Facilitation of murine cardiac L-type Ca_v1.2 channel is modulated by Calmodulin kinase II–dependent phosphorylation of S1512 and S1570

Abstract

Activity-dependent means of altering calcium (Ca²⁺) influx are assumed to be of great physiological consequence, although definitive tests of this assumption have only begun to emerge. Facilitation and inactivation offer two opposing, activity-dependent means of altering Ca²⁺ influx via cardiac Ca_v1.2 calcium channels. Voltage- and frequency-dependent facilitation of Ca_v1.2 has been reported to depend on Calmodulin (CaM) and/or the activity of Calmodulin kinase II (CaMKII). Several sites within the cardiac L-type calcium channel complex have been proposed as the targets of CaMKII. Here, we generated mice with knockin mutations of α₁1.2 S1512 and S1570 phosphorylation sites [sine facilitation (SF) mice]. Homocygote SF mice were viable and reproduced in a Mendelian ratio. Voltage-dependent facilitation in ventricular cardiomyocytes carrying the SF mutation was decreased from 1.58- to 1.18-fold. The CaMKII inhibitor KN-93 reduced facilitation to 1.28 in control cardiomyocytes. SF mutation negatively shifted the voltage-dependent inactivation and slowed recovery from inactivation, thereby making fewer channels available for activation. Telemetric ECG recordings at different heart rates showed that QT time decreased significantly more in SF than in control mice at higher rates. Our results strongly support the notion that CaMKII-dependent phosphorylation of Cav1.2 at S1512 and S1570 mediates Ca²⁺ current facilitation in the murine heart.

Calcium (Ca²⁺) current (I_Ca) through L-type channels is an important determinant of intracellular Ca²⁺ transients that trigger transmitter release, secretion, and contraction (1). In the heart, the size of the intracellular Ca²⁺ transient is determined by the release of Ca²⁺ from intracellular stores and by the size of the L-type current (2). The availability of L-type channels to open is regulated by the membrane potential and other factors that include protein kinases, phosphatases, and Ca²⁺ binding proteins (3, 4). A train of repetitive depolarizations (frequency-dependent) (5, 6) or a strong depolarizing prepulse (voltage-dependent) (7) drives the L-type calcium channels from their normal gating pattern into a mode of gating characterized by long opening and high open probability (8, 9), a process that has been termed “facilitation” and that represents one of the rare positive feedback mechanisms in signal transduction. The present consensus is that facilitation requires elevated intracellular calcium, or [Ca²⁺], in the vicinity of the channel (“Ca²⁺-dependent facilitation”) (10). Because excessive calcium influx is toxic to cells, these cells have evolved voltage- and Ca²⁺-dependent inactivation mechanisms. Both Ca²⁺-dependent facilitation (CDF) and Ca²⁺-dependent inactivation require high-affinity binding of calmodulin (CaM) to the IQ motif located in the carboxyl-terminal tail of the Ca_v1.2 channel (11–15). Several additional peptide sequences may be involved in the complex formation of CaM with the channel (reviewed in ref. 16).

In addition or alternatively to these CaM-dependent mechanisms, CDF of Ca_v1.2 has been attributed to the action of Calmodulin kinase II (CaMKII) (17–20). Constitutively active CaMKII facilitated L-type current in excised patches (21). CaMKII is tethered to the C terminus in close proximity to the IQ motif, and CDF was abolished by mutating the putative interaction site (22). CaMKII was reported by us and by others to phosphorylate S1512 and S1570 (23) and S1517 (corresponding to S1512 in the present study) (24) of the rabbit Ca_V1.2 subunit or T498 (25) of the neuronal β_2a subunit of the calcium channel. Mutation of these amino acids to alanine prevented calcium-dependent facilitation in different expression systems.

Thus far, the significance of these phosphorylation sites for frequency- or voltage-dependent facilitation in vivo has remained unclear. The neuronal β_2a subunit has not been found in the murine heart (26, 27), whereas the α₁ subunit of the Ca_V1.2 L-type calcium channel is present in many cell types, including smooth muscle cells, neurons, and cardiomyocytes (CM) (3, 4). The distinct localization of these subunits suggests that facilitation of I_Ca may be triggered by phosphorylation of different subunits that are expressed in a tissue-specific manner.

