Gαi2- and Gαi3-Specific Regulation of Voltage-Dependent L-Type Calcium Channels in Cardiomyocytes

Background Two pertussis toxin sensitive Gi proteins, Gi2 and Gi3, are expressed in cardiomyocytes and upregulated in heart failure. It has been proposed that the highly homologous Gi isoforms are functionally distinct. To test for isoform-specific functions of Gi proteins, we examined their role in the regulation of cardiac L-type voltage-dependent calcium channels (L-VDCC). Methods Ventricular tissues and isolated myocytes were obtained from mice with targeted deletion of either Gαi2 (Gαi2 −/−) or Gαi3 (Gαi3 −/−). mRNA levels of Gαi/o isoforms and L-VDCC subunits were quantified by real-time PCR. Gαi and Cavα1 protein levels as well as protein kinase B/Akt and extracellular signal-regulated kinases 1/2 (ERK1/2) phosphorylation levels were assessed by immunoblot analysis. L-VDCC function was assessed by whole-cell and single-channel current recordings. Results In cardiac tissue from Gαi2 −/− mice, Gαi3 mRNA and protein expression was upregulated to 187±21% and 567±59%, respectively. In Gαi3 −/− mouse hearts, Gαi2 mRNA (127±5%) and protein (131±10%) levels were slightly enhanced. Interestingly, L-VDCC current density in cardiomyocytes from Gαi2 −/− mice was lowered (−7.9±0.6 pA/pF, n = 11, p<0.05) compared to wild-type cells (−10.7±0.5 pA/pF, n = 22), whereas it was increased in myocytes from Gαi3 −/− mice (−14.3±0.8 pA/pF, n = 14, p<0.05). Steady-state inactivation was shifted to negative potentials, and recovery kinetics slowed in the absence of Gαi2 (but not of Gαi3) and following treatment with pertussis toxin in Gαi3 −/−. The pore forming Cavα1 protein level was unchanged in all mouse models analyzed, similar to mRNA levels of Cavα1 and Cavβ2 subunits. Interestingly, at the cellular signalling level, phosphorylation assays revealed abolished carbachol-triggered activation of ERK1/2 in mice lacking Gαi2. Conclusion Our data provide novel evidence for an isoform-specific modulation of L-VDCC by Gαi proteins. In particular, loss of Gαi2 is reflected by alterations in channel kinetics and likely involves an impairment of the ERK1/2 signalling pathway.


Introduction
G protein-mediated signalling plays a central role in regulation of cardiomyocyte function. Heterotrimeric G proteins consist of three subunits, Ga, Gb, and Gc. Agonist-occupied receptors induce dissociation of GDP from and binding of GTP to the G protein a subunit, resulting in G protein activation. Activated Ga and Gbc subunits couple to a plethora of effectors, including enzymes and ion channels, and hence are involved in many regulatory processes [1,2]. The role of stimulatory G s and inhibitory G i proteins in cardiac signalling pathways is well studied [3,4]. Alterations of Ga i protein expression levels are found in heart disease [5], and heart failure in humans leads to upregulation of Ga i2 and Ga i3 [6,7,8,9]. Whether the upregulation of Ga i2 and Ga i3 in cardiomyocytes is causative, adaptive, or maladaptive still remains unclear.
Cardiac calcium channels are key components in complex signal transduction pathways and play an essential role in cardiac excitability and in coupling excitation to contraction [10]. One major pathway regulating calcium channels is mediated via G protein-mediated signalling. In the heart, the main sarcolemmal calcium channel is the voltage-dependent L-type calcium channel (L-VDCC). This channel is composed of three different subunits. The a 1 subunit represents the pore forming subunit which contains the voltage sensor and the binding sites for calcium channel modulators [11]. It associates with two auxiliary subunits, b, and a 2 d [12]. The functional properties of the pore forming subunit are differentially modified due to interaction with various b subunit isoforms [13,14,15,16]. Furthermore, receptor activated Ga s protein stimulates L-VDCCs via adenylyl cyclase-mediated increases in cAMP levels and protein kinase A (PKA) activity [3]. Activation of G i or G o modifies channel function via diverse signal cascades [17]. Thus, G protein signalling pathways are crucial in determining and balancing cardiomyocyte function in vivo.
In a previous study we addressed the role of Ga i2 in b 2adrenergic receptor-mediated signalling. Gene deletion of Ga i2 in mice reduced single L-VDCC activity in b 2 -adrenergic receptortransgenic mice [18], whereas pertussis toxin (PTX) treatment reversed this effect. We speculated that this unexpected effect of PTX may have been caused by inhibiting an upregulated Ga i3 . Recently, Zuberi et al. [19] showed that Ga i2 knockout leads to increased L-VDCC mRNA expression and a propensity towards ventricular arrhythmia. Muscarinic receptor-mediated inhibition of L-VDCC activity has been reported to depend on Ga i2 but not Ga i3 [20]. Though strongly suggested by these data, subtypespecific effects on cardiac L-VDCC by the highly homologous Ga i2 and Ga i3 isoforms remain unclear so far. Therefore, the present work was undertaken to elucidate whether the effects of these Ga i proteins are redundant or distinct. Using cardiomyocytes from mice lacking Ga i2 or Ga i3 and wild-type (WT) control animals, we determined structural and functional changes. Further, we examined specific signalling pathways implicated in cardiac L-VDCC modulation by Ga i protein. In this work, we provide evidence that the L-VDCC activity and kinetics are regulated in a non-redundant manner and we support this idea by demonstrating subtype-specific activation of the extracellular signal-regulated kinases 1/2 (ERK1/2) signalling cascade.

