nature allows a functional coupling between transcription and splicing, in which properties of the splicing reactions are affected by the transcriptional process. Chromatin landscapes influence both transcription and splicing. In this review, we highlight, summarize and discuss recent progress in the field to gain a comprehensive insight on the cross-regulation between chromatin state, RNAPII transcription and splicing decisions in plants, with a special focus on light-triggered responses. We also introduce several examples of transcription and splicing factors that could be acting as coupling factors in plants. Unravelling how these connected regulatory networks operate, can help in the design of better crops with higher productivity and tolerance.

Introduction

Plants are sessile organisms restricted to their site of germination. To compensate for this lifestyle, plants have a high level of developmental plasticity that allows them to respond and adapt to changes in the environment. As a result of this flexibility, plants move each of their organs to adjust their positions with respect to a source of light among other environmental cues. Charles Darwin, together with his son Francis, studied these movements and expressed “we were often struck with the accuracy with which seedlings pointed to a light” [1]. This accuracy is understandable since plants rely on light as their main source of energy and also as a key source of information about their surroundings [2]. In other words, plants’ sessile and photoautotrophic nature obliges them to be extremely sensitive to this environmental signal [3].

Light controls almost every aspect of development throughout a plant’s life cycle, including seed maturation, seed germination, seedling de-etiolation, phototropism, shade avoidance responses, circadian rhythms and flowering time [4–7]. At least five distinct families of photosensory proteins capture these light signals that provide spatial and temporal information and regulate the expression of many different genes [8,9]; these are: phytochromes, primarily absorb red (R) and far-red (FR) wavelengths (600–750 nm); cryptochromes, members of the Zeitlupe family and phototropins, absorb blue (B) and ultraviolet (UV) A (320–500 nm); and UV RESISTANCE LOCUS 8 (UVR8), that perceives UV-B (282–320 nm). These photoreceptors absorb, interpret, and transduce light signals to generate a wide range of responses, controlling gene expression and facilitating plants’ physiological responses and adaptation. One key event under the control of these photoreceptors is photomorphogenesis: the acquisition of photosynthetic capacity and the re-direction of growth, which takes place when dark-grown seedlings become exposed to light [10]. This allows plants to generate energy and release oxygen. Therefore, photomorphogenesis has dramatic implications to all life on Earth and, in fact, to the planet itself [11,12].

Seedlings must reach the autotrophy state after germination before the nutrients stored in the seed are exhausted [13]. Driving chloroplast biogenesis and function is one of the most important functions of light signalling networks [14]. Regulation of nuclear gene expression plays a major role in chloroplast biogenesis and maintenance; however, the modulation of these processes relies on a fine coordination of gene expression between the plastid and the nuclear compartments achieved by bidirectional communication. Synthesis and import of nuclear-encoded proteins to the plastids serves as anterograde signals, while the signals plastids send to the nucleus, mostly to control de novo synthesis of nuclear-encoded plastid proteins, are known as retrograde signals [15–18]. The relevance of this communication can be properly dimensioned when considering that plastids retain just a fragment of their ancestral genomes [19–21]. In fact, from circa 3000 different proteins that can be found in these organelles, more than 95% are encoded by the nuclear genome [22].

Light signals induce massive reprogramming of gene expression in plants. In 2001 Ma and collaborators evaluated changes in gene expression during light-regulated seedling development in A. thaliana, using an expressed sequence tag (EST)-based microarray. Of the 9216 ESTs (representing ~6120 unique genes) included in the array, one-third showed significant (two-fold or greater) differential expression between white light- and dark-grown seedlings [23]. More recently, in an RNA-seq experiment, Schneider and colleagues reported that almost every expressed gene in A. thaliana shows changes greater than 30% in expression in response to fluctuating light conditions [24].

The principles and mechanisms underlying transcription on naked DNA are remarkably similar between eukaryotes and prokaryotes. A typical RNA polymerase II (RNAPII) transcription cycle in a eukaryote begins with the binding of activators upstream of the core promoter (including the TATA box and transcription start site) followed by the recruitment of the adaptor complexes that facilitate binding of general transcription factors. The preinitiation complex (PIC), consisting of RNAPII and transcription factors, is then positioned at the core promoter. RNA synthesis initiates and the carboxy-terminal domain (CTD) of RNAPII is phosphorylated by the transcription factor IIH (TFIIH) subunit during the first 30 nucleotides of transcription. The phosphorylated CTD begins to recruit the factors that are important for productive elongation and mRNA processing (beautifully reviewed in [25]). Genetic information in eukaryotes is arranged in chromatin, a repeating unit of histones and DNA. Though this structure is advantageous for genome compaction, its functions go far beyond this “simple” task. As discovered quite early, the packaging of the DNA template into nucleosomes affects every step of transcription [26,27]. Remarkably, a key factor in plant photomorphogenesis, DE-ETIOLATED-1 (DET1), associates with nucleosomes and affects chromatin compaction in response to light [28–30]. Furthermore, chromatin landscape in plants is now recognized to be dynamic in response to environmental changes, in particular in the case of seedlings grown in light in comparison to dark [31].

