Akt regulates L-type Ca2+ channel activity by modulating Cavalpha1 protein stability.

The insulin IGF-1-PI3K-Akt signaling pathway has been suggested to improve cardiac inotropism and increase Ca(2+) handling through the effects of the protein kinase Akt. However, the underlying molecular mechanisms remain largely unknown. In this study, we provide evidence for an unanticipated regulatory function of Akt controlling L-type Ca(2+) channel (LTCC) protein density. The pore-forming channel subunit Ca(v)alpha1 contains highly conserved PEST sequences (signals for rapid protein degradation), and in-frame deletion of these PEST sequences results in increased Ca(v)alpha1 protein levels. Our findings show that Akt-dependent phosphorylation of Ca(v)beta2, the LTCC chaperone for Ca(v)alpha1, antagonizes Ca(v)alpha1 protein degradation by preventing Ca(v)alpha1 PEST sequence recognition, leading to increased LTCC density and the consequent modulation of Ca(2+) channel function. This novel mechanism by which Akt modulates LTCC stability could profoundly influence cardiac myocyte Ca(2+) entry, Ca(2+) handling, and contractility.

Introduction

The IGF-1 (insulin growth factor 1)–PI3K (phosphatidylinositol 3-kinase)–Akt pathway plays a crucial role in a broad range of biological processes involved in the modulation of local responses as well as processes implicated in metabolism, cell proliferation, transcription, translation, apoptosis, and growth. In the heart, the IGF-1–PI3K–Akt pathway is involved in the regulation of contractile function, and impairment of this signaling pathway is considered an important determinant of cardiac function (Condorelli et al., 2002; McMullen et al., 2003; Ceci et al., 2004; McMullen et al., 2004; Catalucci and Condorelli, 2006; Sun et al., 2006).

The Akt (also called PKB) family of Ser/Thr kinases consists of three isoforms (Akt-1, -2, and -3) that are activated by IGF-1 and insulin through PI3K, which is a member of the lipid kinase family involved in the phosphorylation of membrane phosphoinositides (Ceci et al., 2004). The product of PI3K binds to the pleckstrin domain of Akt and induces its translocation from the cytosol to the plasma membrane, where Akt becomes accessible for phosphorylation by PDK1 (phosphoinositide-dependent kinase 1), resulting in its activation (Ceci et al., 2004; Bayascas et al., 2008). The Ca2+ current (ICa,L) in both cardiomyocytes and neuronal cells has been shown to be increased by Akt activation (Blair et al., 1999; Viard et al., 2004; Catalucci and Condorelli, 2006; Sun et al., 2006) and decreased by Akt inhibition (Viard et al., 2004; Catalucci and Condorelli, 2006; Sun et al., 2006), suggesting a pivotal role of Akt in regulating L-type Ca2+ channel (LTCC) complex function.

In cardiomyocytes, the LTCC is composed of different subunits: the pore-forming subunit Cavα1 and the accessory β and α2δ subunits (Catterall, 2000; Bourinet et al., 2004). The opening of the LTCC is primarily regulated by the membrane potential and by other factors, including a variety of hormones, protein kinases, phosphatases, and accessory proteins (Bodi et al., 2005). In healthy cardiomyocytes, electrical excitation starting during the upstroke of the action potential leads to cytosolic Ca2+ influx through opening of the LTCC (Bers and Perez-Reyes, 1999; Richard et al., 2006). This triggers the CICR (Ca-induced Ca release) of intracellular Ca2+ from the sarcoplasmic reticulum (SR) through activation of the ryanodine receptor (Ryr), eventually leading to cardiomyocyte contraction (Bers, 2002).

The importance and ubiquity of Ca2+ as an intracellular signaling molecule suggests that altered channel function could give rise to widespread cellular and organ defects. Indeed, a variety of cardiovascular diseases, including atrial fibrillation, heart failure, ischemic heart disease, Timothy syndrome, and diabetic cardiomyopathy, have been related to alterations in the density or function of the LTCC (Mukherjee and Spinale, 1998; Quignard et al., 2001; Bodi et al., 2005; Pereira et al., 2006). However, the molecular basis for dysregulation of LTCC function and the possible involvement of Akt in ICa,L regulation remain unresolved.

Recently, a seminal study in neuronal cells revealed the importance of Akt-dependent phosphorylation of the Cavβ2 subunit in promoting the chaperoning of the Cav2.2 pore-forming unit to the plasma membrane (Viard et al., 2004). In this study, we identify a novel posttranslational mechanism by which Akt modulates LTCC function under physiological conditions, highlighting the pivotal role of this kinase in cardiac function. Interestingly, our results show that the pore-forming channel subunit Cavα1 contains highly conserved PEST sequences that direct rapid protein degradation and demonstrate that Akt-mediated phosphorylation of the Cavβ2 LTCC chaperone subunit prevents PEST site recognition, thereby slowing or preventing Cavα1 degradation. This mechanism of action might be an essential process for Ca2+ channel functional regulation, thus contributing to normal or better cardiomyocyte contractile function.

Results

To gain insight into the mechanism of action by which Akt regulates ICa,L and Ca2+ handling in the heart, we studied a mouse line with tamoxifen-inducible (Sohal et al., 2001) and cardiac-specific deletion of PDK1, the upstream activator of all three Akt isoforms. Mice in which exons 3 and 4 of the pdk1 gene were flanked by loxP excision sequences (Lawlor et al., 2002) were crossed with transgenic (Tg) mice expressing an inducible and cardiac-specific MerCreMer α-MHC promoter driving the cre recombinase gene (Sohal et al., 2001), resulting in MerCreMer α-MHC PDK1 mice (knockout [KO]). As opposed to the previously described muscle creatine kinase–Cre PDK1 mouse model (Mora et al., 2003) in which PDK1 is embryonically deleted in all striated muscles, this model allows for specific deletion of PDK1 in the adult heart. A further advantage of this model is the inducible cardiac-specific deletion that was necessary to circumvent the embryonic lethality we observed in a mouse model with constitutive α-MHC–Cre cardiac deletion of PDK1 (unpublished data). Similar to the muscle creatine kinase–Cre PDK1 mouse model (Lawlor et al., 2002), PDK1 gene deletion in the adult mouse heart (KO; Fig. S1, A and B) resulted in a lethal phenotype with a mortality that reached 100% at 10 d after tamoxifen injection (Fig. S1 C). Age-matched littermate control mice without cre (wild type [WT]) were unaffected by tamoxifen treatment. Consistent with findings from the previously reported analysis of the PDK1 KO mouse model (Lawlor et al., 2002), cardiac function evaluated by echocardiography at 7 d after tamoxifen injection revealed dramatically impaired systolic function with severe dilated cardiomyopathy and an abrupt drop in fractional shortening in KO but not in WT mice (Fig. S1 D, Table S1, and not depicted). Histological examination substantiated the echocardiographic findings, revealing dilatation of both ventricles and atria (Fig. S1 E) with apparently no evidence of significant apoptosis or interstitial fibrosis (Fig. S1, F and G). These observations indicate that PDK1/Akt activity plays a major role in maintaining adult heart function.

