Splicing of an automodulatory domain in Ca_v1.4 Ca²⁺ channels confers distinct regulation by calmodulin

Voltage-gated Ca_v Ca²⁺ channels are essential regulators of Ca²⁺ signaling and are composed of a pore-forming α₁ subunit and auxiliary β and α₂δ subunits. The pharmacological and biophysical properties of Ca_v subunits are largely determined by the α₁ subunit (Simms and Zamponi, 2014). In contrast to the tremendous diversity of genes encoding voltage-gated K_v K⁺ channels, the Ca_v α₁ subunit in mammals is encoded by only 10 different genes (Yu et al., 2005). However, each α₁ gene can give rise to thousands of splice variants, the exact complement of which depends on the types of splicing factors that are expressed (Lipscombe et al., 2013). Combined with the splice variation affecting β and α₂δ subunits (Buraei and Yang, 2013; Dolphin, 2013), the vast molecular diversity of α₁ splice isoforms helps tailor Ca_v channel properties according to a defined cellular and/or developmental context.

Among the cytoplasmic regions of the Ca_v α₁ subunit, the C-terminal domain (CTD) is the largest and subject to significant splice variation that can alter channel function. For example, the inclusion of either of two mutually exclusive exons (37a and 37b) in the proximal C-terminal domain (pCTD) of Ca_v2.2 affects the voltage dependence of channel inhibition by G protein–coupled receptors (Raingo et al., 2007), and the role of Ca_v2.2 in mediating the analgesic effects of morphine in spinal nociceptors (Andrade et al., 2010). Alternative splicing of the analogous exons in Ca_v2.1 (P/Q-type) channels as well as an exon in the distal CTD transforms the Ca²⁺-sensitivity of Ca²⁺/calmodulin (CaM)-dependent facilitation of these channels (Chaudhuri et al., 2004). The expression of these Ca_v2.1 splice variants is developmentally regulated in ways that may support the maturation of the firing properties of cerebellar Purkinje neurons (Chaudhuri et al., 2005).

In the retina, Ca_v1.4 L-type channels are localized in the synaptic terminals of rod and cone photoreceptors, where they mediate the tonic release of glutamate necessary for the encoding of light responses (McRory et al., 2004; Mansergh et al., 2005; Specht et al., 2009; Liu et al., 2013). Unlike Ca_v1.2 and Ca_v1.3 channels, Ca_v1.4 exhibits little Ca²⁺-dependent inactivation (CDI) in heterologous expression systems (Baumann et al., 2004; McRory et al., 2004; Singh et al., 2006; Wahl-Schott et al., 2006). CDI is a hallmark feature of most Ca_v1 and Ca_v2 channels and depends on CaM binding to a consensus IQ domain in the pCTD of the channel (Ben-Johny and Yue, 2014). CDI of Ca_v1.4 is blunted by a CTM that competes with CaM interactions with the channel (Singh et al., 2006; Liu et al., 2010; but see Wahl-Schott et al., 2006; Griessmeier et al., 2009).

Numerous Ca_v1.4 splice variants have been identified in human retina with variations affecting the CTM. For example, Ca_v1.4 variants containing the alternately spliced exon 43 (Ca_v1.4 + ex43) are prematurely truncated, resulting in removal of the CTM. Like channels in which the entire CTM is deleted (Singh et al., 2006; Wahl-Schott et al., 2006), Ca_v1.4 + ex43 exhibits robust CDI as well as a negative shift in the half-maximal voltage (V_h; Tan et al., 2012) of channel activation. Strong CDI and a hyperpolarized V_h are also caused by a mutation (K1591X) associated with congenital stationary night blindness type 2 (CSNB2; Strom et al., 1998), which deletes the CTD distal to the IQ domain, including the CTM (Singh et al., 2006). The physiological significance of naturally occurring splice variants with properties similar to disease-causing mutant channels is unclear.

