The neuronal splicing factor Nova controls alternative splicing in N-type and P-type Ca_V2 calcium channels

Abstract

Many cellular processes are involved in optimizing protein function for specific neuronal tasks; here we focus on alternative pre-mRNA splicing. Alternative pre-mRNA splicing gives cells the capacity to modify and selectively re-balance their existing pool of transcripts in a coordinated way across multiple mRNAs, thereby effecting relatively rapid and relatively stable changes in protein activity. Here we report on and discuss the coordinated regulation of two sites of alternative splicing, e24a and e31a, in P-type Ca_V2.1 and N-type Ca_V2.2 channels. These two exons encode 4 and 2 amino acids, respectively, in the extracellular linker regions between transmembrane spanning segments S3 and S4 in domains III and IV of each Ca_V2 subunit. Recent genome-wide screens of splicing factor-RNA binding events by Darnell and colleagues show that Nova-2 promotes inclusion of e24a in Ca_V2.2 mRNAs in brain. We review these studies and show that a homologous e24a is present in the Ca_V2.1 gene, Cacna1a, and that it is expressed in different regions of the nervous system. Nova-2 enhances inclusion of e24a but represses e31a inclusion in Ca_V2.1 and Ca_V2.2 mRNAs in brain. It is likely that coordinated alternative pre-mRNA splicing across related Ca_V2 genes by common splicing factors allows neurons to orchestrate changes in synaptic protein function while maintaining a balanced and functioning system.

Introduction

Genes that encode the Ca_Vα₁ pore-forming subunits of voltage-gated calcium channels are large. Each Ca_Vα₁ gene includes 40–45 constitutive exons that are invariable in processed mRNA, as well as several alternative exons that are optional and cell-specific. Alternatively spliced (AS) exons serve two major functions. First, inclusion or exclusion of certain AS exons can shift the reading frame of mRNAs and result in early termination of protein translation, which may result in mRNA degradation by nonsense mediated decay.^1–3 Cells use these exons to regulate protein expression levels downstream of transcription.^4,5 Second, AS exons can modify protein sequence and thereby modify protein function. This is achieved by the inclusion, exclusion or substitution of alternatively spliced exons that encode as few as one and up to ∼100 amino acids in key domains of the channel protein.⁶ In this paper we focus on the regulation of alternatively spliced exons that give rise to protein splice isoforms of Ca_V2 channels with unique expression patterns and functions.

The decision to include, skip or substitute a given AS exon usually occurs in the cell's nucleus (although see ref. 7) and depends on interactions among the spliceosome, splicing factor proteins and the transcribed pre-mRNA. The spliceosome, a highly specialized macromolecular complex, is composed of five small nuclear RNA proteins (snRNPS) and hundreds of associated proteins and is common to all cell's nucleus (reviewed in ref. 8). Components of the spliceosome bind at intron-exon boundaries and catalyze intron removal and exon joining. Additionally, cis elements in pre-mRNAs and the cell-specific splicing factors that bind to them direct cell-specific AS. Collectively they define the cell's unique mRNA isoform profile.

Cell-specific splicing factors enhance or repress AS exons, and their expression patterns are tissue-, activity- and cell-specific.⁹ Our discussion here focuses on existing and new evidence that Nova, one of the earliest discovered tissue-specific splicing factors, regulates the pattern of AS exons in Ca_V2 channels.¹⁰ Our emphasis on Nova is driven by two important facts. First, Nova is the first—and to date the only—splicing factor validated as a regulator of AS splicing of Ca_V2 pre-mRNAs.¹¹ Second, Nova-2 knockout mice exist, providing a way to test directly the consequences of Nova-2 loss on AS exon composition in Ca_V2 mRNAs.¹²