Deficiency in Akt activity leads to a reduction in the Cavα1 protein level

Using the cardiac-specific PDK1 KO mouse model, we investigated whether deficiency in Akt activity affects the expression or activation of signaling molecules that are implicated in Ca2+ handling and cardiac function. A time course analysis of extracts from WT and KO mouse ventricles revealed striking changes in protein expression upon induction of the PDK1 KO (Fig. 1, A and B). Notably, KO mice had decreased protein levels of the pore-forming Ca2+ channel subunit, Cavα1, which progressed as PDK1 protein expression gradually declined. No change in the protein level of the regulatory Cavβ2 subunit was observed. As PDK1 expression decayed, levels of Akt activation also dramatically decreased (assessed by phosphorylation of Akt at the PDK1 phosphorylation site Thr308) despite unaltered expression of total Akt protein (Fig. 1, A and B). Furthermore, Akt activity (assessed using GSK-3β as a substrate) was virtually absent in KO hearts (Fig. 1 C). Based on this evidence, we decided to perform further experiments at day 6 after the beginning of treatment.

Figure 1.

Alteration of Ca (A and B) Western blot (A) and densitometric (B) analyses of ventricular homogenates along a time course of tamoxifen inductions (day 1–6 treatment is indicated by the line in A) using various antibodies. A representative experiment is shown (n = 3). Error bars show SD. (C) Total Akt activity in WT and KO cardiomyocyte lysates assayed using a GSK-3β/α Akt-specific substrate. IP, immunoprecipitation.

Although the main physiological action of PDK1 is on Akt activation, PDK1 can potentially influence other members of the cAMP-dependent, cGMP-dependent, and PKC (AGC) kinase protein family such as PKC and PKA, which could also affect cellular Ca2+ handling (Williams et al., 2000; Mora et al., 2004). However, PKC activity was unchanged in KO mice (1.15 ± 0.05–fold greater than WT; not statistically significant; assessed by an assay using a PKC-specific peptide as substrate). There was no apparent effect of PDK1 deletion on SERCA2a (Fig. S2 A) as well as PKA activity because the phosphorylation of specific PKA regulatory sites in two SR Ca2+ regulatory proteins, Ryr (Ryr2-P2809) and phospholamban (PLN; PLN-P16), were unchanged in KO mice (Fig. S2 B), although it cannot be excluded that typical changes associated with heart failure and secondary to adrenergic receptor hyperactivation may take place at subsequent time points. Collectively, these data suggest that an acute reduction in Akt activation affects expression of proteins involved in the Ca2+ influx into the cell.

Deficiency in Akt activity affects ICa,L

Ca2+ handling and inotropism were examined in adult cardiomyocytes freshly isolated from WT and KO mice. Using the whole cell voltage-clamp technique, we recorded and analyzed LTCC ICa,L properties. No difference in cell size was observed between WT and KO cells as deduced from membrane capacitance measurements. Membrane capacitance was 116 ± 6 pF in WT cells (n = 18) and 115 ± 6 pF in KO cells (n = 18). However, the density of ICa,L (picoampere/picofarad) was decreased in KO versus WT (Fig. 2 B). At 0 mV, the density of ICa,L was −9.08 ± 0.96 pA/pF in KO cells (n = 12) versus −16.26 ± 0.96 pA/pF in WT cells (n = 12; P < 0.001). In addition, there was no significant difference in either steady-state activation or inactivation curves (unpublished data). Indeed, mean half-activation occurred at −12.97 ± 0.53 mV in WT cells versus −15.07 ± 0.66 mV in KO cells, and mean half-inactivation occurred at −31.11 ± 0.48 mV in WT cells versus −30.77 ± 0.42 mV in KO cells. The absence of a shift in the voltage dependence of these properties (Fig. 2 B) was consistent with the absence of modification in gating properties of the LTCC, suggesting that a reduction in the number of functional LTCCs can account for the observed decrease in ICa,L in KO mice. Of note, the decay kinetics of ICa,L were slower in KO cells compared with WT cells with a decrease in the early fast inactivating component (Fig. 2 A). Consistent with previous observations by us and others regarding the role of Akt in cardiac function (Blair et al., 1999; Condorelli et al., 2002; Kim et al., 2003; Sun et al., 2006), both contraction (Fig. 2 C) and systolic Ca2+ amplitudes (Ca2+ transients; Fig. 2 D and Fig. S3 A) were significantly depressed (by ∼35% and 30%, respectively; P < 0.05) in KO cardiomyocytes compared with WT littermates.

Figure 2.

Impaired intracellular Ca (A and B) Smaller Ca2+ current in KO cardiomyocytes. (A) Whole cell representative ICa,L currents normalized for difference in cell size. (B) ICa,L I-V current/voltage relationships (n = 12; *, P < 0.05; **, P < 0.01). (C and D) Cardiomyocyte contraction and Ca2+ transients at different stimulation frequencies. (C) Cardiomyocyte shortening is decreased in KO compared with WT cardiomyocytes (*, P < 0.05; ANOVA). (D) Ca2+ frequency relationship indicates smaller peak systolic but not diastolic Ca2+ in KO compared with WT cells (*, P < 0.05; ANOVA). Error bars show SEM.

