C-Terminal Modulatory Domain Controls Coupling of Voltage-Sensing to Pore Opening in Ca_v1.3 L-type Ca²⁺ Channels

Abstract

Activity of voltage-gated Ca_v1.3 L-type Ca²⁺ channels is required for proper hearing as well as sinoatrial node and brain function. This critically depends on their negative activation voltage range, which is further fine-tuned by alternative splicing. Shorter variants miss a C-terminal regulatory domain (CTM), which allows them to activate at even more negative potentials than C-terminally long-splice variants. It is at present unclear whether this is due to an increased voltage sensitivity of the Ca_v1.3 voltage-sensing domain, or an enhanced coupling of voltage-sensor conformational changes to the subsequent opening of the activation gate. We studied the voltage-dependence of voltage-sensor charge movement (Q_ON-V) and of current activation (I_Ca-V) of the long (Ca_v1.3_L) and a short Ca_v1.3 splice variant (Ca_v1.3_42A) expressed in tsA-201 cells using whole cell patch-clamp. Charge movement (Q_ON) of Ca_v1.3_L displayed a much steeper voltage-dependence and a more negative half-maximal activation voltage than Ca_v1.2 and Ca_v3.1. However, a significantly higher fraction of the total charge had to move for activation of Ca_v1.3 half-maximal conductance (Ca_v1.3: 68%; Ca_v1.2: 52%; Ca_v3.1: 22%). This indicated a weaker coupling of Ca_v1.3 voltage-sensor charge movement to pore opening. However, the coupling efficiency was strengthened in the absence of the CTM in Ca_v1.3_42A, thereby shifting I_Ca-V by 7.2 mV to potentials that were more negative without changing Q_ON-V. We independently show that the presence of intracellular organic cations (such as n-methyl-D-glucamine) induces a pronounced negative shift of Q_ON-V and a more negative activation of I_Ca-V of all three channels. These findings illustrate that the voltage sensors of Ca_v1.3 channels respond more sensitively to depolarization than those of Ca_v1.2 or Ca_v3.1. Weak coupling of voltage sensing to pore opening is enhanced in the absence of the CTM, allowing short Ca_v1.3_42A splice variants to activate at lower voltages without affecting Q_ON-V.

Introduction

Ca²⁺ influx through voltage-gated Ca²⁺ channels (VGCCs) in the plasma membrane of electrically excitable cells causes membrane depolarization and triggers intracellular Ca²⁺-dependent signaling processes. Ca_v1.3 and Ca_v1.2, members of the so-called L-type Ca²⁺ channel family (LTCCs, Ca_v1), are widely expressed in many tissues, including muscle and neurons, sensory tissue, and endocrine cells (1–5). Work with genetically modified mice revealed different physiological roles for these two channel isoforms, even when expressed in the same cell (6–8). Differences in their voltage- and Ca²⁺-dependent gating properties underlie this functional diversity (6–8). Ca_v1.3 channels activate faster and at more negative voltages than Ca_v1.2 (9–11) and therefore sustain Ca²⁺ inward currents also at threshold voltages. This allows them to control autonomous pacemaking in the sinoatrial node (2,6) and adrenal chromaffin cells (7) and support upstroke potentials in neurons (12). Ca_v1.3 channels are also expressed in the substantia nigra pars compacta dopaminergic neurons (13) where they seem to contribute to dendritic Ca²⁺ signals linked to mitochondrial stress and the selective vulnerability of these neurons in Parkinson’s disease (14). Selective Ca_v1.3 channel block is currently pursued as a therapeutic option for neuroprotection in Parkinson’s disease. Although Ca_v1.3 channels activate at voltages that are more negative than all other Ca_v1 (Ca_v1.1, Ca_v1.2, and Ca_v1.4) and Ca_v2 high VGCCs (15), they cannot be classified as low VGCCs, such as T-type channels (16). Indirect comparisons of published data suggest that T-type channels activate and inactivate at voltages that are more negative than Ca_v1.3 (16), but a direct comparison of their gating properties, to our knowledge, does not exist.

Within the past few years, we have discovered that the activation voltage range (I_Ca-V) of Ca_v1.3 Ca²⁺ inward currents (I_Ca) can be shifted to potentials that are even more negative by alternative splicing within the C-terminus of its pore-forming α₁-subunit (17,18). Alternative splicing generates Ca_v1.3 α₁-subunits with shorter C-termini (e.g., Ca_v1.3_42A, Ca_v1.3_43S (18–20)) thereby removing a C-terminal modulatory domain (CTM) from long isoforms (Ca_v1.3_L). This causes two apparently independent changes of channel gating:

• 1.It moderates Ca²⁺-induced inactivation (CDI), an important autoinhibitory feedback mechanism of the channel. The molecular basis of this effect is well understood. As in other VGCCs, CDI is induced by Ca²⁺ binding to calmodulin (CaM) preassociated with the proximal C-terminus of α₁-subunits (21). The CTM of Ca_v1.3 competitively interferes with CaM binding, and thereby reduces CDI (17,21).
• 2.The presence of the CTM in long Ca_v1.3 splice variants reduces open probability at negative voltages and shifts the half-maximal activation voltage (V_0.5(I_Ca)) by ∼10 mV to potentials that are more positive. This modulation has been reported by different laboratories, and occurs independently of CaM (11,18,19).