Nova Proteins Regulate Alternative Splicing of Synaptic Proteins

Nova proteins, originally discovered as the target antigens in the autoimmune neurological disorder paraneoplastic opsoclonus myoclonus ataxia (POMA), were the first mammalian neuron-specific splicing factors identified.^10,13,16 Two genes encode Nova proteins: Nova1 and Nova2, although there is some evidence that other members of the Nova family await discovery.¹⁴ Nova-1 and Nova-2 proteins bind similar motifs, but their expression patterns differ. Nova-1 proteins are limited to subcortical CNS structures in developing and adult mice. Nova-2 proteins, by contrast, are expressed throughout the CNS and particularly in cortex, olfactory bulb, thalamus, inferior colliculus, inferior olive, and the internal and external granule cell layers of the cerebellum.¹⁰ Like other splicing factors, Nova-1 and Nova-2 bind to consensus sequences close to target AS exons on pre-mRNAs. The three KH domains mediate Nova binding to clusters of YCAY motifs on target pre-mRNAs.^10,15

Nova-1 and Nova-2 knockout mice die 2–3 weeks after birth, but tissue from early postnatal animals shows that Nova is especially important for correct alternative splicing of two inhibitory receptor mRNAs in brain: glycine alpha 2 and GABA_A.^16,17 Nova-1^-/- mice and POMA patients share a phenotypic similarity—weakness and tremulousness due to motor neuron atrophy—that led Darnell and colleagues to hypothesize a role for disrupted AS of inhibitory synaptic proteins in POMA disease pathology.¹⁶ Evidence is accumulating that misregulation of alternative splicing of synaptic proteins, including voltage-gated calcium channels, can result in disease.^18–24

Ca_V2.2 is a Nova Target

Darnell and colleagues have transformed our understanding of the extent, mechanisms, and potential functional importance of Nova-regulation of AS exons in the mammalian brain. By combining protein-RNA crosslinking and Nova-immunoprecipitation, in a technique called CLIP (cross-linking and immunoprecipitation) they were able to identify new Nova RNA targets.^25,26 Their studies generated the most comprehensive view to date of the extent and the mechanism of action of a cell-specific splicing factor. Nova binds YCAY clusters in introns generally within 500 nucleotides of target AS exons. When Nova binds clusters upstream of a target AS exon, it generally represses inclusion by blocking U1 snRNP binding; when Nova binds downstream of the target AS exon, it generally enhances inclusion (Fig. 1).¹¹ Subsequent studies have indicated that this appears to be a general feature of alternative splicing regulators, evident for example with not only Nova but Fox2, MBNL, hnRNP L, hnRNP C, TIA1 and PTB.²⁷ Microarray analyses showed that the majority of Nova-regulated exons are located in genes encoding proteins with important synaptic functions.¹² Furthermore, they combined the CLIP technique with high throughput sequencing (HITS-CLIP) and mapped more than 16,000 putative sites of Nova-2 binding onto the mouse genome.²⁸

Figure 1.

Nova acts as a splicing repressor or enhancer depending on its binding site relative to the target exon. Nova binds to YCAY motifs in introns upstream of e31a in Ca_V2 pre-mRNAs to repress its inclusion, whereas Nova binds to YCAY motifs in introns downstream of e24a in Ca_V2 pre-mRNAs to enhance its inclusion, during pre-mRNA splicing.

While the importance of these landmark studies rests in the genome-wide scale of the analyses, our interest was piqued by one target: e24a in Cacna1b, the gene that encodes the Ca_V2.2 channel. With analyses that compared e24a expression in wild-type and Nova-2^-/- mice, Darnell and colleagues confirmed that Nova-2 enhances e24a expression in mouse brain, likely by binding YCAY clusters in the 3′ intron downstream of e24a (Figs. 1 and 2 for YCAY motifs located downstream of e24a in mouse Cacna1b).¹¹ Splicing regulation of e24a by Nova is conserved in the homologous Cacna1b genes of zebrafish and chicken, strongly suggesting that this AS exon plays a critical role in optimizing neuronal function in specific regions of the nervous system.²⁸ Additional Nova-2 binding sites have been identified in Cacna1b, motivating us to explore whether Nova-2 coordinates inclusion of other AS exons in Ca_V2.2 and homologous exons in the Ca_V2.1 paralog.

