Calmodulin-dependent gating of Ca_v1.2 calcium channels in the absence of Ca_vβ subunits

Abstract

It is generally accepted that to generate calcium currents in response to depolarization, Ca_v1.2 calcium channels require association of the pore-forming α_1C subunit with accessory Ca_vβ and α₂δ subunits. A single calmodulin (CaM) molecule is tethered to the C-terminal α_1C-LA/IQ region and mediates Ca²⁺-dependent inactivation of the channel. Ca_vβ subunits are stably associated with the α_1C-interaction domain site of the cytoplasmic linker between internal repeats I and II and also interact dynamically, in a Ca²⁺-dependent manner, with the α_1C-IQ region. Here, we describe a surprising discovery that coexpression of exogenous CaM (CaM_ex) with α_1C/α₂δ in COS1 cells in the absence of Ca_vβ subunits stimulates the plasma membrane targeting of α_1C, facilitates calcium channel gating, and supports Ca²⁺-dependent inactivation. Neither real-time PCR with primers complementary to monkey Ca_vβ subunits nor coimmunoprecipitation analysis with exogenous α_1C revealed an induction of endogenous Ca_vβ subunits that could be linked to the effect of CaM_ex. Coexpression of a calcium-insensitive CaM mutant CaM₁₂₃₄ also facilitated gating of Ca_vβ-free Ca_v1.2 channels but did not support Ca²⁺-dependent inactivation. Our results show there is a functional matchup between CaM_ex and Ca_vβ subunits that, in the absence of Ca_vβ, renders Ca²⁺ channel gating facilitated by CaM molecules other than the one tethered to LA/IQ to support Ca²⁺-dependent inactivation. Thus, coexpression of CaM_ex creates conditions when the channel gating, voltage- and Ca²⁺-dependent inactivation, and plasma-membrane targeting occur in the absence of Ca_vβ. We suggest that CaM_ex affects specific Ca_vβ-free conformations of the channel that are not available to endogenous CaM.

In l-type Ca_v1.2 calcium channels, calmodulin (CaM) plays a central role in Ca²⁺-dependent inactivation (CDI), a physiologically important negative feedback regulated by permeating Ca²⁺ ions causing acceleration of the Ca²⁺ but not Ba²⁺ current decay. Considerable progress has been made in identifying and characterizing interactions between the pore-forming α_1C subunit and CaM (for the most recent review, see ref. 1). It is generally accepted that CDI is mediated by a single CaM molecule that is preassociated with the α_1C subunit C-terminal tail (2). The structure-functional analysis revealed two CDI-related CaM-binding sites on the Ca_v1.2 α_1C subunit C tail; one located within the segment 1572–1604 (LA) and the other confined to amino acids 1617–1636 (IQ). The mode of CaM binding to these sites depends on free Ca²⁺ concentration and hence on the occupancy of CaM by Ca²⁺. Of particular interest is that CDI does not solely depend on the presence of these Ca²⁺ sensors and requires a number of other channel structures. These determinants crucial for CDI include a Ca_vβ subunit (3), the α_1C subunit N-tail (4), the determinant of slow inactivation in the pore inner region (5), and possibly the putative EF-hand locus residing in the α_1C subunit C-tail upstream of LA/IQ sites (6). Specific folding of these determinants supporting CDI is mediated by voltage-gated rearrangements between α_1C and Ca_vβ (7) and between the α_1C subunit cytoplasmic N- and C-tails (8). It is also known that Ca_vβ subunits bind to the α₁-interaction domain (AID) in the linker between repeats I and II of the α_1C subunit (9) and, in a Ca²⁺-dependent manner, to the IQ region of the α_1C subunit C tail (3). The complexity of the structure and dynamics of these multifaceted determinants is not well understood. Our report provides evidence that is crucial for better understanding the functional links between these determinants, because it shows that gating of the Ca_vβ-free recombinant Ca_v1.2 channel can be rendered by coexpression of endogenous CaM (CaM_ex).

Results

A number of important studies of CDI relied on coexpression of dominant-negative CaM mutants lacking Ca²⁺ binding but retaining affinity to apo-CaM sites of α_1C (10–12). These studies were carried out in an assumption that coexpression of CaM does not markedly change electrophysiological properties of the channel, mainly because endogenous CaM is an abundant protein reaching micromolar concentrations in the cell (13). Our study shows that CaM_ex does affect gating of the Ca_v1.2 calcium channel and facilitates it in the absence of Ca_vβ subunits.

Effect of CaM_ex on Electrophysiological Properties of the Recombinant Ca_v1.2 Calcium Channel Containing Ca_vβ Subunits.

