C-Terminal Alternative Splicing Changes the Gating Properties of a Human Spinal Cord Calcium Channel α1A Subunit

Abstract

The calcium channel α_1A subunit gene codes for proteins with diverse structure and function. This diversity may be important for fine tuning neurotransmitter release at central and peripheral synapses. The α_1A C terminus, which serves a critical role in processing information from intracellular signaling molecules, is capable of undergoing extensive alternative splicing. The purpose of this study was to determine the extent to which C-terminal alternative splicing affects some of the fundamental biophysical properties of α_1A subunits. Specifically, the biophysical properties of two alternatively spliced α_1A subunits were compared. One variant was identical to an isoform identified previously in human brain, and the other was a novel isoform isolated from human spinal cord. The variants differed by two amino acids (NP) in the extracellular linker between transmembrane segments IVS3 and IVS4 and in two C-terminal regions encoded by exons 37 and 44. Expression inXenopus oocytes demonstrated that the two variants were similar with respect to current–voltage relationships and the voltage dependence of steady-state activation and inactivation. However, the rates of activation, inactivation, deactivation, and recovery from inactivation were all significantly slower for the spinal cord variant. A chimeric strategy demonstrated that the inclusion of the sequence encoded by exon 44 specifically affects the rate of inactivation. These findings demonstrate that C-terminal structural changes alone can influence the way in which α_1A subunits respond to a depolarizing stimulus and add to the developing picture of the C terminus as a critical domain in the regulation of Ca²⁺ channel function.

Multiple types of high-voltage-activated Ca²⁺ channels (L, N, P, Q, and R) coordinate a variety of Ca²⁺-dependent processes, including gene expression, signal propagation, and neurotransmitter release (Tsien et al., 1991; Zhang et al., 1993; Dunlap et al., 1995). These channels can be distinguished by their biophysical and pharmacological properties. L- and R-type channels, found on cell bodies and proximal dendrites, regulate gene transcription and signal propagation (Westenbroek et al., 1990; Murphy et al., 1991; Yokoyama et al., 1995), whereas synaptic N-, P-, and Q-type channels regulate neurotransmitter release (Turner et al., 1992; Takahashi and Momiyama, 1993; Wheeler et al., 1994). The Ca²⁺ channel complex consists of four subunits, α₁, α₂/δ, β, and γ. With the exception of P- and Q-type channels, the different neuronal Ca²⁺ channel phenotypes arise primarily from the expression of five unique α₁ subunit genes (Tsien et al., 1991; Jun et al., 1999). These genes (A–E) are differentially distributed throughout brain and spinal cord (Murphy et al., 1991; Takahashi and Momiyama, 1993; Westenbroek et al., 1998) and encode large proteins consisting of four homologous domains (I–IV) containing six transmembrane segments each (S1–S6) (Tsien et al., 1991; Zhang et al., 1993).

Alternative splicing of the α_1A gene results in the expression of multiple Ca²⁺ channel phenotypes (Sutton et al., 1998; Bourinet et al., 1999; Hans et al., 1999; Jun et al., 1999). Splicing of two amino acids (NP) in the α_1A IVS3–IVS4 linker affects rates of activation and inactivation, the voltage dependence of inactivation, and the affinity of α_1A for ω-Aga IVA (Sutton et al., 1998; Bourinet et al., 1999; Hans et al., 1999; Lin et al., 1999). The biophysical and pharmacological properties of α_1A Ca²⁺channels are also influenced by intracellular signaling molecules. Coexpression of α_1A with different β subunit subtypes alters inactivation rates and ω-Aga IVA affinity (Moreno et al., 1997). Generally, α_1A subunit function is inhibited by interactions with G-proteins or syntaxin-1A (Zhang et al., 1996; Qin et al., 1997; Sutton et al., 1999) and enhanced by protein kinase C (Zamponi et al., 1997; Bourinet et al., 1999). Calcium-activated calmodulin has dual effects on α_1A subunit function (Lee et al., 1999).

Determination of genomic exon–intron boundaries (Ophoff et al., 1996) and isolation of several α_1A C-terminal splice variants (Zhuchenko et al., 1997) suggest that there are four exons between IVS3–IVS4 and the C-terminal stop codon that undergo alternative splicing. This implies that 16 combinations of these four exons are possible. Eight combinations have been isolated previously from human brain (Ophoff et al., 1996; Zhuchenko et al., 1997; Hans et al., 1999). In this study, a ninth is isolated from human spinal cord. Biophysical characterization of this α_1A variant reveals that its rates of activation, inactivation, deactivation, and recovery from inactivation are all significantly slower than that of an α_1A variant from cerebellum. The rate of inactivation was especially affected by sequence encoded by a single exon. These findings suggest that variation in α_1A C-terminal structure provides an added mechanism for functional diversity among neuronal Ca²⁺ channel subtypes.