The observed reduction in Ca2+ transient amplitude and cardiac contractility could be explained by reduced Ca2+ entry into cells via the LTCC, but decreased intracellular Ca2+ release from the SR may also contribute. However, although the Ca2+ transient amplitude between the systolic and diastolic phase (twitch) was smaller in KO cardiomyocytes (Fig. S3 B, left bars), no difference in total SR Ca2+ content was found (Fig. S3 B, right bars), suggesting that the decrease in Ca2+ transient amplitude is only caused by reduced Ca2+ entry. This is consistent with the observed slowing of the early fast inactivation of ICa,L (Fig. 2 A), which is highly dependent on CICR-triggered SR Ca2+ release during the action potential (Richard et al., 2006). Therefore, we conclude that the reduced ICa,L may contribute to the reduced contractility in KO hearts.

Akt regulates the Cavα1 protein level at the plasma membrane

The properties of the Cavα1 subunit are known to be markedly affected by LTCC accessory subunits (Catterall, 2000; Bourinet et al., 2004). Among the LTCC accessory subunits expressed in the heart, Cavβ2 is known to act as a chaperone for the Cavα1 subunit, both as a positive modulator of channel opening probability and for its trafficking from the ER to the plasma membrane (Yamaguchi et al., 1998; Viard et al., 2004). Therefore, supported by previous results (Viard et al., 2004) as well as corroborated by unchanged Cavα1 mRNA levels in KO compared with WT hearts (Fig. 3 A), we hypothesized that in the heart, an Akt-mediated phosphorylation of the LTCC accessory subunit would mainly affect trafficking of Cavα1 protein to the plasma membrane. However, because the amount of Cavα1 was reduced in both microsomal and membrane fractions from KO extracts compared with WT (Fig. 3 B), we hypothesized that the reduced Cavα1 level observed in KO mice was caused by enhanced protein degradation in addition to impaired protein translocation to the plasma membrane. To assess the pathway involved in the Akt-dependent Cavα1 protein degradation, three sets of specific cell degradation system inhibitors were examined for their ability to prevent the decrease in Cavα1 protein elicited by Akt inhibition. Treatment of Cavα1- and Cavβ2-cotransfected cells with bafilomycin-A1, an inhibitor of the lysosomal degradation system responsible for the degradation of many membrane proteins (Dice, 1987), prevented the decrease in Cavα1 protein induced by Akt inhibition (Fig. 3 C, top). Conversely, a ubiquitin/proteasome inhibitor, MG132, failed to protect Cavα1 from protein degradation. Similar results were obtained by inhibiting calpain, the intracellular Ca2+-dependent Cys protease known to be involved in membrane protein degradation (Belles et al., 1988; Romanin et al., 1991). Intriguingly, the bafilomycin-A1–dependent protection effect was abolished in the absence of Cavβ2 cotransfection, a condition under which Cavα1 is retained in the ER (Fig. 3 C, bottom). All together, these results confirm that Akt activity is regulating Cavα1 protein density and reveal that in the absence of Akt function, Cavα1 is susceptible to lysosome-mediated membrane protein degradation.

Figure 3.

Akt mediates regulation of Ca (A) RT-PCR analysis of Cavα1 mRNA expression from WT and KO ventricular extracts. GAPDH served as a loading control. (B) Western blot analysis of whole lysate, membrane, and microsomal fractions from WT and KO ventricular extracts. CSQ, calsequestrin. (C) YFP-Cavα1–transfected COS-7 cells alone or in combination with Cavβ2 expression vector were serum starved and treated with Akt inhibitor (Akt inh.) and 1 µM bafilomycin-A1, 25 µM MG132, or 25 µM calpeptin. 6 h after drug administration, cell lysates were prepared and subjected to Western blot analysis for YFP. GAPDH served as a loading control. (D and E) Cavα1 protein levels in KO cardiomyocytes infected with empty (mock) or active E40K-Akt (AdAkt)–expressing adenoviral vector (D) and in whole lysates of WT and E40K-Akt (Tg Akt) hearts (E). Representative experiments are shown (n = 4).

Because Cavβ2 is the only LTCC accessory subunit containing an Akt phosphorylation consensus site (Viard et al., 2004), we hypothesized that Cavα1 protein degradation at the plasma membrane might result from loss of Cavβ2 chaperone activity in the absence of Akt-induced phosphorylation. In support of this hypothesis, forced expression of the active E40K-Akt mutant (AdAkt) restored Cavα1 protein levels in isolated cardiomyocytes from KO mice (Fig. 3 D). Similarly, cardiomyocytes from Tg mice expressing constitutively active HA–E40K-Akt (Tg Akt; Condorelli et al., 2002) showed increased Cavα1 levels compared with WT controls (Fig. 3 E).

Akt is a determinant for Cavα1 protein level regulation by direct phosphorylation of the Cavβ2 chaperone subunit

To assess whether Akt is directly involved in modulation of Cavβ2 chaperone activity in the heart, we first confirmed the interaction between Akt and Cavβ2. Ventricular homogenates derived from either WT or Tg Akt mice were immunoprecipitated with anti-HA antibody and assayed for Cavβ2, which revealed association of the Cavβ2 subunit with active Akt (Fig. 4 A). Similarly, Cavβ2 was found to coimmunoprecipitate with insulin-stimulated endogenous Akts (Fig. S4 A).

Figure 4.

Akt interacts with and phosphorylates Ca (A) Coimmunoprecipitation assay of Akt and Cavβ2. Ventricular homogenates from WT and HA–E40K-Akt Tg mice (Tg Akt) immunoprecipitated with antibodies against HA and immunoblotted for Cavβ2 as well as HA as a control. (B) Examination of Cavβ2 phosphorylation by Akt. In vitro kinase assays were performed with immunoprecipitated Cavβ2 incubated with recombinant active Akt and 32P-labeled ATP (left) or immunoprecipitated Cavβ2 from WT and KO cardiac extracts from mice treated or not treated with 1 mU/g insulin using phospho-Akt substrate (PAS) antibody (right). (C) Back phosphorylation assay of Cavβ2 from WT and KO hearts. Immunoprecipitated Cavβ2 from solubilized membranes was in vitro back phosphorylated using recombinant active Akt and [γ32]ATP. Precipitate amounts were assayed for [32P]Cavβ2 and total Cavβ2. Representative experiments are shown (n = 4). IP, immunoprecipitation.