Ga i2 deficiency decreases, while Ga i3 deficiency increases L-VDCC current density
To assess consequences of selective deletion of Ga i2 or Ga i3 genes, we measured whole-cell L-VDCC currents in cardiomyocytes from WT, Ga i2 2/2 , and Ga i3 2/2 mice. Currents recorded at different test potentials are shown as representative original recordings and current-voltage diagrams of summarized data in Fig. 1A and B, respectively. In cardiomyocytes of Ga i2 2/2 mice, the calcium current density at 0 mV was slightly but significantly reduced (27.960 gating account for the observed differences in current density, we examined kinetic and steady state properties of activation and inactivation.
L-VDCC kinetics are altered by Ga i2 deletion, but not by Ga i3 deletion The steady-state inactivation properties (Fig. 1C) in Ga i2 2/2 cardiomyocytes were altered compared to WT cells as reflected by a significant leftward shift of V 0.5 (Ga i2 2/2 : 223.461.0 mV, n = 11, WT: 219.260.7 mV, n = 18) and an increased slope factor (Table 1). In addition, recovery from inactivation was slowed in cardiomyocytes from mice lacking Ga i2 (t: 287621 ms, n = 9, p,0.05) in comparison to WT animals (t: 215614 ms, n = 16; Fig. 1D). In contrast, whole-cell currents in cells form Ga i3 2/2 mice were indistinguishable from WT regarding both steady state inactivation (V 0.5 : 218.260.9 mV, n = 13, Table 1) and recovery from inactivation (t: 203624 ms, n = 9). Thus, currents from Ga i2 2/2 mice show altered kinetic properties and this might explain the decreased current density described above (Fig. 1). Despite increased current density no alteration of current kinetics in Ga i3 2/2 myocytes was visible. Thus we analyzed singlechannel activity to elucidate the opposing effects seen on wholecell currents in Ga i2 2/2 and Ga i3 2/2 cardiomyocytes.
No major changes in gating properties of single L-VDCC in Ga i3 2/2 cardiomyocytes Single-channel current recordings in Ga i3 2/2 cardiomyocytes revealed a trend towards increased peak ensemble average currents (Fig. 2B) and higher open probability (Fig. 2C) when compared to WT cardiomyocytes ( Table 2). These effects are based on a significant reduction of the mean closed time (3.860.5 ms vs. 6.660.9 ms, n = 6-7, p,0.05; Fig. 2D). Together with a significantly decreased slow time constant of the closed state and by trend a reduced first latency ( Table 2) our single-channel data suggest that in Ga i3 2/2 exit from deeper closed states of L-VDCC is facilitated. Given that single-channel activity in Ga i2 2/2 mice was decreased by trend [18] our findings presented here suggest that Ga i2 2/2 or Ga i3 2/2 knockout leads to distinct changes of cardiac L-VDCC properties. Yet, these only slight functional changes alone do not elucidate the more remarkable augmentation of calcium current density in Ga i3 2/2 . Since the remaining Ga i isoform might have compensated for effects of Ga i3 deficiency we next checked the expression levels of Ga i2 in Ga i3 knockouts and vice versa.