MATERIALS AND METHODS

Human spinal cord library screening. Calcium channel subunit cDNAs were isolated from an oligo-dT and random-primed human spinal cord λgt11 5′ stretch cDNA library (Clontech, Palo Alto, CA) by the use of a nonradioactive digoxigenin-labeling and colorimetric detection system (Roche Molecular Biochemicals, Indianapolis, IN). The library was constructed from mRNA isolated from whole spinal cords pooled from 26 male or female Caucasians, ages 16–75, who died of sudden death syndrome. The insert size range of the library was 0.8–7.0 kb (average size of 1.7 kb). Plaque-purified phage DNAs were isolated using a Lambda Prep Kit (Qiagen, Santa Clara, CA) and digested with the restriction endonuclease EcoRI (Roche Molecular Biochemicals, supplier of all endonucleases used). Southern blot analysis was used to assess insert size. cDNA isolates were ligated into pBluescriptII (Stratagene, La Jolla, CA) for sequencing. The inserts were subjected to exonuclease III/SI nuclease digestion (Erase-A-Base; Promega, Madison, WI) before sequencing. PCR-based cycle sequencing (FS chemistry; PE Biosystems, Foster City, CA) with universal primers and custom internal primers (Genosys, The Woodlands, TX) was used for each clone. Sequence was obtained using an ABI Prism 310 Genetic DNA analyzer, and data were analyzed using ABI Prism DNA Sequencing Software (version 2.12; PE Biosystems). Sequence comparisons, alignments, and restriction maps were performed using Lasergene Software (DNAStar, Madison, WI). 5′ RACE (Roche Molecular Biochemicals) was used with human spinal cord poly(A⁺) RNA (Clontech) to create a 5′ probe for the α_1A screening. This RNA came from a pool of 92 male or female Caucasians, ages 16–75, who died from sudden death syndrome.

The library-screening process was initiated with a cDNA probe (α₁-EST) obtained from the NCBI dbEST database (1.8 kb human fetal brain α_1A fragment; GenBank number H14053). Clone α₁-9, extending from nucleotide 3416 to the 3′-untranslated region, was the longest of 24 3′ cDNAs isolated in the first rounds of screening. Five α₁ cDNAs were isolated in a second round, in which a 1 kb 5′ EcoRI fragment of clone α₁-9 was used as the probe. The longest clone isolated in this round, extending from nucleotides 2399 to 4551, was subsequently labeled and used as a probe to identify α₁-38 (which contained anEcoRI/EcoRI fragment from 1800 to 4551). To isolate the 5′ portion of the α_1A sequence, a probe containing nucleotides 946 to 1800 was created by reverse transcription-PCR with human spinal cord RNA. Library screening with this probe yielded 14 cDNAs, one of which, clone α₁-80, extended from the 5′-untranslated region to a region beyond the EcoRI site at nucleotide 1800.

Clone construction and sequencing. Full-length spinal cord α_1A cDNAs were assembled in several steps. Initially, PCR was performed using PFU Polymerase (Stratagene) and custom primers (Genosys) to truncate the 5′-untranslated region of α₁-80 and to insert an idealized Kozak (Kozak, 1991) sequence into the shortened 5′ end of the construct. The resulting 1.1 kb fragment was then ligated into the pT-Adv vector (Clontech) and sequenced to ensure that there were no polymerase errors. This PCR product was cut with the restriction endonucleaseNotI, and the resulting fragment was ligated intoEcoRV/NotI-prepared pBluescriptII. The untruncated α₁-80 was cut with NotI, and the resulting 900 bp fragment was then ligated into the shortened 5′ clone by the use of NotI. The resulting clone (referred to as 5′short) spans from the shortened 5′-untranslated sequence to theEcoRI site at nucleotide1800.