In contrast to modulation of CDI, it remains unclear how the CTM can affect voltage-dependent gating. Contemporary homology models of the pore-forming α₁-subunit propose four voltage-sensing domains (transmembrane segments S1–S4 of each of the four homologous repeats), which undergo conformational changes upon de- and repolarization of the channel. The intramembrane movement of the four positively charged S4 helices represents the main moving part (22,23). The cytoplasmic linkers between S4 and S5 segments are tightly packed into the pore-forming segments (in particular S6 helices), and thereby couple the voltage-sensor movements (recorded as nonlinear charge movements Q_ON and Q_OFF) to pore opening and closing (22). In the case of Ca_v1.3, the CTM must therefore interfere with this gating apparatus to induce the positive shift in I_Ca-V. Mechanistically, two possibilities exist: the CTM could affect the voltage sensitivity of the voltage-sensing mechanism itself such that higher voltages are required to move the charged S4 helix (inducing a shift in Q_ON voltage-dependence (Q_ON-V)). Alternatively, it could decrease the efficiency of coupling between voltage-sensor movements and subsequent pore opening, evident as a change in I_Ca-V but not Q_ON-V parameters (24).

To address these questions, we established experimental conditions that allowed us to directly determine differences in gating properties that account for the more-negative I_Ca-V of Ca_v1.3 in comparison to Ca_v1.2 and of short Ca_v1.3 splice variants in comparison to long ones. We also systematically compared the gating properties of Ca_v1.3 with low-voltage-activated Ca_v3.1 T-type channels.

We found that Ca_v1.3 Q_ON is significantly more voltage-sensitive than Q_ON of Ca_v3.1 and Ca_v1.2. The coupling of voltage sensing to pore opening is less efficient in Ca_v1.3 but its negative Q_ON-V still leads to I_Ca-V that is more negative than in Ca_v1.2. The CTM did not affect the voltage-dependence of Q_ON of Ca_v1.3, indicating an inhibitory action on the transmission of voltage-sensor movements to pore opening. Surprisingly, intramembrane charge movement of Ca_v3.1 occurred at voltages more positive than that of Ca_v1.3, but the sensitive coupling of the voltage sensors to pore opening accounted for low voltage activation of its I_Ca-V. In the course of our work, we also discovered a strong effect of intracellular organic cations on the voltage-sensing machinery of all three VGCCs.

Methods

Cell culture and transient transfection

HEK 293 (human embryonic kidney) cells were grown in Dulbecco’s modified Eagle’s medium supplemented with 2 mM L-glutamine (Cat. No. 25030-032; Gibco, Life Technologies, Carlsbad, CA), 10 units/mL penicillin (Cat. No. P-3032; Sigma Aldrich, St. Louis, MO), 10 μg/mL streptomycin (Cat. No. S-6501; Sigma), and with 10% v/v fetal calf serum (Cat. No. 10270-106; Gibco). Cells were grown under 5% CO₂ and 37°C until they reached 80% confluency. They were split with 0.05% trypsin for cell dissociation, and passage did not exceed 20. Transient transfection was achieved as described in Bock et al. (19), using equimolar cDNA ratios encoding Ca_v1.3_L, Ca_v1.2, or Ca_v3.1 (generously provided by Norbert Klugbauer) together with auxiliary subunits β₃ and α₂δ₁. Cells were visualized by cotransfection of 1 μg GFP. Twenty-four hours after transfection, cells were plated on 35-mm polystyrene dishes, pretreated with 10 μg/mL poly-L-lysine. At 48–72 h after cell transfection, whole cell patch-clamp experiments were performed.

Whole-cell patch-clamp recordings

GFP-positive HEK 293 cells were recorded using the whole-cell patch-clamp configuration. Borosilicate glass electrodes, having a final resistance of 2–5 MΩ, were pulled with a micropipette puller (Sutter Instruments, Novato, CA) that was fire-polished (MF-830 microforge; Narishige, Tokyo, Japan). Data were digitized (Digitizer 1322A; Axon Instruments, Novato, CA) and recorded in the whole-cell patch-clamp configuration (Axopatch 200B; Axon Instruments).

