Alternative splicing: Functional diversity among voltage-gated calcium channels and behavioral consequences

Abstract

Neuronal voltage-gated calcium channels generate rapid, transient intracellular calcium signals in response to membrane depolarization. Neuronal Ca_V channels regulate a range of cellular functions and are implicated in a variety of neurological and psychiatric diseases including epilepsy, Parkinson’s disease, chronic pain, schizophrenia, and bipolar disorder. Each mammalian Cacna1 gene has the potential to generate tens to thousands of Ca_V channels by alternative pre-mRNA splicing, a process that adds fine granulation to the pool of Ca_V channel structures and functions. The precise composition of Ca_V channel splice isoform mRNAs expressed in each cell are controlled by cell-specific splicing factors. The activity of splicing factors are in turn regulated by molecules that encode various cellular features, including cell-type, activity, metabolic states, developmental state, and other factors. The cellular and behavioral consequences of individual sites of Ca_V splice isoforms are being elucidated, as are the cell-specific splicing factors that control splice isoform selection. Altered patterns of alternative splicing of Ca_V pre-mRNAs can alter behavior in subtle but measurable ways, with the potential to influence drug efficacy and disease severity. This article is part of a Special Issue entitled: Calcium channels.

1. Introduction

Voltage-gated calcium channel (Ca_V) currents originate from the activity of multiple classes of channels with different pharmacological sensitivities and unique functional properties. Neuronal Ca_V channels are implicated in a variety of neurological and psychiatric diseases including epilepsy, Parkinson’s disease, chronic pain, schizophrenia and bipolar disorder [1–9]. Relatively large differences in the voltage-dependence of channel activation between Ca_V3 (T-type) channels on the one hand, and Ca_V1 (L-type) and Ca_V2 (N, P/Q, R-types) channels on the other offered the first clues that many cells expressed more than one type of voltage-gated calcium channel [10–12]. Genome and transcriptome sequencing showed that mammalian neurons can express up to 9 of 10 different Ca_V α1 subunit genes [13,14]. Low threshold activation and slow deactivation characteristics are features of all Ca_V3 channels that set them apart from Ca_V1 and Ca_V2 family members [15].

With the exception of the skeletal muscle Ca_V1.1 that functions primarily as a voltage sensor, the unifying function of Ca_V channels is rapid control of voltage-dependent calcium entry [16,17]. Calcium entering through different Ca_V channels activates distinct signaling pathways depending on the unique sub-cellular localization, protein associations, and functional properties of Ca_V channels [13]. Thus, certain Ca_V channels are commonly associated with certain cellular functions, but new roles for Ca_V channels continue to be discovered. Ca_V1 channels expressed in neurons (encoded by Cacna1c, Cacna1d, and Cacna1f genes) support a range of calcium-dependent processes including modulation of gene expression, long and short-term changes in synaptic plasticity (Ca_V1.2, Ca_V1.3) [18–20], transmitter release at sensory nerve terminals (Ca_V1.3, Ca_V1.4) [21,22], and intrinsic spiking (Ca_V1.3) [23]. Ca_V2 channels (encoded by Cacna1a, Cacna1b, and Cacna1e genes) are primarily located at presynaptic nerve terminals where they control voltage-dependent calcium entry that triggers transmitter release [24,25]. Ca_V3 channels (Cacna1g, Cacna1h, Cacna1i genes) underlie pacemaking in many neurons including thalamic relay neurons [4,26].

This review is focused on the even finer granulation of structural and functional diversity among Ca_V channels that originates from each major Ca_V channel gene. Distinct Ca_V channels expressed in a given cell, distinguished by relatively small discrete differences in amino acid sequence, may number in the tens to hundreds depending on the extent of alternative pre-mRNA splicing [15,27]. Several sites of alternative splicing are present in Cacna1 genes predicting variations in amino acid sequence (e.g. see Fig. 1 for Cacna1b). Assuming each exon is regulated independent of the others, the number of discrete mRNA isoforms possible from each Cacna1 gene is potentially staggering (2^N for N sites of alternative splicing). Analyses of different brain regions at different stages of development show that the composition of the pool of Ca_V mRNA splice isoforms varies with cell-type, state of development, and possibly neuronal activity [15,28]. This suggests that anticipated subtle functional differences among splice isoforms within a given Ca_V family are either individually or collectively contributing in important ways to neuronal processes. Pre-mRNA splicing is also implicated in many neurological diseases. Disease-causing mutations can disrupt splicing such as inherited frontotemporal dementia and Parkinsonism linked to chromosome 17, amyotrophic lateral sclerosis, spinocerebellar ataxia 8, and myotonic dystrophy [29]. Alternative splicing can play a role in modifying disease such as in the case of Timothy syndrome [30]. The cellular and behavioral consequences of only a few Ca_V splice isoforms are known (discussed below), but this should change as new methods are developed.