We previously characterized a splice variant of Ca_v1.4 L-type channels lacking exon 47 (Ca_v1.4Δex47), which encodes part of the CTM. Ca_v1.4Δex47 is highly expressed in human retina and exhibits prominent CDI as well as a negative shift in V_h (Haeseleer et al., 2016). These properties suggest that deletion of exon 47 might enable CaM regulation by altering or preventing the CTM interaction with the pCTD, but mechanistic details are lacking. Moreover, it is not known how deletion of exon 47 promotes voltage-dependent activation of Ca_v1.4Δex47. Considering that alternative splicing of Ca_v channels can affect the functional impact of disease-causing mutations (Adams et al., 2009), a thorough analysis of the properties of Ca_v1.4Δex47 is important for understanding how CACNA1F mutations may cause vision impairment in humans.

Here, we report a dual role for CaM in regulating both CDI and voltage-dependent activation of Ca_v1.4Δex47. In addition, we demonstrate distinct routes whereby the individual Ca²⁺ binding lobes of CaM regulate K1591X and Ca_v1.4Δex47. Our findings highlight the versatility of CaM as a regulator of Ca_v channels, and how modulation by CaM can be affected by naturally and pathologically occurring variations in the channel protein.

The following cDNAs were used: Ca_v1.4 (GenBank no. AF201304), β_{2 × 13} (GenBank no. AF465485), and α₂δ₄ (GenBank no. NM_172364) in pcDNA3.1 (Lee et al., 2015); CaM deficient in Ca²⁺ binding to the N-terminal lobe (N lobe; CaM₁₂), CaM deficient in Ca²⁺ binding to the C-terminal lobe (C lobe; CaM₃₄), CaM deficient in Ca²⁺ binding to both the N lobe and C lobe (CaM₁₂₃₄; GenBank no. NM_017326.3) in pcDNA6V5-His (Lee et al., 2003); β_2a (GenBank no. M80545.1). In Ca_v1.4Δex47, alanine substitutions at residues I1588A, Q1589A, D1590A, Y1591A, and F1592A (5A), and F1582A, Y1583A, and F1586A (3A), were generated with primers incorporating the mutations (5′-CTACGCCACATTTCTGGCCGCGGCCGCTGCCCGCAAATTCCGG-3′ for 5A, 5′-GTCACCGTGGGCAAAGCCGCCGCCACAGCTCTGATCCAGGACTATTTCCGC-3′ for 3A, respectively) using the QuikChange Lightning Multi Site-directed Mutagenesis kit (Agilent) according to the manufacturer’s protocol. The 3A mutations were generated in K1591X-3A with the HiFi DNA Assembly Cloning System (New England Biolabs) using K1591X as a PCR template and a gBlock containing the 3A mutations (Integrated DNA Technologies). All constructs were verified by DNA sequencing before use.

Human embryonic kidney 293 T cells (CRL-3216, Research Resource Identifier; CVCL_0063, American Type Culture Collection) were cultured in Dulbecco’s modified Eagle’s medium (Life Technologies) with 10% FBS (Atlantic Biologicals) at 37°C in 5% CO₂. At 70–80% confluence, the cells were cotransfected with cDNAs encoding human Ca_v1.4 α₁ (1.8 µg; Ca_v1.4 + ex47, Ca_v1.4Δex47, K1591X, or 5A or 3A mutants), β_2×13 (0.6 µg), α₂δ₄ (0.6 µg), and enhanced GFP in pEGFP-C1 (0.1 µg) using FuGENE 6 transfection reagent (Promega) according to the manufacturer’s protocol. For some experiments, cells were cotransfected with cDNAs encoding CaM₁₂, CaM₃₄, or CaM₁₂₃₄ (1 µg each) or pcDNA3.1 (1 µg) as a control. Cells treated with the transfection mixture were incubated at 37°C for 24 h. After 24 h, cells were incubated at 30°C for at least 24 h before whole-cell patch clamp recordings.

Whole-cell patch clamp recordings were performed at room temperature between 48 and 72 h after transfection with an EPC-9 patch clamp amplifier operated by either Patchmaster software (HEKA Elektronik). External recording solutions consisted of (in mM) Tris (140), CaCl₂ or BaCl₂ (20), and MgCl₂ (1). Internal recording solution consisted of (in mM) NMDG (140), HEPES (10), MgCl₂ (2), Mg-ATP (2), and EGTA (5). The pH of external and internal recording solutions was adjusted to 7.3 with methanesulfonic acid. Pipette resistances were typically 2–4 MΩ in the bath solution, and series resistance compensated up to 70%. Leak subtraction was conducted using a P/−4 protocol.

