Emerging mechanisms and consequences of calcium regulation of alternative splicing in neurons and endocrine cells

Abstract

Alternative splicing contributes greatly to proteomic complexity. How it is regulated by external stimuli to sculpt cellular properties, particularly the highly diverse and malleable neuronal properties, is an underdeveloped area of emerging interest. A number of recent studies in neurons and endocrine cells have begun to shed light on its regulation by calcium signals. Some mechanisms include changes in the trans-acting splicing factors by phosphorylation, protein level, alternative pre-mRNA splicing, and nucleocytoplasmic redistribution of proteins to alter protein–RNA or protein–protein interactions, as well as modulation of chromatin states. Importantly, functional analyses of the control of specific exons/splicing factors in the brain point to a crucial role of this regulation in synaptic maturation, maintenance, and transmission. Furthermore, its deregulation has been implicated in the pathogenesis of neurological disorders, particularly epilepsy/seizure. Together, these studies have not only provided mechanistic insights into the regulation of alternative splicing by calcium signaling but also demonstrated its impact on neuron differentiation, function, and disease. This may also help our understanding of similar regulations in other types of cells.

Introduction

Alternative splicing through the selective exclusion or inclusion of pre-mRNA sequences is an important mechanism for the regulation of gene expression in metazoans. This in most cases leads to the generation of multiple protein isoforms from single genes, greatly expanding the coding capacity of the genome. In humans, about 95 % of protein-coding genes undergo alternative splicing, suggesting that it affects nearly every process in cells [1, 2]. The regulation is particularly prominent in neurons where tens of thousands of variant proteins could be produced from single genes for the highly diverse neuronal properties [3–6]. Not surprisingly, aberrant splicing accounts for up to 50 % of the defects caused by mutations that result in human diseases [7–9].

Precise recognition of splice sites and removal of RNA sequences (introns) are performed by the spliceosome, a multi-subunit molecular machine assembled stepwise on specific RNA sequence elements at the 5′ and 3′ splice sites [10]. In higher eukaryotes, these sites are highly degenerate, and are often not sufficient for accurate exon recognition. To enhance splicing fidelity and/or impose control, cis-regulatory RNA elements are utilized. These elements are usually classified into intronic and exonic splicing enhancers or silencers promoting or inhibiting, respectively, the inclusion of a regulated exon [11]. Many of these elements are bound by trans-acting protein or RNA factors. Identified neuron-specific or -enriched factors include RNA binding feminizing gene on X-1 (Rbfox-1/Fox-1/A2BP1) and -2, neural polypyrimidine tract-binding protein (nPTB/PTBP2), HuB, HuC, HuD, and neuro-oncological ventral antigen-1 (Nova-1) and -2 [12–22]. Post-translational modifications of the trans-acting proteins are extensive according to the protein modifications database (http://www.phosphosite.org/) with the effect of phosphorylation characterized intensively in some cases [23–27]. For a particular exon, its inclusion level is determined by the combinatorial effect of the positive/negative elements and factors in mammalian cells [11, 28].

Of the splicing-regulating signals, Ca²⁺ is critical for the control of various physiological processes including muscle contraction, hormone secretion, and synaptic transmission [29, 30]. An important way of this control is through specific transcriptional and post-transcriptional programs. While much has been learned about its regulation of transcription [31–34], little is known about the molecular mechanisms of its control of alternative splicing and the implications of this control in cell physiology and diseases [4, 35]. However, this situation has begun to change; a number of recent studies have shed light on these issues, which will be the focus of our current review.

Molecular mechanisms of calcium signal-regulated alternative splicing

Studies on the control of a number of alternative exons by calcium signaling in several labs have identified similar as well as distinct ways for the regulation.