Enhanced expression levels of remaining Ga i isoform
If the generally accepted assumption holds true that Ga i2 and Ga i3 proteins are functionally redundant in the heart, we could expect a compensatory upregulation of the remaining Ga i subunit after knockout of the other. We first determined mRNA expression levels for Ga i1 , Ga i2 , Ga i3, and Ga o in samples from WT, Ga i2 2/2 , and Ga i3 2/2 mice using real-time PCR. In ventricular tissue from WT mice, transcripts for all Ga i protein isoforms and Ga o were found in different amounts (Fig. 3A). While Ga i2 and Ga i3 are known to be expressed in cardiomyocytes -with Ga i3 mRNA being clearly less abundant -, Ga i1 and Ga o are likely transcribed in non-cardiomyocyte ventricular cells [21]. Deletion of Ga i2 enhanced the mRNA level of Ga i3 (to 187621%, n = 3, p,0.05 vs. WT), as expected [6,22]. In cardiac tissue of Ga i3 2/2 mice Ga i2 mRNA levels were upregulated to only 12765% (n = 4, p,0.05 vs. WT). No significant changes in Ga o or Ga i1 mRNA expression levels were detected, which is in line with the assumption, that these G proteins are not expressed in cardiomyocytes (see above and Fig. 3B).
Next, protein expression of Ga i isoforms was analyzed in cell membrane preparations from WT, Ga i2 2/2 , and Ga i3 2/2 ventricles by probing with Ga common antibodies subsequent to high resolution urea/SDS-PAGE-separation of proteins. WT mice expressed both Ga i2 and Ga i3 subunits with the protein level of Ga i2 being much higher than that of Ga i3 (Fig. 3C), in line with previous studies (e.g. [22]). As expected, in hearts from Ga i2 2/2 mice Ga i3 was found and Ga i2 was absent while in hearts from Ga i3 2/2 mice, only Ga i2 was detectable (Fig. 3C). The specificity of the detected Ga i2 and Ga i3 protein bands in Ga i2 2/2 , Ga i3 2/2 and WT cardiac tissue was confirmed by analyzing the expression of Ga i isoforms by PTX-mediated [ 32 P]ADP ribosylation [23] (data not shown). These experiments prove that gene deletion indeed led to loss of the Ga subunit. Statistical analysis of relative expression levels obtained from immunoblots ( Fig. 3D) show that Ga i2 protein is significantly upregulated in ventricles obtained from Ga i3 2/2 mice (to 131610%, n = 8, p,0.05 vs. WT). However, Ga i3 protein is much more markedly upregulated upon Ga i2 deficiency (to 567659%, n = 8, p,0.05 vs. WT). Taken together, knockout models examined in this study feature a protein upregulation of the respective other Ga i isoform. These data, together with quantitative mRNA data reported above, suggest that increased levels of Ga i2 or Ga i3 may partially compensate for the loss of the deleted other Ga i isoform. Hence, any effect seen so far could be caused by the loss of one Ga i protein and/or compensatory signalling exerted by the other. Unfortunately, double-deficient (Ga i2 2/2 /Ga i3 2/2 ) mice are not viable [22] and hence cannot be used to directly address this question. However, all G i proteins can acutely be inactivated by PTX treatment.