Intracellular recording solutions used

• NMDG_int: NMDG (150 mM n-methyl-D-glucamine), 10 mM EGTA, 1 mM MgCl₂, 10 mM HEPES, and 4 mM ATP-Mg, adjusted to pH 7.3 with MS (Methanesulfonate);

• Cs_int: 135 mM CsCl, 10 mM Cs-EGTA, 1 mM MgCl₂, 10 mM HEPES, and 4 mM ATP-Na₂, adjusted to pH 7.3 with CsOH;

• TRIS_int: 164 mM Tris, 10 mM EGTA, 1 mM MgCl₂, 10 mM HEPES, and 4 mM ATP-Mg, adjusted to pH 7.3 with MS; and

• TEA_int: 160 mM TEA (triethanolamine), 10 mM EGTA, 1 mM MgCl₂, 10 mM HEPES, and 4 mM ATP-Mg, adjusted to pH 7.3 with MS.

Extracellular solution used for Q_ON recordings

• Choline-Cl(Mg²⁺)_ext: 150 mM choline-Cl, 16 mM MgCl₂, 10 mM HEPES, 0.5 mM CdCl₂, and 0.2 mM LaCl₃, adjusted to pH 7.3 with CsOH.

Extracellular solutions used for I_Ca recordings

• Choline-Cl_ext: 150 mM choline-Cl, 15 mM CaCl₂, 1 mM MgCl₂, and 10 mM HEPES, adjusted to pH 7.3 with CsOH;

• Cs_ext: 150 mM CsCl, 15 mM CaCl₂, 1 mM MgCl₂, and 10 mM HEPES, adjusted to pH 7.3 with CsOH; and

• NMDG_ext: 146 mM NMDG, 15 mM CaCl₂, 1 mM MgCl₂, and 10 mM HEPES, adjusted to pH 7.3 with HCl.

Cells were maintained at a holding potential of −80 mV, before a 25-ms- (I_Ca-V), or a 10-ms-long (Q_ON-V) square-pulse depolarization (2-s interpulse interval). P/4 leak subtraction was applied. Q_ON recordings were performed as described in Baig et al. (3).

To compare the Q_ON-V of Ca_v1.3_L and Ca_v1.3_42A, we quantified Q_ON (Q_{ON, max}) at V_rev after applying conditional prepulses in 5-mV steps to different potentials as previously described in McDonough et al. (24) and Baig et al. (3). During depolarizations to V_rev, we measured the Q_ON not already moved during the prepulse. This allowed calculation of Q_ON-V from the remaining Q_ON at V_rev after the indicated prepulses (Q_{ON, post-pre}). Using this protocol, we have previously also demonstrated that for Ca_v1.3_L, Q_ON-V is the same when Ca²⁺ in the extracellular solution is replaced by equimolar Mg⁺² (+ 0.5 mM Cd²⁺ + 0.2 mM La³⁺, see solutions above). This revealed no differences in surface charge effects and ruled out that the voltage-dependence of Q_ON is affected by the Mg²⁺-based extracellular solution used to block ionic current for Q_ON-measurements (3).

Steady-state activation (G-V) relationships were derived from I_Ca-V curves. We did not analyze tail current-voltage relationships due to contamination of tail currents by off-gating current in Ca_v1.2 and Ca_v1.3 channels. I_Ca-V of individual experiments was fitted to a modified Boltzmann equation,

$$I=G_{\max }\left(V-V_{\text{rev}}\right)/\left\{1+\exp \left[\left(-\left(V-V_{0.5}\right)\right)/k_{\text{act}}\right]\right\},$$

where V_rev is the extrapolated reversal potential, V is the test potential, I is the peak current amplitude, G_max is the maximum slope conductance, V_0.5 is the half-maximal activation voltage, and k_act is the slope factor. For fitting I_Ca-V curves recorded in the presence of intracellular organic cations (no current reversal observed due to the lack of outward current), data points at test pulses to voltages >30–40 mV were excluded from fitting. Q_ON-V and steady-state activation curves (G-V) were fitted to a Boltzmann equation,

$$G\left(V\right)=G_{max}/\left\{1+\exp \left[\left(-\left(V-V_{0.5}\right)\right)/k_{\text{act}}\right]\right\},$$

where G_max is the saturating value and k_act is the slope factor. Junction potentials were individually calculated for every solution combination, using the software included in the PCLAMP 10.2 software suite (Molecular Devices, Sunnyvale, CA), and offline-subtracted. (Missing ion mobility values in PCLAMP 10.2 were collected from http://web.med.unsw.edu.au/phbsoft/mobility_listings.htm (accessed November 27, 2013)).