Control of phosphorylation or protein level of splicing factors

A number of signaling kinases are sensitive to the elevation of intracellular calcium level. These include the calcium/calmodulin-dependent protein kinases (CaMKs), protein kinase C (PKC), cAMP-dependent protein kinase A (PKA), and MAPK/ERK [36–39]. They have been implicated in the regulation of splicing [40–46]; however, in most cases, a direct link between the kinases and splicing factors and/or other molecular details of the regulation are not known. By far, the most detailed example of calcium signal-regulated splicing is through the CaMKIV pathway.

CaMKIV belongs to the family of Ser/Thr protein kinases and requires the Ca²⁺/calmodulin complex as well as phosphorylation of Thr200 by CaMKK1 or 2 for full activation [47, 48]. It regulates the splicing of a number of alternative exons through various RNA elements (Fig. 1). For the underlying mechanisms, the most detailed example comes from studies on the activity-dependent splicing of the stress-axis regulated exon (STREX) of the Slo1 gene for the α subunit of the voltage- and Ca²⁺-activated big potassium (BK) channels [4, 49]. The inclusion of STREX-encoded peptide confers the channels a variety of distinct properties, including higher calcium and voltage sensitivity [49, 50]. The exon is repressed in rat GH₃ pituitary cells or cerebellar neurons by membrane depolarization/CaMKIV, through a CA-rich CaMKIV-responsive RNA element (CaRRE1) within its upstream 3′ splice site [51, 52]. Trans-acting factors critical for this regulation have been identified to be the heterogeneous nuclear ribonucleoprotein L (hnRNP L) and its paralog L-like (LL) [26, 53]. hnRNP L is phosphorylated at Ser513 by CaMKIV, resulting in enhanced hnRNP L-CaRRE1 interaction and reduced binding of U2AF65, an essential component of the early spliceosome, to the immediate upstream polypyrimidine tract [26]. These studies indicate that membrane depolarization/CaMKIV increases the phosphorylation of a splicing factor to regulate a critical step of early spliceosome assembly to control splicing. The regulation of hnRNP LL and the interplay among these factors remain to be elucidated.

Fig. 1.

Mechanisms of Ca²⁺-regulated alternative splicing. Stimulation of various Ca²⁺ channels in the cellular membrane or inhibition of pumps in the endoplasmic reticulum (ER) increases intracellular Ca²⁺ concentration. This activates the CaMKIV or unknown Ca²⁺-dependent pathways or factors (question mark) to regulate alternative splicing of pre-mRNA transcripts by changing protein level, phosphorylation status (pink box), nucleocytoplasmic distribution (green box) of various splicing factors, or by modulating chromatin structural state around the regulated exon (blue box). Experimentally characterized links are in solid arrows, whereas those inferred from observations are dashed. The heavy black arrow after the cytoplasmic Rbfox1 indicates translocation of the depolarization-induced, exon 19-skipped Rbfox1 protein product to the nucleus. For the pre-mRNA targets, exons are shown as heavy bars and introns as lines. H3 histone 3; Pol RNA polymerase; Ac acetyl; Me3 tri-methyl. See text for further details

Another example of CaMKIV-controlled splicing comes from the study of the N-methyl-d-aspartate receptor 1 (NMDAR1) exon 21 [54, 55]. Depolarization/CaMKIV has been shown to increase the nuclear level and binding of hnRNP A1 to the UAGG motifs of exon 21 to promote its skipping [54]. How the nuclear level of hnRNP A1 is increased and whether the protein–RNA interaction is modulated by phosphorylation like in the case of hnRNP L are interesting questions to be addressed.

In addition to hnRNP L and A1, another splicing regulatory protein SAM68 (Src-associated in mitosis, 68 kDa) is also a target of the CaMKIV pathway. In cerebellar granule neurons, SAM68 mediates depolarization-induced skipping of exon 20 of neurexin 1 (Nrx1) transcripts through cooperative AU-rich sequence elements in the flanking introns [56]. SAM68 phosphorylation at Ser20 within a CaMKIV consensus sequence is increased upon membrane depolarization without changes at the protein level or localization (Fig. 1). The effect of the phosphorylation on SAM68 and its constitutive spliceosomal target remain to be uncovered.

