CACNA1A ∆e24a and + e24a transcripts are differentially expressed in the human nervous system.

Absolute quantification of splice variant transcripts was performed employing a standard curve and related to copy number for comparison of +/Δe24a in human brain samples (Fig. 1B; Table 5). The expression of reference genes, TBP and CYC1, was stable across samples (Table 6). In individual brain regions, the expression of +e24a and Δe24a transcripts varied. The Δe24a was found more abundantly expressed in cerebellum and spinal cord while cerebral cortex and hippocampus expressed significantly higher levels of + e24a transcripts. These results suggest the expression of +SSTR and ΔSSTR Cav2.1 channels varies across different brain regions, consistent with previous studies showing the differential expression of Cav2.1 splice variants in human brain [5, 38].

Fig. 1

Expression of two CACNA1A isoforms generated by alternative splicing of exon 24a in human nervous tissue. Upper panel A shows a schematic representation of the cassette exon 24a (left) and the amino acid sequences of the isoforms ΔSSTR and + SSTR (right), resulting from the alternative splicing event. The sequence shown corresponds to residues 1326 to 1358; the inclusion of the tetrapeptide (underlined red) occurs at the limit of the terminal portion of S3 and the extracellular loop connecting with S4 in domain III. Highlighted residues in yellow correspond to the S4 gating charges. The topology diagram of the hCav2.1 α1A subunit (middle) shows the location of the FHM-1 S218L mutation (blue diamond in the S4-S5 intracellular loop of domain I) and the tetrapeptide SSTR encoded by exon 24a (red circle domain III). Lower panel A: amino acid sequence of the S4-S5 linker of domain I (residues 191–237), with residue Ser218 labeled in blue, is shown below the topology diagram. Amino acid numbers correspond to the sequence of the clone used for generating the Cav2.1 cryo-EM structure (PDB: 8X90) used in our study for molecular modeling. B shows the comparison of copy number between transcript isoforms Δe24a and + e24a, obtained using qRT-PCR. Bar graphs represent the copy number values from different human tissue samples (see Table 1), derived from standard curves using qPCR CTs in triplicate. Error bars indicate the variability among replicates. The Δe24a variant is more abundant in the cerebellum and spinal cord, whereas + e24a is the predominant variant in the cerebral cortex

Table 5 Copy number for ∆e24a and + e24a in human brain samplesTable 6 Mean PCR cycle threshold (Ct) values for reference genes CYC1 and TBP across brain regionsDistinct splice-variant dependent effects of the FHM-1 S218L mutation on Cav2.1 biophysical properties

In order to compare the functional effects of S218L mutation on alternatively spliced variants, we examined macroscopic currents in HEK293 cells expressing ΔSSTR or +SSTR wild type isoforms (upper panels Fig. 2A) as well as their corresponding mutant isoforms (lower panels Fig. 2A). Examples of representative currents recorded over a range of test potentials using Ca2+ as the charge carrier are shown in Fig. 2A, and current–voltage profile was analyzed by plotting the average current density as a function of membrane potential (Fig. 2B). As previously described, S218L mutation leads to a reduction in maximal current density (Fig. 2C): wt ΔSSTR 30.96 ± 3.73 pA/pF, n = 15; S218L ΔSSTR 12.46 ± 1.16 pA/pF, n = 16; wt +SSTR 37.20 ± 4.07 pA/pF, n = 17; S218L +SSTR 12.68 ± 2.09 pA/pF, n = 13. One-way ANOVA F(3,57) = 16.86, p < 0.0001, post hoc Tukey test wt ΔSSTR vs. S218L ΔSSTR p = 0.0005 and wt +SSTR vs. S218L +SSTR p < 0.0001. Close examination of the I-V curves revealed higher current density at membrane potentials negative to -20 mV (blue arrows, Fig. 3B), particularly for the S218L +SSTR mutant channel.