Junction potential corrections

• Cs_int versus choline-Cl_ext (−9.3 mV);

• NMDG_int versus choline-Cl_ext (−8.5 mV);

• TEA_int versus choline-Cl_ext (−4 mV);

• Tris_int versus choline-Cl_ext (−5.6 mV);

• Cs_int versus Cs_ext (−2.4 mV);

• Cs_int versus NMDG_ext (−13 mV);

• NMDG_int versus Cs_ext (1.1 mV),

• NMDG_int versus choline-Cl(Mg²⁺)_ext (−8.8 mV); and

• Cs_int versus choline-Cl(Mg²⁺)_ext (−9.5 mV).

Statistics

Data were analyzed with the softwares CLAMPFIT 10.2 (Axon Instruments) and SIGMA PLOT 12 (Systat Software, Chicago, IL). For statistical analysis GRAPHPAD PRISM 5.1 software (GraphPad Software, La Jolla, CA) was used, performing either one-way ANOVA with Bonferroni post-hoc test or Student’s t-test as given. Data are presented as mean ± SE. Significance level was set to α-error lower than p < 0.05 (^∗), p < 0.01 (^∗∗), and p < 0.001 (^∗∗∗).

Results

Ca_v1.3 voltage sensors are highly sensitive to membrane potential changes but are weakly coupled to activation of ion conductance

To compare the voltage-dependence of charge movement (Q_ON-V) with the I_Ca-V of different VGCCs we had to establish experimental conditions for reproducible recordings of Q_ON over a large voltage range. This was achieved by using rat Ca_v1.3 α₁-subunits that express at higher levels than human channels (but with indistinguishable current properties (25)) and by replacing Cs⁺ in our intracellular standard solution with NMDG to prevent contaminating outward currents. Inward currents were blocked by replacing Ca²⁺ with isomolar concentrations of Mg²⁺ and addition of 0.5 mM CdCl₂ and 0.2 mM LaCl₃. Using a prepulse protocol (Methods; see also below), we have previously shown that the voltage-dependence of Q_ON is not affected by this solution exchange (3) and direct comparisons of I_Ca-V and Q_ON-V are possible without corrections for surface charge shifts, as discussed in previous studies (24,26). Representative recordings of I_Ca (left panel) and Q_ON (right panel) for Ca_v1.3_L, Ca_v1.2, and Ca_v3.1 at different test potentials are illustrated in Fig. 1.

Figure 1.

Representative current traces of I_Ca and Q_ON. Representative current traces of I_Ca (left) and Q_ON (right) are shown for Ca_v1.2, Ca_v1.3_L, and Ca_v3.1 at different depolarizing voltages.

As in previous studies with Cs⁺-solution (3,18,25), the long splice variant of Ca_v1.3 (Ca_v1.3_L) became activated at a voltage range that is more negative than Ca_v1.2 under identical experimental conditions with NMDG as the major intracellular cation (Fig. 2 A). Ca_v1.3_L activated with a half-maximal activation voltage (V_0.5(I_Ca)) of ∼10 mV more negative than Ca_v1.2 (Fig. 2 A; for statistics, see Table 1). To test whether this was due to a more refined voltage sensing, we also measured Q_ON-V. Although the threshold voltage for induction of Q_ON was similar for both channels, Ca_v1.3_L displayed a steeper voltage-dependence as evident from the significantly lower slope factor and more-negative half-maximal activation voltage of Q_ON (V_0.5(Q_ON)) (Fig. 2 B; and see Table 1 for statistics). A comparison of conductance (G-V) and Q_ON–V is shown in Fig. 1 C.

Figure 2.

I_Ca and Q_ON voltage-dependence of Ca_v1.2, Ca_v1.3_L, and Ca_v3.1. The n-numbers and detailed statistics are given in Table 1. (A) Normalized I_Ca-V recorded with standard extracellular solution (choline-Cl_ext) and intracellular NMDG (NMDG_int). I_Ca-V was fitted to a modified Boltzmann function, as described in Methods. (B) Normalized Q_ON-V of Ca_v1.2, Ca_v1.3_L, and Ca_v3.1 recorded with choline-Cl(Mg²⁺)_ext/NMDG_int. (C) Comparison of the voltage-dependence of Q_ON (solid line, Boltzmann fit from data in panel B) and conductance (G, calculated from data shown in panel A). (D) Normalized G/G_max in comparison to normalized Q_ON/Q_{ON, max} measured at similar voltages (difference = 0.3 mV) to illustrate the fraction of observed gating charge required for activation of conductance. (Dashed line) Half-maximal conductance. Data points were obtained from the experiments given in Table 1. Pooled data were fitted to a sigmoid function to determine the fractional Q_ON at half-maximal G. Values were significantly different among the three channels (determined by sum-of-squares F-test from GRAPHPAD PRISM, GraphPad Software).

Table 1.

