|Home | About | Journals | Submit | Contact Us | Français|
Calcium ion channels coordinate an astounding number of cellular functions. Surprisingly, only 10 Cavα1 subunit genes encode the structural cores of all voltage-gated calcium channels. What mechanisms exist to modify the structure of calcium channels and optimize their coupling to the rich spectrum of cellular functions? Growing evidence points to the contribution of post-translational alternative processing of calcium channel RNA as the main mechanism for expanding the functional potential of this important gene family. Alternative splicing of RNA is essential during neuronal development where fine adjustments in protein signaling promote and inhibit cell-cell interactions and underlie axonal guidance. However, attributing a specific functional role to an individual splice isoform or splice site has been difficult. In this regard, studies of ion channels are advantageous because their function can be monitored with precision, allowing even subtle changes in channel activity to be detected. Such studies are especially insightful when coupled with information about isoform expression patterns and cellular localization.
In this paper we focus on two sites of alternative splicing in the N-type calcium channel Cav2.2 gene. We first describe cassette exon 18a that encodes a 21 amino acid segment in the II-III intracellular loop region of Cav2.2. Here we show that e18a is up-regulated in the nervous system during development. We discuss these new data in light of our previous reports showing that e18a protects the N-type channel from cumulative inactivation. Second, we discuss our published data on exons e37a and e37b, which encode 32 amino acids in the intracellular C-terminus of Cav2.2. These exons are expressed in a mutually exclusive manner. Exon e37a-containing Cav2.2 mRNAs and their resultant channels express at higher density in dorsal root ganglia and, as we showed recently, e37a increases N-type channel sensitivity to G protein-mediated inhibition, as compared to generic e37b-containing N-type channels.
Human, rat, and mouse genomes contain ten genes that encode Cavα1 subunits. The nervous system expresses nine of these genes. Cavα1 genes are large, spanning 100 - 800 kb of human genome sequence and containing upwards of 50 exons [1-4]. The large size of these genes provides ample opportunity for alternative splicing. Based on already known splice sites, the theoretical number of splice isoforms possible from a single Cavα1 subunit exceeds 1,000 [3, 4]. Other genes, far smaller than Cavα1 genes, have the potential to generate a similar or greater number of splice isoforms , and in some genes the number of possible isoforms is in the tens of thousands .
Alternative splicing occurs at sites important for controlling calcium channel activity, creating an array of functionally distinct calcium channels. Thus each neuron, in theory, can optimize calcium channel activity for the specific task at hand by orchestrating appropriate exon selection. Placing a specific channel isoform within a specific cell type can in turn provide information about the requirements of particular calcium signaling cascades. Furthermore, studies of alternative splicing can and have revealed the location of key domains on calcium channels that regulate their activity.
Investigators have shown that the structural changes generated by alternative splicing can affect channel biophysics, density, targeting, post-translational modification, and coupling to downstream signaling pathways. Functional sites of alternative splicing involve regulation that depends on tissue type, cell type, stage of development, and neuronal activity [7-22]. Importantly, the two sites of alternative splicing we discuss here, in the II-III intracellular loop region (e18a) and in the C-terminus (e37a and e37b), are conserved across a number of mammalian Cav2.2 genes including human, rat, and mouse (see Fig.1 for e18a sequence comparisons and see reference  for e37a and e37b comparisons). We reason that such sites of alternative splicing, that are under cellular control and that are evolutionarily conserved, must be located in key domains essential for regulating N-type channel function and likely evolved to contribute functional advantage to the organisms in which they occur [23, 24].
Our interest in the N-type calcium channel stems from its established role in the control of transmitter release from many different types of neurons [25, 26]. The N-type calcium channel is also implicated in synaptogenesis and regulation of gene expression [27, 28]. N-type calcium channels differ functionally across cell types and sub-cellular compartments within a given cell, suggestive of molecular and structural heterogeneity. Differences in N-type channel inactivation, single-channel conductance, gating, sensitivity to G-protein-mediated modulation, and ability to couple to transmitter release have been documented [29-39]. Several factors support functional diversity among voltage-gated calcium channels, including association with different auxiliary subunits, post-translational modifications, and association with target proteins and signaling molecules [40-47]. These processes are also controlled by alternative splicing [2, 3, 47-50].