Fig. 2

Inclusion of exon 24a enhances the functional impact of S218L mutation on Cav2.1 activation gating. A. Macroscopic Ca2+ currents evoked by 90 ms voltage steps between -60 and + 35 from a holding potential of -90 mV, recorded from HEK293F cells expressing wt hCav2.1 channel splice isoform ΔSSTR (top left), wt +SSTR (top right), or S218L mutant channels in the corresponding splice variant background (bottom panels). Current density vs voltage relationships (B) show that alternatively spliced wild type +SSTR channel displays a negative shift in activation threshold, relative to the wild type ΔSSTR. I-V curves show the distinctive effects of the S218L mutation, such as reduced current density and a hyperpolarizing shift of the activation voltage. Mutation-induced effects were similar between the two splice isoforms; however, S218L +SSTR mutant channel (Panel B, right) displayed a shallower I-V curve slope factor and a further shift towards more negative potentials (see blue arrows), compared to the S218L ΔSSTR (panel B, Left). Bar graph (C) shows a comparison of the averaged current density (pA/pF) values at the peak of the I-V curve. Both S218L mutant channels display reduced current density, regardless of channel splice isoform. The number of cells recorded is given in parentheses; the asterisk indicates statistical significance < 0.0001 of mutant channel values, relative to their respective wt. Voltage dependence of activation (D) and inactivation (E) plots revealed that activation half-point (V50) and slope factor (k) are significantly different between the two wild type splice isoforms (blue dots D), whereas steady-state inactivation curves were nearly identical (black dots vs blue dots, E) (Table 7). The activation curve is shallower for the mutant channels with a prominent hyperpolarizing shift of the activation voltage to even more negative potentials (D). S218L mutant channels display a similar difference (~ 15 mV) in half-maximal inactivation (E), relative to their corresponding splice isoform

Fig. 3

S218L FHM-1 mutation alters the kinetic properties of P/Q-type currents mediated by + /Δ SSTR isoforms. Superimposed normalized recordings of peak currents from SSTR (A) and +SSTR (B) isoforms show the difference in the activation time course between wild type and S218L channels, with the onset of activation distinctively faster in the mutant channels (S218L ΔSSTR, red trace in A; S218L +SSTR purple trace in B). The average activation time constant values at the peak of the I-V curve is shown next to the corresponding traces. Mean activation time constants were plotted against membrane potential (C), showing that mutant channels display a shallower voltage dependence than the wild types. To examine channel deactivation, currents were evoked by a brief depolarizing test pulse from a holding of −100 to −5 mV, and membrane potential was repolarized to various membrane potentials to record tail currents. Exemplary recordings corresponding to repolarization to −40 mV from each pair of wt (black trace, D; blue trace, E) vs mutant isoforms (red trace, D; purple trace, E) were normalized and overlapped, showing that both mutant channels not only activate faster, but also take longer to close when the membrane is repolarized, relative to wild type. The decay phase of the tail currents was fitted with a single exponential to obtain deactivation time constants, plotted as a function of the repolarization potential (F). Deactivation voltage dependence was more pronounced in the wild type isoforms (see text for details)

Voltage-dependence gating was analyzed using the normalized conductance plots (Fig. 2D) and the steady state inactivation curves (Fig. 2E). In wildtype channels, the main consequence of +SSTR inclusion was to shift the voltage dependence of channel activation to more negative potentials. Parameters obtained from the best fit of a single Boltzmann equation to the experimental data (see Table 7) revealed that half-maximal activation (V50) values were significantly different between the two wild type splice isoforms (wt ΔSSTR V50: −11.80 ± 0.46, n = 13; wt +SSTR V50: −15.44 ± 0.30, n = 17, one-way ANOVA F(3,54) = 114.7 p < 0.0001, post hoc Tukey test wt ΔSSTR vs. wt +SSTR p < 0.0001), no significant difference in the slope factor (k) (wt ΔSSTR k: 3.65 ± 0.08, n = 13; wt +SSTR k: 3.99 ± 0.06, n = 17), and no significant changes were observed in the steady-state inactivation (wt ΔSSTR V50: −45.06 ± 0.73, n = 12; wt +SSTR V50: −47.16 ± 0.64, n = 8; wt ΔSSTR k: 6.42 ± 0.12, n = 12; wt +SSTR k: 6.08 ± 0.18, n = 8).

