|Home | About | Journals | Submit | Contact Us | Français|
Phosphoinositide 3-kinase (PI3K) p110α plays a key role in insulin action and tumorigenesis. Myocyte contraction is initiated by an inward Ca2+ current (ICa,L) through the voltage-dependent L-type Ca2+ channel (LTCC). The aim of this study was to evaluate if p110α also controls cardiac contractility by regulating the LTCC.
Genetic ablation of p110α (also known as Pik3ca), but not p110β (also known as Pik3cb), in cardiac myocytes of adult mice reduced ICa,L and blocked insulin signaling in the heart. p110α-null myocytes had a reduced number of LTCCs on the cell surface and a contractile defect that decreased cardiac function in vivo. Similarly, pharmacological inhibition of p110α decreased ICa,L and contractility in canine myocytes. Inhibition of p110β did not reduce ICa,L.
PI3K p110α but not p110β regulates the LTCC in cardiac myocytes. Decreased signaling to p110α reduces the number of LTCCs on the cell surface and thus attenuates ICa,L and contractility.
The force generated by a heart beat results from the coordinated contraction of individual myocytes and is regulated by changes in the intracellular Ca2+ concentration. Each contraction-relaxation cycle is initiated by cell membrane depolarization elicited by action potentials, resulting in a small inward Ca2+ current through the LTCC located on the cell surface. This ICa,L triggers a larger release of Ca2+ from the sarcoplasmic reticulum, resulting in myocyte contraction. Thus, ICa,L is an important determinant of contractile force.
Studies suggest that class I phosphoinositide (PI) 3-kinases (PI3Ks) modulate ICa,L in cardiac myocytes1–3. These enzymes phosphorylate PI 4,5-bisphosphate (PI(4,5)P3) to generate PI 3,4,5-trisphosphate (PI(3,4,5)P3) in vivo. Cardiac myocytes contain at least two class IA PI3Ks, p110α and β, and class IB p110γ. Recent evidence indicates that both class IA PI3K isoforms play a role in tumorigenesis4, 5. Indeed, upregulated PI3K signaling is seen in many human malignancies and a number of PI3K inhibitors are currently being tested in clinical trials5, 6. However, rodent models of diabetes and other models in which PI3K signaling is compromised in the heart suggest that this pathway is needed to maintain normal LTCC function in cardiac myocytes1, 2, 7–9. The PI3K isoform that maintains normal ICa,L in the heart has not been identified. Here we show that ablation of p110α in the heart of adult mice and pharmacological inhibition of p110α in isolated canine ventricular myocytes leads to a decrease in ICa,L and contractility defects. Ablation of p110β or inhibition of this isoform did not reduce ICa,L. Our results suggest the potential for adverse effects on cardiac contractility with cancer therapies that target p110α but not p110β. In addition, inhibition of p110α may provide an explanation for the ventricular dysfunction seen in some diabetic patients.
Generation of MerCreMer/p110αFlox/Flox, p110αFlox/Flox, MerCreMer/p110βFlox/Flox and p110βFlox/Flox mice is described in the Methods section of the online Data Supplement. At 8 weeks of age, all four groups of mice of both sexes were injected intraperitoneally with 1 mg tamoxifen daily for 28 days. These mice were analyzed between 4–6 months of age. MerCreMer/p110Flox/Flox mice were paired with age and sex-matched p110Flox/Flox littermates for all experiments.
Mixed-breed dogs of either sex over 12 months old were purchased from R&R Research. All animal-related experimental protocols were approved by the Stony Brook University Institutional Animal Care and Use Committee.
Mice were killed by intraperitoneal injection of 100 mg/kg body weight sodium pentobarbital and ventricular myocytes were isolated as previously described1. Canine ventricular cells were isolated using a modified Langendorf procedure by perfusing a wedge of left ventricle through a coronary artery with 0.5 mg/ml collagenase (Worthington Type 2) and 0.08 mg/ml protease (Sigma Type XVI) for 12–15 min followed by tissue mincing as described previously10. The solutions used and the method of whole-cell patch clamping of isolated cardiac myocytes were described earlier1. Phosphoinositides (all di-C8 and used at l μM; Echelon Biosciences), p110α/p85α (Millipore) and inactive and activated Akt1 (both from Millipore) were diluted in internal solution and infused through the patch pipette. Heat inactivation of p110α/p85α was performed by incubation at 50 °C for 30 min. Where indicated, cells were pretreated in bath solution with PI-103 (Cayman Chemical), Akt inhibitor VIII (Calbiochem), TGX-221 (Cayman Chemical) or vehicle for 2 h at room temperature before patch clamping. Isoproterenol (Sigma) was freshly prepared in distilled water.
Cell surface biotinylation of the LTCC was performed as previously described with some modifications2. Briefly, isolated mouse myocytes were washed with K-reversal Tyrode buffer (35 mmol/l HEPES, 140 mmol/k KCl, 8 mmol/l KHCO3, 2 mmol/l MgCl2 and 0.4 mmol/l KH2PO4, pH 7.5). Equal numbers of cells were incubated with EZ-Link Sulfo-NHS-LC-Biotin (Pierce Chemicals, Rockford, IL) to biotinylated cell surface proteins. After the reaction was quenched with 100 mmol/l glycine, cells were lysed with 1% Triton X-100 and 0.5% CHAPS. Biotinylated proteins were pulled with neutrAvidin agarose beads (Pierce Chemical) and biotinylated LTCC detected by western blotting with an anti-Cav1.2 antibody (Alomone Labs, Jerusalem, Israel). Total Cav1.2 expression was measured by western blot analysis of membranes prepared from mouse hearts as previously described2.