To measure current density, Ca²⁺ and Ba²⁺ currents (I_Ca, I_Ba) were evoked by 50-ms pulses from −80 mV to various voltages and normalized to the cell capacitance. To characterize voltage-dependent activation, currents were evoked by 10-ms steps from −80 mV to various voltages. Tail currents were measured upon repolarization to −60 mV for 2 ms. For I_Ca, tail current amplitudes were normalized to that at +60 mV. Because of a decline in the amplitude of I_Ba tail currents at positive voltages, tail currents for I_Ba were normalized to that at +20 mV. I–V data were fitted with single or double Boltzmann equations: I = I_max / [1 + exp (V − V_h) / k)] or I = I_max / [1 + exp (V − V_h1) / k₁] + I_max / [1 + exp (V − V_h2) / k₂], where I_max is the maximal current, V is the test voltage, V_h is the half-maximal activation, and k is the slope factor. To measure CDI, currents were evoked by 1-s depolarizations from −80 mV to various voltages. Fractional I_Ca or I_Ba was defined as the residual current amplitude at the end of the pulse normalized to the peak current amplitude. Fractional CDI was the difference in fractional I_Ca at −20 mV and mean fractional I_Ba at −20 mV, except where indicated. Kinetic parameters for I_Ca inactivation were obtained by fitting with a double exponential function (y₀ + A_fast [exp (−t/τ_fast)] + A_slow [exp (−t/τ_fast)), where y₀ is the offset (asymptote), τ_fast and τ_slow are the time constants, and A_fast and A_slow are the relative amplitudes of fast and slow components of inactivation.

Data were analyzed offline with Igor Pro software (Wavemetrics). Statistical analysis was performed using SigmaPlot (Systat Software Inc.) or GraphPad Prism software. The data were initially analyzed for normality using the Shapiro–Wilk or D’Agostino–Pearson omnibus test. For parametric data, significant differences were determined by Student’s t test or ANOVA with post hoc Dunnett or Tukey test. For nonparametric data, Kruskal–Wallis and post hoc Dunn’s tests were used. Data were incorporated into figures using GraphPad, SigmaPlot, and Adobe Illustrator software. All averaged data represent the mean ± SEM from at least three independent transfections.

Fig. S1 shows that CaM mutants do not impact I_Ca density or voltage-dependent activation of Ca_v1.4 + ex47. Fig. S2 shows the lack of CDI in a subset of cells transfected with Ca_v1.4Δex47-3A in which very slow activation was observed.

Because Ca_v1.4Δex47 is expressed in human but not mouse retina (Lee et al., 2015), we focused on Ca_v1.4 channels containing auxiliary Ca_vβ and α₂δ subunits that predominate in the human retina—β_2×13 (a splice variant of Ca_vβ₂) and α₂δ₄ (Lee et al., 2015). We compared the properties of Ca_v1.4 + ex47 and Ca_v1.4Δex47 (Fig. 1 A) cotransfected with β_2×13 and α₂δ₄ in whole-cell patch clamp recordings of human embryonic kidney 293 T cells. Due to our interests in comparing CDI, voltage protocols were performed with either Ca²⁺ or Ba²⁺ as the permeant ion. In agreement with our previous study (Haeseleer et al., 2016), I_Ca mediated by Ca_v1.4Δex47 activated at more negative voltages than Ca_v1.4 + ex47. While this was also seen for Ba²⁺ currents (I_Ba), the doubling in current density for I_Ca in cells transfected with Ca_v1.4Δex47 was not observed for I_Ba (Fig. 1, B and C; and Table 1). Since parameters of voltage-dependent activation obtained from I–V relationships may be affected by the change in driving force that occurs with different test voltages, we also analyzed the voltage dependence of the tail current measured upon repolarization from the test voltage to a fixed voltage (−60 mV). The amplitude of the tail current reflects the number of channels open during the test pulse and rises sigmoidally with the test voltage. Boltzmann fits of these data indicated a significant hyperpolarizing shift in the activation curve, and steepening of the slope (k), for Ca_v1.4Δex47 compared with Ca_v1.4 + ex47. While these alterations in tail I–V curves were found for both I_Ca and I_Ba (Fig. 1, B and C; and Table 2), the difference in V_h and k values between Ca_v1.4Δex47 and Ca_v1.4 + ex47 was significantly larger for I_Ca than for I_Ba (ΔV_h = −15 ± 1.0 mV for I_Ca versus ΔV_h = −10 ± 1.1 mV for I_Ba; P = 0.003 by t test; Δk = −3 ± 0.5 for I_Ca versus Δk = −1 ± 0.2 for I_Ba; P < 0.001 by t test). Therefore, in addition to enhancing voltage-dependent activation of I_Ba, deletion of exon 47 causes a Ca²⁺-dependent enhancement of activation (CDEA) of I_Ca.