Table 7 Activation and steady-state inactivation parameters

The S218L mutant shifted the activation curves towards more negative values compared to their wild type splice isoform counterparts (Fig. 2D), with significantly different midpoint values (wt ΔSSTR V50: -11.80 ± 0.46, n = 13; S218L ΔSSTR V50: −16.78 ± 0.31, n = 16; wt +SSTR V50: −15.44 ± 0.30, n = 17; S218L +SSTR V50: −22.05 ± 0.44, n = 12; one-way ANOVA F(3,54) = 114.7 p < 0.0001, post hoc Tukey test wt ΔSSTR vs. S218L ΔSSTR p < 0.0001, wt +SSTR vs. S218L +SSTR p < 0.0001). The conductance-voltage relation was steeper in both wild type splice isoforms compared to the mutants, as indicated by the slope factor (wt ΔSSTR k: 3.65 ± 0.08, n = 13; S218L ΔSSTR k: 5.21 ± 0.16, n = 16; wt +SSTR k: 3.99 ± 0.06, n = 17; S218L +SSTR k: 5.70 ± 0.04, n = 12, one-way ANOVA F(3,54) = 83.52 p < 0.0001, post hoc Tukey test wt ΔSSTR vs. S218L ΔSSTR p < 0.0001, wt +SSTR vs. S218L +SSTR p < 0.0001).

The effect of S218L mutation on the steady-state inactivation of the +SSTR variant paralleled the effect elicited on ΔSSTR (Fig. 2E). The most striking effect was a hyperpolarizing shift (wt ΔSSTR V50: −45.06 ± 0.73, n = 12; S218L ΔSSTR V50: −60.47 ± 0.37, n = 8; wt +SSTR V50: −47.16 ± 0.64, n = 8; S218L +SSTR V50: −62.90 ± 0.43, n = 8, one-way ANOVA F(3,32) = 217.7, post hoc Tukey test wt ΔSSTR vs. S218L ΔSSTR p < 0.0001, wt +SSTR vs. S218L +SSTR p < 0.0001). The effect of the mutation on the inactivation slope factor was more pronounced for the ΔSSTR variant (wt ΔSSTR k: 6.42 ± 0.12, n = 12; S218L ΔSSTR k: 5.28 ± 0.18, n = 8), although the change on the +SSTR was also significant (wt +SSTR k: 6.08 ± 0.18, n = 8; S218L +SSTR k: 5.24 ± 0.31, n = 8, one-way ANOVA F(3,32) = 9.572 p = 0.0001, post hoc Tukey test wt ΔSSTR vs. S218L ΔSSTR p = 0.0009 and wt +SSTR vs. S218L +SSTR p = 0.0364). We observed that, as a consequence of the hyperpolarizing shift in the activation of S218L mutant channels, the membrane potential at which the maximum window current is observed was ~ 13 mV more negative than that of wild type channels in both splice backgrounds (supplementary Fig. 1). Our data is consistent with a study using brainstem slices from transgenic mice harboring the S218L mutation [22], where presynaptic calcium currents recorded from the calyx of Held displayed a hyperpolarizing shift in the window current.

A comparison of the time-dependent current increase at the peak of the IV curve for each isoform is shown in Fig. 3 (A, B). The time course of Ca2+ current activation was similar between the two splice isoforms, and S218L mutation results in a faster current rise, regardless of splice variant (wt ΔSSTR = 2.71 ± 0.12 ms, n = 9; S218L ΔSSTR = 1.37 ± 0.15 ms, n = 9; wt +SSTR = 2.28 ± 0.19 ms, n = 9; S218L +SSTR = 1.19 + 0.09 ms, n = 10, one-way ANOVA F(3,33) = 27.08 p < 0.0001, post hoc Tukey test wt ΔSSTR vs. S218L ΔSSTR p < 0.0001 and wt +SSTR vs. S218L +SSTR p < 0.0001).