The intracellular loops connecting homologous domains II and III (LII-III) of pore-forming Cavα1 subunits are of special interest. This region supports key cellular functions and links calcium channels to downstream effector proteins. In skeletal muscle, the LII-III of Cav1.1 binds directly to ryanodine receptors on the sarcoplasmic reticulum, an interaction necessary and sufficient for excitation-contraction coupling [51-53]. In the case of Cav2 voltage-gated calcium channels, II-III loops connect channels to sites of synaptic vesicle fusion via binding to SNARE (soluble N-ethyl-maleimide-sensitive fusion (NSF) attachment protein receptor) proteins , critical for coupling N-type and P/Q-type calcium channels to transmitter release and for targeting channels to presynaptic nerve terminals [48, 54-61]. The LII-III region of Cav2.2 channels also regulates cumulative inactivation of channels in response to stimulus trains [54, 62, 63].
Isoforms of Cav2.1, Cav2.2, and Cav2.3 exist that differ within the LII-III [16, 64] and differ in their abilities to bind presynaptic SNARE proteins . Our group has focused on cassette exons, e18a, located between exons 18 and 19 in LII-III of mammalian Cav2.2 and Cav2.3 genes that, as described below, are evolutionary conserved (Fig. 1). We showed previously that e18a in Cav2.2 protects the channel from inactivation. Specifically, e18a-containing Cav2.2 channels are less sensitive to closed-state inactivation, and therefore calcium influx during action potential trains is expected to be sustained relative to that through Cav2.2 channels lacking e18a .
Our interest in e18a was motivated from our analyses of various Cav2 gene sequences. Mammalian Cav2.2, and Cav2.3, but not Cav2.1, genomic sequences contain a putative cassette exon between constitutive exons, e18 and e19. In rat, e18a resides within a large 6125 bp intron between e18 and e19 of the Cav2.2 gene . We found a nearly identical cassette exon (61 of 63 bp), flanked by consensus gt-ag splice junctions, within the 9775 bp intron between e18 and e19 of the human Cav2.2 gene (see Fig. 1) . We also identified the location of a 57 nucleotide exon within a 6221 bp intron of the human Cav2.3 gene consistent with previous studies of Cav2.3 mRNAs [65, 66]. Despite similarity in size and position, exons 18a of Cav2.2 and Cav2.3 contain distinct nucleotide sequences with only 25% homology and encode dissimilar amino acid sequences (Fig. 1). A similar analysis of the Cav2.1 human gene revealed a much smaller intron (~ 1 kb) between e18 and e19, and no evidence for a putative cassette exon.
We also examined the genomic sequences of Cav2.2 and Cav2.3 for Pan troglodytes, Bos taurus, Mus musculus, Canis familiaris, and Gallus gallus. All these species contain the alternatively expressed e18a of Cav2.2 and Cav2.3 (Fig. 1B) and exon sequences were highly conserved. By our analysis intracellular regions of the channels, including the C-terminus and loops connecting homologous domains, were 71.8 ± 3.0% similar among species. In comparison, Cav2.2 e18a nucleotide sequences shared 96.2 ± 0.5% similarity among the seven species investigated here, significantly greater than the average sequence conservation within constitutively expressed exons that encode intracellular domains of the channel (p < 0.0001).
We next used RT-PCR to establish that e18a of Cav2.2 and Cav2.3 are expressed in human neuronal tissue, and we also confirmed the lack of alternative splicing in the same region of Cav2.1 (Fig. 2A). We found Cav2.2 e18a expression in human whole brain and spinal cord (Fig. 2A), contrary to reports that this exon is not expressed in human cells . Notably, the proportion of Cav2.2 e18a-containing mRNAs was higher in adult spinal cord as compared to whole brain. Furthermore, Cav2.2 e18a was not detected in fetal whole brain. Thus, Cav2.2 e18a expression in the human is both regionally and developmentally regulated. Cav2.3 e18a is also alternatively spliced in human tissues (Fig. 2A) . Interestingly, as Cav2.2 e18a increased during development, there appeared to be a concomitant decrease in Cav2.3 e18a expression. Inclusion of e18a in Cav2.2 results in N-type calcium channels that are less sensitive to closed-state inactivation . In contrast, Cav2.3 channels that contain e18a exhibit increased sensitivity to calcium-dependent current enhancement [67, 68] and stimulation by phorbol esters .