To investigate the mechanism underlying CDEA of Ca_v1.4Δex47, we considered CaM because of its role in Ca²⁺-dependent modulation of Ca_v channels (Ben-Johny and Yue, 2014). For Ca_v1.2, Ca²⁺ binding to the pairs of EF-hand motifs in the C lobe and N lobe of CaM is required for CDI (Peterson et al., 1999) and CDF (Van Petegem et al., 2005), respectively. CaM mutants deficient in Ca²⁺ binding to the N lobe (CaM₁₂) or C lobe (CaM₃₄) have been instrumental in revealing CaM lobe-specific modulation of Ca_v1 and Ca_v2 channels (Ben-Johny and Yue, 2014). To determine if CaM is involved in CDEA, we cotransfected Ca_v1.4Δex47 with either CaM₁₂ or CaM₃₄. While CaM₃₄ did not affect Ca_v1.4Δex47 current density, or cause alterations in I–V or tail I–V parameters that were specific for I_Ca (Fig. 2; and Tables 1 and 2), there was a significant decrease in current density and positive shift in V_h in the I–V curve in cells cotransfected with Ca_v1.4Δex47 and CaM₁₂ for I_Ca, and these alterations were not observed for I_Ba (Fig. 3, A and B; and Table 1). Coexpression of CaM₁₂ also caused a positive shift in the tail I–V relationship only for I_Ca and not I_Ba (Fig. 3 C). Unlike for Ca_v1.4Δex47 transfected alone or with CaM₃₄ (Table 2), the tail I–V for I_Ca in cells cotransfected with CaM₁₂ was best fit with a double Boltzmann function (Fig. 3 D and Table 3) with a decline in channel availability over voltages that paralleled the enhanced activation of I_Ca mediated by Ca_v1.4Δex47 compared with Ca_v1.4 + ex47 (Fig. 1 B). There was also no impact of either CaM₁₂ or CaM₃₄ on the I–V or tail I–V relationship for Ca_v1.4 + ex47 (Fig. S1), demonstrating that this effect of CaM₁₂ is specific for channels lacking exon 47. Collectively, our results show that deletion of exon 47 not only enhances voltage-dependent activation of I_Ba but also enables CDEA that involves the N lobe of CaM.