Activation time constants, obtained from exponential fits to the rising phase of macroscopic currents, decrease sharply with an increase in depolarizing steps (Fig. 3C). The voltage dependence of the activation kinetics were less pronounced in the mutant channels than their respective wild type splice isoform: wt ΔSSTR 11.66 ± 0.86 mV per e-fold change n = 9, S218L ΔSSTR 16.56 ± 1.22 mV per e-fold change n = 9; wt +SSTR 12.05 ± 0.85 mV per e-fold change n = 9, S218L +SSTR 20.33 ± 1.53 mV per e-fold change n = 10, one-way ANOVA F(3,33) = 12.58 p < 0.0001, post hoc Tukey test wt ΔSSTR vs. S218L ΔSSTR p = 0.0311 and wt +SSTR vs. S218L +SSTR p < 0.0001).

An impact of the S218L mutation on current kinetics was also observed concerning the rate of deactivation. To examine the kinetics of channel closing, a 15 ms depolarization to −5 mV was followed by repolarization to a range of membrane potentials between −60 to −20 mV; the decay of repolarization-induced tail currents was analyzed by fitting a single exponential to obtain time constants. Representative traces showing the slowed deactivation of mutant channels, compared to their respective wild type splice isoforms, are shown on Fig. 3D, E; decay time constants of the tail currents evoked by repolarizing to −40 mV were significantly different (wt ΔSSTR = 0.35 ± 0.02, n = 7; S218L ΔSSTR = 0.50 ± 0.03, n = 10; wt +SSTR = 0.29 ± 0.01, n = 9; S218L +SSTR = 0.62 ± 0.04, n = 6; one-way ANOVA F(3,28) = 28.34 p < 0.0001, post hoc Tukey test wt ΔSSTR vs. S218L ΔSSTR p = 0.0035 and wt +SSTR vs. S218L +SSTR p < 0.0001).

The plots of deactivation time constant against repolarization membrane potential (Fig. 3 F) show that the rate of change as a function of voltage was smaller in the S218L mutant channels (ΔSSTR: 24.28 ± 3.08 mV per e-fold change, n = 8; +SSTR: 19.75 ± 1.91, n = 6) than in the wt splice isoforms, reaching significance only for the ΔSSTR splice variant (wt ΔSSTR: 12.7 ± 1.26 mV, n = 7; wt +SSTR: 15.00 ± 1.10 per e-fold change, n = 9 one-way Welch’s ANOVA W(3.0,13.12) = 5.54 p = 0.0112, post hoc Dunnett’s T3 test wt ΔSSTR vs S218L ΔSSTR p = 0.0359, wt +SSTR vs. S218L +SSTR p = 0.2774). Intuitively, the slow deactivation of mutant channels would allow a sustained Ca2+ entry after action potential repolarization, yet a shallower voltage dependence of channel closing might disrupt the effective contribution to the spike-triggered Ca2+ transients.

Together, our biophysical analyses indicate that the alternatively spliced SSTR tetrapeptide mainly affects voltage-dependence of Cav2.1 gating and further distinctively modifies the effect of S218L mutation on Cav2.1 activation, with nearly insignificant changes in voltage dependent inactivation.

Molecular modeling of the S218L FHM-1 mutation in the + /Δ SSTR splice variants.

Conformational changes associated with voltage-driven gating occur concomitantly with changes in interactions between residues from adjacent segments. Mutations that strengthen or create new interactions are thought to stabilize particular conformational states resulting in altered voltage-dependent sensitivity [39]. Here, we find that the SSTR splice insertion or the S218L missense mutation results in a hyperpolarizing shift in the V50 of activation. A further strikingly shift in the voltage dependence of activation occurs when both changes are combined (e.g., +SSTR and S218L). Therefore, we wanted to explore whether the + / ΔSSTR alternative splicing event and the S218L missense mutation altered interactions between residues that might impact the transition between gating states. To investigate this, we examined the structural impact of the S218L mutation and SSTR variation using the recently published cryo-EM structure of Cav2.1 [32].