We have now investigated the functional significance of CaMKII-dependent phosphorylation of S1512/1570 for CDF of cardiac L-type calcium channels by generation of mutant mice using gene knockin techniques. Mutation of these serines to alanines affected facilitation in vitro and in vivo.

Results

Serine 1512 and serine 1570 were mutated to alanines using transgenic gene knockin techniques to test the physiological significance of these amino acids for CaMKII-dependent facilitation (Fig. 1A). The resulting sine facilitation (SF) mice (genotype Ca_V1.2^{S1512A/S1570A} on both alleles) were compared with litter-matched control (Ctr) mice (Ca_V1.2^wt/wt). The mutation in SF mice was confirmed by genomic sequencing (Fig. 1B). Western blot analysis of cardiac ventricles with an antibody specific for the Ca_V1.2 protein detected similar protein levels in Ctr and SF hearts (Fig. 1D). Mutation of serine 1512 and 1570 reduced significantly the phosphorylation of the Cav1.2 protein by CaMKII (Fig. 1 C and E). As expected from previous reports (22, 28), CaMKII still phosphorylated weakly sites not related to the mutated serines in the Cav1.2 protein from SF mice. SF mice were viable, fertile, and reproduced in a 1:2:1 Mendelian ratio (WT, 26.7%; heterozygous, 48.7%; SF, 24.7%; n = 384). SF mice behaved normally in open-field, beam-walking, and footprint tests. The heart rate was not different between Ctr and SF mice when averaged over 1 week by telemetry (Fig. S1A). Fractional shortening measured in the absence and presence of isoproterenol by echocardiography was identical in Ctr and SF animals (Fig. S1B).

Fig. 1.

Generation of SF phosphomutant mice. (A) (Top) genomic DNA structure of CACNA1C. Boxes represent exons 35–38 encoding part of the carboxy-terminus of Ca_v1.2. (Middle) Targeting vector. Neo, neomycin-resistance gene; TK, thymidine kinase gene with loxP sequence (triangles) at both sides. S1512A/S1570A alanine substitutions for serines are shown. (Bottom) Knock-in locus after homologous recombination and Cre-mediated deletion of resistance markers. B, BamHI; C, ClaI; K, KpnI; kb, kilobases; N, NotI; X, XhoI. (B) Sequence analysis of genomic DNA in the region coding for S1512 and 1570 from two Ca_v1.2^SF mice. (C) (Top) Autoradiogram of WT (Ctr) and SF mutant Ca_v1.2 phosphorylated by preactivated CaMKII. Strong phosphorylation was detectable only in the WT protein, whereas Ca_v1.2 expression levels were unchanged. (Middle) Western blot of amount of expressed Ca_v1.2 protein. (Bottom) Quantification of the autoradiography signals (n = 3) normalized to Ctr levels. ***P < 0.001, t test. (D) Immunoblots of cardiac membrane preparations demonstrating normal ventricular expression levels of Ca_v1.2 in hearts of Ca_v1.2^SF (SF) compared with WT (Ctr) mice. β₁-Adrenoreceptor was used as loading control. (E) Absence of phosphorylation at Ser¹⁵¹² in SF phosphomutants. Heart samples from WT (Ctr) and homozygote (SF) mice were prepared for immunoblot analysis using antibodies against phosphorylation site Ser¹⁵¹² (top lane) or total Ca_v1.2 (bottom lane).

Isolated CM of either genotype had normal size (Ctr, 120 ± 6.5 pF; n = 16; SF, 117 ± 13 pF; n = 11) and I_Ca (Ctr, –4.0 ± 0.8 pA/pF at +10 mV; SF, –3.7 ± 0.44 pA/pF at +10 mV) (Table S1A and Fig. S2A) with a normal I–V relation (Fig. S2A). The derived parameters such as half-maximal activation potential, reversal potential, and maximal Ca²⁺ conductance did not differ between genotypes (Table S1A). Half-maximal inactivation under steady-state condition was shifted in the SF CM from −24.9 ± 0.2 mV to −29 ± 1.6 mV (Fig. S2B and Table S1B). The shift in inactivation curve was significant (P < 0.05) and expected, because it was reported that inactivation is the major determinant of facilitation (29). Furthermore, it has been shown that inhibition of CaMKII cause a negative shift of voltage-dependent current inactivation (30, 31).