To determine whether Cavβ2 can be phosphorylated by Akt, Cavβ2 immunoprecipitates from cardiac homogenates were incubated with recombinant active Akt and γ-[32P]ATP. A band corresponding to phosphorylated Cavβ2 was detected only in the presence of the kinase (Fig. 4 B, left). To determine whether the Cavβ2 subunit was phosphorylated by Akt in vivo, we treated overnight-starved mice with 1 mU/g insulin to induce activation of Akt (Bayascas et al., 2008). 20 min after treatment, Cavβ2 was immunoprecipitated from ventricular homogenates, subjected to Western blot analysis, and probed for phosphorylated Akt consensus sites using phospho-Akt substrate antibody. This revealed insulin-stimulated phosphorylation of Cavβ2 in WT but not in KO hearts (Fig. 4 B, right). Furthermore, a back phosphorylation assay, which is used to assess the basal state of Cavβ2 phosphorylation, revealed a reduction of the basal phosphorylation level of Cavβ2 by 36% (P < 0.05) in KO mouse ventricle compared with WT (Fig. 4 C). Collectively, these data demonstrate that active Akt binds to and phosphorylates Cavβ2, the chaperone for Cavα1.

To directly assess whether Akt phosphorylation of Cavβ2 protects Cavα1 from protein degradation, we constructed a mutant of Cavβ2 in which Ser625, which is contained in the putative Akt consensus site (R-X-X-R-S/T), was replaced by glutamate (Cavβ2-SE) to mimic phosphorylation. Cotransfection of 293T cells with Cavα1 and Cavβ2-SE resulted in Cavα1 protein levels that were increased compared with those found when cotransfected with Cavβ2-WT (Fig. 5 A). Similarly, Cavα1 expression was increased in insulin-treated Cavβ2-WT–cotransfected cells (Fig. 5 A). Notably, the active phosphomimic Cavβ2-SE also counteracted the down-regulation of Cavα1 induced by an Akt inhibitor (Fig. 5 B). Opposite results were obtained with a dominant-negative (DN) Cavβ2 mutant in which Ser was replaced by Ala (Cavβ2-SA) to prevent Akt phosphorylation. Indeed, Cavα1 protein levels were reduced when coexpressed with Cavβ2-SA (Fig. 5 C). In addition, insulin stimulation failed to increase Cavα1 in the presence of the DN Cavβ2-SA mutant (Fig. 5 C). Consistent with the hypothesis that Cavα1 protein down-regulation relies on Akt kinase activity, overexpression of a DN form of Akt (AdAktDN) resulted in a significant reduction in Cavα1 protein levels, whereas forced expression of AdAkt was sufficient to counteract Cavα1 reduction in a serum-free condition, in which Akt is not phosphorylated (Fig. S4 B). Furthermore, suppression of Akt expression in 293T cells by siRNA (small interfering Akt [siAkt]) resulted in reduction of the Cavα1 protein level (Fig. 5 D).

Figure 5.

Akt phosphorylation of Ca (A–C) YFP-Cavα1–cotransfected 293T cells with the indicated mutant variant of Cavβ2. Cells were serum starved overnight and treated with 100 µM insulin (A and C) or 5 µM Akt inhibitor (Akt inh; B) as indicated. The expression of YFP-Cavα1 in lysates was monitored by Western blot analysis with anti-YFP antibody and normalized based on transfection efficiency (Cavβ2) and protein amount (tubulin; n = 3). (D) Cavα1- and Cavβ2-cotransfected 293T cells were treated with siAkt-expressing vector as indicated. 3 d after transfection, cell lysate was tested by Western blot analysis. Protein loading was normalized to GAPDH levels. Representative experiments are shown (n = 3).

To support the evidence that Akt-dependent phosphorylation of Cavβ2 is a determinant for Cavα1 stability and functionality, we measured the effect of the Cavβ2 mutants on Ca2+ current. Although cotransfection of cells with Cavα1 and Cavβ2-WT resulted in significant depressed ICa,L in serum-free medium compared with serum-containing medium in which Akt is phosphorylated (not depicted), cotransfection of Cavα1 and Cavβ2-SE mutant but not Cavβ2-SA mutant completely counteracted this reduction (Fig. 6).

Figure 6.

Akt phosphorylation of Ca Ca2+ currents recorded in cotransfected tsA-201 cells with YFP-Cavα1 and Cavβ2-WT, Cavβ2-SE, or Cavβ2-SA mutant cultivated for 36 h in the presence or absence of 10% fetal bovine serum. Currents were recorded 1–2 min after the whole cell configuration was achieved (i.e., after stabilization of the current) and were elicited by a 0-mV depolarization of 200-ms duration applied from a holding potential of −80 mV. Currents are normalized to cell capacitance (current density, picoampere/picofarad). Representative current traces are shown. n > 35 at each condition (*, P < 0.05 compared with YFP-Cavα1; ANOVA). Error bars show SEM.