Figure 2.

Nova binding motifs are located in the intron downstream of e24a of Ca_V2.2. The genomic sequence in this region of Cacna1b is shown, and the amino acids encoded by e24a indicated (SFMG). The Nova binding motifs (black squares) are shown below the genomic sequence and were mapped to the UCSC genome browser (July 2007 mouse assembly). Nova binding motifs are shown within 70 nucleotides and downstream of e24a in Cacna1b. The Nova binding site track was provided by SFmap.⁶² SFmap is a computational tool that uses an algorithm to predict splicing factor binding sites by identifying binding site motif clusters and evaluating the conservation of these clusters across species.

AS Exons in Ca_V2 Genes

Figure 3 highlights some of the known AS exons in three closely related Ca_V2 genes—Cacna1a, Cacna1b and Cacna1e—that encode Ca_V2.1, Ca_V2.2 and Ca_V2.3 proteins, respectively. These proteins (in association with accessory subunit proteins) in turn underlie P-type, N-type and R-type currents in neurons. We and others have mapped the tissue distribution and functional consequences of AS exons on channel activity and drug action in mice in vivo.^{20,21,30–49} Several AS exons are conserved among the three Ca_V2 genes, and in some cases, for example the homologous exons e18a of Cacna1b and Cacna1e, their inclusion appears to be coregulated.³¹ Here we review published data from the Darnell lab and show new evidence that the same splicing factor, Nova-2, regulates the splicing of homologous exons, e24a and e31a in Ca_V2.1 and Ca_V2.2 channels.

Figure 3.

Alternatively spliced exons in the Ca_V2 genes. The locations of constitutive exons (black) and conserved alternative exons (colored) in mouse Cacna1a, Cacna1b and Cacna1e genes. Gene diagrams are adapted from the UCSC genome browser (Feb. 2006 mouse assembly). Alternative exons 37a and 37b are found in all three Ca_V2 channel genes. An alternative exon e18a is present in Cacna1b and Cacna1e, but a homologous exon has not been identified in Cacna1a. Alternative exons e24a and e31a are conserved in Cacna1a and Cacna1b, but equivalent exons have not been identified in Cacna1e.

Nova-2 Regulates e24a and e31a in Ca_V2 Channels

E24a and e31a encode short sequences in the S3–S4 extracellular linkers of domains III and IV, respectively, of Ca_V2.1 and Ca_V2.2. In separate studies, the Snutch and Williams labs showed that inclusion of e31a in Ca_V2.1 decreases the affinity of ω-agatoxin IVA for the channel ∼10-fold, while inclusion of e31a in either channel slows channel activation and deactivation kinetics.^33,49,50 We analyzed gating currents from Ca_V2.2 channels containing e31a and showed that it affects the putative S4 voltage sensor.⁴¹ Most relevant to this discussion, we found e31a is absent (Ca_V2.2) or at lower levels (Ca_V2.1) in rat brain compared to peripheral ganglia.³⁶ E24a of Cacna1b encodes four amino acids, SFMG, in domain IIIS3–IIIS4 of mouse and rat Ca_V2.2 and SFVG in human Ca_V2.2 (S. Schorge and D. Lipscombe).⁴⁵ E24a is found in the majority of Ca_V2.2 mRNAs in brain but is found at lower levels in ganglia. Although it is not currently present in EST or cDNA databases, we also recently discovered a corresponding e24a in Cacna1a (see below).

We recently used RT-PCR to compare the expression pattern of Ca_V2 channel mRNAs including and lacking e24a and e31a in brains and ganglia of wild-type and Nova-2^-/- mice to determine whether Nova-2 regulates tissue-specific expression of these exons.