Besides CaMKIV, several studies have suggested that calcium can also act through other signaling pathways to regulate splicing. For example, pharmacological evidence suggests that the PKC/Rho-associated protein kinase II (ROCK II) pathway is essential for the depolarization-induced repression of the neurexin 2α exon 11 and neurexin 3α exons 11 and 20 in rat cortical neurons [57]. Interestingly, one of the essential factors for the activity-dependent repression is also hnRNP L [58]. It will be important to identify direct links between PKC/ROCK II and hnRNP L or the downstream spliceosomal targets in future investigations.

Nucleocytoplasmic redistribution of splicing factors

Besides phosphorylation- or protein level-regulated protein–RNA interactions, subcellular redistribution has also been demonstrated for a number of splicing regulatory proteins upon stimulation of calcium signaling [59–61]. For example, increasing intracellular calcium concentration by blocking sarco-endoplasmic reticulum Ca²⁺-ATPases with thapsigargin leads to the hyperphosphorylation and accumulation of the splicing factor transformer 2 homolog-beta 1 (Tra2-beta 1) in the cytoplasm [60]. This translocation is accompanied by the change of alternative splicing of ICH-1 transcripts in primary cortical neurons (Fig. 1) [60].

Particularly interesting is the subcellular redistribution of Rbfox1 upon depolarization. This is induced by the skipping of its exon 19 (E19) after 6 h of depolarization of neurons differentiated from mouse P19 cells (Fig. 1) [59]. The E19 skipping changes the amino acid composition of the C-terminus of Rbfox1, leading to its re-localization from the cytoplasm to the nucleus. The accumulated nuclear Rbfox1 promotes the inclusion of a number of alternative exons that are initially repressed by depolarization/CaMKIV. Therefore, the re-localization of Rbfox1 has a homeostatic effect on the initially repressed exons upon prolonged depolarization [59].

How the Tra2-beta 1 translocation and skipping of the Rbfox1 E19 are controlled by calcium signals in these neurons remain unknown.

Control of chromatin status

Depolarization increases chromatin accessibility around the variable exon by inducing histone modifications (H3K9 acetylation and H3K36 tri-methylation) in N2a neuroblastoma cells [62]. This change is associated with the enhanced processivity of RNA Polymerase II, likely allowing the spliceosome less time to recognize exon 18 of the NCAM transcripts (Fig. 1) [62]. Thereby, epigenetic chromatin control and transcription machinery also affect the choice of alternative exons upon depolarization, as also demonstrated in other systems [63, 64]. However, it remains unknown how the signaling pathway downstream of depolarization directs the selective modification of the histones within/around the regulated exon region. One possibility is that the specificity is derived from particular features of this exon and/or its flanking sequences or their directly bound factors.

Regulation of protein–protein interaction

A number of regulatory protein–protein interactions are mediated by calcium ions. For example, binding of Ca²⁺ to the EF-hand motif of calmodulin promotes its interaction with a variety of targets. Another member of the EF-hand domain protein family, apoptosis linked gene-2 (ALG-2) protein, has been identified as an interaction partner of the RNA-binding factor RBM22, which is involved in the splicing regulation of the Dscam gene in Drosophila [65, 66]. Moreover, RBM22 facilitates nuclear translocation of ALG-2 in a Ca²⁺-dependent way [35, 65]. Calcium signals thus also regulate splicing factors by modulating protein–protein interaction [67]. The effect of this interaction on the splicing of alternative exons awaits tests in future experiments.

Taken together, although many details remain to be learned, these recent studies have demonstrated that calcium signals control splicing factors at different levels to regulate splicing. These include posttranslational modification (phosphorylation), protein level, alternative pre-mRNA splicing, nucleocytoplasmic distribution of proteins and chromatin status (Fig. 1).