Akt regulates Cavα1 protein stability

PEST sequences have been suggested to serve as signals for rapid proteolytic degradation through the cell quality control system (Rechsteiner, 1990; Smith et al., 1993; Krappmann et al., 1996; Sandoval et al., 2006). Notably, PEST-mediated protein degradation has recently been suggested to play an essential role in modulating neuronal Ca2+ channel function through regulation of the Cavβ3 accessory subunit (Sandoval et al., 2006). Our findings raise the possibility that processing of the Cavα1 protein may be affected in a similar way. To test this hypothesis, we used the web-based algorithm PESTfind (Rogers et al., 1986) in a search for potential Cavα1 PEST sequences and found several putative motifs (amino acids 435–460, 807–820, 847–858, 1,732–1,745, and 1,839–1,865). Intriguingly, the highest scored potential PEST sequences obtained are highly conserved among species (Table I), with one located in the I–II linker of the Cavα1 subunit and overlapping with the α1-interacting domain (AID), which is the primary binding region for Cavβ2 (Fig. 7 A; Bodi et al., 2005). To determine whether these PEST sequences are involved in Cavα1 degradation control, we generated two in-frame deletion mutants encompassing either the I–II (Cavα1-ΔP) or II–III (Cavα1-ΔH) cytosolic linker region (Fig. 7 A). Western blot and immunofluorescence analyses of serum-starved 293T cells transfected with these mutants revealed higher protein expression levels for both Cavα1-ΔP and Cavα1-ΔH mutants compared with Cavα1-WT, which is consistent with the hypothesis that these motifs determine Cavα1 protein stability (Fig. 7 B). Furthermore, a pulse-chase analysis, with a chase starting 36 h after cell starvation, revealed markedly increased protein stability of Cavα1-ΔP and Cavα1-ΔH compared with Cavα1-WT (Fig. 7 C). In particular, Cavα1-WT showed a short half-life typical of proteins containing PEST sequences (Dice, 1987), with a rapid and progressive degradation starting 4 h after the chase and reaching 50% of degradation 25 h after the chase. In contrast, Cavα1-ΔP and Cavα1-ΔH mutants were less sensitive to degradation and were degraded by only 23% and 15% after 25 h, respectively (P < 0.001). Notably, cotransfection of Cavβ2-SE with Cavα1-WT resulted in a considerable increase in the half-life of Cavα1-WT (Fig. 7 C). In addition, transfection of 293T cells with Cavα1 PEST sequences fused in frame with GFP resulted in marked instability of GFP, as shown by both Western blot and immunofluorescence analyses (Fig. 7 D), providing further evidence that these motifs are determinants for Cavα1 protein stability. Consistent with the hypothesis that Akt-mediated protection of Cavα1 degradation acts through PEST sequences, overexpression of AdAktDN or siAkt had no significant effect on protein levels of either Cavα1-ΔP or Cavα1-ΔH mutants (Fig. S4, B and C). To assess whether the observed PEST mechanism is caused by a direct Akt-dependent interaction between Cavβ2 and Cavα1, we performed in vitro binding assays using in vitro–translated [35S]Met-labeled Cavα1 cytosolic domains and a GST-fused Cavβ2 C-terminal coiled-coil region. Notably, direct interaction took place between the Akt-phosphorylated Cavβ2 C-terminal coiled-coil region and the Cavα1 C-terminal domain (Fig. 7 E). No interactions were found with other Cavα1 cytosolic domains (unpublished data), although it cannot be excluded that other binding sites may exist.

Table I.

PEST sequences are highly conserved in Cavα1

Species	Fragment	Sequences	PESTfind score
Mouse	PEST I	435-KGYLDWITQAEDIDPENEDEGMDEDK-460	8.45
Rat	PEST I	476-KGYLDWITQAEDIDPENEDEGMDEDK-501	8.45
Human	PEST I	446-KGYLDWITQAEDIDPENEDEGMDEEK-471	8.66
Mouse	PEST II	807-KSITADGESPPTTK-820	9.45
Rat	PEST II	848-KSITADGESPPTTK-861	9.45
Mouse	PEST III	837-HSNPDTAGEEDEEEPEMPVGPR-858	19.51
Rat	PEST III	878-HSNPDTAGEEDEEEPEMPVGPR-899	19.51
Human	PEST II	845-KSPYPNPETTGEEDEEEPEMPVGPR-869	20.26
Mouse	PEST IV	1,732-KTGNNQADTESPSH-1,745	5.5
Rat	PEST IV	1,772-KTGNNQADTESPSH-1,785	5.5
Mouse	PEST V	1,839-RMSEEAEYSEPSLLSTDMFSYQEDEH-1,865	5.86
Human	PEST IV	1,937-HDTEACSEPSLLSTEMLSYQDDENR-1,961	7.54
Human	PEST V	2,214-RGAPSEEELQDSR-2,226	7.71

Occurrence of PEST sites within the amino acid sequence of Cavα1 from the mouse, rat, and human. Amino acid identity is underlined.

Figure 7.

Rapid protein degradation PEST sequences determine Ca (A) Schematic representation of Cavα1 mapping the AID and PEST sequences in the I–II and II–III cytosolic loops. Deleted PEST sequences (P and H) are highlighted in red. (B) Western blot and immunofluorescence analyses showing relative levels of WT and PEST-deleted mutants of YFP-Cavα1 (n = 3). The asterisk indicates a YFP-Cavα1 degradation fragment. (C) Half-lives of WT Cavα1 subunit (alone or cotransfected with Cavβ2-SE) and its in-frame ΔPEST mutants (Cavα1-ΔP and Cavα1-ΔH) were determined in COS-7 cells. After overnight starvation, transfected cells were pulse chased and analyzed along a time course (*, P < 0.001 compared with Cavα1; ANOVA; n = 3). Error bars show SD. (D) Western blot and immunofluorescence analyses showing relative levels of WT GFP and N-terminal fusion PEST mutants (n = 3). (E) The Cavα1 C terminus interacts with the Akt-phosphorylated GST-Cavβ2 coiled-coil region. Bacterially expressed GST or GST-C-Cavβ2 (Cavβ2, amino acids 480–655) fusion protein and glutathione–Sepharose beads were incubated with equal amounts of in vitro–translated [35S]Met-labeled C-Cavα1 (Cavα1, amino acids 1,477–2,169). Binding occurred only with Akt-phosphorylated GST-C-Cavβ2. Bound proteins were resolved by SDS-PAGE (4–12%). 10% of the input protein in each binding reaction is shown. Coomassie staining of SDS-PAGE is shown in the bottom panel. (F) Ca2+ currents recorded in tsA-201 cells cotransfected with Cavβ2-WT and either Cavα1-WT or Cavα1-ΔH and cultivated for 36 h in the presence or absence of 10% fetal bovine serum. Current densities (picoampere/picofarad) are normalized to the control condition. n > 35 at each condition (*, P < 0.05 compared with Cavα1; ANOVA). Error bars show SEM. Bars: (B) 5 µm; (D) 20 µm.

To assess whether PEST-deleted Cavα1 channels are still functional, traffic appropriately to the membrane, and associate with the Cavβ2 subunit, we measured Ca2+ current in Cavα1-ΔH mutant–transfected cells. No significant differences in ICa,L were found in cells transfected with Cavα1-WT compared with Cavα1-ΔH (Fig. 7 F). Conversely, although serum deprivation resulted in ICa,L reduction in Cavα1-WT–transfected cells, no significant changes were observed in Cavα1-ΔH mutant–transfected cells (Fig. 7 F). This confirms that PEST-deleted Cavα1-ΔH is resistant to rapid protein degradation and maintains its integrity and physiological function. Furthermore, current-voltage analysis (I-V curves) revealed that neither serum deprivation nor PEST-H deletion modifies steady-state activation parameters (Fig. S5). Also, all electrophysiological experiments were performed at a holding potential of −80 mV, which is a value far away from the potential for half steady-state inactivation (V0.5) of ICa,L, indicating that a change in the macroscopic current properties of Cav1.2 is unlikely.