We anticipated that the SF mutation might also affect recovery from inactivation, because recovery from inactivation is another important determinant of channel availability (32). Recovery from inactivation could be fitted by a single time constant that was identical for both genotypes, when tested at a holding potential of –80 mV (Fig. S2C and Table S1B). The SF mutation significantly prolonged the time course of recovery from inactivation at an interpulse potential of –40 mV (Fig. S2D). This slowing of recovery from inactivation occurred without any significant alteration of the time constants of inactivation during depolarization (Table S1B). The results support the notion that the SF mutation not only affected inactivation but also recovery from inactivation.

We next compared prepulse facilitation of I_Ca in CMs of both genotypes (Fig. 2). After a prepulse to +80 mV for 200 ms, the holding potential was returned to intervals of different length at −80 mV. I_Ca increased almost 2-fold in Ctr CM, when the return to −80 mV was limited to 200 ms. Prolongation of this interval decreased I_Ca almost to normal values. Prepulse facilitation was reduced significantly at the early time points in the SF mice (Fig. 2), supporting the notion that phosphorylation of S1512/1570 enabled maximal voltage-dependent facilitation.

Fig. 2.

Prepulse facilitation of I_Ca was reduced significantly at the early time points in SF compared with Ctr CM. (Upper) Pulse protocol. (Lower) Representative current traces before (1) and after (2) prepulse facilitation for both genotypes at 100 ms. Facilitation was calculated as the ratio of the peak current (I₂/I₁) during V2 and V1. Ctr, n = 10; SF, n = 9. (Scale bars, 20 ms, 200 pA.) *P < 0.05, ** P < 0.01.

It is widely accepted that CDF is caused by activation of CaMKII followed by phosphorylation of a component of the calcium channel complex (6, 21–23, 25). We tested this hypothesis using the specific CaMKII inhibitor KN-93 (Fig. 3). Voltage-dependent facilitation of I_Ca in Ctr CM was reduced by KN-93 to the same level as that of SF CM. In agreement with the postulate that voltage-dependent facilitation depends on the phosphorylation of S1512/S1570 of Ca_V1.2, KN-93 did not further decrease the facilitation of I_Ca in SF CM. This finding makes it highly unlikely that CaMKII-dependent phosphorylation of additional sites is needed to allow voltage-dependent facilitation in murine CM.

Fig. 3.

After pretreatment with 0.5 μM KN-93, prepulse facilitation of Ctr I_Ca is diminished to the same extent as in SF CM. Same pulse protocol as in Fig. 1, with a fixed pulse interval of 100 ms. *P = 0.02, **P < 0.01 compared with Ctr without pretreatment with KN-93.

Calcium channels also show facilitation with an increase in depolarization frequency (5, 6). In this paradigm, facilitation depends on Ca²⁺ influx and an intact sequence around the IQ motif in the carboxy-terminus of the Ca_V1.2 channel (12–14, 19–22, 33). A widely used measure of this type of facilitation is the amount of calcium passing through the channel in consecutive depolarizations (34, 35). An increase in the amount of Ca²⁺ passing through the open channel can be visualized by a change in peak current, inactivation time constants, or both. I_Ca evoked by the first depolarization after a >1-min resting phase at −80 mV was smaller in SF CM than in Ctr CM (P < 0.1) (Fig. S3 A and B). Inactivation time constants were not different between both genotypes (Table S1B and Fig. S3C). These results support the notion that the SF mutation did not significantly affect I_Ca inactivation during a single pulse. Integration of the amount of Ca²⁺ passing through open Ca_V1.2 channels during repetitive depolarization (0.5 Hz) showed that the amount of Ca²⁺ entering through Ca_V1.2 channels was slightly diminished in SF CM compared with Ctr CM (Fig. 4). The effect of the SF mutation on this type of facilitation was smaller than on the voltage-dependent facilitation described above. As expected, this type of facilitation depended on Ca²⁺ as charge carrier and was abolished by intracellular perfusion with 10 mM 1,2-bis(o-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid (Fig. S4).

Fig. 4.