Acute G i inactivation induces L-VDCC kinetic alterations in Ga i3
2/2 cardiomyocytes We performed experiments with PTX for acute inactivation of G i/o proteins in cardiomyocytes of WT and Ga i -deficient mice. Freshly isolated cardiomyocytes were incubated at 37uC with or without PTX for 3 hours and afterwards maintained and examined at room temperature. It has been shown that this protocol completely ablates G i -mediated signalling, e.g. when triggered by adenosine receptors [24]. Of note, the 3 hours incubation phase -even in the absence of PTX -reduced current density in Ga i3 -deficient cells (Fig. 4A). Notably, similar to WT cells, PTX treatment in Ga i3 2/2 myocytes induced a shift of V 0.5 for steady state inactivation to more negative potentials (to 222.761.1 mV, n = 11; Fig. 4B). Likewise, the recovery from inactivation in Ga i3 2/2 cardiomyocytes was slowed by PTX treatment (to t: 350627 ms, n = 8 p,0.05 vs. WT; Fig. 4C). In contrast, the calcium current phenotype in Ga i2 2/2 cells was preserved with time and resistant to treatment with PTX. To summarize, PTX alters channel regulation in cells from Ga i3 2/2 but not from Ga i2 2/2 mice. In Ga i3 2/2 , acute PTX treatmentand thus inactivation of Ga i2 2/2 -mimics the kinetic changes induced by chronic gene deletion of Ga i2 2/2 . In contrast, the effect of Ga i3 gene deletion appears to be transient and does not interfere with channel kinetics. Because some of the calcium current changes reported here may be caused by (long-term) structural alterations rather than (acute) functional modulation of calcium channels, we next examined channel composition.

No significant structural modification of L-VDCC
To obtain insight into the effects of specific and constitutive Ga ideficiency on L-VDCC structure and expression, we determined RNA expression levels of the pore forming Ca v a 1 subunit and the predominant murine cardiac Ca v b subunit, Ca v b 2 [14,16,25], which is involved in calcium channel trafficking and gating [26,27,28]. The Ca v b 2 subunit mRNA expression in ventricles from Ga i2 2/2 and Ga i3 2/2 animals is not altered compared to WT expression profile (Fig. 5B). In Ga i2 2/2 and Ga i3 2/2 ventricles, the pore forming Ca v a 1 subunit mRNA (Fig. 5B) and membrane protein ( Fig. 5C and D) expression levels exhibit only slight and insignificant changes. Thus, these data do not suffice to explain the Ga i isoform-specific regulation of the current density. Therefore, we next switched to the posttranslational level and addressed signalling pathways that are involved in G i proteinmediated action. Akt activation is unaltered by deletion of Ga i2 or Ga i3 Cardiac Ga i is known to activate the PI3-kinase Akt/PKB pathway [29], and Akt-mediated b-subunit phosphorylation prevents Ca v a 1 degradation [30]. A G i -isoform-specific regulation of Akt could explain the calcium current increase in the case of Ga i3 2/2 and decrease in case of Ga i2 2/2 . To examine the functional significance of Ga i -dependent activation of Akt in vivo, animals were treated with either saline or the muscarinic receptor agonist carbachol (CCh, 0.5 mg/kg body weight) by i.p. injection; 15 min later animals were killed, and signalling activity was assessed in heart preparations. Phosphorylation of Akt and its downstream effector glycogen synthase kinase-3a/b (GSK3a/b), were increased in cardiac tissue from all mouse strains after treatment with CCh (Fig. 6A). Although the basal phosphorylation of Akt in Ga i2 2/2 and Ga i3 2/2 cardiac tissue was slightly increased compared to WT (to 138644% and 143658% of WT, respectively), we could not observe statistically significant differences between CCh stimulated WT (233614% to basal), Ga i2 2/2 (201627%) and Ga i3 2/2 (195625%) mice (each n = 3). Importantly, the total amount of Akt was not changed in all mice models. Thus, G i -subtype specific channel regulation seems to be independent of Akt phosphorylation in the investigated knockout models.