Similar mechanisms are also used in non-neuronal cells mediated by second messengers other than Ca²⁺ [11, 23, 68], suggesting that different cells and signaling pathways adopt mechanistically related strategies to regulate alternative splicing. It is thus likely that the mechanisms of Ca²⁺-regulated splicing found in neuronal and endocrine cells also work in other tissues. The neuron, endocrine cell or tissue specificity could be from the upstream protein kinase (e.g., CaMKIV) or the splicing factor (e.g., Rbfox1).

Beyond the control of individual splicing factors in these model studies, the regulation of alternative splicing by calcium signaling in cells is likely far more complex. Consistent with this notion, exon- or cell-dependent regulation by calcium signals has also been observed. For example, the STREX can be either inhibited or enhanced by depolarization depending on the cells used [52, 54]. Such context-dependence is likely due to the different combinations of RNA elements among genes or of splicing factors/signaling pathways among different cells, as in other cases [44, 69]. More detailed studies regarding this aspect are needed in future investigations.

Sculpting protein and cellular functions through calcium-regulated alternative splicing

In excitable cells, calcium signaling regulates the alternative splicing of a large number of genes associated with different functional categories. Investigations of individual exons, database search for CaMKIV-responsive RNA elements, as well as microarray analyses have identified calcium-regulated alternative exons of genes involved in calcium homeostasis, ion or vesicular transport, intracellular signaling and synaptic functions, as well as metabolism and cell apoptosis [4, 51, 52, 54, 56, 57, 59, 70–73]. These genes encode ion channels, pumps, receptors, protein kinases/phosphatases, neuroendocrine secretory proteins, synaptic proteins, metabolic enzymes, and DNA/RNA binding proteins. Their regulation has important implications in cellular functions.

Temporal control of alternative splicing of different genes by calcium signaling

The regulation of NMDAR1 exon 5 by relocalized Rbfox1 upon prolonged depolarization is apparently not the only example of temporal control of splicing (Fig. 1). Data from an exon array analysis has shown that depolarization/calcium signals induce splicing changes of many genes involved in various pathways in a temporally controlled manner [70]. In this study using human IMR-32 neuroblastoma cells depolarized with potassium chloride (KCl), genes displaying changes throughout the whole course of KCl treatment (0.5–24 h) encode mainly Ca²⁺-ion binding proteins. Genes affected only at the early time points (0.5–3 h) encode kinases and their regulators. Those changed at the intermediate time of depolarization (6–24 h) are involved in metabolism and transcription, whereas genes regulating the cell cycle, apoptosis, and RNA splicing change their splicing pattern only at later time points (12–24 h). Like the regulation by Rbfox1 [59], these temporal splicing alterations are likely part of a homeostatic control mechanism employed by excitable cells to adapt to prolonged stimulation. Its underlying molecular mechanism and impact on particular cellular functions are interesting for further studies.

How could the regulated splicing of specific exons/genes by calcium signaling translate into the alteration of protein and cellular functions? Two recent studies in animal models have provided important insights into its role in sculpting neuronal functions (Fig. 2).

Fig. 2.

Cellular functions modified by Ca²⁺-regulated alternative splicing. a Critical role of Ca²⁺-regulated NRX1 variants in synaptic maturation during development. Depolarization switches NRX1 post-synaptic ligand binding preference from NL-1B to Cbln1/GluD2 complex by inhibiting exon 20 (E20) inclusion in cerebellar neurons in an l-type Ca²⁺-channel dependent way. b Role of Ca²⁺-regulated AMPAR subunit isoforms in homeostatic synapse response. Inhibition of sodium channels with tetrodotoxin (TTX) or l-type Ca²⁺-channels with nifedipine results in the skipping of AMPAR subunits GluA1 and A2 Flip and inclusion of Flop exon in hippocampal CA1 neurons. Flip-included mRNA isoform GluA1i decays faster than GluA2i. Moreover, GluA1 turns over more rapidly than GluA2 protein variants in the endoplasmic reticulum, where AMPAR channel is assembled. Thus, TTX-induced activity blockage leads to the enrichment of newly synthesized GluA1o relative to GluA1i. GluA1o pairs preferentially with GluA2i variant producing GluA1o/A2i heterotetramer, which is associated with reduced depression of AMPAR-mediated excitatory post-synaptic potentials evoked by electrical stimulation of pre-synaptic neuron. For simplicity, nuclei of pre- (a) and post- (b) synaptic neurons are depicted close to synaptic membrane