To test the effects of CaM_ex on the properties of Ca²⁺ channels, we used Ca²⁺ channel-free COS1 cells (4, 14). In our first set of experiments, we coexpressed EYFP_N-α_1C, α₂δ, and β_2d subunits in the presence (+CaM_ex) or absence (−CaM_ex) of ECFP_N-CaM. Epifluorescent images of expressing cells (Fig. 1 A and B) revealed distinct plasma membrane (PM) targeting of EYFP_N-α_1C (Fig. 1 Aa and Ba, arrows). ECFP_N-CaM was abundantly expressed in the cell [supporting information (SI) Fig. S1A] exhibited some PM targeting because it binds to the channels (Fig. 1Ab). Representative traces of I_Ca shown in Fig. 1 A and B were evoked by 600-ms test pulses in the range of 0 to +60 mV (10-mV increments). First, we found that all traces were better fitted by a single exponential function except the three traces on Fig. 1B (−CaM_ex) recorded at test potentials +10, +20, and +30 mV. These traces required double-exponential fitting revealing an apparent slow component of inactivation that, on average, accounted for 10–19% of the total I_Ca amplitude (n = 5). We also noticed that CaM_ex reduced ≈3-fold the fraction I_o of the Ca²⁺ current remaining at the end of a 600-ms test pulse (Fig. 1C, open circles). Although the nature of these changes is not yet clear, these data suggest that CaM_ex promoted inactivation of the channel. Analysis of current–voltage (I–V) relationships (Fig. 1D) showed that CaM_ex increased the density of I_Ca 2.4 ± 0.1-fold (n = 8). Independently on this increase, CaM_ex affected channel gating by shifting the maximum of I–V curve and V_0.5 to more negative potentials (open circles, Fig. 1D) and increasing k_I–V as compared with the control channel expressed in the absence of CaM_ex (closed circles). However, the kinetics of I_Ca inactivation was not significantly changed by CaM_ex (Fig. 1E) and in both cases exhibited a U-shaped dependence of the time constant of inactivation (τ) on membrane voltage characteristic for CDI that is accelerated with larger I_Ca (15). There was a 10-mV shift of the maximum of the τ-V relation to more negative potentials (open circles) corresponding to the shift of the I–V maximum caused by CaM_ex. Finally, the CaM_ex-modulated channel was fully inhibited by a specific l-type Ca²⁺ channel blocker PN200–110 (2 μM, Fig. 1F). Taken together, these results revealed that CaM_ex modified channel gating and augmented I_Ca through the Ca_v1.2 calcium channel.

Fig. 1.

Effects of CaM_ex on the properties of the Ca_v1.2 channel. The EYFP_N-α_1C, α₂δ and β_2d subunits were expressed in COS1 cells in the presence (A) or absence (B) of ECFP_N-CaM. Shown are representative traces of I_Ca recorded in response to 600-ms steps to indicated test potentials (V_t) from the holding potential V_h = −90 mV. (a and b) Epifluorescent images of the expressing cells showing distribution of EYFP_N-α_1C and ECFP_N-CaM and obtained with the YFP and CFP filters, respectively. (Scale bars, 4 μm.) Arrows point to PM targeting of EYFP_N-α_1C. Inactivation time constants (τ) were determined from the fitting of I_Ca decay by an exponential function: I(t) = I_∞ + I × exp(−t/τ), where I_∞ is the steady-state amplitude of the current, I is apparent inactivating component of the initial current. I_o is the sustained current component determined as the ratio of steady state to peak current amplitudes. (C) Voltage dependence of the sustained component I_o of the I_Ca inactivation. (D) The averaged I–V curves for I_Ca recorded in the absence (filled circles) or presence of coexpressed CaM (open circles). Currents were measured with 30-s intervals between 0.6-s test pulses in the range of −60 to +80 mV applied with 10-mV increments from V_h = −90 mV. Smooth lines represent fitting by equation I_Ca = G_max (V − E_rev)/(1 + exp[(V − V_0.5)/k_I–V]), where G_max is maximum conductance, E_rev is the approximated reversal potential, V_0.5 is voltage at 50% of I_Ca activation, and k_I–V is slope factor. Ca_v1.2: V_0.5 = 14.6 ± 0.6, k_I–V = −9.9 ± 0.3, E_rev = 105.2 ± 3.5 mV (n = 5); Ca_v1.2 + CaM_ex: V_0.5 = 2.4 ± 0.6, k_I–V = −5.0 ± 0.5, E_rev = 93.9 ± 1.3 mV (n = 8). (E) Voltage dependence of τ for I_Ca recorded in the absence (filled circles) or presence of CaM_ex (open circles). All error bars reflect SEM. (F) Inhibition of I_Ca through the EYFP_N-α_1C/α₂δ/β_2d/CaM_ex channel by 2 μM (+)PN200–110. V_h = −90 mV, V_t = +30 mV.