We first modeled the Ser218 to Leu218 substitution in the published Cav2.1 cryo-EM structural template that lacks the SSTR insertion [32]. The S218L mutation is located in the centre of the S4-S5 helical linker of domain I (Fig. 4A). In the Cav2.1 structure, the domain I voltage sensor (VSD I) is in an ‘up’ state with a single S4 gating charge located below the occluding Phe residue. The S4-S5 linker is oriented parallel to the membrane (Fig. 4B), in accordance with the sliding-helix model where the position of the S4-S5 linker is determined by the movement of the S4 during gating. This is consistent with previous work using the structure of a voltage gated sodium channel in a resting and activated states showing that in the up position, the S4 pulls the S4-S5 linker towards the inner surface of the membrane [40]. We first analyzed whether the S218L mutation impacted interactions between residues in close proximity and in neighbouring segments. In the wildtype Cav2.1 structural template, the position of the domain I S4-S5 linker is stabilized by numerous interactions with neighboring residues in other segments from domain I and II that range from van der Waals interactions to hydrogen bonds. While approximately 65% of residues in the domain I S4-S5 linker are involved in various intersegment bonding interactions, the wildtype Ser218 residue does not appear to form any intersegment interactions in the conformational state captured by the cryo-EM structure (Fig. 4C). In our S218L model, Leu218 similarly did not form any intersegment interactions, nor did the mutation modify interactions between neighbouring residues (Fig. 4D), suggesting that the voltage dependent effects of S218L mutation do not stem from the strengthening of interactions between amino acids that stabilize the S4 ‘up’ conformational state.

Fig. 4

Molecular modeling of Cav2.1 S218L mutation depicts increased hydrophobicity in domain I S4-S5 linker. A Membrane view of Cav2.1 cryo-EM structure. VSD I is highlighted with S1 in yellow, S2 green, S3 blue and S4 pink, S4-S5 linker cyan, S5 red and S6 in purple. The location of Ser218 residue (*) is shown in orange. B Location of the Ser218 residue (*) with respect to modeled membrane in the isolated domain I. In a depolarized state, S4-S5 linker runs parallel to the inner surface the membrane. C–D) Orientation of WT Ser218 and Leu218 residues. Both Ser218 and Leu218 do not interact with any residues in neighboring domains. E The hydrophobic residues of the amphipathic domain I S4-S5 linker in WT are aligned on one side of the alpha helix, forming a hydrophobic patch (grey). F S218L mutation introduces a hydrophobic amino acid on the surface of the alpha helix facing the cytoplasm, increasing hydrophobicity with Val214 and Leu218

As the S218L mutation involves the substitution of a polar amino acid to a hydrophobic Leu, we next generated surface maps of hydrophobic regions to investigate whether the mutation impacts hydrophobicity of the domain I S4-S5 linker region (Fig. 4E–F). The S4-S5 linker is amphipathic [41]; in domain I the hydrophobic residues Leu212, Val215, Leu216, Ile 219, Met220 and Met223 face the inner plasma membrane, while the residues Ser211, Val214, Lys217, Ser 218, Lys221 and Ile224 face the cytoplasm (Fig. 4E). Our model predicts that the S218L mutation leads to the formation of a larger hydrophobic region facing the cytoplasm formed by Val214, Leu218, Lys221 and Ile224 (Fig. 4F). During the transition from resting state to activated, as the domain I S4-S5 linker moves from the hydrophilic cytoplasm to the hydrophobic membrane [40], this enhanced hydrophobicity is predicted to make the transition more energetically favourable. Overall, our modeling indicates the S218L mutation likely stabilizes the domain I S4-S5 linker at the interphase between the inner membrane and the cytosol, potentially reducing the threshold for activation, and accounting for the major observed electrophysiology effect of a hyperpolarizing shift in the V50 of activation.