Lack of Ga i2 protein abolishes ERK1/2 activation
Recently, a marked increase of L-type calcium channel density that involved PKC-dependent activation of the ERK1/2 pathway was reported [31]. To determine its involvement, we measured the Analysis of single-channel gating parameters in ventricular myocytes from WT (n = 7) and Ga i3 2/2 (n = 6) mice. Recordings with more than one channel were excluded from the analysis.   mice demonstrated a significantly blunted increase of ERK1/2 phosphorylation in CCh-stimulated animals (12865% to basal, n = 3) compared to WT (26866%, n = 3) and Ga i3 2/2 (192620%, n = 3) cardiomyocytes. Of note, in Ga i2 2/2 and Ga i3 2/2 cardiac tissues, the basal ERK1/2 phosphorylation levels were increased compared to WT basal phosphorylation (Fig. 6B), while the total amount of ERK1/2 in the heart was the same in all mouse models tested. These results indicate a Ga i2 -dependent ERK1/2 phosphorylation and strongly suggest that ERK1/2 plays an important role in isoform-specific Ga i protein signalling.

Discussion
The two inhibitory G protein isoforms G i2 and G i3 are both upregulated in heart failure [6,7,8]. One functionally important target of G i protein signalling is the L-VDCC, the crucial trigger of cardiac excitation-contraction coupling. G i -protein-mediated inhibition of L-VDCC has been demonstrated for b 2 -adrenergic [4] and muscarinic [20] receptor signalling. In this context, we previously provided single-channel evidence that Ga i2 does not confer the L-VDCC inhibition observed in mice with chronic overexpression of the b 2 -adrenergic receptor [18]. On the other hand, cardiac Ga i2 (but not Ga i3 ) seems necessary and sufficient to mediate the muscarinic receptor-mediated L-VDCC inhibition [20], presumably through the classical adenylyl cyclase pathway. So far, no isoform-specific function could be assigned to cardiac Ga i3 ; however, G i3 has been shown to be an exclusive and specific regulator of autophagy in the liver [22,32].
The different behaviour of L-VDCC currents obtained with isolated myocytes from Ga i2 2/2 and Ga i3 2/2 mice shown here and previously [18] demonstrates contrasting functional roles of these two Ga i isoforms. Although the issue has also been addressed by others [19,20,33], our study is the first to demonstrate small but significant changes of basal whole-cell current density: a reduction in myocytes from Ga i2 2/2 mice and an increase in myocytes from Ga i3 2/2 mice. The altered steady-state inactivation and recovery observed with Ga i2 2/2 under basal conditions ( Fig. 1C and D), and with Ga i3 2/2 myocytes after PTX treatment ( Fig. 4B and C), point to a modulation of gating properties specific to Ga i2 . In contrast to data presented here, Nagata et al. [20] did not detect a significant difference in L-VDCC activity between myocytes from WT, Ga i2 2/2 and Ga i3 2/2 mice. With respect to Ga i2 2/2 , this finding can be explained by the different prepulse potentials: the prepulse voltages (250 mV used by Nagata and co-workers vs. 240 mV in our case) -intended to inactivate primarily sodium currents -lie within the descending part of steady-state inactivation (Fig. 1C). This likely translates into the more reduced peak in the current voltage plot in our study. Interestingly, Zuberi et al. [19] compared Ga i2 2/2 mice with Ga i1 2/2 /Ga i3 2/2 double knockout mice and found distinct effects on surface ECGs: in Ga i2 2/2 animals (but not in the double knockout mice), the effective refractory period was reduced and ventricular arrhythmias were induced more easily. In summary, Ga i2 protein deletion showed dramatic consequences on channel regulation in vivo and ex vivo.
There are no changes of cardiac L-VDCC composition regarding the main cardiac L-VDCC subunits, Ca v a 1 and Ca v b 2 ( Fig. 5B and D) that would explain the obtained effects on current density. Furthermore, because of the compensatory upregulation of the remaining Ga i isoform in case of either Ga i2 or Ga i3 deficiency ( Fig. 3B and D), it is difficult to attribute the observed changes in L-VDCC regulation/function to the higher expression of one Ga i isoform or to the loss of the other or to both. Therefore, given the novel functional effects reported here for cardiac Ga i2 and Ga i3 , we have to consider a number of molecular pathways. For