CaM_ex Supports Calcium Channel Gating on Coexpression with α_1C/α₂δ in COS1 Cells in the Absence of Ca_vβ Subunits.

In the absence of Ca_vβ subunits, coexpression of α_1C with either CaM_ex or α₂δ generated silent channels (Fig. 2A a and b). This is believed to be a result of poor PM targeting by the Ca_vβ-deficient Ca_v1.2 channel and an inhibition of the channel by the α_1C subunit N tail (for details, see ref. 4). Lack of significant PM targeting was confirmed by the quantitative analysis of distribution of α_1C between PM and the cytoplasm (Fig. 2B, columns 1 and 2). Expression of β_2d in the absence of α₂δ stimulated PM targeting of α_1C (Fig. 2B, column 3), but the channel remained silent (Fig. 2Ac) unless CaM_ex was coexpressed (Fig. 2Ad). Thus, CaM_ex facilitates voltage gating of the Ca_v1.2 channel.

Fig. 2.

Effects of CaM_ex on the properties of the β-deficient Ca_v1.2 channel. (A) Ca²⁺ channel activity in COS1 cells expressing EYFP_N-α_1C and CaM_ex (a), α₂δ (b), β_2d (c), or β_2d +CaM_ex (d). Shown are representative traces (n = 5–10) of maximal I_Ca evoked by 600-ms test pulses to +20 mV (a and c), +30 mV (b), or +50 mV (d) applied from V_h = −90 mV. (B) Relative distribution of EYFP_N-α_1C in PM over the cytoplasm in the presence of CaM_ex (1), α₂δ (2), β_2d (3), β_2d + CaM_ex (4), α₂δ + β_2d (5), α₂δ + CaM_ex (6), or α₂δ + β_2d + CaM_ex (7). The ratio of fluorescence intensity in PM over the area underneath PM was averaged after background subtraction in each cell. The ratio <1.0 indicates lack of significant α_1C PM targeting. ANOVA statistical analysis with Tukey–Kramer multiple comparison test was applied. The number of tested cells is shown in the bars. *, P < 0.05. (C) Effect of CaM_ex on PM targeting and activity of Ca²⁺ channels in COS1 cells expressing EYFP_N-α_1C, α₂δ. (a) Whole-cell EYFP fluorescence. (Scale bar, 4 μm.) (b) Representative traces of I_Ca recorded in response to the indicated 600-ms test pulses (V_h = −90 mV). (D) Average I–V relationship for I_Ca through the α_1C/α₂δ/CaM_ex channel (filled circles) coplotted with the voltage dependence of τ (open circles). V_0.5 = 15.8 ± 0.8, k_I–V = −9.1 ± 0.5, E_rev = 110.3 ± 2.2 mV (n = 5). (E) Voltage dependence of activation of EYFP_N-α_1C/α₂δ coexpressed with β_2d (filled circles, n = 7) or CaM_ex (open circles, n = 9). Ca²⁺ tail currents (I_tail) were recorded after repolarization for 10 ms to −50 mV following V_t from −40 to +90 mV applied from V_h = −90 mV for 20 ms. I_tail were normalized to the peak I_tail (I_tail,max) and fitted with a Boltzmann equation: I_tail/I_tail,max = (A₁ − A₂)/[ 1 + exp(V − V_a,50)/k_a] + A₂, were V_a,50 is the half-maximal voltage for current activation, k_a is the slope factor, A₁ and A₂ represent proportion of fully activated and nonactivated current. (F) Representative trace of I_Ca through the CaM_ex-activated β-deficient Ca_v1.2 channel evoked by V_t to +40 mV applied from V_h = −90 mV for 30 s. (G) Averaged steady-state inactivation curve for I_Ca through the EYFP_N-α_1C/α₂δ/CaM channel (n = 5). One-second conditioning prepulses were applied from V_h = −90 mV (up to +50 mV, 10-mV increments) followed by a 100-ms V_t to +40 mV. The intervals between each cycle were 15 s. The peak current amplitudes in each curve were normalized to the maximum value determined in the range of −40 to +50 mV. The curves were fitted (smooth line) by Boltzmann function: I = A + B/(1 + exp[(V − V_0.5,in)/k]), where A (0.50 ± 0.01) and B are fractions of noninactivating and inactivating currents, respectively, V is the conditioning prepulse voltage, V_0.5,in = 10.3 ± 0.6 mV is the voltage at half-maximum of inactivation, and k = 5.4 ± 0.5 is a slope factor.