Biophysical properties of VGCCs α₁-subunits

choline-Clext / NMDGint	V0.5(ICa) [mV]	Slope (ICa) [mV]	n
Cav1.2	−3 ± 2	8.6 ± 0.6	10
Cav1.3L	−13 ± 1 ∗∗∗	6.4 ± 0.3 ∗∗	14
Cav1.342A	−20 ± 2 ∗∗∗, †	5.6 ± 0.3 ∗∗∗	17
Cav3.1	−40 ± 2 ∗∗∗, †††, ‡‡‡	3.8 ± 0.3 ∗∗∗, †††, ‡‡	12
choline-Clext/ Csint
Cav1.2	14 ± 2 §§§	11.2 ± 0.5 §§	28
Cav1.3L	−4 ± 1 §§§	8.8 ± 0.2 §§§	19
Cav3.1	−35 ± 1 §	5.3 ± 0.2 §§§	11
choline-Cl(Mg2+)ext/ NMDGint	V0.5(QON) [mV]	Slope (QON) [mV]	n
Cav1.2	−3 ± 2	22.6 ± 1.3	10
Cav1.3L	−24 ± 2 ∗∗∗	11.6 ± 0.9 ∗∗∗	13
Cav3.1	−6 ± 2 †††	19.9 ± 0.9 †††	10
choline-Cl(Mg2+)ext/ NMDGint	V0.5(QON) at Vrev[mV]	Slope (QON) at Vrev[mV]
Cav1.3L	−29 ± 3	10.6 ± 1.9	12
Cav1.342A	−27 ± 2	11.1 ± 0.7	10

Parameters (mean ± SE) were obtained by fitting data of I_Ca-V relationships or by fitting data of Q_ON-V relationship, as described in methods. Q_ON-V was either determined by measuring Q_ON during pulses to different voltage steps or by measuring Q_ON at V_rev after prepulses to different voltages (V_0.5(Q_ON) at V_rev). Statistical significances are indicated for comparison vs. Ca_v1.2 (^∗, ^∗∗, ^∗∗∗), vs. Ca_v1.3_L (†, ††, †††), and vs. Ca_v1.3_42A (‡, ‡‡, ‡‡‡) (one-way ANOVA with Bonferroni post-hoc test). §, §§, §§§ indicate statistical significance for intra-construct comparisons of parameters obtained with intracellular NMDG vs intracellular Cs (e.g. Ca_v1.2 choline-Cl_ext/NMDG_int vs. Ca_v1.2 choline-Cl_ext/Cs_int) (Student’s t-test).

As a measure for the efficiency of voltage-sensor charge movement to pore opening, we plotted the fraction of total observed Q_ON (Q_ON/Q_{ON, max}), required for the activation of G (G/G_max) at each voltage (Fig. 2 D). Fitting the data to a sigmoid function revealed that significantly less total Q_ON (52%, n = 10) was required to activate half-maximal conductance of Ca_v1.2 than of Ca_v1.3_L (68%, n ≥ 13) (see legend to Fig. 2 D). These data indicate weaker coupling of voltage-sensor movements to pore opening in Ca_v1.3 as compared to Ca_v1.2. However, the more-sensitive Q_ON charge movement still results in activation at voltages that are more negative than Ca_v1.2. Low-VGCCs (T-type; Ca_v3 family (27)) are known to activate at significantly lower voltages than Ca_v1.3_L (28,29), also shown in our direct comparison with Ca_v1.3_L (Fig. 2 A). This can be explained by the finding that only a small fraction of total Ca_v3.1 Q_ON (22%, n ≥ 9; Fig. 2, B–D; and see Table 1) was required for half-maximal activation of conductance. This allows Ca_v3.1 I_Ca to activate at much lower voltages than Cav1.3 despite its more positive Q_ON-V, which is similar to Ca_v1.2 (Fig. 2 B).

Our data show that the voltage-sensing machinery of Ca_v1.3 responds more sensitively to depolarizing stimuli than Ca_v1.2 and even Ca_v3.1. Despite its weaker coupling to pore opening, this allows Ca_v1.3_L to carry I_Ca at voltages that are more negative than for Ca_v1.2.

Alternative splicing affects Q_ON coupling to pore opening of Ca_v1.3 LTCCs

We next investigated whether the more-negative I_Ca-V previously observed for naturally occurring C-terminally short splice variants (such as Ca_v1.3_42A) as compared to the long Ca_v1.3_L splice variants (18,19) is attributable to more refined voltage sensing or more efficient pore coupling. A negative shift was also observed when intracellular NMDG was used instead of Cs⁺ (Fig. 3, A and C) as in previous studies (18,19). The V_0.5(I_Ca) for Ca_v1.3_42A was 7 ± 2 mV more negative (Fig. 3 A; for statistics, see Table 1) and inactivation of I_Ca was faster (due to more pronounced CDI as demonstrated in previous work (18)). Measuring Q_ON for this short splice variant was more difficult than for Ca_v1.3_L due to small (presumably Mg²⁺) inward currents contaminating the measurement of Q_ON at voltages at ∼V_max, but not at V_rev. We therefore determined Q_ON-V at V_rev (Fig. 3 B) by measuring Q_ON that remained after applying conditioning prepulses to different voltages (see Methods).