instance, activation of Stim1 [34] may lead to altered L-VDCC function and subcellular distribution. Enhanced endocytosis and degradation of calcium channels can also be mediated by activation of PIKfyve [35] and subsequent Ca v a 1 targeting to lysosomes. Further, the RGK proteins Rad and Rem expressed in the heart [36] are appealing candidates, because they negatively regulate both membrane expression and gating of L-VDCC [37,38] while little is known about how these small GTP-binding proteins are regulated themselves [39]. In our present study, we focused on two other important molecular pathways since they are significantly regulated by G i -signalling. First, PI3-kinase Akt/PKB signalling is known to be activated by cardiac Ga i [29], and Catalucci et al. [30] revealed a mechanism through which the PI3kinase Akt/PKB pathway modulates Ca 2+ entry in cardiac cells via L-VDCC. Our data showed a CCh-induced Akt phosphorylation independent of the deletion of either Ga i isoform (Fig. 6A). Second, it was demonstrated that deactivation of G i leads to a significant reduction in ERK1/2 phosphorylation and that this effect was G s -and G q -independent [29,40]. In the present study, we have shown that Ga i2 deletion prevents phosphorylation of ERK1/2. Recently, Smani et al. [31] found a leftward shift and a marked increase in L-VDCC density induced by urocortin, which involved PKC-dependent activation of the MAPK-ERK1/2 pathway. Based on these findings we propose that loss of activation of L-VDCC by ERK1/2 might be a mechanism involved in functional regulation of calcium current in Ga i2 deficient mice.
A change of Ga s mediated signalling might account for altered calcium currents when a Ga i is lacking. With this caveat in mind, we demonstrated in our previous study that Ga s protein expression remained unaffected in hearts from mice deficient in the major isoform Ga i2 [18]. Yet, possible changes in associated proteins like Gbc subunits might indirectly affect Ga mediated signalling [41]. Indeed, we observed that expression of Gb 1/2 was slightly reduced in Ga i2 2/2 , but not Ga i3 2/2 (data not shown). In any case, our single-channel analysis (in particular, the decrease in open time) does not support the idea of enhanced Ga s mediated, cAMPmediated signalling in Ga i3 -deficient hearts ( Table 2).
The data presented here could not elucidate all effects seen in the knockout animals. Thus, eventually, G i3 's role in L-VDCC regulation remains unclear, mainly in light of the absence of acute PTX effects in Ga i2 2/2 mice. However, due to the effects seen by incubation without PTX, our findings suggest that the increment in channel activity observed in the absence of Ga i3 might be driven by an in vivo mechanism, which is not preserved ex vivo (Fig. 4A). It has also to be pointed out that the immunoblot data reveal total cardiac membrane channel protein levels, which does not necessarily match up the fraction of functional channels located in the sacrolemma. Given our currently available methods to analyze the subcellular localization of calcium channels and their regulation, all of these ideas require further work, which for technical reasons has to be done in recombinant systems.
Taken together, our data reported here and in a previous paper [18] point to the (patho-) physiological importance of subtypespecific G i protein signalling in the heart. In particular, in terminal heart failure, Ga i2 upregulation now appears as an attractive mechanism linked to remodelling of L-VDCC [13,14,15,16]. Therefore, the present study provides new insights into potential mechanisms linking modulation of L-VDCC to the inhibitory G protein isoform G i2 in cardiomyocytes, and highlights G i2 -specific signalling via ERK1/2. Further research needs to focus on detailed signalling pathways involving ERK1/2.

Animals
Generation, breeding and characterization of Ga i2 -or Ga i3deficient mice have been described previously [18,22,42,43,44,45]. All Ga i -deficient mouse strains used were backcrossed onto a C57Bl6 background for .10 generations. Knockout and WT control mice were maintained at the animal facilities of the Heinrich-Heine-University, Düsseldorf, and of the Department of Pharmacology at the University of Cologne. Mice analyzed in this study were of both sexes, 3-9 months of age and weighted 20-35 g.