Collectively, our results suggest that Akt-mediated phosphorylation of Cavβ2 regulates Cavα1 density through protection of Cavα1 PEST motifs from the cell protein degradation machinery. Impairment of this mechanism is expected to result in dysregulation of cardiomyocyte contractile function.

Discussion

This study reveals a mechanism through which the insulin IGF-1–PI3K–PDK1–Akt pathway can sustain or modulate Ca2+ entry in cardiac cells via the voltage-gated LTCC and eventually affect cardiac contractility. Using a mouse model with an inducible and cardiomyocyte-specific deletion of the upstream activator PDK1, we showed that Akt is of key importance for the structural organization and functionality of the LTCC complex at the plasma membrane. This regulation of LTCC activity is directly related to the Akt-mediated phosphorylation of the accessory subunit Cavβ2, which in turn results in increased protein density of the pore-forming Cavα1 subunit through protection of PEST sequences from the proteolytic degradation system. In the absence of phosphorylated Akt, the Ca2+ current is reduced, resulting in a decreased Ca2+ transient and contractility. Therefore, it is tempting to speculate that the Akt-mediated phosphorylation of Cavβ2 and the consequent direct association of the Cavβ2 C-terminal tail with the Cavα1 C-terminal coiled-coil region (Fig. 7 E) may induce conformational changes that prevent PEST sequences from being recognized by the cell degradation system (Fig. 8). In addition, one cannot exclude the possibility that phosphorylated Cavβ2 might also act indirectly through other, as of yet unknown, LTCC protein partners.

Figure 8.

Proposed mechanism. Akt, followed by PDK1 activation, phosphorylates Cavβ2 at the C-terminal coiled-coil domain. The phosphorylation allows association of the C-terminal portion of Cavβ2 with the Cavα1 C-terminal domain. In turn, a conformation shift prevents PEST sequence recognition, stabilizing Cavα1 protein levels. The blue and red ribbons in Cavα1 represent AID and PEST sequences, respectively.

The identified mechanism alone is unlikely to be responsible for the detrimental cardiac defects observed in the PDK1 KO mouse model. To assess whether a reduction in the Akt antiapoptotic activity could lead to increased cell death, we measured caspase 3 activation (Fig. S1). However, consistent with previous evidence reported by Mora et al. (2003), our results failed to prove any significant involvement of this mechanism in the PDK1 KO phenotype. Our PDK1 KO mouse model does not appear to progress through slow transitional states, which are typical of heart failure, but rather progresses directly to a dilated cardiac phenotype, which eventually leads to premature death (Fig. S1). Therefore, we hypothesize that the lethal phenotype is caused by activation of more complex systems that rapidly remodel the extracellular matrix and cell to cell contacts and change the energy metabolism. Further studies are required to unravel the complex mechanisms that contribute to the establishment of the observed PDK1 KO mouse heart phenotype.

Several findings have shown the importance of the insulin IGF-1–PI3K–Akt pathway in heart function. Our group has previously demonstrated that overexpression of an active form of Akt-1 results in improved cardiac inotropism both in vivo (Condorelli et al., 2002) and in vitro (Kim et al., 2003), augmenting ICa,L. Similar results were recently obtained in a mouse model with cardiac-specific Akt-1 nuclear overexpression (Rota et al., 2005) and in mice deficient for PTEN (phosphatase and tensin homologue deleted on chromosome 10), an antagonizer of PI3K activity (Sun et al., 2006). In addition, short-term administration of IGF-1 in animal experiments has also been reported to increase cardiac contractility (Duerr et al., 1995). However, the mechanism through which the insulin IGF-1–PI3K–Akt pathway affects Ca2+ current has remained elusive. In an elegant in vitro study, Viard et al. (2004) demonstrated that a region of the Cavβ2a subunit is involved in the PI3K-induced chaperoning of Cav2.2α in neurons. This PI3K-induced regulation was shown to be mediated by Akt phosphorylation of the Cavβ2a subunit, which in turn regulates Cav2.2α trafficking from the ER to the plasma membrane. Notably, the C-terminal region containing the putative Akt phosphorylation consensus site is conserved in all variants of the Cavβ2 subunit both in neurons and the heart (Viard et al., 2004), thus illustrating the importance of this site. Interestingly, two very short human cardiac splice isoforms, Cavβ2f and Cavβ2g, with preserved Akt sites have been shown to be essential for modulating Ca2+ channel function and Cavα1 channel density (De Waard et al., 1994; Kobrinsky et al., 2005). Strikingly, the same two Cavβ2 variants do not contain the PKA phosphorylation site (Kamp and Hell, 2000), which is consistent with our data suggesting no PKA involvement in the modulation of LTCC density (Fig. S2 B). As a corollary, the presence of this conserved C-terminal region in all Cavβ2 splice isoforms corroborates the relevance of identifying new functional motifs that may give important insights into LTCC modulation. Consistent with an important functional role of the conserved Cavβ2 C-terminal region, Lao et al. (2008) recently showed that, in the absence of the main Cavβ2 protein domain, the selected C-terminal essential determinant is sufficient for ICa,L stimulation. All together, this evidence supports the notion that this region is a potential pharmacological target.

In conclusion, we show that the insulin IGF-1–PI3K–PDK1–Akt pathway regulates Cavβ2 chaperone activity through phosphorylation by Akt and suggest that, in turn, this controls Cavα1 channel density by protection of Cavα1 from PEST-dependent protein degradation (Fig. 8). This paradigm highlights an unanticipated regulatory function for Akt in modulating LTCC function and provides evidence for an essential role of Akt in the control of cardiomyocyte Ca2+ handling and contractility. Interestingly, the high level of conservation of PEST sequences in the Cavα1 subunit throughout evolution (Table I) indicates that our proposed mechanism may play a universal role in regulating cell Ca2+ handling and survival. Because pathophysiological states are often accompanied by alterations in LTCC function (Mukherjee and Spinale, 1998), the elucidation of this novel regulatory pathway may open new therapeutic perspectives.