Gene expression in plants, as in any other eukaryote, can be modulated at different levels: transcription initiation, mRNA processing, splicing, export, translation, and degradation, among others. The splicing process is performed by a complex RNA-protein cellular machine, the spliceosome, that recognizes and removes some regions – so called introns – while joins others – so called exons. Alternative splicing produces multiple mRNAs from a single gene through variable and regulated selection of splice sites, significantly modulating the transcriptome, and to some extent also the proteome, during development and in response to environmental signals [32,33]. This requires a precise regulation, to guarantee plasticity while still displaying high specificity and fidelity. Different regulatory sequences along introns, as the 5´ and 3´ splice sites, the branching point and the polypyrimidine tract, must be recognized by a set of small nuclear ribonucleoprotein particles (snRNPs) and other accessory proteins, in a fixed sequence of steps that leads to the assembly of the catalytically active spliceosome. These cis regulatory sequences and their recognition by trans auxiliary splicing factors, such as serine/arginine -rich (SR) proteins and heterogeneous nuclear ribonucleoproteins (hnRNPs), modulate the ability of the spliceosome to recognize and use particular splice sites [34]. In addition, since splicing reactions occur mainly while transcription takes place (cotranscriptionally), the regulation of RNAPII transcription affects the splicing outcomes [35–37]. In this framework, chromatin compaction and modifications also affect transcription and, then, splicing decisions [38,39]. Although neglected for a long time, alternative splicing occurrence in plants seems to be comparable to that in animals. In A. thaliana, the observed alternative splicing frequency is above 61% for intron-containing genes [40,41]. However, the accumulation of evidence on the functional roles of alternative splicing and its regulation in plants is much delayed in comparison to their animal counterparts [32].

The splicing machinery and mechanisms are shared across kingdoms to some extent, such as the RNA sequences that define exon/intron boundaries, spliceosome component proteins, and splicing factors, which are indeed conserved across all eukaryotes [42–44]. There are, nevertheless, some striking differences in splicing among eukaryotes, for example, the average length of introns and exons changes dramatically between plants and animals, and between different species. The same occurs with the frequency of each splicing event. Intron retention is the most common alternative splicing event in plants while exon skipping is the most frequent in animals [32,40,41]. These similarities and differences make plants ideal systems to study deeply conserved strategies for alternative splicing regulation, and to find novel mechanisms that have evolved in plants and help them endure constant environmental vicissitudes.

In this review, we explore the accumulated knowledge on the interactions and cross-regulation between chromatin state, RNAPII transcription and splicing decisions in plants, with focus on light-triggered responses in A. thaliana’s life cycle.

Light regulates nuclear size and chromatin state throughout plants’ development

Plant perception of environmental cues triggers most of the developmental transitions in A. thaliana. Light is a powerful environmental cue regulating many aspects of plant development and phenotypic plasticity. As introduced before, plant photoreceptors absorb, interpret, and transduce light signals to generate a wide range of responses, that include modulating the expression of light-regulated nuclear genes, facilitating plants’ physiological responses and adaptation. In the model plant A. thaliana, photoreceptors tightly communicate different light signalling pathways generating synergistic or antagonistic cross-talks between them. This signal integration network befalls mainly at the gene expression level. Transcriptional regulation occurs through the action of multiple protein complexes (activators, repressors, remodelling enzymes, adaptors, polymerases) as well as the deposition of chemical modifications on histones and DNA itself [45].