In the absence of Ca_vβ, CaM_ex enhanced PM targeting of α_1C/α₂δ (Fig. 2Ca, arrows) as effectively as that by β_2d ± α₂δ (Fig. 2B, columns 3, 5, 6), but there was no synergy between CaM_ex and β_2d with or without α₂δ (columns 4, 7). Coexpression of ECFP_N-CaM_ex with α_1C/α₂δ recovered gating of the Ca_vβ-deficient channel that retained sensitivity to PN200–110 (Fig. S2A). ECFP_N-tagging does not interfere with this effect of CaM_ex (see Fig. S2 B and C). Fig. 2C shows a collection of representative traces of I_Ca elicited by 600-ms test pulses to indicated voltages applied from V_h = −90 mV. The corresponding averaged I–V relationship and deduced voltage-dependent characteristics are presented in Fig. 2D. The threshold of activation of I_Ca was ≈ −40 mV, and the voltage that elicited the maximal I_Ca (+40 mV) was shifted by 10 mV to more positive potentials as compared with β_2d (Fig. 1D). Analysis of tail currents (16) showed a significant difference in the activation parameters with β_2d. Half-activation potential V_a,0.5 was shifted from 14.6 ± 0.3 mV (n = 7) for β_2d to 42.5 ± 1.1 mV (n = 9) for CaM_ex without notable change in the slope factor [k_a = 17.1 ± 0.3 (β_2d) and 17.4 ± 0.7 (CaM_ex)] (Fig. 2E). This result confirms that CaM_ex affects Ca_v1.2 channel gating in the absence of Ca_vβ. An intriguing parallel between the effects of Ca_vβ and CaM_ex on channel gating is that both critically depend on the presence of α₂δ. Neither β_2d nor CaM_ex supported appreciable currents in the absence of α₂δ (Fig. 2A).

An interesting feature of the Ca_vβ-deficient channel activated by CaM_ex is its notably slow inactivation. When the test pulse duration was prolonged to 30 s (Fig. 2F), approximately one-third of the maximal I_Ca did not show appreciable decay. This result is in agreement with the steady-state inactivation analysis (Fig. 2G) indicating there is a large fraction of noninactivating channels. A monoexponential fitting of the inactivation time course (Fig. 2D, open circles) revealed a markedly slower inactivation (τ = 117 ± 8 ms for the peak current at +30 mV, n = 5) as compared with the α_1C/α₂δ/β_2d channel (59 ± 6 ms at +20 mV, n = 5) and a distinct U-shaped voltage dependence of τ reflecting CDI. Thus, lack of Ca_vβ is not crucial for CDI on coexpression of α_1C and α₂δ with CaM_ex. However, CDI accounts for only a fraction of I_Ca decay in the Ca_vβ-free channel.

Ca²⁺ Dependence of the Channel Modulation by CaM_ex.

To further explore Ca²⁺ dependence of the CaM_ex effects, we inhibited Ca²⁺-induced molecular rearrangements of CaM by replacing Ca²⁺ for Ba²⁺ as the charge carrier. Fig. 3A shows a representative trace of I_Ba recorded in response to a 600-ms test depolarization to +30 mV corresponding to the maximum of the I–V curve (Fig. 3B). We found that I_Ba decays with kinetics slower than that of I_Ca through this channel (Fig. 3A; see the superimposed gray trace scaled to the same amplitude), confirming that CDI is responsible in part for inactivation of I_Ca in this channel. Accordingly, the steady-state inactivation curve (Fig. 3C) showed that the voltage-dependent availability of Ca²⁺ channels (0.67 ± 0.01, n = 5) increased by ≈34% in the Ba²⁺ bath medium as compared with Ca²⁺ (Fig. 2G). Thus, the Ba²⁺ experiment showed that CaM_ex-induced gating of the channel does not require Ca²⁺ and is not due to enhanced Ca²⁺ buffering by CaM_ex.

Fig. 3.

Effect of the replacement of Ca²⁺ for Ba²⁺ as the charge carrier through the EYFP_N-α_1C/α₂δ channel modulated by CaM_ex. The EYFP_N-α_1C and α₂δ subunits were coexpressed in COS1 cells with ECFP_N-CaM. (A) Representative trace of the maximum I_Ba evoked by V_t = +30-mV applied for 600 ms from V_h = −90 mV. For comparison, gray line shows the decay portion of the I_Ca trace (+30 mV, see Fig. 2C) scaled to the same amplitude. (B) The averaged normalized I–V curve: V_0.5 = 11.4 ± 1.5, k_I–V = −9.4 ± 0.7, E_rev = 93.5 ± 3.7 mV (n = 18). (C) Averaged steady-state inactivation curve for I_Ba: A = 0.67 ± 0.01, V_0.5,in = 15.4 ± 1.6 mV; k = 6.9 ± 1.4 (n = 5).