Materials and methods

Generation of genetically modified mice

Cardiac-specific PDK1-inducible KO mice (MerCreMer α-MHC PDK1) were generated by breeding PDK1floxed/floxed Tg mice (provided by D.R. Alessi, Medical Research Council Protein Phosphorylation Unit, University of Dundee, Dundee, Scotland, UK; Williams et al., 2000) with mice expressing the cardiac-specific MerCreMer α-MHC promoter-driven cre recombinase gene (provided by J.D. Molkentin, University of Cincinnati, Cincinnati, OH; Sohal et al., 2001). The resulting background strain of the MerCreMer mice was C57BL/6-SV129 and was unchanged throughout all experiments. Control animals used in this study were PDK1floxed/floxed littermates not expressing the cre recombinase gene and were treated with the same tamoxifen regiment. Tamoxifen dissolved in maize oil was injected intraperitoneally once a day at a dose of 75 mg/kg body weight. Male animals 7–8-wk old were used. All animal procedures were performed in accordance with the Guide for the Care and Use of Laboratory Animals and approved by the Institutional Animal Care and Use Committee.

Culture and treatment of mouse cardiomyocyte cells

Isolation of ventricular myocytes was performed as previously described (Care et al., 2007). Cells were infected with an adenovector expressing either no transgene (mock), HA–E40K-Akt (AdAkt), or Akt-K179M (AdAktDN) at MOI 100 and harvested 48 h after infection. The viral vector was amplified and purified in 3% sucrose/PBS by ViraQuest, Inc.

Cell culture and cDNA mutagenesis

Cell transfection was performed in serum-starved medium using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. 5 µM Akt-XI inhibitor (EMD), insulin (Sigma-Aldrich), 1 µM bafilomycin-A1 (Sigma-Aldrich), 25 µM MG132 (EMD), and 25 µM calpeptin (EMD) were used as described in Results. Cacnb2 cDNA (complete coding sequence, cDNA clone MGC:129335, IMAGE:40047531; American Type Culture Collection) was cloned in the pcDNA3 vector. Site-directed mutagenesis was performed using the QuikChange Site-Directed Mutagenesis kit (Agilent Technologies). Cavα1 PEST deletion mutants and GFP fusion proteins were generated by PCR. YFP-Cavα1 expression plasmids were provided by N. Soldatov (National Institute on Aging, National Institutes of Health, Baltimore, MD). A lentivirus vector was generated and used as an expression vector for siRNA-mediated silencing of the akt gene (siAkt). The sequence used (5′-TGCCCTTCTACAACCAGGATT-3′) was chosen in a conserved region between rat, mouse, and human and has been validated for targeting Akt-1 and -2 (Katome et al., 2003). All constructs were confirmed by DNA sequencing.

Ca2+ current measurement

Macroscopic ICa,L was recorded at room temperature (∼22°C) using the whole cell patch-clamp technique in native cells as previously described (Maier et al., 2003; Aimond et al., 2005). External recording solution contained 136 mM tetraethylammonium (TEA)-Cl, 2 mM CaCl2, 1.8 mM MgCl2, 10 mM Hepes, 5 mM 4-aminopyridine, and 10 mM glucose, pH 7.4, with TEA-OH. Pipette solution contained 125 mM CsCl, 20 mM TEA-Cl, 10 mM EGTA, 10 mM Hepes, 5 mM phosphocreatine, 5 mM Mg2-ATP, and 0.3 GTP, pH 7.2, with CsOH. Myocytes were held at −80 mV, and 10-mV depolarizing steps from −50 to 50 mV for 300 ms were applied. Analysis was performed using a microscope (Diaphot 200; Nikon) equipped with 10× NA 20 objective lenses (CFWN; Nikon). pCLAMP 9 (MDS Analytical Technologies) was used as acquisition software. For electrophysiological recordings of recombinant Cavα1 currents, tsA-201 cells were transfected in OptiMEM (Invitrogen) with a DNA mix containing plasmids encoding YFP-Cavα1, Cavβ2 subunit (either Cavβ2-WT, Cavβ2-SE, or Cavβ2-SA), Cavα2δ1 subunit, and CD8 (in a ratio of 1:2:0.5:0.1). After 24 h, cells were cultured in DME with or without serum for 36 h, and electrophysiological recordings were performed on cells expressing both YFP-Cavα1 and CD8, which is identified using anti-CD8–coated beads (Dynabeads; Invitrogen). The ∼330-mosM extracellular solution contained 135 mM NaCl, 20 mM TEA-Cl, 5 mM CaCl2, 1 MgCl2, and 10 mM Hepes (pH adjusted to 7.4 with KOH). Borosilicate glass pipettes have a typical resistance of 1.5–3 MW when filled with an ∼315-mosM internal solution containing 140 mM CsCl, 10 mM EGTA, 10 mM Hepes, 3 mM Mg-ATP, 0.6 mM GTP-Na, and 2 mM CaCl2 (pH adjusted to 7.2 with KOH). Analysis was performed using a microscope (x71; Olympus). Data acquisition was performed with pCLAMP 9 software.

Fluorescent measurement of [Ca2+]i

Isolated myocytes were loaded with 5 µM Fura-PE3 acetoxymethyl (TefLabs) and analyzed as previously described (Bassani et al., 1994; DeSantiago et al., 2002). Analysis was performed using a Diaphot 200 microscope. Data acquisition and analysis were performed using pCLAMP software (Clampex and Clampfit version 8.2; MDS Analytical Technologies).

Akt and PKC kinase assay

Myocardial tissue lysates were tested using the Akt Kinase Assay kit (Cell Signaling Technology) and PKC (Millipore) according to the manufacturer's instructions.