In the next step, a 3′ cDNA extending from nucleotide 3826 to the 3′-untranslated region of the clone, α₁-10, was cut with NotI to eliminate the EcoRI site located in the polylinker of pBluescriptII. The construct was then cut withSacI, blunted using the Klenow fragment (Roche Molecular Biochemicals), and cut with EcoRI to yield a 3.5 kb fragment spanning from the EcoRI site at 4551 to the 3′-untranslated region. This fragment was ligated intoEcoRI/SmaI-prepared pBluescriptII, removed withEcoRI/BamHI, and ligated into theEcoRI/BamHI-cut vector containing 5′short. Thus, the resulting construct (5′short + 3′) extended from the shortened 5′-untranslated region to the EcoRI site at 1800 and continued from the EcoRI site at 4551 to the 3′-untranslated region. In the final step, an EcoRI fragment of α₁-38 extending from 1801 to 4550 was cloned into EcoRI-cut 5′short + 3′. The fully constructed clone α_1A-C1 was then sequenced to ensure the fidelity of the construction process. Exchanging the C-terminal domain of α_1A-C1 with that of α₁-EST and α₁-9 created α_1A-C2 and α_1A-C16, respectively. Thus, both α_1A-C2 and α_1A-C16 are constructed of the same cDNAs up to nucleotide 4550 but then contain segments of different cDNAs beyond this point.

The rabbit Chβ_1a and α_2a/δ clones used in this study were provided by T. Tanabe (Tokyo Medical and Dental University, Tokyo, Japan). The Chβ_1b, Chβ₃, and Chβ₄ subunits were cloned from the same human spinal cord library described above and are nearly identical to previously reported sequences (Chβ_1b, GenBank number M923303; Chβ₃, GenBank number U07139; and Chβ₄, GenBank number U95020).

Electrophysiology and data analysis. Standard Xenopus laevis oocyte expression methods were used to characterize the α_1A splice variants. Briefly, full-length α_1A subunit cDNA was in vitrotranscribed (Ambion, Austin, TX). The resulting cRNA was injected into defolliculated X. laevis oocytes (stage V–VI) along with equimolar ratios of rabbit α_2a/δ and Chβ cRNAs (0.40 μg/ml α_1A; 0.16 μg/ml α_2a/δ; 0.08 μg/ml Chβ in a total of 46 nl). Ca²⁺ channel currents were recorded by standard two-microelectrode voltage-clamp techniques using a Warner amplifier (OC-725B) at room temperature (20–22°C), and data were collected using pCLAMP6 software (Axon Instruments, Foster City, CA). The bath solution contained the following: 40 mmBa(OH)₂, 40 mm TEA-OH, 2 mm KOH, and 5 mm HEPES, with pH adjusted to 7.4 with methanesulphonic acid. Microelectrodes were filled with 3 m KCl, and the resistances of the current and voltage electrodes were 0.3–1.5 MΩ. Data were filtered at 2 kHz and sampled at 10 kHz. Currents were recorded 6–18 d after injection. The holding potential of all experiments was −100 mV unless otherwise noted. Control oocytes (α_1Aalone or uninjected oocytes) did not yield currents >50 nA. Because of lower expression levels when Chβ subunits other than the rabbit Chβ_1a were used, currents as low as 150 nA were included in the initial portion of the study. Only currents >0.5 μA were analyzed for the remainder of this work. To diminish the contamination caused by the Ca²⁺-activated Cl⁻ channels, currents that exhibited slow deactivation (τ > 10 msec) were excluded from analysis. The leak current and capacitive current transients were subtracted on-line by a standard P/4 protocol. Data were analyzed using pCLAMP6 and Excel 7.0 (Microsoft, Redmond, WA). For statistical analysis, a one-way ANOVA test was used followed by a Fisher's PLSD test using Statview software (SAS Institute, Cary, NC).

RESULTS

Two α_1A C-terminal splice variants, α_1A-C2 and α_1A-C16, are the focus of this study (Fig.1A). These variants were assembled from a parent spinal cord construct, α_1A-C1, and thus are identical up to homology domain IV. Sequencing of the α_1A-C1 cDNA through domain IV revealed only three differences when compared with an α_1A cDNA isolated from human cerebellum (GenBank number AF004883). A deletion (ΔG419) identified previously in rat brain was found in the region encoding the intracellular linker between homology domains I and II (Bourinet et al., 1999). A mutation resulting in a charge change (M537R) was found in the region encoding transmembrane segment IIS2. A nine nucleotide deletion resulting in the loss of amino acids 726–728 (VEA) was found in the region encoding the intracellular linker between homology domains II and III. This deletion has been identified in another α_1A variant from human brain (GenBank number U79666). The spinal cord α_1A-C1 cDNA did not contain a six nucleotide insertion that codes for the amino acid sequence NP in the extracellular linker between IVS3 and IVS4.

Fig. 1.