In Ca_v1.4, CaM binds to a consensus IQ domain in the pCTD of Ca_v1 channels (see Fig. 5 A). Mutation of the initial 5 amino acids within this domain (IQDYF) to alanine abolishes the binding of CaM to Ca_v1.4 (Griessmeier et al., 2009). Assuming that the 5A mutations would also disrupt CaM modulation of Ca_v1.4Δex47, we introduced these mutations (I1588A, Q1589A, D1590A, Y1591A, and F1592A) into Ca_v1.4Δex47 (Ca_v1.4Δex47-5A; Fig. 4 A) and tested their involvement in CDEA. In Boltzmann fits of the tail I–V curves for Ca_v1.4Δex47-5A, V_h was significantly more positive, and k shallower, than that for Ca_v1.4Δex47. These alterations were seen for I_Ca and not for I_Ba mediated by Ca_v1.4Δex47-5A (Fig. 4, B and C; and Table 2), indicating the importance of the IQ domain for CDEA of I_Ca but not voltage-dependent activation of I_Ba. To further probe the molecular determinants underlying CDEA, we focused on three aromatic residues that were shown to mediate CaM N lobe interactions with Ca_v1.2 (F1618A, Y1619A, F1622A; Van Petegem et al., 2005). Although there are differences in the residues surrounding the IQ domain of Ca_v1.4 and Ca_v1.2, these residues are conserved in Ca_v1.4 (F1582A, Y1583A, and F1586A; Fig. 5 A). Therefore, we reasoned that mutations of these residues might disrupt CaM N lobe regulation of CDEA. Consistent with this possibility, the 3A mutations in Ca_v1.4Δex47 (Ca_v1.4Δex47-3A) reversed the negative shift in V_h for I_Ca (Fig. 5 B; and Tables 2 and 3). Moreover, the tail I–V relationship of I_Ca for Ca_v1.4Δex47-3A was best fit by a double Boltzmann equation (Fig. 5 C) with no significant differences in fit parameters compared with those obtained for Ca_v1.4Δex47 cotransfected with CaM₁₂ (Table 3). Compared with Ca_v1.4Δex47, the positive shift in the tail I–V curve for I_Ca mediated by Ca_v1.4Δex47-5A (Fig. 4 B) was weaker than that for Ca_v1.4Δex47-3A (Fig. 5 B), perhaps due to residual CaM N lobe modulation of Ca_v1.4Δex47-5A. There was no effect of either 5A or 3A mutations on I_Ba, which further suggested that the enhanced voltage-dependent activation of I_Ba relies on distinct molecular determinants as compared with CDEA. Collectively, our results demonstrate that CaM N lobe mediates CDEA of Ca_v1.4Δex47, which requires F1582A, Y1583A, and F1586A.

In contrast to Ca_v1.4 + ex47, Ca_v1.4Δex47 undergoes CDI that is evident as stronger inactivation of I_Ca as compared with I_Ba (Haeseleer et al., 2016). Previous analyses of Ca_v1.3 channels revealed the importance of the C lobe of CaM in mediating a rapid phase of CDI, whereas the N lobe mediates a slower component (Yang et al., 2006). To determine if this was also true for Ca_v1.4Δex47, we tested the impact of CaM mutants on CDI of Ca_v1.4Δex47. Inactivation of I_Ca and I_Ba was measured as the current amplitude at the end of a 1-s pulse normalized to the peak current amplitude (fractional I). Fractional Ca²⁺-dependent inactivation was defined as the difference in fractional I_Ca and fractional I_Ba (F_CDI). In cells transfected with Ca_v1.4Δex47 alone, I_Ca inactivation was greatest (smallest fractional I) at voltages on the rising phase of the I–V relationship (from −20 to 0 mV; Fig. 1 C), consistent with a role for Ca²⁺ influx during the test pulse in promoting CDI. In contrast, I_Ba showed little inactivation across all voltages tested (Fig. 6 A). In cells cotransfected with CaM₁₂ or CaM₃₄, Fractional I_Ca was still smaller than that for I_Ba, but to a lesser extent than in cells transfected with Ca_v1.4Δex47 alone, such that F_CDI was significantly weaker with CaM₁₂ and CaM₃₄ than with Ca_v1.4Δex47 alone (Fig. 6, B and C; and Table 4). To verify the importance of both lobes of CaM in regulating CDI, we analyzed the effects of a CaM mutant with both N and C lobes disabled (CaM₁₂₃₄). There was no difference in fractional I_Ca and I_Ba in cells cotransfected with CaM₁₂₃₄, indicating that CDI was abolished (Fig. 6 D and Table 4). These results show that CDI of Ca_v1.4Δex47 is mediated by both the N and C lobe of CaM.