Fig. 1.

Alternatively spliced exons in Cacna1b and patterns of expression. Cacna1b pre-mRNA undergoes extensive alternative splicing generating 10 s to 100 s of unique Ca_V2.2 proteins in neurons. Alternatively spliced exons e18a, e24a, and e31a are cassette exons. E37a and e37b are mutually exclusive exons. Alternatively spliced exons are expressed in different regions of the nervous system, in different types of cells, and at different stages of development. Examples of different expression patterns of splice isoforms are shown for embryonic and adult brain, superior cervical ganglia (SCG), cortex, and nociceptors. The tissue distribution and functional consequences of these exons on Ca_V2.2 channel properties have been described in a series of publications [15,28,85,86,88,119,120].

1.1. Alternative pre-mRNA splicing

Alternative pre-mRNA splicing is particularly prevalent in the mammalian brain [31,32], consistent with the theory that alternative splicing evolved in parallel with biological complexity [33]. Alternative splicing is essential for normal neuronal development, axon targeting, neuronal excitability, and neural circuit formation [27,34–38]. Several excellent reviews outline the molecular interactions involved in alternative pre-mRNA splicing [39–42]. In brief, this form of pre-mRNA processing occurs in the cell’s nucleus and it is controlled by the concerted actions of cell-specific splicing factors (SFs). These cell-specific SFs bind to consensus motifs on pre-mRNAs and influence the action of the spliceosome by promoting or repressing inclusion of alternatively spliced exons, and promoting or repressing the use of alternative splice acceptor or donor sites at intron/exon boundaries. The cell-specific features of alternative pre-mRNA splicing are controlled by the collective action of cell-specific SFs that bind to elements encoded in each gene [40,43–49].

SFs known to control alternative pre-mRNA splicing in neurons include Nova1/2 [45], nPTB [50–52], rbFox1/2/3 [53–55] and SF2/AF [56]. Networks of genes targeted by specific SFs have been generated from genome-wide analyses of pre-mRNAs that bind them. Based on data from these types of studies, certain SFs are shown to associate with genes that control particular aspects of neuronal function. For example, Nova appears to preferentially regulate alternative splicing of pre-mRNAs encoding proteins found at inhibitory synapses [45,57,58]. Current knowledge of SFs that regulate alternative splicing of Ca_V pre-mRNAs is summarized below.

1.2. Alternative splicing and voltage-gated calcium channels

The large number of Ca_V channel splice isoforms expressed in the nervous system, combined with their distinct expression patterns according to region and cell type, is consistent with the remarkable range of neuronal functions and behaviors regulated by Ca_V channels. Experiments aimed at establishing the contributions of Ca_V channel splice isoforms to neuronal function have been hindered by the lack of isoform-specific tools to tease their contributions apart at the protein level. Nonetheless, experimental approaches that selectively target mRNA splice isoforms and that directly modify alternatively spliced exons at the gene level have proven valuable. Similarly, knowledge of cell-specific expression patterns of splice isoform mRNAs gives important clues about function (see below).