Alternative splicing of α_1A domain IV and C-terminal exons. A, Alternatively spliced exon patterns of α_1A-C2 (middle bar) and α_1A-C16 (bottom bar) cDNAs relative to a consensus protein that is a composite of known exons (top bar). This figure is drawn on the basis of our data and that of others (Ophoff et al., 1996; Zhuchenko et al., 1997; Hans et al., 1999). Note that the NP sequence is located in the extracellular loop between transmembrane domains IVS3 and IVS4. B, The amino acid sequences of exons 37a, 37b, and 44. Regions of variation are highlighted in black. Charged amino acids are denoted by an asterisk.

The α₁-EST and α₁-9 C-terminal cDNAs used to construct α_1A-C2 and α_1A-C16, respectively, differed from the C terminus of α_1A-C1 and other α_1Aclones in sequences that could be traced to specific exon–intron boundaries (Ophoff et al., 1996). The α₁-EST cDNA did not extend to domain IV, and therefore α_1A-C2, like α_1A-C1, lacks NP (Fig. 1A). The α₁-9 cDNA contained the NP exon, and this region is therefore present in α_1A-C16. The α₁-EST cDNA contained one version of exon 37 (37a), whereas α₁-9 contained a second (37b). There are nine amino acid changes in this region including one charge change (Fig.1B, L1851K). The α₁-EST cDNA contained 12 amino acids encoded by exon 44, whereas α₁-9 did not. Interestingly, five of these amino acids are positively charged (Fig. 1B). A 3′ GGCAG insert in exon 46 that would extend the open reading frame to exon 47, although present in α_1A-C1 (data not shown), was not found in either α₁-EST or α₁-9 (Fig. 1A). Thus, neither α_1A-C2 nor α_1A-C16contains the ∼240 amino acid domain encoded by exon 47. The end result is that α_1A-C2 is identical to the human cerebellar clone described above, and α_1A-C16represents a novel spinal cord clone containing a unique arrangement of NP and C-terminal exons.

To determine the functional consequences of C-terminal splicing, both α_1A-C2 and α_1A-C16 were coexpressed with rabbit α_2a/δ and a rabbit Chβ_1a or a human Chβ_1b, Chβ₃, or Chβ₄ subunit in Xenopus oocytes. Representative current traces demonstrate that α_1A-C2 inactivates more rapidly than does α_1A-C16 when associated with Chβ_1a, Chβ_1b, or Chβ₃ subunits (Fig.2A–C; see below). When coexpressed with the Chβ₄ subunit, the α_1A-C2 complex appears to inactivate at approximately the same rate as α_1A-C16 (Fig.2D) although the low expression levels of this complex make interpretation difficult. As seen in Figure2A, the deactivation rate of the α_1A-C2 variant was slightly faster than that of α_1A-C16 [τ_deactivation(α_1A-C2) = 5.6 ± 0.3 msec at −80 mV (n = 7); τ_deactivation (α_1A-C16) = 6.6 ± 0.3 msec at −80 mV (n = 6)]. When expressed with each of the Chβ subunits, the current–voltage relationships of α_1A-C2 and α_1A-C16 were essentially indistinguishable (Fig. 2). With all Chβ subunits, both α_1A variants had peak inward currents at +15 to +20 mV. Additionally, the voltages at which 1% of the maximal current was obtained were nearly identical (Chβ_1a/α_1A-C2 = −19.2 ± 4.4 mV, whereas Chβ_1a/α_1A-C16 = −19.5 ± 5.0 mV; this measurement was not obtained with the other Chβ subunits because of poor expression levels). Coexpression of the rabbit Chβ_1a subunit yielded currents two to five times higher than that with any of the human Chβ subunits (Fig.2). The goal of this study was to determine the effect of C-terminal α_1A splicing on the function of the channel, independent of Chβ binding to the region. Therefore, the remainder of the study was performed with the Chβ_1a subunit because it does not appear to bind to this region of the α_1A subunit and has the most robust expression levels (Walker et al., 1998).

Fig. 2.

Biophysical properties of the α_1A-C2and α_1A-C16 subunits coexpressed with rabbit α_2a/δ and Chβ_1a, Chβ_1b, Chβ₃, or Chβ₄ subunits. A–D, Representative current traces of α_1A-C2(top) and α_1A-C16(middle) between −40 and +40 mV with the Chβ_1a (A), Chβ_1b(B), Chβ₃(C), or Chβ₄(D) subunit. Normalized current–voltage relationships (bottom) of α_1A-C2(open diamonds) and α_1A-C16(filled diamonds) with the Chβ_1a(n = 14 and 20, respectively), Chβ_1b (n = 21 and 14, respectively), Chβ₃ (n = 10 and 10, respectively), or Chβ₄ (n = 26 and 22, respectively) subunit.