Figure 3.

I_Ca and Q_ON voltage-dependence of Ca_v1.3_L in comparison to Ca_v1.3_42A. The n-numbers and detailed statistics are given in Table 1. (A) I_Ca-V of Ca_v1.3_L and Ca_v1.3_42A recorded with standard recording solutions (choline-Cl_ext/NMDG_int). Fits were generated as described in Fig. 2A. (B) Q_ON-V measured for Ca_v1.3_L (n = 12) and Ca_v1.3_42A (n = 10) using the prepulse protocol (inset) as described in Methods with Mg²⁺-containing solution. (Inset) Example traces for both constructs obtained by depolarization to the reversal potential after a prepulse to −28.8 mV, which causes partial movement of Q_ON. The value Q_ON after the prepulse (Q_{ON, post-pre}) was normalized to Q_{ON, max}. For Ca_v1.3_L the same gating parameters were obtained using this and the protocol in Fig. 2B. (C) Comparison of the Q_ON-V and G-V relationships of Ca_v1.3_L and Ca_v1.3_42A. (Solid and dashed lines) Boltzmann fits of Q_ON-V data obtained from experiments illustrated in panel B (dotted line is the same as in Fig. 2C (Ca_v1.3_L) for comparison). (D) Data representation as in Fig. 2D. Data points were obtained from the experiments given in Table 1. Pooled data were fitted to a sigmoid function to determine the fractional Q_ON at half-maximal G. Values were significantly different between the two splice variants (determined by sum-of-squares F-test using GRAPHPAD PRISM, GraphPad Software).

Such prepulses moved part of Q_ON in a voltage-dependent manner and allowed calculation of Q_ON-V from the remaining Q_ON at V_rev after the indicated prepulses (Q_{ON, post-pre}). This protocol has originally been described by McDonough et al. (24) and was validated by us previously (3) and in this study (see legend to Fig. 3 B). Despite the more-negative V_0.5(I_Ca) for Ca_v1.3_42A, no significant difference in the Q_ON-V was observed between long and short Ca_v1.3 constructs (Fig. 3 B; for statistics, see Table 1). G-V curves for both splice variants are shown in Fig. 3 C in relation to their Q_ON-V (for analysis, see Fig. 2 D). Half-maximal G of Ca_v1.3_42A required significantly less Q_ON than the long isoform (68 vs. 79%, n ≥ 10) (Fig. 3 D). Taken together, these findings demonstrate that the presence of the intramolecular protein interaction forming the CTM in Ca_v1.3 LTCCs modulates the I_Ca-V activation of Ca_v1.3 by reducing the coupling efficiency between voltage sensing and pore opening.

Cation composition strongly affects voltage-dependent VGCC gating

Comparison of our I_Ca-V relationships measured using intracellular NMDG (Figs. 2 and 3) with our previously published data employing intracellular Cs⁺ revealed an ∼10 mV shift of V_0.5(I_Ca) toward voltages that were more negative for Ca_v1.3_L (Fig. 4 A). In these experiments, an identical extracellular solution with choline as the major extracellular cation, and 15 mM Ca²⁺ as the charge carrier, was employed (3,9,18,19,25) (Fig. 4 A). We corrected all data for junction potentials precisely calculated for all our solutions as described in the Methods, ruling out differences in junction potentials as an explanation for this difference. To further characterize this unexpected finding, we also recorded Ca_v1.2 and Ca_v3.1 I_Ca-V relationships with NMDG (NMDG_int) or Cs⁺ containing intracellular solution (Cs_int). This revealed a negative shift also for these channel types (Table 1) and thus ruled out a Ca_v1.3-specific effect of NMDG. Next, we tested whether the effect was mimicked by other large organic cations. Tris and TEA caused a similar negative shift of I_Ca-V like NMDG as shown for Ca_v1.3_L (Fig. 4 D). We also quantified effects on Q_ON for Ca_v1.3_L. Intracellular NMDG also shifted Q_ON-V by approximately the same extent as I_Ca-V (Fig. 4 B, inset; Ca_v1.3_L: n = 12; Ca_v1.3_42A: n = 9; p = 0.02, Student’s t-test), suggesting that the NMDG effect is due to voltage-sensing that is more refined rather than by pore-coupling (Fig. 4 C).