Nuclear and chromatin architecture are both dynamic processes that undergo several rearrangements throughout a plant life’s cycle. This is mostly dependent on light signals, which elicit – directly or indirectly – (a) specific histone modifications, (b) global changes in chromatin architecture and (c) chromatin decondensation in the later stages of seedling growth during the transition to photomorphogenesis and during seed development and germination [29,30]. As an example of this, the “life” of a seed is characterized by two major phase transitions: from embryogenesis to seed maturation, and from dry seed to seed germination. These different stages display specific transcriptomes and require activation or repression of diverse sets of genes. In addition, cells of fully mature seeds possess very small nuclei with highly compacted chromatin, that are established during seed maturation. These unique features require extensive epigenetic signalling mechanisms that tightly coordinate the phase transitions and control chromatin accessibility [46,47]. During the first phase, when the embryo development is coming to an end, seeds are not fully prepared to survive outside the mother plant and still attain high moisture levels. During the seed maturation phase, seeds dehydrate and become desiccation tolerant, storage compounds accumulate, and seed dormancy is induced [48]. This mechanism prevents mature seeds from germinating under conditions otherwise favorable [49]. In this sense, seed dormancy plays a central role in the adjustment of plant populations to their environment; and environmental signals that modulate dormancy define the germination timing of a seed population and are, consequently, of the utmost adaptive importance [50]. It is well established that chromatin integrates a variety of endogenous and exogenous signals, facilitating the activation or repression of specific sets of genes involved in different developmental programs, integrating them with the response to the environmental signals. These signals are thought to direct distinct local and global functional states of chromatin, therefore controlling the capacity of a cell’s genome to store, release and inherit biological information. Moreover, transcriptional changes that take place during seed maturation, after-ripening and germination, are likely to be reflected in changes of chromatin organization and are associated with chromatin remodelling [51]. Chromatin remodelling factors alter DNA-histone interactions as well as the accessibility of genomic regions to the transcriptional machinery or transcription factors, regulating gene expression [52–54]. Interestingly, there is an extensive DNA demethylation (passive loss of DNA methylation of most of the differentially methylated regions) occurring at the seed-to-seedling transition stage, corresponding with the onset of DNA replication and cell division [55]. Moreover, several studies have also shown how light shapes chromatin architecture and nuclear size throughout the seed-to-seedling transition. Light (and dark conditions too) elicits moderate nuclear expansion and relaxation of heterochromatin upon seed imbibition [30]. From the third day of germination, heterochromatin is recompacted and nuclear size increases [56–58], a process exclusively dependent on light signalling, since under dark conditions, heterochromatin is further decondensed. During de-etiolation, this arrest is rapidly released [30].

Different photoreceptors play diverse roles in chromatin re-organization and compaction, and their interactions are also relevant during plant development. Phytochrome B (phyB) has a prominent role as the main photoreceptor regulating the R/FR reversible seed germination response [59,60]. Upon R light irradiation, the Pfr-PHYB activates the degradation of PHYTOCHROME-INTERACTING FACTOR3-LIKE5 (PIL5) protein to promote seed germination through gibberellins (GA), leading to an increase in JUMONJI C DOMAIN-CONTAINING PROTEIN20 (JMJ20) and JMJ22 expression, two histone arginine demethylases [61]. As a result, the GA-pathway is activated and germination promoted [62]. This clearly shows that light and hormone signalling pathways co-ordinately regulate histone methylation during phyB-mediated seed germination.

Changes in phyA transcript levels, the main photoreceptor for FR light perception [63–65], are correlated with specific modifications in histone marks; and the presence of opposing histone marks under dark/light transitions, enables rapid activation and inactivation of phyA in response to changing light conditions, suggesting that histone deacetylation regulates the light-induced changes in gene expression of phyA, thus regulating its abundance. Moreover, the activating and repressive histone marks on phyA locus suggests a mechanism that ensures rapid turn-off in the light during de-etiolation and transcriptional reactivation when plants are returned to dark conditions [66].

As we said before, light positively regulates chromatin compaction. Under standard light conditions, phyB appears as the main photoreceptor controlling this response [67]. Interestingly, shading by low light intensity results in a decrease in chromatin compaction, a response mediated by phytochromes and cryptochromes, and there is a negative correlation between CRY2 level and chromatin compaction [68, and also exquisitely reviewed in ²⁹]. Even more interesting is the fact that natural genetic variation in light-signalling components affects large-scale chromatin compaction phenotypes in A. thaliana, and this variation can be explained due to local alterations in light intensity [67].

UV-B light also triggers changes at the transcriptional level, which are mostly regulated by the UV-B photoreceptor UVR8 [69–71]. UVR8 can constitutively associate with chromatin by histone binding [72] and could potentially enhance the recruitment of chromatin-modifying enzymes and transcriptional regulators [72,73].

The transition to flowering is another key developmental change in the life cycle of plants. Chromatin remodellers play an important role in the regulation of flowering time and flower development, integrating both internal and external signals [74]. Molecular analysis of vernalization, the acceleration of flowering that occurs following exposure of seeds or seedlings to low temperatures, in a wide variety of plant species have uncovered temperature-dependent epigenetic silencing of genes involved in floral repression. Chromatin remodelling during flowering acts at several loci, including the MADS AFFECTING FLOWERING (MAF) genes, even though it has been best characterized at the key integrators FLOWERING LOCUS C (FLC, a central flowering repressor) and FLOWERING LOCUS T (FT, a flowering inducer) [75]. Several studies have shown that several chromatin-modifying components are involved in activation or repression of FLC expression: (a) activation is associated with various chromatin modifications whereas (b) repression is associated with various histone modifications. Interestingly, the blue light photoreceptor CRY2 also controls chromatin compaction during floral transition [76].