To investigate how molecular determinants in the IQ domain may influence CDI, we compared CDI in cells transfected with Ca_v1.4Δex47-5A and Ca_v1.4Δex47-3A. For Ca_v1.4Δex47-5A, there was no CDI in that there was no difference in inactivation of I_Ca and I_Ba, similar to Ca_v1.4 + ex47 (Fig. 7, A and B; and Table 4). For Ca_v1.4Δex47-3A, the results were more complex. In a subset of cells, there was no CDI, and I_Ca activated very slowly, likely reflecting the loss of CaM N lobe regulation of CDEA (Fig. S2). The lack of inactivation of both I_Ca and I_Ba in these cells suggests that the 3A mutations weaken CaM C lobe as well as CaM N lobe contributions to CDI. However, in some cells, there was residual CDI (Fig. 7 C), perhaps due to higher concentrations of CaM (Liu et al., 2010). In these cells, the fast phase of I_Ca inactivation was similar to that in cells cotransfected with Ca_v1.4Δex47 and CaM₁₂ (Fig. 6 B), suggesting that it was mediated by the CaM C lobe. Consistent with this possibility, cotransfection of Ca_v1.4Δex47-3A with CaM₃₄ eliminated CDI in all cells analyzed (Fig. 7 D and Table 4). We interpret these findings to mean that the 3A mutations in Ca_v1.4Δex47 disrupt CaM N lobe–mediated CDI, but also weaken CaM C lobe–mediated CDI under some conditions.

Although it responds to more global elevations in cytosolic Ca²⁺ with respect to CDI of Ca_v2 channels (DeMaria et al., 2001; Lee et al., 2003), the N lobe of CaM is a local Ca²⁺ sensor in the context of Ca_v1.2 and Ca_v1.3 CDI due to CaM binding to an N-terminal spatial Ca²⁺ transforming element (NSCaTE) in these channels. Thus, CaM N lobe–triggered CDI can be measured even with strong Ca²⁺ buffering (i.e., 5–10 mM EGTA) when Ca_v1.2 and Ca_v1.3 are coexpressed with CaM₃₄ (Dick et al., 2008). The NSCaTE is not conserved in Ca_v1.4, and yet CaM N lobe–mediated CDI of Ca_v1.4Δex47 (i.e., +CaM₃₄; Fig. 6 C) was still evident under our experimental conditions with 5 mM EGTA in the intracellular recording solution. We hypothesized that while deletion of exon 47 might disrupt the CTM’s ability to blunt CaM interactions with the channel, the CTM and/or the rest of the C-terminal domain might enable CaM N lobe–mediated CDI even in the presence of high intracellular Ca²⁺ buffering. If so, then deletion of these C-terminal regions should prevent CaM N lobe–driven CDI.

To test this, we turned to the K1591X mutation associated with CSNB2 (Strom et al., 1998). This mutation truncates the entire CTD of Ca_v1.4 distal to the IQ domain (Fig. 8 A); the loss of the CTM enables CaM-dependent CDI (Singh et al., 2006). In contrast to the suppression of CDI of Ca_v1.4Δex47 caused by both CaM₁₂ and CaM₃₄ (Fig. 6, B and C), I_Ca inactivation in K1591X mutant channels was suppressed by CaM₃₄ but unaffected by CaM₁₂. Moreover, CaM₃₄ completely abolished inactivation of I_Ca mediated by K1591X channels (Fig. 8 A), whereas CaM₃₄ had a more modest impact in this respect for Ca_v1.4Δex47 (Fig. 6 C). I_Ca in cells transfected with K1591X alone or together with CaM₁₂ showed a rapid initial phase of inactivation that was abolished in cells cotransfected with K1591X and CaM₃₄ (Fig. 8 A), similar to the CaM C lobe–mediated CDI of Ca_v1.4Δex47 (i.e., +CaM₁₂; Fig. 6 B). For both K1591X and Ca_v1.4Δex47, I_Ca could be fit with a double exponential function. However, the time constant (τ) for fast inactivation was significantly shorter for K1591X than for Ca_v1.4Δex47 (Fig. 8 B). There was no significant difference in the amplitude of slow (A_slow) and fast (A_fast) inactivation for Ca_v1.4Δex47, but A_fast was generally larger than that for A_slow for K1591X, although the difference did not reach statistical significance. These results show that K1591X is incapable of CaM N lobe–mediated CDI, and that the CaM C lobe mediates a fast phase of CDI that is more prominent in K1591X than in Ca_v1.4Δex47.

We also analyzed the impact of the 3A mutations in K1591X (K1591X-3A). In contrast to the mixed effects of these mutations in weakening CDI of Ca_v1.4Δex47 (Fig. 7, C and D; and Fig. S2), CDI was completely abolished in all recordings of cells transfected with K1591X-3A (Fig. 9, A and B; F_CDI = 0.47 ± 0.04 for K1591X versus −0.04 ± 0.03 for K1591X-3A; P = 0.001 by t test). These results reveal a more prominent role for F1582A, Y1583A, and F1586A in CaM C lobe–mediated CDI of K1591X as compared with Ca_v1.4Δex47.