The voltage dependency of activation, as determined from tail current measurements, was similar between α_1A-C2 and α_1A-C16 [Fig.3A, V_1/2(α_1A-C2) = 9.5 ± 1.4 mV; V_1/2 (α_1A-C16) = 7.3 ± 1.1 mV], as was the voltage dependency of steady-state inactivation (Fig. 3B, V_1/2(α_1A-C2) = −37.9 ± 1.4 mV; V_1/2 (α_1A-C16) = −34.3 ± 1.5 mV]. The time constants of activation and inactivation were described with a single exponential (Fig.3C,D). The time constants of activation of the α_1A-C16 variant, however, were much slower than that of α_1A-C2 and varied with membrane potential (Fig. 3E). The τ values of activation of α_1A-C16 ranged from 6.15 ± 0.40 msec at 0 mV to 1.24 ± 0.06 msec at +30 mV, whereas those of α_1A-C2 ranged from 2.07 ± 0.51 msec at 0 mV to 0.75 ± 0.12 msec at +30 mV. The time constants of inactivation of the α_1A-C16 variant were also slower than that of α_1A-C2 (Fig.3F)_. The τ values of inactivation of α_1A-C2 ranged from 347 ± 69 msec at −10 mV to 175 ± 11 msec at +50 mV, whereas those of α_1A-C16 ranged from 984 ± 325 msec at −10 mV to 338 ± 46 msec at +50 mV. Analysis of the correlation between peak current and the τ of inactivation demonstrates that slow activation of the Cl_Ca current did not contaminate the apparent rates of inactivation (r = 0.304).

Fig. 3.

Voltage dependencies and rates of inactivation and activation of α_1A-C2 and α_1A-C16 subunit complexes. A, Voltage dependence of activation of α_1A-C2 and α_1A-C16 as measured from tail current measurements. Inset, Tail currents that were measured after the capacitive current (dashed vertical line) at 0 mV after a 20 msec depolarization to the test potential (−40 to +75 mV in 5 mV increments). Tail currents were normalized to the largest tail current in each series of test pulses. These data were fit with a Boltzmann equation: % I= 1/[1 + exp(−(V_test − V_1/2)/k)]. B, Isochronal inactivation of α_1A-C2 and α_1A-C16. A 20 sec conditioning pulse ranged from −100 to +40 mV in 10 mV increments. The conditioning pulse was followed by a test pulse to +20 mV for 300 msec. Data were fit with a Boltzmann equation: % I= 1/[1 + exp((V_test − V_1/2)/k)]. C, Single-exponential fits of activation of α_1A-C2 and α_1A-C16 at +20 mV. Fits are shown as solid black lines. D, Single-exponential fits of inactivation of α_1A-C2 and α_1A-C16 at +20 mV. Fits are shown as solid black lines.E, Average τ_activation of α_1A-C2 and α_1A-C16 between the voltages 0 and +30 mV. Traces were fit with a single exponential from the onset of the inward current to the time of peak current. For this and all following figures, the asterisks denote statistical significance (p < 0.01) by the use of an ANOVA test. F, Average τ_inactivation between −10 and +50 mV for α_1A-C2 and α_1A-C16. Eachpoint represents a minimum of six recordings. The SEM for each point is shown unless the values were smaller than the symbol. Traces were fit with a single exponential from the peak inward current to the end of the depolarization.

Because α_1A subunits are expressed primarily at synapses and high-frequency trains of action potentials govern neurotransmitter release, recovery from inactivation was characterized as a means of predicting how α_1A-C2 versus α_1A-C16 might respond to repetitive stimulation. Interestingly, at −40 and −60 mV, the rates of recovery from inactivation of α_1A-C2 and α_1A-C16 are not significantly different (Fig.4A,D). However, at more negative potentials, the α_1A-C2 subunit recovers from inactivation more rapidly than does the α_1A-C16 subunit (Fig. 4B–D). A comparison of the fits of these recovery rates with a single-exponential function reveals that the differences in recovery rates become more apparent at increasingly negative potentials (Fig.4D), showing that the rate of recovery from inactivation of the α_1A-C2 subunit is more voltage dependent than that of α_1A-C16. Although this difference is greater at potentials not likely to be seen physiologically (−100 to −120 mV), the recovery rate is still significantly different between −70 and −90 mV: at −70 mV, τ_recovery α_1A-C2 = 398 ± 19 msec, and τ_recovery α_1A-C16 = 469 ± 23 msec; at −80 mV, τ_recovery α_1A-C2 = 234 ± 11 msec, and τ_recoveryα_1A-C16 = 321 ± 19 msec; and at −90 mV, τ_recovery α_1A-C2 = 178 ± 6 msec, and τ_recovery α_1A-C16 = 251 ± 15 msec. The voltage dependence of the recovery from inactivation is shifted by ∼10 mV in the physiological range (Fig.4D).