Sites of alternative splicing in Cacna1 genes are typically, although not exclusively, located in regions encoding hyper-variable domains of Ca_V proteins that presumably can accommodate changes in protein structures. These include the intracellular C-termini and the II–III intracellular linker (see Fig. 1 and see [59]). The location of certain alternatively spliced exons is conserved among Cacna1 genes. For example, all but one Cacna1 gene contain an alternatively spliced exon that encodes a peptide in the putative extracellular linker between transmembrane spanning helices S3 and S4 in domain IV of Ca_V channels. In some genes such as Cacna1a and Cacna1b, a homologous alternatively spliced exon also encodes a short peptide sequence in the S3–S4 linker of domain III [58,60,61]. The composition and/or length of the S3–S4 extracellular linker exons influences voltage-dependent gating of Ca_V channels, perhaps unsurprisingly given their proximity to putative S4 voltage-sensors [62,63] (Fig. 2). The modulation of Ca_V channel activation induced by these exons can be relatively small (5–7 mV) but such changes can significantly impact the total movement of calcium during action potential-like stimuli [62]. Thus, different Ca_V splice isoforms might be selected to fine-tune the coupling efficacy between membrane depolarization and calcium entry according to cell needs (Fig. 2). Ca_V1.1 channel splice isoforms with different IVS3–IVS4 linkers are also expressed in the skeletal muscle and these originate from different mRNAs that either contain or lack exon 29 of Cacna1s. The major form of Ca_V1.1 early in muscle development lacks e29, but it is included with greater frequency during pre-mRNA splicing with maturation [64]. Ca_V1.1 splice isoforms lacking e29a generate larger currents that activate at more negative voltages compared to clones that contain e29a [64].

Fig. 2.

The potential impact of alternative splicing of Ca_V channels on the voltage range over which each channel subtype can operate. Boltzmann activation curves are plotted using V_1/2 and k values from the literature from recordings using 1–2 mM calcium as charge carrier, except for data for Ca_V1.1 obtained from recordings using 10 mM calcium. A, Ca_V1.1 (V_1/2 =6.2 mV, k=5.3 mV; [121]), Ca_V1.2 (V_1/2 = −17 mV, k=8 mV; [122]), Ca_V1.3 (V_1/2 = −36 mV, k=8 mV; [123,124]), Ca_V2.2 (V_1/2 = −0.1 mV, k=7.5 mV; [125]), and Ca_V3.1 (Ca_V3.1, V_1/2 = −46 mV, k=4.11; [126]). B, Alternative splicing of exons can modify voltage-dependence of activation and this will increase the operating voltage range for each Ca_V channel family. Approximate locations of regions encoded by alternatively spliced exons that influence the voltage-dependence of channel activation are shown. Ca_V2.2 data are from [125]; Ca_V1.1 data are from [64]; and Ca_V3.1 data are from [127]. Ca_V1.2 [128] and Ca_V1.3 [67,129] isoforms that have different activation properties were compared using 11 mM barium as charge carrier, this will shift the voltage-dependence of activation relative to recordings with 2 mM Ca. For comparison with other data, we illustrate an approximate operating range for Ca_V1.2 and Ca_V1.3 (dotted lines) based on the different V_1/2 values from [67,128,129] but with reference to the activation curves shown in A obtained with 2 mM Ca.

Regardless of their location relative to the voltage sensors and putative gating domains, many alternatively spliced Cacna1 exons impact Ca_V channel gating. For example, alternatively spliced cassette exons encoding peptide sequences and the use of alternative splice junctions in intracellular regions of Ca_V2.2, Ca_V1.2, and Ca_V1.3 channels influence the voltage dependence of channel activation and inactivation, channel gating kinetics, and calcium-dependent inactivation [65–69]. The different C-termini splice isoforms of Ca_V1.3 channels have distinct tissue distributions with a short C-terminus isoform devoid of tonic inhibitory control exerted by full length C-termini [68]. This short C-terminus isoform is expressed in the brain but not in the heart and passes substantially more current during short burst action potential-like stimuli [68].