We then coexpressed α_1C and α₂δ in COS1 cells with the Ca²⁺-insensitive mutant CaM₁₂₃₄ (17). This dominant-negative CaM mutant was shown to inhibit CDI of Ca_v1.2 calcium channels (10, 12) while retaining ability to bind to the CDI site of the α_1C subunit (11). Similar to CaM_ex, coexpression of CaM₁₂₃₄ enhanced PM targeting of EYFP_N-α_1C (Fig. 4A, arrows). Both I–V (Fig. 4B) and steady-state inactivation curves (Fig. 4C) for I_Ca measured with CaM₁₂₃₄ were not significantly different from those obtained with CaM_ex (compare statistics in legends to Figs. 2 and 4). The activation curve (Fig. 4D) was shifted from that for β_2d by ≈15 mV to more positive potentials. Differences of the activation parameters for CaM₁₂₃₄ (V_a,0.5 = 31.0 ± 0.4 mV, k_a = 15.2 ± 0.3, n = 4) with CaM_ex (Fig. 2E) may be due to the CaM₁₂₃₄-induced inhibition of CDI that is known to affect voltage dependence of the channel (17). Indeed, experiment with CaM₁₂₃₄ did not reveal a U shape of τ-V dependence (Fig. 4B). Respectively, inactivation of I_Ca through the α_1C/α₂δ/CaM₁₂₃₄ channel recorded at the peak of the I–V relationship (Fig. 4E) was slower than that with CaM_ex (gray trace) and matched closely inactivation of I_Ba (Fig. 3A). Taken together, the results of Ba²⁺ and CaM₁₂₃₄ experiments suggest that the ability of CaM_ex to support the Ca_vβ-free Ca_v1.2 channel gating is not associated with the Ca²⁺-binding property of CaM and its role in CDI.

Fig. 4.

Ca²⁺-insensitive CaM₁₂₃₄ mutant supports gating of the Ca_vβ-subunit-deficient Ca_v1.2 calcium channel. The EYFP_N-α_1C and α₂δ subunits were coexpressed in COS1 cells with CaM₁₂₃₄. (A) Epifluorescent image of an expressing cell showing PM targeting of EYFP_N-α_1C (arrows). (Scale bar, 4 μm.) (B) The averaged I–V curve (filled circles) coplotted with voltage dependence of τ for I_Ca (open circles): V_0.5 = 15.8 ± 1.0, k_I–V = −9.1 ± 0.6, E_rev = 120.8 ± 3.5 mV (n = 7). (C) The averaged steady-state inactivation curve for I_Ca: A = 0.52 ± 0.01, V_0.5,in = 14.5 ± 0.4 mV; k = 8.8 ± 0.4 (n = 6). (D) Averaged normalized voltage dependence of activation of I_Ca through α_1C/α₂δ coexpressed with β_2d (filled circles; n = 4) or CaM₁₂₃₄ (open circles; n = 4). (E) Representative trace of the maximum I_Ca activated by V_t to + 40 mV applied for 600 ms from V_h = −90 mV. For comparison, the gray line shows a decay portion of I_Ca through the β-deficient α_1C/α₂δ/CaM_ex channel evoked by V_t = +40-mV (for original trace, see Fig. 2C).

Molecular Correlates of the CaM_ex-Dependent Gating of Ca_vβ-Deficient Channels.

Coimmunoprecipitation analyses have shown that CaM_ex pulled down with α_1C, whereas endogenous CaM was not detectable in coimmunoprecipitated protein mixture (Fig. S1B). Because binding of CaM_ex to α_1C was inhibited in the presence of Ca_vβ with 0 or 10 μM Ca²⁺ (Fig. 5A), we focused on the role of known Ca_vβ determinants of α_1C (AID and IQ) in the effect of CaM_ex. LA and IQ are the primary binding sites of CaM supporting CDI. It was shown that mutation or deletion of LA and/or IQ deprives the channel of CDI but does not inhibit the modulation of Ca_v1.2 gating by Ca_vβ (18–20), as can be seen in Fig. 5 Ba–Da. However, the LA- (α_1C,L), IQ- (α_1C,K), or LA+IQ-deficient (α_1C,ΔLK) channels showed no activity in the absence of Ca_vβ irrespectively of coexpression of CaM_ex (Fig. 5 Bb–Db). Thus, determinants of CDI are crucial for the CaM_ex-dependent gating of the Ca_vβ-deficient channel.

Fig. 5.