Chromatin interactions with the transcription and processing machinery during seed dormancy

There are several examples of genes influencing dormancy levels that are directly involved in chromatin remodelling. A mutagenesis screen for seed dormancy in A. thaliana yielded reduced dormancy (rdo) mutants [77,78]. Interestingly, rdo4 has a mutation in the H2B MONOUBIQUITINATION1 (HUB1) gene, an ortholog of the human BRE1 [79]. HUB1 encodes a C3HC4 RING finger protein, which functions as the E3 ligase responsible for monoubiquitination of histone H2B [79]. In the work by Liu and colleagues (2007), the authors present vast evidence sustaining that HUB1 is necessary for histone H2B monoubiquitination in vivo and influences gene expression of dormancy related genes (including DELAY OF GERMINATION1 -DOG1-, LYSOPHOSPHATIDIC ACID ACYLTRANSFERASE1 -ATS2-, NINE-CIS-EPOXYCAROTENOID DIOXYGENASE9 -NCED9-, 1-CYSTEINE PEROXIREDOXIN 1 -PER1-, and CYP707A2), suggesting a role for chromatin remodelling in the regulation of seed dormancy. Interestingly, hub1-2 mutant not only showed reduced seed dormancy and longevity but also other defects during seedling establishment suggesting that HUB1 could play a role in several processes in the plant life´s cycle [79]. Among these various pleiotropic phenotypes of the hub1-2 mutant, the effect on seed dormancy is relatively strong, implying that histone H2B monoubiquitination is a key factor in the induction and/or maintenance of dormancy levels (see Figure 1). With the identification of HUB1, Liu and colleagues (2007), demonstrated the involvement of chromatin remodelling in the seed dormancy mechanism [79].

Figure 1.

Different factors at the interphase of chromatin remodelling, transcription, and alternative splicing regulation determine seed dormancy levels. HUB1 is responsible for histone H2B monoubiquitination and influences gene expression of dormancy related genes, such as DOG1. TFIIS (RDO2) modulates DOG1 gene expression and is required for RNAPII processivity, stimulating RNAPII to reassume elongation after pausing. The spliceosome disassembly factor NTR1, present at the promoter region and in the gene body of DOG1, is required for correct expression and splicing of DOG1. Activation of DOG1 through chromatin remodelling and transcriptional elongation constitutes an important regulatory mechanism during seed dormancy. Variations in expression of the dormancy-related genes above mentioned correlate with phenotypic expression in this trait. Grey and yellow boxes at the DOG1 gene represent the exons. lgDOG1 and shDOG1 represent the two forms of DOG1 transcripts, described in [146]

Another interesting example is REDUCED DORMANCY 2 [RDO2). Several years ago, Liu and colleagues (2011) cloned the mutant reduced dormancy 2–1 (rdo2-1], characterized by reduced seed dormancy levels [80]. RDO2 is ubiquitously expressed throughout all plant tissues and encodes an elongation factor known as TRANSCRIPTION FACTOR IIS (TFIIS), required for RNAPII processivity [81,82]. More interestingly, a significant overlap in differentially expressed genes during seed maturation was found between hub1 and rdo2, confirming their involvement in the same process. Both genes are also up-regulated during seed maturation [80]. These results open the question whether RDO2 and HUB1 might share common targets. Interestingly, DOG1, a master gene controlling seed dormancy, appears as one of the genes commonly down-regulated in both mutants [80]. Thus, activation of DOG1 through chromatin remodelling and transcriptional elongation might be an important mechanism of seed dormancy. Moreover, mutations in POLYMERASE II-ASSOCIATED FACTOR 1 COMPLEX (PAF1C) and related factors (VERNALIZATION INDEPENDENCE4 -VIP4-, VIP5, EARLY FLOWERING7 -ELF7-, ELF8 and ARABIDOPSIS TRITHORAX-RELATED7 -ATXR7-) reduce seed dormancy levels in a similar manner to those of rdo2-1 and hub1-2, suggesting the importance of PAF1C and transcription elongation for seed dormancy. Furthermore, the up-regulation of these PAF1C associated genes at the end of the seed maturation stage might be indicating that they could be particularly important during this phase, possibly by counteracting the negative effects of desiccation on gene expression [80]. The previous examples evidence the tight interaction between chromatin and transcription in the regulation of seed dormancy. In addition, transcriptional efficiency is also involved in this regulation. Transcriptional efficiency is determined by recruitment of RNAPII to the DNA template and the rate of transcription elongation after its binding to DNA. During transcription, release of paused RNAPII is mediated by different transcription elongation factors, which facilitate efficient mRNA synthesis in the chromatin context [83]. Interestingly, Grasser and colleagues (2009) demonstrated that a mutation in TFIIS results in a reduced seed dormancy phenotype [84]; and TFIIS modulates DOG1 expression [85]. Additionally, all these processes and molecular events underlying the seed dormancy response are connected when using an enhanced TFIIS mutant in A. thaliana [81,86]. RNAPII elongation affected by this mutation impacts the alternative splicing outcomes of DOG1, one of the key regulators affecting seed dormancy (see Figure 1). These results imply that chromatin state, transcription elongation and (alternative) splicing, are a critical part of the dormancy regulatory mechanisms. The connection and cross-regulation between these mechanisms will be further explained in the next sections.