We next examined the developmental expression of Cav2.2 exon 18a in the rat superior cervical ganglia, a tissue that expresses relatively high levels of Cav2.2 and that contains a relatively homogenous class of neurons (Fig. 2B) . We quantified e18a splice isoforms by ribonuclease protection from P1, P5, P18, and adult superior cervical ganglia. We used an antisense probe that was complementary to constitutively expressed exons 15-18 and the alternatively spliced e18a. Cav2.2 mRNAs that contain e18a fully protected the probe (upper band, Fig. 2B), while Cav2.2 mRNAs that lack the alternatively spliced exon yielded a partially protected probe (lower band, Fig. 2B). In the adult ganglia, e18a was present in 67 ± 4% of all Cav2.2 mRNAs (Fig. 2B). By comparison, the relative abundance of e18a was 54 ± 5% at P18, 25 ± 4% at P5, and 16 ± 4% at P1 (n = 3 separate experiments for all ages). The adult expression level of e18a is significantly different than that in P1 and P5 SCG (P < 0.0003), and the proportion of e18a mRNA transcripts increased >4-fold between P1 and adulthood. The overall expression level of Cav2.2 mRNA decreased between P1 and P18, in comparison to GAPDH expression levels (Fig. 2B, bottom gels). This is consistent with an increase in the number of non-neuronal cells in the maturing SCG.
During the first three weeks of postnatal life, while e18a expression continuously increased in our assay (Fig. 2), SCG neurons undergo a variety of developmental changes. By birth, the majority of SCG neurons have differentiated after migrating into the ganglia, and they begin to form synaptic connections with target structures soon after they arrive. The vast majority of these connections have formed by P21 . Rat SCG neurons rely exclusively on the Cav2.2 N-type calcium channel to control transmitter release . Our data, therefore, indicate that Cav2.2 N-type channels that contain the 21 amino acids encoded by e18a likely control transmitter release from mature SCG terminals and are specifically upregulated during the period of exuberant ganglion cell synaptogenesis for this purpose.
We note that alternatively spliced exons e18a within Cav2.2 and Cav2.3 show a high level of amino acid sequence conservation across mammalian and avian species (Fig. 1) when compared to constitutively expressed exons in this region of the genes. Others have also reported an exon homologous to mammalian e18a in the pufferfish Cav2.2 ortholog . These data signify that additional evolutionary pressure exists to preserve e18a sequences and suggest an important functional role for Cav2.2 and Cav2.3 isoforms that contain the additional 21 and 19 amino acids encoded by their respective exons. Interestingly, our data point to the possibility that a common mechanism regulates expression of e18a exons in Cav2.2 and Cav2.3 based on their approximate reciprocal expression patterns. Specifically, we find that e18a containing Cav2.2 mRNAs are more abundant in the mature brain (Fig. 1), peripheral ganglia , and spinal cord (Fig. 1), whereas, Cav2.3 mRNAs containing e18a are more abundant in fetal brain (Fig. 1). Further, Cav2.3 e18a mRNAs are virtually absent from the peripheral nervous system (unpublished observations, ACG and DL). One tantalizing possibility is that e18a controls targeting of Cav2.2 and Cav2.3 channels to sites of transmitter release given that this exon modifies amino acid composition in the II-III linker region. Although e18a is absent from Cav2.1, others have reported distinct LII-III isoforms of this channel that might arise from alternative splicing of a different exon. These isoforms of Cav2.1 differ in their interactions with SNARE proteins and sub-cellular expression patterns [48, 64]. Regulated expression of LII-III isoforms could represent a mechanism to limit overall Cav2 channel density at the synapse during maturation.
The C-termini of Cav2 genes coordinate various aspects of calcium channel function, including inactivation, modulation by G-proteins, modulation by calmodulin, and protein-protein interactions that regulate activity and/or target the channel to specific cellular compartments [9, 37, 47, 49, 73-76]. Alternative splicing in exon 46 of Cav2.2 affects the ability of G-proteins to modulate the N-type calcium channel  and alters sub-cellular targeting [47, 73]. Mutually exclusive e37a and e37b are found in Cav2.1, Cav2.2, and Cav2.3 genes (Figure 3A). The amino acid compositions of these exons differ among the Cav2 genes, and the expression pattern of Cav2.2 e37a and e37b  is distinct from the reported expression pattern of the equivalent exons in Cav2.1 , suggesting that they serve different functions. Notably, we have shown that Cav2.2e[37a] is preferentially expressed in capsaicin-responsive neurons of the dorsal root ganglia, implying a role in nociception .
Cav2.2 channels localize to synapses of nociceptive neurons in the dorsal horn laminae I/II  and dominate in the control of transmitter release from these presynaptic terminals [78-80]. Mice lacking Cav2.2 have impaired nociception [81-84], and N-type channel blockers administered intrathecally, are powerful analgesics in a variety of animal pain models [85-89]. Some of the analgesic properties of morphine and other opiates are mediated via G-protein-dependent inhibition of N-type channels in the dorsal horn. A sub-set of nociceptive neurons preferentially express Cav2.2e[37a] mRNAs [7, 90]. This novel Cav2.2 splice isoform supports larger N-type currents in nociceptive neurons, as compared to the more prevalent splice isoforms of Cav2.2 that contain the alternative e37b [7, 91]. Larger currents associated with e37a-containing channels impact the amount of calcium entering cells in response to a variety of action potential waveforms . More recently, we showed that e37a increases the sensitivity of the N-type channel to the voltage-independent form of G-protein modulation . Collectively, our findings provide potentially important insights into the cellular mechanisms that control N-type channel activity in nociception and indicate that a sub-set of nociceptive neurons express a form of the N-type channel that is highly sensitive to inhibition by neurotransmitters and hormones.