Fig. 4.

Voltage dependency of recovery from inactivation of the α_1A-C2 and α_1A-C16 subunit complexes. A–C, Average recoveries of α_1A-C2 and α_1A-C16 at −60 mV (A), −80 mV (B), and −100 mV (C). A two-pulse protocol was used with an initial test pulse of 600 msec to +30 mV followed by a conditioning pulse (ranging from −40 to −120 mV) of intervals ranging from 20 to 1100 msec. This was followed by a second 200 msec test pulse to +30 mV,I₂. The percentage recovery was measured as % Recovery = (I₂ −I_{end of pulse I1})/(I₁ −I_{end of pulse I1}). Each pointshown is the average of six to nine different recordings. The SEM for each point is shown unless the values were smaller than the symbol. The solid lines are the single-exponential fits of the % Recovery. D, Voltage dependency of the rate of recovery from inactivation of α_1A-C2 and α_1A-C16. τ_recoveryis the average value derived from single-exponential fits of individual experiments. The asterisks (p< 0.01) and cross (p < 0.05) denote statistical significance by the use of an ANOVA test.

Because the most obvious structural difference between α_1A-C2 and α_1A-C16 is the presence of the highly charged 12 amino acid segment encoded by exon 44 (Fig. 1A), a chimera was constructed to characterize the influence of this segment on channel gating. The chimera α_1A-C14 is identical to α_1A-C16 except that it contains the 12 amino acids encoded by exon 44. As expected, the voltage dependencies of activation and steady-state inactivation of α_1A-C14 were similar to those of α_1A-C16 [V_{1/2activation}(α_1A-C14) = 6.7 ± 1.3 mV; V_{1/2inactivation} (α_1A-C14) = −33.0 ± 1.1 mV]. The differences between the rates of activation (Fig.5C) and deactivation [at −80 mV, τ_deactivation(α_1A-C14) = 6.3 ± 0.4 msec] and the rates of recovery from inactivation (at −100 mV, τ_recovery = 216 ± 19 msec; at −80 mV, τ_recovery = 316 ± 31 msec; at −60 mV, τ_recovery = 520 ± 63 msec) were not statistically significant. The most striking difference between the two variants was the enhancement of the rate of inactivation of α_1A-C14 relative to that of α_1A-C16 (Fig.5A,B,D). Moreover, these rates were similar to those of α_1A-C2, ranging from 198 ± 14 msec at 0 mV to 133 ± 16 msec at +50 mV.

Fig. 5.

Exon 44 enhances the rate of inactivation.A, Current traces of α_1A-C14, a clone that contains the 12 amino acids encoded by exon 44, between −50 and +40 mV. B, Currenttraces of α_1A-C16, a clone that does not contain exon 44, between −50 and +40 mV. C, Average τ_activation of α_1A-C14 and α_1A-C16 between 0 and +30 mV. D, Average τ_inactivation between 0 and +50 mV for α_1A-C14 and α_1A-C16. Eachpoint represents a minimum of five recordings. The SEM for each point is shown unless the values were smaller than the symbol. Traces were fit with a single exponential from the peak inward current to the end of the depolarization.

To determine whether the effects of the amino acid segment encoded by exon 44 on the rate of inactivation were background dependent, this exon was removed from α_1A-C2 to create α_1A-C4. Thus, α_1A-C4 is identical to α_1A-C2 except that it lacks the 12 amino acids encoded by exon 44. The effect of the removal of exon 44 was indeed to slow the rate of inactivation (Fig.6A,B,D) without significantly affecting the voltage dependency of activation [V_1/2 (α_1A-C4) = 7.7 ± 1.0 mV] or inactivation [V_1/2(α_1A-C4) = −31.4 ± 1.2 mV]. Interestingly, this construct had a slower rate of activation (Fig.6A–C).

Fig. 6.