Integrated I_Ca evoked by repetitive depolarization is slightly smaller in SF than in Ctr CM (Ctr, n = 11; SF, n = 10). SF CM exhibit diminished peak I_Ca during the first depolarization (Fig. S3B). Command steps were from –80 mV to 0 mV, 200 ms, 0.5 Hz. I_Na was inhibited by a prepulse to –40 mV.

The experiments carried out so far did not show marked effects of the SF mutation on the in vivo performance of the heart. Wu et al. showed that CaMKII affects the L-type I_Ca of rabbit ventricular myocytes under non–steady-state condition (36), a condition that prevails in vivo. It was reported that the L-type calcium channel contributed to the action potential duration (APD) of CM (37). To test the effect of the SF mutation under non–steady-state condition, action potentials (AP) were recorded in isolated CMs from Ctr and SF mice (Fig. S5). APD33 (in milliseconds) was decreased from 3.6 ± 0.6 (9) in Ctr to 2.3 ± 0.4 (8) in SF CMs at 1 Hz. As expected (38) at 5 Hz, APD33 (in percent) increased to 147.0 ± 7.3 (9) and 131.4 ± 7.0 (8) of the value at 1 Hz in Ctr and SF CMs, respectively. These differences are significant at P < 0.05, and suggest that the SF mutation affected the frequency-dependent modulation of the cardiac excitation.

In the next series of experiments, we analyzed the in vivo ECG of Ctr and SF mice. The ECG is an established in vivo surrogate parameter for APs, and reflects ventricular myocyte depolarization and repolarization in the QRST complex. It was suggested that the “physiological role of I_Ca facilitation is not entirely clear, but it may partly offset reduced Ca channel availability at high heart rates” (39). With this notion in mind, we speculated that SF mice should have a shortened QT time because, in SF mice, the CDF was reduced, whereas Ca²⁺-dependent inactivation was unchanged (Fig. S3). We tested this hypothesis by recording the ECG of freely moving animals at different heart rates. As is evident from Fig. 5A, QT time decreased significantly in SF mice when the frequency changed from 440 to 700 BPM. QT time corrected for frequency (QTc) was 0.096 s and 0.088 s in Ctr and SF CM, respectively. This difference in QTc was highly significant at P < 0.01(n = 40). At the same BPM change, TP time increased in the SF mice (Fig. 5B), resulting in very similar heart rates in Ctr and SF mice.

Fig. 5.

ECG intervals are changed in SF mice at high heart rate. (A) Comparison of QT interval between Ctr and SF mice at 440, 630, and 700 bpm. At 630 and 700 bpm, QT interval is significantly reduced in SF mice (*P < 0.05, ***P < 0.001). (B) Comparison of TP interval shows significant increase at 630 and 700 bpm in SF mice. TP reflects the time between repolarization of the ventricle and the next depolarization of the atria. (**P < 0.01, *** P < 0.001). n = 40 ECG per frequency of four mice from each genotype.

Discussion

[Ca²⁺]_i regulates Ca²⁺ entry through L-type Ca²⁺ channels in the myocardium in an activity-dependent manner. Increased [Ca²⁺]_i results in dual and opposing signals to Ca_V1.2; increased [Ca²⁺]_i enhances L-type I_Ca through a process termed facilitation (40), but also leads to channel inactivation (41), which causes a reduction of I_Ca.

CaM can facilitate I_Ca by activation of CaMKII to induce a modal gating shift that favors prolonged channel openings (21). Emerging data in recent years suggest that mechanisms underlying CaMKII modulation of L-type calcium channels in vivo are complex. Phosphorylation of multiple sites in the Ca_V1.2 α₁ subunit by CaMKII has been proposed as the mechanism for CDF based on overexpression experiments in heterologous cells or virally transfected isolated CM (20, 23, 24). In addition, interactions of CaMKII with multiple intracellular domains of the Ca_V1.2 α₁ subunit have been implicated in calcium channel facilitation (22). The importance of a direct role of the ubiquitous Ca²⁺-binding protein CaM in determining I_Ca facilitation and inactivation in vitro is also well established (13–15, 42). CaM interacts with distinct CaM-binding domains on the C terminus of the Ca_v1.2. The first CaM-binding region identified was an “IQ-like” domain (IQ) (14, 43, 44). IQ is implicated in [Ca²⁺]_i-dependent modulation of I_Ca, because point mutations within IQ result in marked and distinct effects on Ca²⁺–CaM–dependent facilitation and inactivation (12, 13, 15) of heterologously expressed Ca_v1.2 channels. However, the mechanism of CaMKII regulation of endogenous Ca_v1.2 in vivo and the relationship between IQ and CaMKII in the modulation of I_Ca remain unknown to date.