1.3. Cell-specific factors that control alternative splicing of exons of Cacna1 genes

Comparative analyses of Ca_V mRNAs expressed in different regions of the nervous system provide strong evidence of cell-specific and development-specific splicing events [15,27,28,70–72]. Splicing factors bind their unique motifs which are often located in introns that flank target AS exons, but these motifs can also exist within the target exon itself [73]. SF binding location relative to the target exon is often predictive of the overall impact of the SF. For example, Nova and rbFox proteins typically repress exon inclusion when they bind their respective consensus sequences upstream of the target exon, and typically promote exon inclusion when they bind downstream of the target exon [74,75]. The coordinated expression and/or activity of available SFs determines the composition of the pool of Ca_V channel isoforms in a given cell type [27]. Splicing factors are in turn regulated by cell-specific molecules including miRNAs [76], kinases, and phosphatases [77,78]. The potential exists to approximate which Ca_V mRNA isoforms are expressed when, and in which cell-type, based on SF binding sites. Cacna1-derived sequences are present in recent genome-wide, high throughout sequence analyses of the sites of splicing factor binding, and certain SF binding sites have been functionally validated. For example Nova-2, a brain-specific splicing factor, binds Cacna1b and Cacna1a pre-mRNAs to promote inclusion of e24a and repression of e31a thereby influencing the composition of both Ca_V2.1 and Ca_V2.2 mRNAs [58,60]. E24a and e31a encode short peptide sequences in the IIIS3–IIIS4 and IVS3–IVS4 linkers, respectively, and as discussed above, they modulate the voltage-dependence of Ca_V2 channel activation (Fig. 2). RbFox proteins, a family of splicing factors implicated in the control of alternative splicing during development, were recently shown to repress e9* and promote inclusion of e33 of Cacna1c, influencing the composition of Ca_V1.2 mRNAs expressed in the cortex [79,80]. At a different splice site, polypyrimidine tract-binding protein (PTB) and its neuronal homolog (nPTB) control the developmental switch from exon 8 to exon 8a-containing Ca_V1.2 mRNAs in neurons [81]. Many more cell-specific splicing events involving Ca_V pre-mRNAs are described for which the corresponding SFs have not yet been identified. Identifying these SFs is necessary to understand how cells control Ca_V calcium signaling.

1.4. Mutually exclusive splicing

The alternatively spliced exons that encode short peptide sequences in the S3–S4 linkers of Ca_V channels are cassette exons, they are either included or excluded during pre-mRNA processing. In either form, exon included or skipped, the reading frames of the resultant mRNAs are typically preserved and translated into functional Ca_V isoforms. An exception to this involves cassette exons that shift the reading frame, when included or when excluded, leading to non-functional proteins. This form of alternative splicing regulates expression levels of critical proteins, including SFs, and is known to be critical for early cellular differentiation [82]. By contrast, mutually exclusive splicing involves the selection of one of a pair of exons that encode slightly different sequences. In mutually exclusive splicing, the mechanism of exon selection must incorporate a form of steric hindrance to ensure exon selection is strictly mutually exclusive. If neither one of the pair of mutually exclusive exons e37a and e37b of Cacna1b is included during pre-mRNA splicing, this results in a frame shift which creates an early stop and a non-functional truncated Ca_V protein [66]. The same is true for the homologous exons e37a and e37b of Cacna1a [83,84] and e8 and e8a of Cacna1c [81]. Thus, aberrant splicing involving mutually exclusive exons is expected to be deleterious resulting in non-functional protein.

The splicing factors that regulate mutually exclusive splicing of e37a and e37b of Cacna1b and Cacna1a are not known. E37a of Cacna1b is expressed in a limited number of cell types including nociceptors, whereas 37b is abundant throughout the nervous system [28,85]. Several mechanisms could explain the observed cell-specific expression pattern but all involve cell-specific splicing factors. Based on our analyses of Ca_V2.2 mRNAs expressed in dorsal root ganglia and brain of mice that contain either tandem e37a or tandem e37b exons, we favor a mechanism involving a splicing repressor that binds to e37a [86]. We predict this putative splicing repressor is expressed in most neurons but is absent or at low levels in nociceptors and other neurons that express e37a-contaiing Ca_V2.2 mRNAs (Fig. 3).

Fig. 3.

Theoretical model showing how the action of splicing factors might regulate the expression of e37a in different neurons. A putative splicing repressor (orange) binds to e37a sequence in Ca_V2.2 pre-mRNA preventing its inclusion during alternative pre-mRNA splicing. The putative splicing repressor is expressed in most neurons except in nociceptors and other neurons that express e37a. In the absence of the splicing repressor, e37a and e37b are equally likely to be included during pre-mRNA splicing. The model is based on data published in [86]. When e37a is moved to the e37b position, e37a is repressed consistent with the presence of an exonic repressor element [86]. When e37b is moved to the e37a position, it is expressed at wild-type levels inconsistent with the presence of a repressor element that resides exclusively in the intron [86].