Ca_V2.1 and Ca_V2.2 e31a

Exons 31a in Ca_V2.1 and Ca_V2.2 map to the extracellular linker between transmembrane spanning segments 3 and 4 in the IV domain of both channel proteins. An AS exon is located in this region of several Ca_Vα₁ subunits and in many species, including human and Drosophila.³¹ E31a in both Ca_V2.1 and Ca_V2.2 encode two amino acids, NP in Ca_V2.1 and ET in Ca_V2.2.⁵⁰ We have shown that e31a is found in Ca_V2.1 mRNAs in dorsal root ganglia (DRG) and superior cervical ganglia, similar to the distribution of e31a of Ca_V2.2.⁵⁰ However, e31a-containing Ca_V2.1 mRNAs are also expressed in hippocampus and cerebellum albeit at levels lower than in ganglia, an expression pattern that is different from e31a of Ca_V2.2 which is at very low levels in all regions of the CNS.⁵⁰

Putative Nova-2 binding sites have been mapped very close to the 31a exons of the Cacna1b and Cacna1a genes. As discussed above, exons 31a of both genes are found at very low levels in brain, suggesting that Nova-2 might repress e31a inclusion. Consistent with this, we found an increase in the percentage of Ca_V2.1[e31a] transcripts (from 30% to 80%) and Ca_V2.2[e31a] transcripts (from not detectable to ∼45%) in RT-PCR amplified products from Nova-2 knockout mouse brain cDNA compared to wild-type brain cDNA (Fig. 4). In cerebellum (CB), the increase in e31a-containing Ca_V2 sequences in Nova-2^-/- was smaller (Ca_V2.1: from 20% to 50% and Ca_V2.2: ∼10% to 30%) compared to the rest of the brain. But in DRG, the proportion of e31a-containing sequences was not different in wild-type compared to Nova-2^-/- mice (∼80–85%); this is consistent with the lack of Nova-2 in this tissue.

Figure 4.

Alternative exons e31a of Ca_V2.1 and Ca_V2.2 are repressed in brain by Nova-2. The approximate locations of e31a of Ca_V2.1 (NP) and Ca_V2.2 (ET) are mapped on a schematic of Ca_V2 channels that shows transmembrane and intracellular domains (A and B). RT-PCR products amplified from mouse brain using primers that flank e31a are shown separated in 8% denaturing polyacrylamide gels. Predicted sizes of PCR products are 100 (+31a) and 94 (Δ31a) nts for Ca_V2.1 (A) and 106 (+31a) and 100 (Δ31a) nts for Ca_V2.2 (B). RT-PCR amplification was carried out using brain lacking cerebellum (Brain), cerebellum (CB) and dorsal root ganglia (DRG) from wild-type (WT) and Nova-2 knockout mice (KO). Percentages of cDNAs containing e31a of the total amplified cDNA for each reaction are shown below each lane. (A) 30% of cDNAs amplified from wild-type brain contain e31a compared to 80% in Nova-2^-/- brain. In cerebellum, 17% of cDNAs amplified from wild-type tissue contain e31a compared to 46% in Nova-2^-/- tissue. cDNA amplified from DRG contains a high percentage of e31a-containing sequence similarly in wild-type and Nova-2 knockout tissue (82–87%). (B) cDNAs amplified from wild-type brain lacked e31a compared to 44% representation in cDNAs from Nova-2^-/- brain. In cerebellum, 13% of cDNAs amplified from wild-type tissue contain e31a compared to 32% in Nova-2^-/- tissue. cDNA amplified from DRG contains a high percentage of e31a-containing sequence (83–84%) similarly in wild-type and Nova-2 knockout tissue. A schematic shown below the gels illustrates how Nova acts to repress e31a inclusion via a binding site in the upstream intron in brain and cerebellum, but not in DRG.

Our data show that Nova-2 is a repressor of e31a in both Ca_V2.1 and Ca_V2.2 pre-mRNAs in the mammalian brain (Fig. 1). Based on the very low to undetectable levels of e31a-containing Ca_V2.2 mRNAs in brain, Nova appears to exert stronger repression of e31a in Ca_V2.2 than in Ca_V2.1. Additional studies are needed to explore the precise mechanism of Nova action. For example, Nova repression of e31a in Ca_V2.2 might be stronger compared to Ca_V2.1, or the inclusion of e31a in Ca_V2.1 may also be influenced by other as yet unidentified splicing factors, such as a splicing enhancer that promotes e31a inclusion in specific brain regions.