Genotyping
Tail-clip analysis was performed on 3-4 weeks old mice. Genomic DNA was prepared and genotyping PCR for Ga i2 and Ga i3 was performed as described previously [18,43].

Real time PCR
Primer for Ga i1 , Ga i2 , Ga i3 isoforms and Ca v a 1 subunit were described previously [25,46]. Specific primers for the Ga o and Ca v b 2 subunit were designed using Primer Express Software v3.0 Figure 6. Signalling events downstream of Ga i stimulation. Representative western blot of cardiac tissues of animals treated either with saline or CCh (0.5 mg/kg body weight) for 15 min. (A) Western Blot and densitometry analyses of ventricular homogenates show that Akt phosphorylation was significantly and to a similar extent increased after CCh stimulation in all genotypes (n = 3). Total amount of Akt protein was unaltered and was used as loading control. (B) Western blot showing ERK1/2 phosphorylation and total ERK1/2 expression and bar graphs of combined results expressed as increase in pERK1/2 normalized to total ERK1/2 and compared to saline treated WT mice. The increase of ERK1/2 activation in CCh treated cells was markedly inhibited by Ga i2 gene deletion compared to WT and Ga i3 2/2 mice (n = 3). *p,0.05 vs. WT NaCl. doi:10.1371/journal.pone.0024979.g006 (Applied Biosystems, Foster City, USA). For Ga o : 59-TGGC-ATCGTAGAAACCCACTT-39 (sense) and 59-CGACGTCAAA-CAGCCTGAAG-39 (antisense) and for Ca v b 2 : 59-GGGAGG-CAGTACGTAGAGAAGCT-39 (sense) and 59-TGCAAATG-CAACAGGTTTT GTC-39 (antisense). Total cellular RNA was extracted from murine heart (ventricle) according to the manufacturer's protocol (Qiagen QIAshredder and RNeasy Mini Kit, Qiagen, Hilden, Germany). For qualitative analysis of RNA integrity, 2 mg of total RNA was separated on a 1% formaldehyde agarose gel. Total RNA was subsequently converted into cDNA by ImProm-II Reverse Transcription Kit (Promega, Mannheim, Germany). Real-time PCR was carried out using the 7500 Real-Time PCR system (Applied Biosystems) under standard conditions with 200 pM PCR primers. Each sample was analyzed in triplets using SYBR green (Applied Biosystems) as fluorescent detector and GAPDH as endogenous control.

Cell membrane preparation
Murine cardiac ventricles were disrupted and homogenised in a lysis buffer containing 10 mM Tris-HCl (pH 7.4), 1 mM EDTA, 0.5 mM DTT, and an EDTA-free protease inhibitor cocktail (Roche, Penzberg, Germany) using an ultra-turrax blender. Cellular membranes were isolated by two steps of centrifugation at 450 g and 30.000 g. Membrane pellets were subsequently dissolved in a buffer consisting of a freezing supplement (70 mM Tris (pH 7.4), 12 mM MgCl 2 , 60% Glycerol, 240 mg/ml DNAse) and the lysis buffer in a ratio of 1:6 in order to stabilize membraneassociated proteins.