Transcription as a bridge between chromatin state and alternative splicing outcomes

We previously introduced several examples revealing relevant interactions between chromatin state and RNAPII transcription. In this section, we will further explore the underlying mechanisms and working models connecting chromatin, transcription, and alternative splicing.

Alternative splicing is widely recognized as a cotranscriptional process and this has been demonstrated in multiple organisms, including plants [87–90]. Moreover, this co-transcriptional nature allows a functional coupling between transcription and splicing, in which properties of the splicing reactions itself are affected by the transcriptional process, not only because they occur at the same time and space: coupling means that the splicing reactions depend on transcription and transcription depends on splicing [91]. Two models have been proposed to explain the underlying mechanisms of the functional coupling between transcription and alternative splicing: recruitment coupling and kinetic coupling. This latter explains how changes in RNAPII elongation rate influence alternative splicing choices [89,92,93], while the recruitment coupling involves the engagement of splicing factors by the transcription machinery. Interestingly, these mechanisms are not mutually exclusive.

RNAPII’s largest subunit contains a CTD. In animals and plants, the CTD is composed of a number of repeats of a consensus heptad, YSPTSPST, although the number of repeats varies among species: the human CTD has 52 repeats, while in plants the CTD is comprised of 34 repeats [94]. This domain undergoes different post-translational modifications in their residues that modulate the affinity for factors that, in turn, regulate mRNA processing [95]. For example, CTD phosphorylation patterns determine the recruitment of splicing factors to transcription sites [96,97].

The kinetic coupling model explains how changes in RNAPII elongation regulate alternative splicing, because RNAPII elongation rate can favor the recruitment of splicing factors. We can illustrate this with an example: a transcriptional unit has two consecutive splice sites of different strength. The promoter-proximal one is weaker than the promoter-distal one. Surrounded by these splice sites is an alternative exon. If RNAPII elongation is slow, the spliceosome components or other splicing factors have more time to be recruited to the proximal, weaker splice site, because it emerges from RNAPII before the distal one is transcribed. This produces the inclusion of the alternative exon. On the other hand, if RNAPII elongation is fast, the two splice sites are presented to the splicing machinery at the same time, and the stronger one would be recognized more efficiently than the weaker one. This, in turn, produces exon skipping.

The first direct evidence of kinetic coupling in vivo was achieved in human culture cells working with an RNAPII mutant that has a slow elongation rate. Transcription by this slow RNAPII mutant, favors exon inclusion in a reporter minigene, compared to wild type RNAPII [89]. More recently, a work performed in A. thaliana shows that RNAPII elongation regulates alternative splicing in the context of a whole organism. Previously, Petrillo and colleagues (2014) demonstrated that light regulates alternative splicing in a subset of A. thaliana transcripts, and that this regulation is achieved through a chloroplast retrograde signal [98]. The light control of alternative splicing responds to the kinetic coupling mechanism [90]. Light promotes transcription elongation, while in darkness RNAPII elongation is lower (see Figure 2(a)). This is demonstrated by different experimental approaches, including RNAPII chromatin immunoprecipitation and a nascent RNA single molecule method. Furthermore, the light control on alternative splicing is abolished in a mutant plant defective in RNAPII elongation. TFIIS elongation factor, introduced previously in the context of seed dormancy, is required for RNAPII processivity and stimulates RNAPII to reassume elongation after pausing [81,82]. Remarkably, a group of analyzed alternative splicing events do not change their splicing patterns in an enhanced TFIIS mutant in response to light/dark transitions [90]. Therefore, the chloroplast control of alternative splicing in plants responds to the kinetic coupling in the context of a whole organism.

Figure 2.