Western blot analysis and antibodies

Protein expression was evaluated in total lysates or cell fractions by Western blot analysis according to standard procedures. Antibodies against the following proteins were used: Cavα1 (Novus Biologicals); Cavα1 and Cavβ2 (provided by H. Haase, Max Delbrück Center for Molecular Medicine, Berlin, Germany); Ryr and Ryr2-P2809 (provided by A. Marks, Columbia University, New York, NY); PDK1 (EMD); Akt-1, -2, and -3, Akt, Akt-P308, and anti–phospho-Ser/Thr-Akt substrate (Cell Signaling Technology); PLN and PLN-P16 (Novus Biologicals); calsequestrin (BD); caspase 3 (Cell Signaling Technology); HA (Roche); GFP/YFP (GeneTex, Inc.); tubulin (Novus Biologicals); GSK-3β (Cell Signaling Technology); and glyceraldehyde-3-phosphate dehydrogenase (GAPDH; Cell Signaling Technology). ImageJ software (National Institutes of Health) was used to perform densitometry analyses.

Tissue preparation, immunoprecipitation, and in vitro phosphorylation

When described, overnight-fasted mice were injected intraperitoneally with 1 mU/g insulin or saline solution. 20 min after injection, the hearts were rapidly extracted, freeze clamped in liquid nitrogen, and homogenized to a powder in liquid nitrogen. In vitro phosphorylation assays on immunoprecipitates were performed as described previously (Haase et al., 1999).

Cell fractionation

Pulverized hearts were homogenized in ice-cold solution 1 (300 mM sucrose, 10 mM Tris-HCl, pH 7.5, 1 mM EDTA, 1 mM EGTA, 50 mM NaF, 1 mM Na3VO4, and protease inhibitors) at 1.5 ml/ventricle by three bursts of 10 s in a homogenizer (PT 3000; Polytron). Homogenates were then incubated for 15 min on ice (whole homogenates). Samples were spun at 1,000 g for 10 min at 4°C. Pellets were washed in solution 1 and spun at 1,000 g for 10 min at 4°C, and supernatants were filtered through four layers of cheese cloth and centrifuged at 10,000 g for 30 min at 4°C. Supernatants were then centrifuged at 143,000 g for 30 min at 4°C, and pellets were resuspended in solution 2 (600 mM KCl, 30 mM Tris-HCl, pH 7.5, 300 mM sucrose, 1 mM EDTA, 1 mM EGTA, 50 mM NaF, 1 mM Na3VO4, and protease inhibitors). Supernatants were saved as cytosolic fractions. Resuspended pellets from a further centrifugation at 143,000 g for 45 min at 4°C were resuspended in solution 3 (100 mM KCl, 20 mM Tris-HCl, pH 7.5, 300 mM sucrose, 1 mM EDTA, 1 mM EGTA, 50 mM NaF, 1 mM Na3VO4, and protease inhibitors) and saved as ER fractions. All aliquots were stored at −80°C.

Histology and confocal microscopy

Fixation, staining, and confocal analysis were performed as previously described (Care et al., 2007). Confocal microscopy was performed using a confocal microscope (Radiance 2000; Bio-Rad Laboratories) with a 60× Plan-Apochromat NA 1.4 objective (Carl Zeiss, Inc.). Individual images (1,024 × 1,024 pixels) were converted to tiff format and merged as pseudocolor RGB images using Imaris (Bitplane AG).

Pulse-chase and immunoprecipitation experiments

36 h after transfection, 293T cells were starved for 30 min in Met- and Cys-free DME (Sigma-Aldrich) and were then labeled for 30 min by adding 500 µCi 35S-labeled L-Met and 2 mM L-Cys. Radioactive media was eventually washed out with PBS (time 0 pulse) and replaced with normal DME. Time points were at 4, 10, and 25 h after pulse. Anti-GFP polyclonal IgG (GTX20290) was used for immunoprecipitation. Radioactivity was quantitated with ImageQuant 5.2 software (GE Healthcare).

GST pull-down assay

Affinity-purified GST fusion proteins were generated using a pGEX system (GE Healthcare) and phosphorylated as described below. GST fusion protein bound to glutathione–Sepharose 4B beads (GE Healthcare) was incubated with 25 µl of 35S-labeled Met protein with moderate shaking at 25°C for 2 h in 200 µl of binding buffer containing 20 mM Hepes, pH 7.9, 1 mM EDTA, 10% glycerol, 0.15 M KCl, 0.05% Nonidet P-40, and 1 mM DTT. 35S-labeled probes were generated from the C-terminal region of Cavα1 cDNA fragments under control of the T7 promoter using the TnT Quick Coupled Reticulocyte Lysate System (L1170; Promega), washed three times with washing buffer (20 mM Hepes, pH 7.9, 1 mM EDTA, 10% glycerol, 250 mM KCl, and 0.1% Nonidet P-40), and centrifuged. Bound proteins were eluted in SDS sample buffer, subjected to SDS-PAGE, and detected by autoradiography. Recombinant GST-Cavβ2 beads or GST beads were phosphorylated by incubation with recombinant Akt (Millipore). In brief, 5 µg GST-Cavβ2 or GST beads were incubated at 30°C for 45 min in a 50-µl solution containing 2 µg activated Akt kinase, 10 mM Hepes-KOH, pH 7.5, 50 mM γ-glycerophosphate, 50 mM NaCl, 1 mM dithiothreitol, 10 mM MnCl2, and 1 mM ATP.

Statistical analysis

Statistical comparison was performed within at least three independent experiments by paired or unpaired Student's t test, whereas comparison between groups was analyzed by one-way repeated-measures analysis of variance (ANOVA) combined with a Newman-Keuls post-test to compare different values using Prism 4.0 software (GraphPad Software, Inc.). Differences with P < 0.05 were considered statistically significant.

Online supplemental material

Fig. S1 shows additional biochemical, histological, and echocardiographic analyses of mice lacking PDK1 expression. Fig. S2 shows SERCA2 level and phosphorylation of specific PKA regulatory sites in two SR Ca2+ regulatory proteins, Ryr (Ryr2-P2809) and PLN (PLN-P16). Fig. S3 shows representative Ca2+ traces and twitch Ca2+ transient amplitude in KO compared with WT cardiomyocytes. Fig. S4 shows coimmunoprecipitation of Cavβ2 with insulin-activated Akt isoforms and the effects of dominant-active and -negative Akt as well as siAkt on the Cavα1 protein level. Fig. S5 shows current-voltage analysis (I-V curves) of cells transfected with Cavα1-WT or Cavα1-ΔH in normal or serum-free conditions. Table S1 shows echocardiography analysis values of WT and KO mice. Online supplemental material is available at http://www.jcb.org/cgi/content/full/jcb.200805063/DC1.