In a different C-terminal combination, the exon 44 slows the rate of activation but still enhances the rate of inactivation. A, Current traces of α_1A-C2, a clone that contains the 12 amino acids encoded by exon 44, between −50 and +40 mV. B, Currenttraces of α_1A-C4, a clone that does not contain exon 44, between −50 and +40 mV. C, Average τ_activation of α_1A-C2 and α_1A-C4 between 0 and +30 mV. D, Average τ_inactivation between 0 and +50 mV for α_1A-C2 and α_1A-C4. Eachpoint represents a minimum of five recordings. The SEM for each point is shown unless the values were smaller than the symbol. Traces were fit with a single exponential from the peak inward current to the end of the depolarization.

DISCUSSION

The gating properties of neuronal Ca²⁺ channels are determined by multiple structural domains within the α₁ subunit (Tsien et al., 1991; Catterall, 1995). Our results indicate that C-terminus alternative splicing can influence some of these properties. The C terminus of the α_1A subunit is encoded by 12 exons, numbered 36–47 (Ophoff et al., 1996), of which 4 exons, 37, 44, 46, and 47, undergo alternative splicing (Zhuchenko et al., 1997). Depending on the pattern of splicing, the C terminus may vary from 436 to 517 amino acids. Binding sites for several intracellular signaling molecules, which affect channel gating, have been identified within this sequence. An EF-hand Ca²⁺-binding motif can be aligned to sequence encoded by exons 36 and 37 (de Leon et al., 1995). Calmodulin has been shown to bind to a region of the C terminus encoded by exon 40 (Peterson et al., 1999) and to a calmodulin-binding domain (CBD) identified in the region encoded by exon 42 (Lee et al., 1999). Calcium-dependent binding of calmodulin to the CBD motif speeds inactivation and recovery from inactivation and produces a long-lasting facilitation of Ca²⁺ current (Lee et al., 1999). Calcium channel β₄ subunit binding to a region encoded by exons 43–47 has been shown to enhance the rate of inactivation (Walker et al., 1998). In addition, the C terminus appears to be essential for modulation by G_βγ-proteins (Zhang et al., 1996), particularly those regions encoded by exons 45 and 46 (Qin et al., 1997). G-proteins modulate the kinetics of channel activation and inactivation, the voltage dependence of activation, and recovery from inactivation (Bean, 1989; Patil et al., 1998; Zamponi and Snutch, 1998). Polyglutamine tract expansions in exon 47 (Zhuchenko et al., 1997) shift the voltage dependence of activation to more negative potentials, dramatically increase current density (Piedras-Rentería et al., 1999), and alter activation and inactivation kinetics (Resituito et al., 1999). These latter effects appear to be dependent on coexpression of the β₄ subunit.

This study focuses on two naturally occurring splice variants of the α_1A subunit gene that differ in nucleotide sequences corresponding to the NP exon of the IVS3–IVS4 extracellular linker and C-terminal exons 37 and 44. One variant, α_1A-C2, has been identified previously in human cerebellum but has not been expressed and characterized (Zhuchenko et al., 1997) (GenBank number U79663). It lacks the NP exon but contains exons 37a and 44. The other variant, α_1A-C16, which has not been identified previously, contains the NP exon and exon 37b but lacks exon 44. Neither α_1A-C2 nor α_1A-C16 contains exon 47, which is present in the well characterized α_1A subunits BI-1 and BI-2 isolated from rabbit brain (Mori et al., 1991; Sather et al., 1993). Because of the central role of the α_1A C terminus in defining channel-gating characteristics, we hypothesized that the difference in splicing pattern between the two variants would be reflected by changes in their biophysical properties.

Expression in Xenopus oocytes demonstrated that α_1A-C2 and α_1A-C16 are similar with respect to current–voltage relationships and the voltage dependency of activation and inactivation. This is different from previously reported results. The presence of the NP exon shifted the voltage dependence of activation and inactivation of a rat brain α_1A subunit (Bourinet et al., 1999). In a study conducted with a human α_1A subunit, the NP exon only shifted the voltage dependence of inactivation (Hans et al., 1999). Taken together, these data suggest that regions in addition to NP are important for determining the voltage dependence of activation and inactivation. However, other differences, such as the isoform of the Chβ subunit coexpressed with the α_1Asubunit, differing expression systems, and the lack of complete inactivation at the end of the conditioning pulse could shift the voltage dependence of inactivation (Hans et al., 1999).