Ca_V2.1 and Ca_V2.2 e24a

Several observations led us to hypothesize the presence of an AS exon in Cacna1a equivalent to e24a of Cacna1b. These include the overall high degree of conservation between Cacna1a and Cacna1b genes, their close functional roles, the conservation of AS exons 31a that are located in the homologous region of the channel in domain IV (IVS3–IVS4), and the location of a putative Nova binding site in the intron between exons 24 and 25 in Cacna1a. We located a region in the intron that is strongly conserved among species (Fig. 5A). We used RT-PCR with primers located in constitutive exons e24 and e25 to amplify Ca_V2.1 mRNA isolated from different regions of the brains of wild-type mice (Fig. 5B). We observed two differently sized bands when we analyzed PCR cDNA products by gel separation, consistent with the presence of two alternative sequences that differ by only a few nucleotides. We confirmed a 12-nucleotide insert encoding the tetrapeptide sequence SSTR in Ca_V2.1-derived products by sequencing (Fig. 5A).

Figure 5.

Alternative exons e24a of Ca_V2.1 and Ca_V2.2 are enhanced in brain by Nova-2. (A) Location and sequence, SSTR, of newly identified e24a (red) in the mouse Cacna1a gene. Constitutive exons e24 and e25 are shown in gray flanking the alternatively spliced e24a (red). Below, the degree of sequence conservation among vertebrate Cacna1a genes is represented. Image is adapted from UCSC genome browser, Feb 2006 mouse assembly. (B) RT-PCR products amplified using primers located in e24 and e25 from RNA isolated from the following regions of adult mouse brain: cortex (CTX), hippocampus (HC), hypothalamus (HP), midbrain (MB), olfactory bulb (OB) and cerebellum (CB). Predicted sizes of products are 152 nts (+24a) and 140 nts (Δ24a). PCR products were separated on 8% denaturing polyacrylamide gels. (C) RT-PCR products amplified from mouse brain using primers that flank e24a are shown separated on a 3% agarose (DRG) or 8% denaturing polyacrylamide gels. Predicted sizes of PCR products are 152 (+24a) and 140 (Δ24a) nts. RT-PCR amplification was carried out using brain lacking cerebellum (Brain), cerebellum (CB) and dorsal root ganglia (DRG) from wild-type (WT) and Nova-2 knockout mice (KO). Percentages of cDNAs containing e24a of the total amplified cDNA pool are shown below each lane. (A) 13% of cDNAs amplified from wild-type brain contain e24a but this isoform was not detectable in Nova-2^-/- brain samples. In cerebellum, 23% of cDNAs amplified from wild-type tissue contain e24a compared to 12% in Nova-2^-/- tissue. cDNA amplified from DRG contains a very low percentage of e24a-containing sequence similarly in wild-type and Nova-2 knockout tissue (4–5%). A schematic shown below the gels illustrates how Nova acts to enhance e24a inclusion via a binding site in the downstream intron in brain and cerebellum, but not in DRG.

We tested whether e24a of Canca1a was regulated by Nova-2 by comparing its expression levels in wild-type and Nova-2^-/- mice. In wild-type brain e24a-containing sequences represented about ∼10% of Ca_V2.1 sequences, but we were unable to detect e24a containing sequences in PCR amplified products derived from RNA from Nova-2^-/- brain (Fig. 5C). Levels of e24a-containing Ca_V2.1 sequences were very low (∼5%) similarly in DRG of wild-type and Nova-2^-/- mice. Our data suggest that Nova has an enhancing action on e24a in brain but not in DRG, where Nova-2 is absent. The analogous e24a in Ca_V2.2 is found at higher levels in brain compared to peripheral ganglia. Furthermore, Darnell and colleagues showed that e24a in Ca_V2.2 is enhanced by Nova.¹¹

We would like to know the function of e24a in Ca_V2.1 and Ca_V2.2 channels. In Ca_V2.2, e24a has a relatively minor influence on channel properties. As described above, we only recently identified e24a in Ca_V2.1 and have not assessed its effects on channel properties.⁵⁰ However, we speculate that the extracellular location of e24a might point to a role in mediating interactions with extracellular proteins. For example, neurexins have been shown to regulate Ca_V2.1 and Ca_V2.2 channel functioning through direct or indirect extracellular coupling.^51–53 It is possible that other extracellular proteins might interact with Ca_V2.1 and Ca_V2.2 channels in a splice isoform-dependent manner, a possibility that needs to be tested.