1.5. Functional consequences of alternative splicing

Alternatively spliced exons are present in >95% of multi-exon genes [87] arguing that cellular control over exon selection must play a critical role in normal development and cell function [43]. By definition, cell-specific splicing events occur in a limited population of cells and tissues making it difficult to locate the products and, most of the time, tools such as antibodies and modulators to selectively identify and specifically control the activity of individual splice isoforms do not exist. The enrichment of the mutually exclusive e37a of Cacna1b in capsaicin-responsive nociceptors of dorsal root ganglia therefore offered a unique opportunity to assess its functional importance. Several other features of this system, as well as knowledge of the functional differences between e37a and e37b Ca_V2.2 splice isoforms, were advantageous. Among these features are: 1) A measurable behavior that is relatively easy to isolate and characterize; nociceptors have a well-defined role in detection of thermal stimuli and standard methods exist to monitor thermal responsiveness in vivo. 2) Ca_V2.2 channels dominate in controlling calcium entry to trigger glutamate release from presynaptic terminals of nociceptors in the spinal dorsal horn. 3) Several neurotransmitters and drugs down regulate synaptic transmission by inhibiting the gating of presynaptic Ca_V2.2 channels at nociceptor terminals through G protein coupled receptor activation. 4) E37a promotes G_i/o protein coupled receptor inhibition of Ca_V2.2 by a mechanism that persists independent of the stimulus. This pathway contrasts with G_i/o protein coupled receptor inhibition of e37b Ca_V2.2 channel isoforms that is reversed by membrane depolarization. Two complementary approaches were used to assess the functional importance of e37a in vivo [88].

1.6. Behavioral significance of e37a of Cacna1b

Without tools to selectively inhibit e37a-Ca_V2.2 and to leave e37b-Ca_V2.2 channels unaffected, isoform-specific siRNAs were used to reduce mRNA levels and protein levels in vivo [88]. Behavior analyses of thermal and mechanical thresholds in mice, following intrathecal application of isoform-selective siRNAs, suggested that e37a-Ca_V2.2 channels preferentially participate in the transmission of basal thermal nociception. E37a-Ca_V2.2 channels also mediated thermal hyperalgesia that accompanied peripheral nerve injury based on the complete reversal of thermal withdrawal thresholds in mice injected with siRNAs targeting e37a-Ca_V2.2 mRNAs [88]. This study showed that e37a-Ca_V2.2 splice isoforms are targeted to nociceptors nerve terminals and that they have a preferred role in mediating synaptic transmission in thermosensing.

An alternative experimental strategy was needed to establish if functional differences in G_i/o protein inhibition between e37 Ca_V2.2 splice isoforms impact behavior. In order to do this, e37a in the mouse Cacna1b gene was replaced with a second copy of e37b pre-serving wild-type Ca_V2.2 levels [86]. No obvious behavioral deficits were observed in e37a-null mice and withdrawal reaction times to thermal stimuli were not different from wild-type mice. However, in mice lacking e37a spinal morphine was a less effective analgesic to thermal stimuli compared to wild-type mice. This report offers evidence that disrupting an individual splicing event can impact animal behavior in subtle but measurable ways, in this case by altering the pharmacological efficacy of morphine as an analgesic [86].

1.7. Alternative pre-mRNA splicing, calcium channels, disease, and therapeutics

The most widely cited example of alternative splicing involves a gene critical in sex determination pathways in the fruit fly Drosophila melanogaster [89,90]. The splicing factor transformer (tra) controls alternative splicing of doublesex (dsx), the master controller of somatic sexual dimorphism in flies [91,92]. The consequences of aberrant splicing in the sex determination pathway are dramatic and global [93]; however, the majority of alternative pre-mRNA splicing events likely participate in delimited cellular processes.