Figure 4.

Effect of intracellular cations on Ca_v1.3_L voltage-dependence. (A) I_Ca-V of Ca_v1.3_L recorded with either Cs_int or NMDG_int. (For comparison, I_Ca-V of Ca_v1.3_L recorded with NMDG_int is shown as a line, as taken from Fig. 2A.) Fits were generated as described in Fig. 2A. For statistics, see panel D. (B) Q_ON-V of Ca_v1.3_L, recorded at V_rev as in Fig. 3B with either Cs⁺ (Cs_int) or NMDG (NMDG_int) as the major intracellular cation. For comparison, the Q_ON-V of Ca_v1.3_L in NMDG_int is illustrated (line taken from Fig. 3B). (Inset) Statistical comparison of the V_0.5 values (Cs_int: −20 ± 1 mV, n = 9; NMDG_int: −29 ± 3 mV, n = 12; p = 0.022, Student’s t-test). (C) Normalized G/G_max to normalized Q_ON/Q_ON,max (replotted from Fig. 3D as 1-Q_{ON, post-pre}/Q_{ON, max}). (D) V_0.5(I_Ca) values for Ca_v1.3_L, recorded with different internal and external cation-based solutions, as indicated (for solution composition, see Methods) (n ≥ 9). For calculation of statistical significance, one-way ANOVA with Bonferroni post-hoc test was performed (p < 0.05, ^∗; p < 0.005, ^∗∗; p < 0.001, ^∗∗∗).

To test whether changes in the voltage-dependence of gating are also observed by corresponding changes of the extracellular solution, we exchanged choline-Cl (in our standard extracellular solution) by either CsCl (Cs_ext) or NMDG-Cl (NMDG_ext), closely resembling the changes in the internal solutions. The V_0.5(I_Ca) values obtained with the various combinations of intra- and extracellular solutions are illustrated in Fig. 4 D. When intracellular Cs⁺ was present, the replacement of extracellular choline by NMDG or Cs⁺ did not cause a change in V_0.5(I_Ca). Intracellular NMDG also caused a negative shift with extracellular Cs⁺, as observed with extracellular choline. Taken together, our data revealed that intracellular but not extracellular organic cations can enhance the coupling efficiency of Ca²⁺ channel voltage sensors to membrane depolarization.

Discussion

Our study was motivated by two previous observations regarding Ca_v1.3 function:

• 1.Ca_v1.3_L channels were previously reported to activate at lower voltages than Ca_v1.2. This special feature has been discovered in heterologous expression studies (9,10) but was subsequently confirmed for native Ca_v1.3 currents in sinoatrial node cells (6), cochlear inner hair cells (2), and adrenal chromaffin cells (7). It allows Ca_v1.3 channels to sustain subthreshold inward currents and thus serve as a pacemaker channel in the sinoatrial node and chromaffin cells and shape firing patterns of neurons (12).
• 2.C-terminal splicing removes a CTM (11,18,19), which can further shift the channel's I_Ca-V to more-negative voltages in short splice variants.

Because these splice variants are expressed in a tissue-dependent manner, it is likely that they contribute to the fine-tuning of Ca_v1.3 channel activity in different tissues. Here we present data showing that the voltage sensors of Ca_v1.3 respond more readily to depolarizing stimuli than those of Ca_v1.2 and Ca_v3.1. This ensures that despite the weaker coupling of voltage sensing to pore opening, Ca_v1.3 currents can activate at lower membrane potentials than Ca_v1.2. Moreover, alternative splicing enhances the efficiency of coupling between charge movement and pore opening, explaining the even lower activation voltage range of naturally occurring short splice variants lacking the CTM.

Based on contemporary structural models of the voltage-gated cation channel family, mainly derived from x-ray structures of voltage-gated K⁺- (30) and bacterial Na⁺-channels (31), a negative shift in the voltage-dependence of channel conductance (G-V) may, in principle, result via two possible mechanisms:

• 1.Values of Q_ON-V that are more negative. Even if the efficiency of voltage-sensor coupling to the pore remains unchanged, this should shift G-V to voltages that are more negative. An example is the deletion of a ‘gating brake’ in T-type channel α₁-subunits. This causes a negative shift and steeper voltage-dependence (slope) of the Q_ON-V relationship (29) paralleled by a corresponding negative shift in G-V (32).
• 2.Q_ON-V is unaltered but the efficiency of coupling to pore opening is enhanced. This has been reported for the LTCC activators FPL64176 and BayK8644, which induce strong changes in the kinetics and gating of Ca_v1.2 currents (24,33), including a shift of the I_Ca-V relationship to potentials that are more hyperpolarized (33). However, they do not affect the voltage-dependence of Q_ON (24,33).