We next examined how the insertion of the splice variant SSTR tetrapeptide might impact the loop structure of the Cav2.1 domain III S3-S4 linker. The SSTR insertion occurs between T1330 and G1331 at the N-terminal region of the S3-S4 linker (Figs. 1A, 5A). Similar to VSD I, the S4 VSD III is in the up state (Fig. 5B). The structures predicted by the top 5 modeled loops show that the SSTR insertion either extends the alpha helix by approximately one and a half turns or increases the length of the domain III S3-S4 linker loop (Fig. 5B). The position of the S3 and S4 helixes were not substantially altered by the insertion, similar to previous modeling of a 19 residue S3-S4 linker splice variant insertion in the Cav1.1 channel [39]. The analysis was performed in all five loops with the results shown corresponding to the representative loop displaying the highest percentage of common features. Given the close proximity to the domain III S4 voltage sensor, we first investigated whether the insertion of SSTR modifies interactions between gating charge residues. The modeling data predicts that the number or strength of interactions between gating charge residues is not significantly impacted by the SSTR insertion. Further, we did not find any significant changes in the strength of interactions in non-gating charge residues.

Fig. 5

Inclusion of SSTR modifies positively charged surfaces surrounding VSD III. A Cav2.1 cryo-EM structure in ΔSSTR background shown in a membrane view. Domain III is highlighted with S1 in yellow, S2 green, S3 blue, S4 pink, S4-S5 cyan, S5 red and S6 purple. Location of SSTR insertion is highlighted by an asterisk. B Isolated VSD III with top 5 models of + SSTR splice variant shown in grey. Gating residues are highlighted. C–D) Hydrophobic (grey) and positively charged (red) patches in ∆SSTR (C) and + SSTR backgrounds (D). E–F) The inclusion of the tetrapeptide SSTR inserts a charged residue (Arg highlighted in red) that creates a positively charged region with Arg1277 in S2 and Arg1345 in S4 (F). The position of Lys1334 and Lys1336 are also shifted by SSTR insertion, changing the distribution of the positively charged region on the extracellular side of the domain III S3-S4 linker

As the + e24a splice variant encodes for the insertion of three polar residues (2 Ser and 1 Thr) and a positively charged Arg, we also generated surface maps of hydrophobic and charged regions to investigate whether the +SSTR insertion indirectly modifies voltage sensing. Besides the gating charge residue Lys1342, in the wildtype ΔSSTR structure there are two positively charged regions that surround domain III S3 and S4: one region formed by Arg1277 and Arg1345 and the second by Lys1334 and Lys1336 (Fig. 5C, E). In our SSTR loop models, both positively charged regions were altered by the insertion. The inserted Arg extends the positively charged region formed by Arg1277 and Arg1345 (Fig. 5D, F). The modification of the domain III S3-S4 linker caused by the SSTR insertion also changed the orientation of several amino acids, including the positively charged extracellular residues Lys1334 and Lys1336, which are oriented away from each other in the ΔSSTR cryo-EM structure (Fig. 5E) and face the same direction in the +SSTR model (Fig. 5F). These newly charged regions are in proximity to key residues directly involved in gating. The inserted Arg appears to directly influence the positive surface formed by the gating charge residue Arg1345; while Arg1277 is separated by two residues from the negative counter charge Asp1280. Further, Lys1336 is separated by five residues from the first S4 gating charge Lys1342. These altered charged regions are likely to influence the local electric field sensed by these residues, suggesting the SSTR insertion impacts voltage sensitivity contributed by VSD III.

Lastly, we modeled the S218L mutation in the +SSTR background to understand how this combination might further shift the voltage dependence of activation. We did not find any additional change in the structure or interactions between residues in this model compared to each separate model. This may be due to the domain swapped arrangement of voltage gated Ca2+ channels [42], in which residues located in domain I and III have limited interactions with each other. As such, the combined functional effect of S218L and +SSTR may result from independent contributions of each respective VSD in shifting the voltage dependence of activation.