Having shown preferential expression of e37a in nociceptors of dorsal root ganglia, we considered what features of N-type channels in these cells are especially prominent. Numerous G-protein-coupled neurotransmitter receptors suppress transmitter release from many neurons by inhibiting N-type calcium channels [30, 44, 45, 93]. G-protein-dependent inhibition can be voltage-dependent and mediated by direct interaction of G-protein βγ subunit dimers with Cav2.2 subunits [43, 44, 94]. This inhibition is membrane delimited, slows activation of the N-type current, and can be relieved by a strong depolarizing pre-pulse. Voltage-dependent inhibition is most effective against calcium entry triggered by single action potentials. It is relieved by a train of action potentials [94-96]. Another form of G-protein-dependent inhibition of the N-type current is voltage-independent and this is an especially prominent feature of N-type channels in nociceptors. This form of inhibition is not relieved by strong depolarizing pre-pulses and may involve kinase-dependent phosphorylation by protein kinase C  and tyrosine kinases . The contributions of voltage-dependent and voltage-independent inhibition vary with transmitter receptor, cell type, and level of syntaxin 1A [43, 44, 100-103]. However, the molecular mechanisms that set the level and type of transmitter-mediated inhibition of the N-type channel are unknown. Our recent findings show that the degree and type of G protein-mediated inhibition of the N-type channel expressed in nociceptors is regulated by the alternative splice site, e37a and e37b, in the C-terminus of the Cav2.2 subunit.
We were motivated to test the role of e37a in G protein-mediated inhibition of the N-type channel because we had shown enrichment of e37a in nociceptors, where N-type channels are highly sensitive to inhibition by G protein-coupled receptors. We found that inclusion of e37a in Cav2.2 created an inhibitory pathway that was independent of voltage and that substantially increased the sensitivity of Cav2.2 channels to modulation by opiates and GABA . Voltage-dependent inhibition was present in e37a and e37b-containing N-type channels and was unaffected by alternative splicing at this site in the C-terminus. We also showed that both forms of inhibition, voltage-dependent and voltage-independent, utilize PTX-sensitive Gi and Go class G proteins, but the two pathways diverge downstream of the trimeric G protein. As described by many investigators, voltage-dependent inhibition requires the Gβγ dimer binding to the Cav2.2 channel and is independent of kinase activation [44, 45] (Fig. 3B). By contrast, voltage-independent inhibition proceeds independent of Gβγ and requires pp60c-src tyrosine kinase activation [92, 99] (Fig. 3B). Interestingly, there is a tyrosine residue encoded in e37a and absent in e37b, that is essential for supporting voltage-independent inhibition of Cav2.2e[37a] channels (Fig. 3A). GABA, opiates and probably other transmitters and drugs can utilize this pathway to inhibit Cav2.2 channels by a mechanism that is independent of stimulus intensity [95, 96, 104] (Fig. 3B).
Our studies of alternative splicing illustrate how investigations of evolutionarily conserved, natural variants of Cav2.2 can uncover critical domains in N-type channels that regulate their activity. The e18a splice site illustrates that alternative splicing of certain exons in Cav2.2 is under developmental control and possibly coordinated with splicing of the homologous site in Cav2.3. Collectively, our studies suggest that N-type channels become less sensitive to inactivation following trains of stimuli during development. Alternative splicing of e37a on the other hand is under cell-specific control. Knowledge that e37a-containing Cav2.2 channels are preferentially expressed in nociceptors provided the functionally relevant context to understand the significance of our findings.
We still know very little of the factors that regulate alternative splicing of voltage-gated calcium channels. Of the proteins that have been identified that regulate neuron-specific alternative splicing, only the Nova2 RNA binding protein has been linked to Cav2.2 . As factors that regulate expression of alternatively spliced exons in Cav2.2 channels are identified, we should begin to gain a better understanding of the mechanisms that exist to regulate calcium channel function and to control synaptic efficacy.
This work was supported by Howard Hughes Medical Institute Predoctoral Fellowship (A.C.G.) and National Institutes of Health Grants: NS29967 and NS055251 (D.L.)
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.