Competition between CaM_ex and Ca_vβ for interaction with α_1C/α₂δ. (A) Western blot analysis (representing two independent experiments) of coimmunoprecipitation (IP) of CaM_ex with α_1C in the absence (lane 1) or presence (2-4) of β_2d. FLAG_N-α_1C and α₂δ were coexpressed in COS1 cells with Venus (ct, control), ECFP_N-CaM (1), Venus-β_2d (2), or ECFP_N-CaM + Venus-β_2d (3, 4) (Right). Cells were lysed in the presence of 10 μM (ct, 1-3) or zero Ca²⁺ (4) and coIP with anti-FLAG Ab (Left). The expressed proteins (see Left) were analyzed with anti-FLAG (Upper) or anti-LC Ab (Lower). Molecular mass standards (in kDa) are indicated on the right. (B–E) Inhibition of CaM_ex modulation of the Ca_vβ-free Ca²⁺ channels by mutation of major CDI- or Ca_vβ-related α_1C functional motifs. (B) α_1C,L (LA-), (C) α_1C,K (IQ-), (D) α_1C,ΔLK (LA+IQ-deficient), or (E) mVenus_N-α_1CAIDM α_1C subunits were coexpressed with α₂δ and either ECFP_N-CaM or β_2d (a) or (b). Shown are representative traces (n = 3–10) of maximal I_Ca recorded in response to V_t = +30 (C and D) or +20 mV (B and E). V_h = −90 mV. (c) Distribution of EYFP_N-α_1CAIDM between PM and the cytoplasm in the presence of CaM_ex or β_2d as compared with that for EYFP_N-α_1C in the presence of β_2d (see Fig. 2B). Number of tested cells is shown in the bars. *, P < 0.05.

We then tested whether the CaM_ex-supported gating depends on AID. The crucial amino acids (Asp⁴³³, Gly⁴³⁶, Tyr⁴³⁷, and Trp⁴⁴⁰) in AID (21–23) were converted to alanines, and the α_1CAIDM mutant was coexpressed with α₂δ and β_2d (Fig. 5E). Western blot analysis unequivocally confirmed the lack of binding between β_2d and the mutated AID (33), but β_2d retained binding to IQ (3), and the channel generated I_Ca in response to V_t = +30-mV (Fig. 5Ea). When α_1CAIDM and α₂δ were coexpressed with CaM_ex in the absence of Ca_vβ, the channel remained silent in response to V_t in the range of −40 to +80 mV (see an exemplar trace on Fig. 5Eb), despite distinct PM localization (Fig. 5Ec). These data suggest that AID is also responsible for the CaM_ex-dependent gating, and its mutation ablates the effect of CaM_ex. Taken together, our results provide evidences that the CaM_ex-dependent gating of the Ca_vβ-deficient channel is mediated by interdependent determinants AID and LA/IQ of α_1C involved in the regulation of the channel by Ca_vβ.

Assessment of Endogenous Ca_vβ Subunits.

Previously, little or no endogenous Ca_vβ immunoreactivity was observed in COS7 cells even when α_1C was expressed (14). To test whether CaM_ex may induce the expression of endogenous Ca_vβ subunits, we carried out a comparative qPCR analysis of the monkey β₁, β₂, and β₃ transcripts in nontransfected COS1 cells (NT) and those coexpressing the α_1C and α₂δ subunits with Venus (−CaM) or ECFP_N-CaM (+CaM) (Fig. 6A). A less common β₄ (known to be expressed in the brain and cochlea) was not analyzed by qPCR, because the structure of monkey β₄ is not known. We found that CaM_ex did not induce mRNA of endogenous β₂ and β₃ subunits, and in the case of β₁, even significantly reduced it. Two independent coimmunoprecipitation analyses with α_1C confirmed this result and showed (Fig. 6B) that Abs to common epitopes of rabbit/rat/human and monkey Ca_vβ subunits did not detect appreciable binding of α_1C to endogenous Ca_vβ in the absence (lanes 1) or presence of CaM_ex (lanes 2), as compared with a respective exogenous Ca_vβ (lanes 3). Endogenous β₄ in COS1 cells was not detectable with anti-β₄ polyclonal Ab (data not shown), thus confirming a similar earlier assessment (14). In conclusion, these data and lack of appreciable Ca²⁺ or Ba²⁺ currents on coexpression of α_1C and α₂δ subunits without CaM_ex (Fig. 3A) strongly suggest that the channel activity rendered by CaM_ex is not due to an induction of endogenous Ca_vβ subunits.

Fig. 6.