A role for calcium signal-regulated splicing in synapse maturation

Various  NRX protein isoforms interact with a number of post-synaptic ligands such as neuroligins, leucine-rich transmembrane proteins (LRRTMs), Cbln1/glutamate receptor delta 2 (GluD2) to establish functional synapses [74]. Three different Nrx genes (Nrx1, 2, and 3) are extensively regulated at the level of alternative promoter usage and splicing, generating protein isoforms with distinct binding preferences [75]. For example, Nrx1 exon 20-included isoform NRX1 (+E20) preferentially binds to the Cbln1-GluD2 complex but only weakly to the neurolignin-1B receptor (NL1B), whereas the exon 20-excluded variant NRX1 (−E20) exhibits the opposite activity. As mentioned above, the alternative splicing of Nrx genes is controlled by membrane depolarization in neurons [56–58]. Depolarization-induced skipping of exon 20 correlates with enhanced pre-synaptic differentiation in neurons induced by Cbln1-GluD2 but not by NL1B, consistent with strong functional coupling between NRX1 (−E20) and Cbln1-GluD2 (Fig. 2) [56]. Importantly, the activity-dependent Nrx splicing was specifically abolished in the cerebellar granule neurons of mice knocked out of its essential splicing regulator SAM68. These mice have altered mossy fiber synapses and motor coordination defects. Taken together with the central role of Nrx alternative splicing in selective synapse formation [75], these results point towards an important role of Ca²⁺-regulated splicing in this process during neuron differentiation.

A role for calcium signal-regulated splicing in modulating synaptic transmission

This was found in a recent study on the regulated assembly of heterotetrameric α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid-type glutamate receptors (AMPARs) [76]. Of the mutually exclusive Flip/Flop exons of the subunits GluA1 and GluA2 of AMPAR, Flip usage was reduced in response to activity deprivation in an l-type calcium channel-dependent way in the CA1 but not CA3 hippocampal subfield. This change led to the expression of compositionally and kinetically distinct AMPARs with reduced depression of their synaptic responses (Fig. 2) [76]. Thus, splicing-driven remodeling of post-synaptic receptors upon stimulation by calcium signals likely plays a role in synaptic plasticity and neuronal homeostasis in a cell-type-specific manner.

Related to the molecular basis of this regulation, notably SRp38 has been shown to regulate these exons in non-neuronal cells [77]. Moreover, the Flip inclusion appears to be also correlated with the A to I RNA editing at the −2nt of the upstream constitutive exon, which adds an additional layer of complexity to the regulation for the fine-tuning of neuronal properties [78].

Role of calcium signal-regulated alternative splicing under abnormal conditions or in diseases

Intracellular calcium homeostasis is tightly controlled through the complex system of ion channels, pumps, ion exchangers, and Ca²⁺-binding proteins to maintain neuronal physiology. Distortion of the Ca²⁺ balance can lead to neuronal injury or death accompanying the development of neurological disorders such as Alzheimer disease, Parkinson disease, Huntington disease, and epilepsy [79, 80]. Studies by disrupting the control of alternative splicing by calcium signaling have pointed to its role in some of the neurological disorders.

Loss of the calcium-regulated splicing factor Rbfox1 confers susceptibility to epilepsy/seizure

Rbfox1 controls the expression of many neuronal transcripts [12, 14, 16]. Consistently, disruption of the human RBFOX1 gene by mutations has been associated with several neuropathies including mental retardation, autism, and idiopathic epilepsy [81–85].