Here we clearly demonstrate that the more-negative activation range of Ca_v1.3 as compared to Ca_v1.2 is not due to a more efficient coupling. Instead, coupling is even weaker, as evident from a higher fractional Q_ON required for V_0.5(I_Ca). However, Q_ON of Ca_v1.3 displayed a steeper voltage-dependence and thereby still permits a more-negative I_Ca-V. We show that this is in contrast to the mechanism imposed by alternative splicing. We found that in the absence of the CTM, the Q_ON-V does not change but more channel activation (i.e., fractional conductance) is observed at a given percentage of maximal Q_ON. Similar to the findings obtained with FPL64176 (24), this can be interpreted as more efficient coupling between voltage-sensor movements and pore opening.

It is unclear how the CTM can moderate this coupling. Molecular studies using mutant and chimeric channel constructs will be difficult to perform because this modulatory domain is part of a larger structure consisting not only of the two noncovalently interacting putative α-helices (PCRD, DCRD) of the modulatory domain itself but also of channel-bound CaM (18,34). Based on our observations, it is most likely that the CTM targets the interaction of pore-forming helices (primarily S6) with the S4-S5 linkers that are considered the main structural determinants of Q_ON-V to pore opening (22,23,35,36). Another possibility is interference of the distal C-terminus with the channels II-III linker. Such an interaction, which is also modulated by A-kinase anchoring proteins, has recently been described in Ca_v1.2 α₁-subunits (37).

What structural differences mediate the steeper voltage-dependence of Ca_v1.3 and the higher coupling efficiency of Ca_v3.1? The amino-acid sequence within the S4 segments including the positive charges of Ca_v1.3 and Ca_v1.2 are highly conserved (see alignment in Fig. S1 in the Supporting Material) and are unlikely to explain the difference in their Q_ON-V relationships. Structural features outside S4 must therefore play a crucial role. Mutations in S6 segments forming the activation gate or the S5-S6 linker, which couples voltage-sensors to the gate, can induce Q_ON-V shifts to voltages that are more negative. This has been reported by us (for Ca_v1.3 (3)) and others (for Ca_v3.2 and Ca_v2.3 (35,38)). Interdomain cytoplasmic linkers (shown for N-terminal regions of the I-II loop in the case of Ca_v3 channels) can also affect the voltage-dependence of Q_ON-V by serving as a ‘gating brake’ (29). The high sequence similarity of Ca_v1.3 α₁ subunits with Ca_v1.2 also outside S4 regions provides an excellent opportunity to identify the structural determinants accounting for its uniquely steep voltage-dependence using chimeric approaches.

The high coupling efficiency of Ca_v3.1 channels is also not readily explained by charge differences in the S4 segments (see Fig. S1). Assuming that all four voltage sensors have to move completely for pore opening, the fact that only ∼25% of Q_ON charge are moved when conductance is already fully activated (Fig. 2) could indicate that the activation of only one of the four voltage sensors is sufficient for activating the channel gate.

During the course of our studies, we also discovered that intracellular organic cations sensitize voltage-responses of all three VGCCs investigated. A shift to more negative I_Ca-V was initially observed when intracellular Cs⁺ was replaced by NMDG, but was also found for Tris and TEA. In contrast to splicing, NMDG affected Q_ON-V with no major change in pore coupling, suggesting that it primarily affects voltage sensing itself. This also demonstrates that ion permeation is not required for this voltage shift. This modulation is unlikely to have been caused by a high affinity interaction with NMDG; it was not observed when only 15 mM of Cs⁺ were replaced by NMDG (n = 4, not shown).

To investigate the possibility of passive charge screening effects, we measured changes in Ca_v1.3_L gating with all possible combinations of equimolar concentrations of NMDG and Cs⁺ in the intra- and extracellular solutions. Considerably less is known about passive-charge-screening effects of organic cations as compared to mono- or divalent inorganic cations (39–43). Therefore, although unlikely, the possibility of passive charge screening effects cannot be completely excluded. Alternatively, intracellular organic cations may somehow more specifically interfere with the gating apparatus. An example has previously been described for K_v1.2 channels (44) by showing that internal cations are able to occupy the inner cavity of the open channel. Thereby they prevent closing of the inner pore gate and stabilize the open state of the voltage sensors (44). This was not associated with a shift in Q_ON-V, but caused a slowing of off-gating currents, especially with such larger cations as NMDG and TEA. However, we were not able to detect a major off-gating current-stabilizing effect of NMDG on Ca_v1.3_L (n ≥ 9, not shown) or a slowing of I_Ca deactivation kinetics.

Independent from the molecular mechanism, our findings clearly emphasize that recording buffer compositions have to be considered when comparing biophysical parameters across different studies examining the voltage-dependent gating properties of VGCCs.