Further characterization reveals differences in the respective rates of activation, inactivation, deactivation, and recovery from inactivation of channels containing the α_1A-C2 or α_1A-C16 subunits. Complexes containing the α_1A-C16 subunit activate more slowly than do those that contain α_1A-C2. This may be explained, in part, by the presence of the NP amino acids in the extracellular IVS3–IVS4 linker, which have been shown to decrease the rate of activation of α_1A and α_1B channel complexes by ∼1.5-fold at +10 mV (Hans et al., 1999; Lin et al., 1999). The α_1A-C2 complex has an apparent rate of activation that is similar to that of other α_1Asubunits that lack NP (1.6 ± 0.2 vs 1.2 ± 0.5 msec) (Hans et al., 1999). The activation rate of α_1A-C16is considerably slower than that of other subunits that contain NP (4.8 ± 0.5 vs 2.2 ± 1.1 msec) (Hans et al., 1999), suggesting that other regions of the C terminus also affect activation. Our results indicate that the region encoded by exon 44 appears to be acting in a modular manner to enhance the rate of inactivation of α_1A complexes twofold to threefold. It is unclear whether this results from a direct conformational change or whether the presence or absence of this charged sequence affects modulation by intracellular signaling molecules. Further experiments are required to resolve this issue. Interestingly, the regions encoded by exons 43–47 have been shown to bind β₄ and β_2a subunits but not β₁or β₃ subunits (Walker et al., 1998). Our results with the β₁ subunit suggest that changes in inactivation imparted by exon 44 are independent of C-terminal β subunit binding. The deactivation rate of the α_1A-C16 complex is slightly slower than that of α_1A-C2. Because the settling time of the oocyte clamp is ∼2 msec, these rates are likely to be faster than we report. Regardless, the α_1A-C16 complexes are consistently slower over a wide voltage range (−60 to −100 mV; data not shown). Thus, as is true for the α_1Csubunit of L-type Ca²⁺ channels (Soldatov et al., 1997, 1998), C-terminal alternative splicing does affect the biophysical properties of channel complexes composed of α_1A subunits.

The slower activation, inactivation, and deactivation of α_1A-C16 splice variants relative to α_1A-C2 could translate to differences in neurotransmitter release at subtype-specific synapses. Although differences in channel number and density, second messenger effects, and Ca²⁺ buffering make it difficult to predict what the combined effects of C-terminal splicing on Ca²⁺ entry will be (Park and Dunlap, 1998;Lin et al., 1999), our results do allow for some generalizations to be made. The slower rate of activation of channels containing the α_1A-C16 subunit should lead to a decrease in the initial rate of calcium influx. However, the slower inactivation rate of the α_1A-C16 subunit would enhance later calcium entry, providing the channel does not recover from inactivation by passing through the open state (Slesinger and Lansman, 1991). The slower deactivation rate of α_1A-C16 channel complexes would increase Ca²⁺ flux through these channels at a time when the driving force is relatively large. Therefore, it is likely that in response to a single action potential, the initial rate of Ca²⁺ entry would be slower through complexes containing the α_1A-C16subunit, but these channels would allow for a greater total Ca²⁺ influx.

The rate of recovery from inactivation at negative potentials plays an important role in determining Ca²⁺ influx and neurotransmitter release in response to a train of action potentials. The voltage dependence of the rate of recovery of channel complexes containing the α_1A-C16 subunit differs from those containing α_1A-C2. At less negative potentials, the two splice variants behave similarly. After 500 msec at −60 mV, ∼30% of channels composed of either variant are fully recovered. However, at more negative potentials, complexes containing the α_1A-C16 subunit recover more slowly. After 500 msec at −80 mV, ∼73% of α_1A-C2 complexes are recovered, compared with ∼62% of α_1A-C16 complexes. Thus, Ca²⁺ entry after a train of action potentials would be less in nerve terminals expressing the α_1A-C16 subunit relative to those expressing the α_1A-C2 subunit. These differences may be important from the standpoint of Ca²⁺-mediated synaptic plasticity.

Our results add to a developing picture of neuronal calcium channels that suggests that α₁ subunit genes express channels with a wide array of structures and functions. These channels may have evolved to meet the specific needs of highly specialized synapses. Our results, combined with those of several laboratories (Bourinet et al., 1999; Hans et al., 1999), point to the fact that it is no longer sufficient to refer to α_1Achannels as simply P- and Q-type. Minor splicing events in critical domains of the α_1A subunit alter both pharmacological and physiological properties in ways in which we are only beginning to understand. Determination of where specific α_1A subunit splice variants are expressed, how they respond to trains of action potentials, and how they interact with various signaling molecules should greatly enhance our understanding of the physiology of the synapse.