RNAPII elongation, splicing and possible coupling factors in seedlings. (a). Alternative splicing is regulated by a light-dependent retrograde signal (green arrow, left panel) which modulates RNAPII elongation. In light conditions, RNAPII is faster than in darkness. In this elongation rate, both 3´ splice sites are presented to the splicing machinery almost simultaneously. This influences alternative splicing decisions, favoring the choice for exon exclusion in light. On the other hand, when RNAPII is slower, 3´ splice sites are presented and processed sequentially. Therefore, in this condition, exon inclusion is favored. Adapted from [90]. (b). The functional link between RNAPII elongation and splicing decisions might be achieved by proteins acting in both processes. SKIP interacts with ELF7, a subunit of PAF1 C -chromatin remodeler-, and with the splicing machinery. PRP40 interacts with RNAPII’s CTD, and it is also part of the U1 snRNP, component of the spliceosome. Depicted in bidirectional arrows, known interactions of SKIP and PRP40

Another work shows that an accelerated RNAPII elongation rate increases the polymerase signal in gene bodies [38]. The authors generated a mutant plant with a higher RNAPII transcription activity through point mutations in NRPB2, the second largest subunit of the RNAPII in A. thaliana. Using nascent RNA sequencing techniques, they demonstrated that, in this accelerated mutant, RNAPII accumulates in gene bodies. This, in turn, correlates with an enhanced splicing efficiency and with a trend to shift 5´ splice site usage upstream and 3´ splice site usage downstream [38]. Thus, accelerated transcription can also modulate alternative splicing choices. Evidence indicating a similar mechanism of regulation was previously reported by Dolata and collaborators (2015). These authors studied the case of the spliceosome disassembly factor NTC-Related protein 1 (NTR1), required for correct expression and splicing of DOG1 [81]. Interestingly, ntr1 mutant seeds have reduced dormancy, mainly due to low DOG1 expression and a shift toward downstream splice site selection in this transcriptional unit [81]. NTR1 protein was shown to be present at the gene body and promoter region of DOG1 and it is also acting in a cotranscriptional manner at affected splice sites. The splicing defects observed in the ntr1 mutant could be a consequence of a fast RNAPII elongation. In concordance with this, using the TFIIS mutant with a slower rate of RNAPII elongation, the authors observed the opposite effect on splice site selection, an increase in upstream splice site usage [81].

Transcription and splicing factors that could be acting as coupling factors

RNAPII transcription cycle can be summed up in three main steps: initiation, when the polymerase is recruited to the promoter and starts RNA synthesis; elongation, where the RNAPII transcription elongation complex (TEC) is set up and the polymerase productively transcribes through the gene body; and termination, when the polymerase finally disengages from the DNA template [99]. Interactors needed for each step (and transitions) of this highly dynamic process can then be separated into three categories: initiation factors, elongation factors and termination factors. Historically, it was believed that the initiation of transcription was the most crucial step in regulation of gene expression, that is, the recruitment of basal transcription machinery to form the pre-initiation complex [100]. However, recent work has shown that during RNAPII elongation many factors can be recruited that coordinate RNAPII elongation rate and mRNA processing, thus making the TEC a new field of interest for gene expression regulation [101,102].

RNAPII elongation rate can be modulated by different actors. These can include chromatin remodelling factors [38,103], DNA topology [104,105], or RNAPII elongation factors [106]. Among these last, TFIIS – as mentioned above-, is required for correct RNAPII elongation and is needed for alternative splicing decisions in response to light [90].

Other factors involved in the RNAPII transcription elongation complex have been described in the past years [106]. Interestingly, many of these have been shown to interact physically with the mRNA processing machinery (such as U1, U2 and U5 spliceosomal complexes and subunits of the Cstf complex, implicated in 3´ end processing). This suggests that factors acting both in transcription and in splicing regulation are a possible functional link between these two processes. Recently, a genetic screen in yeast found that a component of the Spt–Ada–Gcn5 Acetyltransferase (SAGA) complex, Spt8, genetically interacts with the spliceosomal RNA helicase Prp5p. This interaction is proposed to mediate a balance between transcription initiation/elongation and spliceosome function [107]. The SAGA complex is an evolutionary conserved, multifunctional transcription co-activator comprising two distinct enzymatic activities that modify histone residues, acetylation and deubiquitination. The histone acetyltransferase (HAT) activity of the complex is required for cotranscriptional recruitment of the U2 snRNP [108]. In yeast, the prp5-1 mutation causes RNAPII accumulation in intron sequences of intron-containing genes [107]. Hence, the SAGA complex is functionally coupling all these mechanisms underlying chromatin state regulation, transcription initiation/elongation and pre-mRNA processing. In the next section we will revisit the evidence pointing to the existence of such coupling complexes and/or factors in plants.