Nova Influences a Subset of Alternatively Spliced Exons in Ca_V2

Other sites of alternative splicing have been studied in detail by our lab (e18a and e37a/e37b in Ca_V2.2) and by other labs (e37a/e37b in Ca_V2.1).^{30,33–35,38,48,54–58} None of these AS exons appears to be regulated by Nova-2. For example, we find no difference in expression patterns of exon 37a of Ca_V2.2 in Nova-2^-/- and wild-type mice (data not shown). Exon 37a of Ca_V2.2 is enriched in nociceptors of DRG but is expressed at very low levels in other parts of the nervous system. Likewise, the expression pattern of e18a in Ca_V2.2 is similar in wild-type and Nova-2^-/- mice (data not shown). Although we focus on Nova-2 in this report, splicing factors other than Nova must regulate cell-specific inclusion of e18a and e37a/e37b. Furthermore, exon selection in vivo is likely to depend on the concerted actions of splicing enhancers and repressors (Fig. 6).^4,58,60 The Darnell group recently concluded that ∼15% of Nova-regulated exons might also be regulated by Fox proteins, another family of neuronal splicing factors.⁶²

Figure 6.

Cellular control of alternative splicing likely involves the net contribution of more than one splicing factor. Schematic shows hypothetical, combinatorial control of an alternatively spliced exon by Nova binding downstream and acting as a repressor, by another serine rich (SR) splicing factor binding downstream and acting as an enhancer, and by a third splicing factor of the hnRNP family binding to the target exon and acting as a repressor of exon inclusion. The relative importance and contribution of each splicing factor depends on several factors, including their expression levels. Thus, tissue specific splicing can arise from the tissue-specific expression profiles of splicing factors.

Summary

The nervous system operates with small error margins and adaptive responses are essential to normal neuronal function. These impact neuronal networks, individual neurons, and individual synapses and involve changes in synapse number, morphology and efficacy. Coordinated and temporal changes in protein activity and function underlie all these adaptive responses. Alternative pre-mRNA splicing has an impressive ability to generate a vast array of functionally distinct proteins. Alternative pre-mRNA splicing gives cells the capacity to modify and selectively re-balance their existing pool of transcripts in a coordinated way across multiple mRNAs, thereby effecting relatively rapid and relatively stable changes in protein activity.

Alternative pre-mRNA splicing events are predicted to occur in ∼95% of multi-exon human genes.^63,64 Genome-wide screens of splicing factor-RNA binding events such as HITS-CLIP Nova screens and those looking at binding sites for other splicing factors such as Fox and PTB are providing us the information we need to define cell-specific mechanisms that orchestrate alternative splicing across multiple genes, including Ca_V2.^27,65,66 In this short report we review limited published data and in addition show evidence that four alternative exons in two closely related calcium channel Ca_V2 subunit genes are regulated by Nova-2. Thus, the ability of Nova to function both as repressor (e31a) and enhancer (e24a), depending on binding location imprinted in each gene, provides an elegant yet simple mechanism to coordinate splicing of homologous exons in related Ca_V2 genes (Fig. 1). This duality of function, based on RNA binding location relative to the target exon, is a property shared by several splicing factors.^27,59 Cells control and coordinate patterns of alternative splicing by regulating the expression levels and activity of individual splicing factors. Coordinated alternative pre-mRNA splicing across related Ca_V2 genes allows neurons to orchestrate changes in synaptic protein function while maintaining a balanced and functioning system.