Phosphorylation assay
Animals were injected i.p. with 0.5 mg/kg carbachol (CCh) diluted in 0.9% normal saline. Sham injections were performed with 400 ml 0.9% saline per 30 g weight. After 15 min ventricular tissue was harvested and lysed in buffer (50 mM HEPES, 1% Triton, 50 mM NaCl, 10 mM Na 3 VO 4 , 0.1% SDS, 0.1 M NaF, 10 mM EDTA, complete mini protease inhibitor, pH 7.4) and left for 30 min at 4uC. Total cell lysates were extracted from the supernatant by centrifugation at 13.500 g for 20 min.
Immunoblotting Ga i and Ca v a 1 proteins isolated from cell membranes were separated on 6 M urea/9% SDS-PAGE gels (protein content per lane was 90 mg) and on 8% SDS-PAGE gels (protein content per lane was 100 mg), respectively [22]. Protein kinase B/Akt and ERK1/2 in total cell lysates were separated on 10% SDS-PAGE gels (100 mg protein content per lane). Separated proteins were blotted onto nitrocellulose membranes (Hybond C extra; Amersham Bioscience). Ga i proteins were detected with an anti-Ga common antibody [45] (1:1000) and Ca v a 1 with an anti-Ca v a 1 antibody (1:200; Sigma Aldrich). For detection of phosphorylation, membranes were incubated with phospho-Akt (Ser473), phospho-GSK3a/b (Ser21/9) and phospho-ERK1/2 (Thr202/Tyr204) antibodies. Membranes were reprobed with Akt and ERK1/2 antibodies (each 1:1000; Cell Signalling) after stripping. Emitted light of stained membranes were captured on films and developed in different expositions. Protein densities were calculated using Aida Image Analyzer (Raytest, Straubenhardt, Germany) software. The Ras-GTP-activating protein (RasGAP; [47]) was used as a loading control for Ca v a 1 . Equal loading on blotting membrane for Ga i proteins was controlled by a non-specific protein staining using Ponceau S. Only blots with equal loading were analyzed. To confirm Ga i band specificity, we performed ADP ribosylation of PTX-sensitive G proteins as described [23].

Cardiomyocyte isolation
Single ventricular myocytes were isolated from hearts of 3-9 months old mice by enzymatic dissociation using a method described previously [48]. Only rod shaped cardiomyocytes were used for the experiments. Cells were maintained at room temperature and subjected to patch-clamp analysis. If indicated, a fraction of isolated ventricular myocytes was incubated without or with 1.5 mg/ml of PTX (Sigma Aldrich, St. Louis, USA) for 3 hours at 37uC [24].

Whole-cell current measurements
Conventional whole-cell patch clamp recordings were performed with cells maintained at room temperature in bath solution containing (mM): NaCl 137, CsCl 5.4, CaCl 2 2, MgCl 2 1, glucose 10, HEPES 10 (pH 7.4 with NaOH). Pipette (2-3 MV) solution was composed of (mM): CsCl 120, MgCl 2 1, Mg-ATP 4, EGTA 10, HEPES 5 (pH 7.2 with CsOH). Giga-Ohm seals (resistance 2-5 GV) were formed by gentle suction. At the beginning of each experiment, membrane capacitance was measured by means of fast depolarizing ramp pulses from 280 to 285 mV over 25 ms. Cells were depolarized from a holding potential of 280 mV to a 50 ms prepulse to 240 mV in order to inactivate sodium channels. This was followed by test pulse voltages ranging from 240 to +50 mV in 10 mV steps (pulse duration 150 ms). For timedependent inactivation, the declining raw currents at 210, 0, +10, +20 and +30 mV were fitted by a double-exponential function, yielding fast and slow time constant. For investigation of gating kinetics, standard two-pulse protocols were used: the voltagedependent inactivation was measured after a prepulse of variable amplitude and 250 ms duration, followed by a test pulse of fixed amplitude for 50 ms. The midpoint voltage V 0.5 was determined by fitting a Boltzmann function to the data. Recovery from inactivation was determined by two 200 ms depolarizing voltage pulses to 0 mV. The interpulse interval at a holding potential of 245 mV was increased from 50 to 375 ms in 25 ms steps. Recovery was fitted by a mono-exponential function, yielding the recovery time constant t.

Statistical analysis
Data are presented as means 6 SEM. Patch-clamp data were analyzed using pClamp software (CLAMPEX 6 and FETCHAN, Axon Instruments). Analysis of L-VDCC kinetics was performed as described previously [49] using GraphPad Prism 4. A one-sample ttest was used to analyze normalized data. Differences between genotypes were analyzed by one-way ANOVA followed by Dunnett's or Tukey's post test, and between treatments within one given genotype by two-tailed Student's t tests. For electrophysiological statistics, number of cardiomyocytes from a minimum of 3 animals, and for molecular biology statistics, number of animals was evaluated. P values,0.05 were considered statistically significant.