Given that K1591X is deficient in CaM N lobe–dependent CDI, we tested if this mutant was also lacking in CaM N lobe regulation of CDEA. If so, then the tail current I–V relationship for I_Ca should not be affected by coexpression with CaM₁₂ or by the 3A mutations. As shown previously with different combinations of auxiliary subunits than those used here (Singh et al., 2006), K1591X exhibited a negative shift in activation of I_Ca compared with wild-type Ca_v1.4 + ex47 (Fig. 10 A and Table 2). Unlike Ca_v1.4Δex47, I_Ca mediated by K1591X was unaffected by CaM₁₂ or the 3A mutations (Fig. 10, B and C); there was no significant difference in the V_h or k for K1591X transfected alone or with CaM₁₂, or for K1591X and K1591X-3A (Table 2). Thus, although F1582A, Y1583A, and F1586A regulate CaM C lobe–dependent CDI, these residues are dispensable for regulating activation of K1591X.

Our study provides new insights into the regulation of Ca_v channels by CaM. First, we identify an unexpected role for CaM in CDEA—a Ca²⁺-dependent enhancement of Ca_v1.4 activation mediated by the N lobe of CaM. CDEA requires the deletion of exon 47 as well as the inclusion of other regions of the CTD in Ca_v1.4Δex47; the absence of these regions in K1591X may account for its lack of CaM N lobe–mediated CDEA. Second, we show that when CDI of Ca_v1.4 is unmasked by partial disruption of the CTM, CaM N lobe mediates a slow component of CDI that is significantly weakened upon full deletion of CTD distal to the IQ domain. This contribution of CaM N lobe is evident in CDI of Ca_v1.4Δex47 but not that of K1591X, which is dominated by the CaM C lobe. Our findings support a model in which variations in the CTD adjust how the N and C lobe of CaM modulate Ca_v1.4 (Fig. 11), providing a context for understanding the functional consequences of naturally occurring and pathological variants of Ca_v1.4 in the retina.

CaM is a dynamic regulator of Ca_v channels and can vary its modulatory impact upon binding Ca²⁺. When bound to the pCTD, Ca²⁺-free (apo) CaM boosts channel open probability (P_o), which then diminishes as Ca²⁺ binding to CaM triggers CDI (Adams et al., 2014). In the context of Ca_v1.4 + ex47 and the long variant of Ca_v1.3 (+exon 42A; Singh et al., 2008), the CTM competes with apoCaM binding (Liu et al., 2010; Adams et al., 2014). Deletion of exon 47 may allow for stronger binding of apoCaM to the pCTD, which may explain the negative shifts in the tail I–V curves for I_Ba mediated by Ca_v1.4Δex47 (Fig. 1 C). While these shifts were not affected by CaM mutants (Figs. 2 C and 3 C) or by the 3A or 5A mutations (Figs. 4 C and 5 C), it is possible that these experimental manipulations incompletely blunted apoCaM interactions with the channel, as some determinants for apoCaM binding may reside outside the IQ domain (Ben Johny et al., 2013). As has been proposed for Ca²⁺/CaM-dependent facilitation of Ca_v2.1 channels (Chaudhuri et al., 2007), Ca²⁺/CaM may promote transitions to a gating mode characterized by even higher P_o than with apoCaM, and this may be disrupted by the 3A and 5A mutations. Finer resolution of the mechanism by which the CTM modulates activation of I_Ca and I_Ba awaits detailed alanine scanning mutagenesis to unearth specific contacts for CaM in the pCTD in combination with single-channel analyses of P_o.