CaM_ex does not induce endogenous Ca_vβ subunits in COS1 cells. (A) Lack of effect of CaM_ex on relative mRNA levels of endogenous Ca_vβ in COS1 cells. Each image represents a real-time PCR assessment (mean ± SEM, n = 5) of the mRNA levels (relative to GAPDH mRNA) of three major Ca_vβ subunits in nontransfected COS1 cells (NT) or those coexpressing α_1C and α₂δ with the EYFP mutant Venus (−CaM) or ECFP_N-CaM (+CaM) under standard conditions used for electrophysiological experiments (Methods). PCR primers were designed to invariant exons of the monkey β₁, β₂, and β₃ subunit genes. *, P < 0.05; **, P > 0.05. (B) Lack of effect of CaM_ex on endogenous Ca_vβ binding to α_1C revealed by coimmunoprecipitation analysis. FLAG_N-α_1C and α₂δ were coexpressed in ≈10⁶ COS1 cells with (lane 1) Venus, (lane 2) ECFP_N-CaM, or (lane 3) human β_1b (GenBank accession no. M92302, a), β_2d (GenBank accession no. AF423191, b), or β₃ subunit (GenBank accession no. X76555, c). α_1C was identified on Western blot by anti-FLAG Ab (Upper). Ca_vβ subunits were identified (Lower) by Abs generated against rat/rabbit/human epitopes common with monkey: β₁ (GenBank accession no. XM_001085813, monkey amino acids 19–34, a), β₂ (GenBank accession no. XM_001092601, 387–410, b), and β₃ (GenBank accession no. XM_001102938, 477–491, c). Molecular mass calibration in kDa is shown at right.

Discussion

Although many details of the mechanism of CDI were understood after the discovery of the LA/IQ determinants in α_1C and the CDI-supporting function of CaM (1), much less is known regarding the role of Ca_vβ subunits in the regulation of the Ca_v1.2 channel by CaM. Our findings give a previously uncharacterized perspective on the role of CaM in Ca_v1.2 channels by establishing that, in the absence of Ca_vβ subunits, CaM_ex exerts Ca_vβ-like functions in the channel, including the stimulation of PM targeting and support of the channel gating (Fig. 2). These effects of CaM_ex do not rely on endogenous Ca_vβ (Fig. 6) and require α₂δ and α_1C with fully functional LA/IQ and AID motifs (Fig. 5). The ability of the dominant-negative CaM₁₂₃₄ to support Ca_vβ-free gating (Fig. 4) indicates that the functional effect is not related to Ca²⁺ binding to CaM_ex.

Perhaps the most surprising result from our study is that the functions, traditionally linked to Ca_vβ (24), are mediated by a ubiquitous, naturally abundant, and structurally different protein, CaM, but only on coexpression with α_1C and α₂δ. Recent image correlation spectroscopy measurements showed that in vivo CaM is sequestered in cells, and its availability for additional targeting is limited (25). This is consistent with our data showing that endogenous CaM is not sufficient for the Ca_vβ-like modulation of the Ca_vβ-free channel. An increase of local availability of CaM on overexpression (Fig. S1) may create conditions when CaM_ex targets specific Ca_vβ-free conformations of the channel at different transient steps of assembly of the channel complex. Given that physiologically relevant variations of CaM expression do occur in vivo (26), a CaM_ex-like modulation of the channel may take place as a compensatory response. For example, this could explain a surprising observation (27) that knockout of the primary Ca_vβ₃ gene in mouse ileum smooth muscle cells had little effect on I_Ca but did not change expression of Ca_v1.2 proteins.

The electrophysiological recordings indicated that, in the presence of Ca_vβ, CaM_ex modulated Ca_v1.2 channels by increasing I_Ca amplitude, shifting maximum of the I–V curve to more negative voltages and facilitating (but not accelerating) inactivation (Fig. 1). In the absence of Ca_vβ, Ca_v1.2 channels are silent, and the PM targeting by the Ca_vβ-deficient complex is inhibited unless CaM_ex is coexpressed (Fig. 2B). The finding that β_2d and CaM_ex/α₂δ are equipotent but not additive in the stimulation of the α_1C PM expression indicates that these different molecular entities may target the same mechanisms of the channel assembly and/or trafficking. However, amplitudes of the maximal I_Ca through α_1C/α₂δ/CaM_ex/1234 channels (Figs. 2D and 4B) were ≈2 times smaller than that through α_1C/α₂δ/β_2d (Fig. 1D), suggesting that either electrophysiological properties or PM conformations of the channels (e.g., interaction with α₂δ) are different. Thus, unlike Ca_vβ or α₂δ, CaM_ex affects both the surface expression and gating of the channel. This bimodal regulation requires the presence of α₂δ or Ca_vβ, but is not associated with Ca²⁺-binding activity of CaM_ex (Figs. 3 and 4). The latter suggests that Ca²⁺/CaM-mediated signaling cascades and CDI are not involved, and that Ca²⁺-dependent conformational changes of CaM_ex are not crucial for (but may affect) the gating of Ca_vβ-free channels.