Assessing the impact of such events on animal behavior, while more challenging, is potentially very relevant to processes that could contribute to neurological and psychiatric diseases [35,94–98]. In some cases altered splicing patterns for a particular gene or a subset of genes explain the pathophysiological manifestations of a disease such as cystic fibrosis [99,100]. In these cases, the presence of certain alternatively spliced mRNAs and proteins serves as disease biomarkers and, in the case of certain cancers, the presence of specific isoforms of MDM2, survivin-2B and CD44 defines cancer type [101]. In a very recent study, the loss of the splicing factor muscleblind-like protein 2 (Mbnl2) disrupts normal splicing of hundreds of exons and produces neurological symptoms similar to those that characterize human myotonic dystrophy [102]. Exon 12a of Cacna1d is among the list of affected splicing events in mice lacking Mbnl2 and, altered levels of e12a-containing Ca_V1.2 mRNAs are also observed in brains of human patients with myotonic dystrophy [102]. In humans, myotonic dystrophy is associated with abnormal CUG expansions in RNAs. A hypothesis presented in this study posits that CUG repeats sequester Mbln2, interfering with target exon splicing, and generating abnormal patterns of splice isoforms, including Ca_V1.2 mRNAs, in the nervous system.

In other cases, alternative splicing acts as a disease modifier. For example, mutually exclusive alternatively spliced exons e8 and e8a of human CACNA1C are mutated in two different types of Timothy’s syndrome respectively [30]. Because these exons are mutually exclusive, the effect of a gain-of-function mutation in one exon might be mitigated to some degree by the activity of the alternative wild-type exon in some tissues [30]. One of a pair of mutually exclusive exons, e8B, in CACNA1D is mutated in the human disorder SANDD syndrome and is associated with deafness and bradycardia [103]. Interestingly, the exons affected in both Timothy syndrome and SANDD map to homologous regions of CACNA1C and CACNA1D genes and encode homologous regions of Ca_V1.2 and Ca_V1.3 channels, respectively. Changes in the relative abundance of Ca_V1.2 channel splice isoforms have also been documented during cardiac disease [104]. Finally, spinocerebellar ataxia type 6 (SCA6) is associated with poly-glutamine repeats in the C-terminus of Ca_V2.1. The choice of an alternative 3′ acceptor during pre-mRNA splicing determines whether e47 of CACNA1A is in frame or out of frame resulting in an early stop and a truncated C-terminus. In patients with SCA6, inclusion of the complete e47 during splicing leads to Ca_V2.1 mRNAs containing expanded CAG repeats; the translation of these mRNAs generates Ca_V2.1 channels with poly-glutamine repeats that are responsible for the severity of SCA6 [65,105,106].

The role of alternative pre-mRNA splicing of Ca_V channels in disease, particularly as a disease modifier, adds credence to therapeutic strategies designed to target splicing factors [107–109] and specific splice isoforms [110,111]. For example, in Timothy’s syndrome, it might be possible to shift the balance of splicing toward the non-mutated exon by targeting the splicing factors that control exon selection or by knocking down the mutated splice isoforms that underlie the gain of function phenotype [30]. Isoform-specific antisense oligonucleotides and interference RNAs have been developed to knockdown mutated mRNA isoforms as a strategy to treat certain types of thalassemias and dystrophies, cystic fibrosis, cancer, and pain [88,112–115]. Other approaches to modify splicing events involve the use of bifunctional oligonucleotides. These hybrid molecules are designed to bind to specific regions of pre-mRNAs and to modify splicing via an antisense-targeting domain. For example, bifunctional oligonucleotides have been used to enhance e7 inclusion in SMN2 to reduce the severity of spinal muscular atrophy [116–118].

2. Concluding remarks

Cell-specific control of alternative pre-mRNA splicing is used to optimize depolarization-dependent calcium signaling in different parts of the nervous system, at different stages of development, and potentially depending on neuronal activity. Comparing unique functional characteristics among splice isoforms points to domains on Ca_V channels that regulate function. Identification of the cell-specific splicing factors that determine the composition of Ca_V splice isoforms is now needed. This information will be valuable not only to understand the coordinated expression of Ca_V splice isoforms and their relationship to other functionally related genes, but also to develop strategies to manipulate splicing patterns. The role that alternative splicing plays in disease particularly as a disease modifier suggests novel therapeutic approaches based on selectively down regulating disease causing isoforms and up-regulating isoforms that have therapeutic benefits. Continued technical advances in transcriptome analyses according to cell-type will greatly assist in identifying functionally relevant Ca_V channel splice isoforms, and ultimately their behavioral significance.