This report supports the notion that CaMKII-dependent phosphorylation of S1512 and S1570 is required to observe voltage- and frequency-dependent facilitation of the cardiac L-type calcium channel. Mutation of the two serines did not affect the I–V relation and the time constants for inactivation, whereas the voltage dependence of inactivation was shifted to more negative potentials, and recovery from inactivation was slowed under certain conditions as has been reported previously (19, 29, 32, 45). The reduced amplitude of the first depolarization after a resting phase at −80 mV in SF CMs suggests that a basal CaMKII activity exists in ventricular myocytes, as is the case in SA node cells (31).

The results of this study confirm that CaMKII enhances the activity of Ca_V1.2 by accelerating recovery from inactivation and positively shifting the voltage dependence of steady-state inactivation (29, 30, 32). Similar to facilitation itself, the effects are not tremendous but consistent. Furthermore, mutation of S1512 and S1570 did not completely abolish voltage- and frequency-dependent facilitation suggesting that other processes contribute to CDF. One possibility is that in addition to the phosphorylation of S1512/S1570 complete CDF requires the modification of the L-type calcium channel at other locations including the β subunit as suggested by reports on the neuronal β_2a subunit (25, 46). However, the results with the specific CaMKII inhibitor KN-93 indicate that this additional mechanism may not involve CaMKII dependent phosphorylation. It might be that the remaining CDF is caused directly by CaM as suggested in previous publications (11, 13–16, 33, 47).

This report supports several conclusions. (i) CaMKII-dependent phosphorylation is involved in facilitation of the cardiac Ca_v1.2 channel in the intact animal, as already suggested by others for isolated CM (21). (ii) CaMKII-dependent phosphorylation of S1512 and S1570 is necessary to observe calcium-dependent facilitation. (iii) Mutation of S1512 and S1570 to alanine affects frequency-dependent changes in the ECG, reflecting a change in ventricular de- and repolarization. (iv) The SF mutation does not affect the time constants of calcium dependent inactivation.

A wide range of in vivo and in vitro studies link AP prolongation, disordered [Ca²⁺]_i, neurohumoral activation, and structural heart disease to increased CaMKII activity as a cause of calcium channel facilitation, early afterdepolarizations, arrhythmias, and sudden cardiac death (reviewed in refs. 10 and 48). Mice with cardiac overexpression of CaMKIV, a CaMK that is not endogenous to heart and that leads to increased endogenous CaMK activity, develop cardiac hypertrophy, marked QT interval and AP prolongation, and arrhythmias (49). Cardiomyocytes from these mice had Ca_v1.2 channels with increased opening probability, as occurs in failing human CM (50). CaMKII inhibition normalized the channel opening probability and APs and eliminated early afterdepolarizations, indicating that Ca_v1.2 CDF was responsible for these electric phenotypes. As seen in Fig. 5, the specific loss of Ca_v1.2 regulation by CaMKII in SF mice counteracts the potentially proarrhythmic QT interval prolongations induced by CAMKII overactivity at high heart rates. Our findings raise the intriguing notion that Ca_v1.2 regulation by CaMKII phosphorylation at S1512 and S1570 may be an important part of the arrhythmogenic circulus vitiosus.

Materials and Methods

Detailed materials and methods are described in SI Materials and Methods. All substances used were of the highest purity available. The Ca_V1.2-specific antibody used in this study has been described previously (51). Amino acid numbering is according to the O. cuniculus Ca_V1.2 sequence (GenBank X60782.1). Generation of mice lacking phosphorylation sites on Ca_V1.2, electrophysiological recordings, telemetric ECG recordings, echocardiography, gene expression, and Western blot analysis are described in SI Materials and Methods. Data were subjected to statistical analysis using OriginPro software, version 6.1 (OriginLab).