Unlike Ca_v1.4Δex47, K1591X does not undergo CDEA in that neither CaM N lobe nor the 3A mutations had any effects on the tail I–V curves for I_Ca (Fig. 10). The absence of CaM N lobe regulation of K1591X is further suggested by our kinetic analyses of I_Ca inactivation, as well as the blockade of CDI by CaM₃₄ (Fig. 8 A). These experiments implicated the CaM C lobe in mediating CDI of K1591X, in stark contrast to the contributions of both CaM lobes to CDI of Ca_v1.4Δex47 (Fig. 6). Deletion of exon 47 removes the initial 47 amino acids of the CTM (Singh et al., 2006), 26 residues according to the CTM region defined by Wahl-Schott et al. (2006). However, Ca_v1.4Δex47 still possesses a 34–amino acid stretch within the CTM shown to be critical for CDI suppression (Singh et al., 2006; Wahl-Schott et al., 2006). We propose that the residual CTM in concert with the post-IQ region unmasks and/or stabilizes CaM N lobe contacts within the IQ domain that support the slow component of CDI in Ca_v1.4Δex47 (Fig. 11), which is blocked by CaM₁₂ (and by 3A mutations; Figs. 6 B and 7 C). In the context of Ca_v1.2, thephenylalanine, tyrosine, and phenylalanine residues affected by the 3A mutations form extensive contacts with CaM N lobe (Van Petegem et al., 2005). Since these residues are conserved in Ca_v1.4, as well as in all Ca_v1 channels (Van Petegem et al., 2005), we propose that they may similarly serve as anchoring points for CaM N lobe in Ca_v1.4Δex47. Although these residues may be free to interact with CaM in K1591X, the absence of both CTM and post-IQ regions in K1591X may prevent the CaM N lobe interaction, or its functional consequences in promoting CDI (Fig. 11). However, our findings also reveal a key role for F1582A, Y1583A, and F1586A in weakening CaM C lobe–mediated CDI in both Ca_v1.4Δex47 and K1591X, which is not wholly unexpected given that CaM C lobe contacts may be intermingled among these residues (Van Petegem et al., 2005). Additional structural studies are needed to resolve how the CTM with and without exon 47 affects CaM interactions with the pCTD.

Although our results suggest differences in CaM regulation of Ca_v1.4Δex47 and K1591X, the impact of CaM on these channels in photoreceptors is likely influenced by CaBP4 (Haeseleer et al., 2004; Lee et al., 2015)—a member of a family of Ca²⁺ binding proteins (CaBPs) that suppress CDI of Ca_v1 channels in part by competing with CaM binding to the IQ domain (Hardie and Lee, 2016). CaBP4 binds to the same sites as CaM within the pCTD of Ca_v1.4 (Shaltiel et al., 2012; Haeseleer et al., 2016), and suppresses but does not completely abolish CDI of Ca_v1.4Δex47 (Haeseleer et al., 2016). This suggests that CaM may still interact with and modulate Ca_v1.4Δex47 despite the presence of CaBP4 in photoreceptor synaptic terminals (Haeseleer et al., 2004; Lee et al., 2015). When bound to Ca²⁺, CaM is expected to compete more effectively for binding to the Ca_v1 IQ domain than CaBPs (Findeisen et al., 2013), and CaBPs can bind to sites other than the IQ domain in regulating CDI (Zhou et al., 2005; Yang et al., 2014). Fluctuations in presynaptic Ca²⁺ levels may promote dynamic regulation of Ca_v1.4Δex47 by CaM and CaBP4, allowing for CDEA and CDI to shape presynaptic Ca²⁺ signals supporting glutamate release from photoreceptor terminals. If K1591X is similarly regulated by CaM, the lack of CaM N lobe regulation of CDEA and CDI may lead to aberrant regulation of Ca²⁺ signals that could degrade the transmission of visual information. Considering that the CTD of Ca_v channels is a hotspot for protein interactions (Calin-Jageman and Lee, 2008), it is also conceivable that K1591X is unable to assemble with key synaptic proteins needed for the proper localization and/or function of the channel in photoreceptors.

This work was supported by the National Institutes of Health (DC009433, NS084190, and EY026817 to A. Lee and EY026477 to B. Williams), the Neuroscience Training Program (T32 NS007421), a Carver Research Program of Excellence Award, and the University of Iowa’s Bioscience Academy (R25GM058939).

The authors declare no competing financial interests.

Author contributions: B. Williams conducted the research. B. Williams, F. Haeseleer, and A. Lee contributed to experimental design. B. Williams and A. Lee wrote the paper.

Richard W. Aldrich served as editor.