A bouquet of possible coupling factors in plants

Amidst the factors mentioned above, PAF1C has been shown to interact with both spliceosome and polyadenylation components [106]. Interestingly, its activity in the regulation of FLC expression and consequent modulation of flowering has been widely studied, having major roles in histone PTMs (post-translational modifications) among others [109–111]. In mammalian cells, Hou and collaborators (2019) have shown that the loss of Paf1 (a subunit of PAF1C) results in many major transcriptional defects, including defects in CTD phosphorylation and accumulation of RNAPII in promoter-proximal regions [112]. Concomitantly with its known roles in histones PTMs [113,114], they show that this complex could act as a chromatin remodeller and impact RNAPII activity during productive elongation. However, it does not affect the recruitment of other elongation factors, such a FACT complex (another chromatin remodeller), suggesting that Paf1 may have a mechanistic role of its own on modulating RNAPII elongation. That would mean that an elongation factor might be directly involved in the modulation of RNAPII elongation rate.

Moreover, ELF7, A. thaliana’s homolog of Paf1 [109], interacts with SKIP (Ski-interacting protein) in the regulation of FLC expression [115–117]. SKIP protein participates in multiple key pathways in plant development and in response to external cues. It is involved in abiotic stress response and regulation of the circadian clock mainly through its association with the spliceosome [116–120]. Consequently, SKIP is a versatile protein acting both in splicing and other layers of gene expression modulation (see Figure 2(b) and Figure 3), suggesting a possible role as a scaffold or adaptor for protein interactions and functional complex formation [121,122].

Figure 3.

Putative coupling factors modulating chromatin remodelling, RNAPII transcription and alternative splicing outcomes in A. thaliana during transition to flowering. PRMT5 symmetrically dimethylates H4 (H4R3sme2) decreasing FLC expression and also methylates some Sm proteins that influence key splicing decisions since they are part of the spliceosome. PAF1C regulates FLC expression by affecting histone PTMs. ELF7, a component of PAF1C, interacts with SKIP to activate FLC expression. In addition, PAF1C and PRMT5 can modulate RNAPII elongation. All these regulatory mechanisms, directly or indirectly, impact on alternative splicing outcomes and, in turn, modulate circadian rhythms and flowering

We can yet add another example to the list. PRE-mRNA-PROCESSING PROTEIN 40 (PRP40) was first discovered in yeast as an essential factor in the early steps of spliceosome complex formation. In yeast as well as in mammals, PRP40 helps identify the bridging interaction that links both ends of the intron. Each spliceosome is assembled around an intron in an ordered stepwise fashion. In the first step, U1 snRNP binds to the 5´ splice site of the intron, and the essential pre-mRNA splicing factors SF1 and the U2AF65/U2AF35 heterodimer accurately and cooperatively recognize the branching point sequence (BPS), the polypyrimidine tract, and the 3´ splice site. In yeast, Prp40 is associated to the U1 snRNP and interacts with the BRANCHPOINT BINDING PROTEIN (BBP) [123], the yeast ortholog of the splicing factor SF1, which in turn interacts with Mud2p, the yeast homolog of animal U2AF65 (U2 small nuclear ribonucleoprotein auxiliary factor 65) [123–125]. This interaction network might help link the ends of an intron forming the required loop [126]. PRP40 family has also relevant interactions with the CTD of the RNAPII through its WW domains (see Figure 2(b)). Furthermore PRP40 was established as the first phospho-CTD-associating protein (PCAP) in plants, linking its function with transcriptional processivity [127,128]. There are two putative mammalian orthologs of Prp40, PRPF40A and PRPF40B. These interact with the transcription and splicing machineries, and at least for PRPF40B, the modulation of alternative splice site selection in apoptosis-related genes has been shown [125,129]. Hernando and colleagues (2019) have recently shown that At-PRP40 C links the regulation of gene expression and pre-mRNA splicing in A. thaliana, and modulates plant growth, development, and stress responses [130]. Moreover, the PRP40 ortholog in Drosophila melanogaster, dPrp40, was described as a component of the Histone Locus Body (HLB) that mediates replication efficiency in vivo. The localization of dPrp40 to the HLB points toward a possible role for this factor in the regulation of histone gene expression [131]. Once again, a factor first described as a splicing regulator, possesses different roles, operating at the interphase between splicing decisions and transcription elongation. These findings further support the notion of functional coupling mechanisms and their conservation in different eukaryotes.