A recent study has demonstrated that loss of Rbfox1 causes dysregulation of Ca²⁺ homeostasis and some neuropathological conditions [14]. Rbfox1 knockout mice are susceptible to kainic acid-induced seizure and display enhanced neuronal excitability associated with altered synaptic function in the dentate gyrus. Exon-junction microarray and RT-PCR analysis identified splicing changes of genes essential for synaptic transmission and Ca²⁺ homeostasis. Mutations of some of the genes have been directly linked to epilepsy, such as the synaptic protein Snap25 and the sodium channel Scn8a [86–88]. The aberrant splice variant(s) that is responsible for the altered intracellular calcium balance and the epileptic phenotype of Rbfox1 ^−/− mice remains to be verified.

Altered splicing of BK and NMDAR1 exons in epilepsy

Traumatic brain injury (TBI) can potentially lead to the development of acquired epilepsy (AE). It has been suggested that alterations of intracellular calcium dynamics triggered by TBI may be a major cause of long-term morphological, biochemical, and physiological plasticity changes underlying AE [80, 89–93]. Even though AE accounts for about 50 % of all cases of epilepsy, there is still a lack of understanding how changes in neuronal molecules contribute to this disease.

Reports have shown that the alternative splicing of STREX and the NMDAR1 exons 5 and 21 are altered by enhanced neuronal excitability during epileptic changes or prolonged toxic depolarization. In rat dentate gyrus following pilocarpine-induced status epilepticus, the Kcnma1 (Slo1) transcripts of the BK potassium channels are reduced but its STREX-containing isoform is increased [94, 95]. In cortical neurons, depolarization-driven rearrangement of NR1 variant subunits (exon 21 skipping) leads to the decrease of NMDAR-mediated inward Ca²⁺ current [96]. This decrease is accompanied in survival neurons by neuroprotection from cell death and calpain activation under excitotoxic stress upon prolonged depolarization [96]. The splicing change is consistent with those in seizures in earlier studies [97–100]. In the context of epileptogenesis, the surviving neurons likely form the pathophysiological basis of epilepsy [80, 101]. The role of these altered alternative splicing events in the progression of epilepsy is thus worth further detailed investigation.

Conclusions and future considerations

Recent studies have provided a number of mechanistic insights into calcium signal-regulated alternative splicing in neuronal and endocrine systems as well as the important implications of this regulation in neuronal functions and diseases. What has been learned from these systems may be helpful to us in understanding the Ca²⁺-regulated splicing in other tissues as well. Yet lying ahead there are still challenges before a comprehensive view could be obtained.

For the molecular mechanisms, current studies have been mostly based on the regulation of one or a few target exons, but the regulation of alternative splicing must be accurate and specific both temporally and spatially in the context of all the other alternative exons to meet the demand for the highly complex structure and function of the brain or other tissues. How the splicing regulation by calcium signaling can be achieved specifically for a group of exons/genes among different cells, during development or in adults is an important question for future investigations.

For the role of this regulation in other cellular functions such as learning, memory, and hormone secretion, as well as in diseases, more examples of the splicing-driven changes in these processes need to be examined and verified. Although this could be challenging in many cases considering the large number of splice variants produced from the alternative exons and complex functions of the “splicing factors”, there have been successes in other systems [102–106]. Ideally, the splicing-driven changes should be confirmed by loss-of-function/rescue of the effect of a regulated splice variant as in other cases [102, 103]. The rescue is particularly important when studying the loss-of-function effect of the commonly known “splicing factors” since most of them also have additional functions other than splicing in RNA metabolism [107, 108]. Another possible approach is to combine large-scale sequencing data with evidence of genetic co-segregation of a regulated variant with a phenotype in disease families, as for the discovery of disease-causing genetic mutations that disrupt splicing [104–106].

Such investigations in the future shall provide a fine map of the regulated exons under different contexts for a better understanding of their regulatory mechanisms and roles in processes such as the fine-tuning of neuronal properties or the development of neurological diseases.