Experiments with expressed LA-IQ (2) have shown that folding and conformation of the LA-IQ region in the absence of Ca_vβ are strongly affected by CaM. It is also known that split LA and IQ motifs of the LA-IQ region bind CaM with different affinities. However, LA-IQ binds a single CaM when LA-IQ/CaM molar ratio is ≥1. This interaction, implicated for CDI, may be more complex in the native channel because of the binding of Ca_vβ to IQ (3) that may affect the CaM-dependent folding of LA-IQ and its affinity to CaM. We speculate that CaM_ex may exert its action on Ca_vβ-free channels via interaction with Ca_vβ-binding sites AID and/or LA/IQ in α_1C. Indeed, Ca_vβ inhibited binding of CaM_ex to α_1C/α₂δ (Fig. 5A). Mutation or deletion of known Ca_vβ sites in α_1C completely eliminated CaM_ex-dependent gating (Fig. 5 B–E), indicating that CaM_ex may target multiple interconnected determinants of α_1C associated with Ca_vβ-subunit modulation of the channel. A mechanism consistent with our findings is that LA/IQ independently mediates both CDI and the effect of CaM_ex in the absence of Ca_vβ, so that lack of the Ca_vβ–IQ interaction in the Ca_vβ-free channel may increase the probability of CDI-unrelated interaction(s) of CaM_ex with LA/IQ. Whether these interactions correspond to those observed in the laboratory of Hamilton and coworkers (28) remains to be seen. Finally, because Ca_vβ, at least in part, inhibits these interactions (Fig. 5A), the augmentation of the current by CaM_ex (Fig. 1) may rely on a different set of interactions.

In conclusion, there is a functional matchup between CaM_ex and Ca_vβ that, in the absence of Ca_vβ, renders PM targeting and gating of Ca_v1.2 channels via interaction with CaM molecule(s) other than the one tethered to LA/IQ to support CDI. Our results challenge the view that Ca_vβ subunits are indispensable for PM targeting and gating of Ca_v1.2 channels and raise the possibility that a similar Ca_vβ-like modulation of PM targeting and gating by CaM may have place in other CaM-dependent Ca_v1 and Ca_v2 calcium channels (29, 30).

Materials and Methods

Expression in COS1 Cells.

β_2d (31) was PCR-amplified from human cardiac polyA(+) mRNA and subcloned into a pcDNA3 vector. Because fusion with FLAG or ECFP/EYFP does not compromise functional properties of CaM (32) (see also Fig. S2) and α_1C,77 (4), we used ECFP_N-CaM and the FLAG_N- or EYFP_N-tagged variants of human vascular α_1C (α_1C,77) throughout experiments to ease detection and visualization of PM targeting by the channel. COS1 cells were grown on poly-d-lysine-coated coverslips 18 h before transfection with cDNAs coding for α_1C,77, α₂δ, β_2d, and/or CaM (17) (1:1.2:1.4:5) using Effectene (Qiagen).

Electrophysiology.

Whole-cell recordings were performed (20°C–22°C) 48–72 h after transfection as described in ref. 4 with an Axopatch200 B amplifier (Axon Instruments). The external solution was: 100 mM NaCl, 20 mM CaCl₂ (when recording I_Ca) or BaCl₂ (I_Ba), 1 mM mMgCl₂, 10 mM glucose, and 10 mM Hepes (pH 7.4), with NaOH. Patch pipettes had resistances of 2.5–4 MΩ when filled with an internal solution containing: 100 mM CsCl, 5 mM MgATP, 0.2 mM cAMP, 20 mM tetraethylammonium, 10 mM 1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetate, and 20 mM Hepes (pH 7.4) with CsOH. Currents were filtered at 1 kHz, sampled at 2.5–5 kHz using pClamp 10 (Axon). Tail currents were filtered at 5 kHz and sampled at 13 kHz. Leak and capacitive transients were subtracted by using P/4 protocol. To achieve complete recovery from inactivation, test pulses were applied with 15-s intervals from V_h = −90 mV. Images were recorded with a Hamamatsu digital camera C4742–95 mounted on the Nikon epifluorescent microscope TE200 (60 × 1.2 N.A. objective) equipped with an excitation 75-W xenon lamp and multiple filter sets (Chroma Technology). Data were acquired and analyzed by using pClamp 10 (Axon) and Origin 7.5 (Microcal). Statistical analysis was performed with a unpaired two-tailed Student's t test. All data are presented as mean ± SEM and considered significant if P < 0.05.

Protein Analyses.

Assay of endogenous Ca_vβ subunits in COS1 cells was carried out as described in SI Methods. Monoclonal anti-β₁ (Neuromab) and polyclonal Ab to β₂, β₃, and β₄ (Millipore) were used for immunoblot analysis as described (7). Expressed proteins were solubilized, precleared with mouse IgG-agarose for 3 h, immunoprecipitated with anti-FLAG m₂ monoclonal affinity gel (Sigma), and resolved by SDS/PAGE. The specificity of our FLAG coIP system for FLAG_N-α_1C and ECFP_N-CaM is shown in Fig. S3.