Another factor modulating FLC expression and controlling flowering and development is PROTEIN ARGININE METHYLTRANSFERASE 5 (PRMT5). In A. thaliana, PRMT5 has been shown to decrease FLC expression by methylation of H4 (H4R3sme2) in the promoter region [132,133]. In general, PRMTs can methylate an increasingly large fraction of the proteome and so regulate transcription factors, spliceosome components and histones. Their ability to methylate histones provides a direct route for input into the epigenetic regulation of gene expression. Indeed, the vast majority of PRMT5 substrates are associated with RNA, so arginine methylation has been implicated in all aspects of RNA metabolism, including mRNA transcription, splicing, transport, translation, and turnover [134]. PRMT5 appears to be one of the best characterized proteins of the family, acting in several types of complexes both in the cytoplasm and the nucleus. In the nucleus, it binds to COPR5 (co-operator of PRMT5) resulting in the preferential methylation of H4R3 over H3R8 [135]. Moreover, PRMT5 also interacts with Switch/Sucrose nonfermentable (SWI/SNF) chromatin remodelling complexes acting as a transcriptional coactivator [134]. PRMT5 also regulates circadian rhythms by influencing PSEUDO-RESPONSE REGULATOR 9 (PRR9) alternative splicing choices (see Figure 3). It does this by methylating Sm proteins, affecting key splicing decisions as reported by Sanchez and collaborators in 2010 in A. thaliana [136]. Another interesting interactor of PRMT5 is SPT5, member of the SPT4/SPT5 heterodimer involved in the regulation of RNAPII elongation. Remarkably, the demonstration that SPT5 and RNAPII are targets for phosphorylation by P-TEFb (CDK9/cyclin T1) indicates that post-translational modifications of these factors are important in regulating RNAPII elongation. Kwak and collaborators (2003) showed in mammalian cells that SPT5 is specifically associated with PRMT1 and PRMT5, and that SPT5 methylation modulates its interaction with RNAPII [137]. Given this plethora of interactors and known functions of PRMT5 it is possible to postulate it as another piece in this bouquet of putative coupling factors between RNAPII, chromatin and splicing regulation in A. thaliana.

Concluding remarks

An elegant experiment showed that chromatin can be established in a naive system without substantially affecting key DNA-templated processes as replication or transcription [138]. The work by Rojec and colleagues (2019) demonstrated that chromatinizing the Escherichia coli bacterial genome with histones from archaea is not substantially changing bacterial gene expression. Their results suggest this is because chromatin state and nucleosome positioning depend on patterns of gene expression. In other words, though chromatin state and histone modifications can alter RNAPII transcription, the transcriptional activity also regulates the deposition of nucleosomes and shape chromatin structure. The fact that a bacterial cell that has neither dedicated nucleosome remodellers nor co-evolved sequence context is able to cope with genome chromatinization, indicates that the cross-regulation between these processes was – most likely – established very early in eukaryotes. It is then easy to imagine that every layer, every process, required for a gene to be expressed would be regulated by, and would also be regulating, every other process and mechanism involved.

Alternative splicing is a source of gene expression diversity. Its regulation is a complex process guided by the functional coupling between transcription and splicing, with chromatin structure influencing both processes. All these molecular features operate together to increase transcriptome diversity and, most likely, protein diversity [32]. Thus, we cannot take them as separated events; in contrast, all these layers of multiple regulations are intrinsically and functionally connected processes sharing common components with different roles at each level of regulation. Noteworthy, while the role of chromatin in transcription has been extensively studied, and its role in splicing has also been widely boarded [139–142], in plants it has gained detail in the last years [36,143–145].

We have gained vast knowledge on the ways plants perceive changes in the environment and the developmental adjustments that allow these organisms to adapt to these variations. Furthermore, in the last decades we started to unravel the underlying molecular mechanisms that modulate these responses and, in turn, allow plant adaptation and shape development. However, we still need to disentangle the main actors that transduce an environmental cue to a change in gene expression that modulates a developmental/morphological response. When studying plant light responses, it becomes evident that we only know the main actors, the photosensory proteins, the retrograde signalling pathways, and some of the key factors that control chromatin state, transcription and/or splicing. As exposed throughout this review, plant development and morphology are exquisitely controlled by these cross-regulated processes. Unraveling how these connected regulatory networks operate and how this knowledge could potentially be translated to plant breeding programs can help in the design of better crops with higher productivity and tolerance to various biotic and abiotic stresses.

Funding Statement

The work was supported by the Agencia Nacional de Promoción Científica y Tecnológica of Argentina (ANPCyT) grant to EP (PICT2017-1343). RST is an ANPCyT postdoctoral fellow, MGK and LS are fellows from Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET). MGH and EP are career investigators from CONICET

Disclosure of interest

The authors report no conflict of interest