Mice were anesthetized with 1–2% isoflurane and transthoracic echocardiography was performed as previously described2. Ventricular measurements were made on 9 cardiac cycles for each animal.
See Methods sectionof the online Data Supplement.
Details are described in the Methods section of the online Data Supplement.
Data are expressed as mean ± S.E.M. Statistical analysis were performed using the OriginPro 8 computer software program (OriginLab, Northhampton MA). The normality assumption was assessed by the Shapiro-Wilk test. Comparisons between two groups were analyzed by 2-sided Student’s t-test. Multiple group data were analyzed by one-way ANOVA with pair-wise comparisons using Tukey’s post hoc test. In cases where normality cannot be assumed, the Wilcoxon signed-rank test was used for paired comparisons.
The authors had full access to the data and take responsibility for its integrity. All authors have read and agree to the manuscript as written.
A breeding program between p110αFlox/Flox or p110βFlox/Flox mice (supplementary Figure 1) and mice expressing a tamoxifen-regulated Cre recombinase (MerCreMer) in cardiac myocytes11 generated 2 pairs of experimental and control strains: MerCreMer/p110αFlox/Flox and p110αFlox/Flox littermates; and MerCreMer/p110βFlox/Flox and p110βFlox/Flox littermates. The animals were born at the expected Mendelian ratios and displayed no obvious abnormalities.
Adult mice from the four groups were injected with tamoxifen to activate MerCreMer. (Tamoxifen-injected MerCreMer/p110Flox/Flox and p110Flox/Flox animals are referred to as α−/− or β−/− and α+/+ or β+/+, respectively.) PCR analysis of heart DNA showed that deletion of the loxP-flanked regions was detected only in the MerCreMer-expressing animals (Figure 1A). Western blot analysis of a phosphotyrosine peptide pull-down of class IA PI3Ks from heart extracts confirmed that the level of p110α was reduced in the α−/− heart as compared to the α+/+ control (Figure 1B). The remaining signal in the α−/− heart is most likely due to p110α protein in other cell types in the heart. There were minimal changes in the amounts of p110β and p85 in the α−/− heart (Figure 1B). Similarly, the amount of p110β was reduced in the β−/− heart as compared to the control, with little effect on protein levels of p110α or p85 (Figure 1B). The level of p85 detected on Western blots of heart lysates also showed little difference between knockout and control hearts (Figure 1B).
The knockout mice appeared normal up to 6 months after tamoxifen treatment. Mean body weights were not significantly different between the knockouts and controls, but the heart weight to body weight ratio was decreased by 15% in the α−/− mice and slightly increased in the β−/− mice (supplementary Table 1). The myocytes in stained sections of α−/− or β−/− hearts appeared normal, with no obvious structural changes or evidence of apoptosis or necrosis (supplementary Figure 2).
PI3K p110α is activated by insulin and other hormones that act through receptor tyrosine kinases, and p110β is activated primarily by G protein-coupled receptors4, 12, 13. The protein kinase Akt is an important downstream effector of PI3K. Insulin activation of Akt was markedly inhibited in α−/− as compared to α +/+ hearts, but no significant difference was seen between β−/− and β+/+ hearts (Figure 1C). The protein levels of Akt were similar in the four groups (Figure 1C).
To determine if compromised PI3K signaling affects LTCC function, whole-cell patch clamping was used to measure ICa,L in isolated myocytes. ICa,L activation was elicited by depolarizing voltage pulses (300 ms duration) from −50 to +50 mV in 10 mV increments from a holding potential of −50 mV. The membrane capacitance was measured and the current amplitudes were normalized to cell capacitance to obtain current densities in picoamps (pA)/picofarads (pF). The mean cell capacitance was similar in the four groups (supplementary Table 2). Representative recordings show that ICa,L amplitude was markedly blunted in α−/− as compared to α+/+ cells (Figure 2A). In contrast, ICa,L amplitude was not affected by p110β ablation (Figure 2C).
The peak inward current densities from these recordings were used to construct current density-voltage (I–V) relationships that show a reduction in ICa,L density in α−/− myocytes as compared to control cells at most of the voltages tested (Figure 2B). In contrast, the I–V curves for β−/− and β+/+ myocytes were quite similar (Figure 2D). For statistical comparisons, the peak ICa,L density was measured at +10 mV after a single depolarizing step of 300-ms duration from a holding potential of −50 mV. The mean ICa,L density at +10 mV was 23% lower in α−/− myocytes than in α+/+ controls, whereas the ICa,L densities were not significantly different between β−/− and β+/+ myocytes (Table 1). Because the effect of p110β ablation on ICa,L was minor, the β−/− and β+/+ groups were not further examined.
To confirm that the lower ICa,L density in α−/− myocytes is due to decreased PI3K activity, we infused the cells with PI(3,4,5)P3 or recombinant p110α/p85α protein through the patch pipette. Both treatments rapidly increased ICa,L density in α−/− myocytes (Figure 2E) and restored the mean ICa,L density at +10 mV to control levels (Table 1). In contrast, infusion of PI(3,4,5)P3 had no effect on ICa,L in α+/+ myocytes (Table 1). Infusion of PI(4,5)P2 or PI(3,5)P2 had no effect on ICa,L density in myocytes of either genotype (Table 1). Infusion of inactive p110α/p85α protein also did not increase ICa,L in α−/− myocytes (Figure 2E). We next investigated if activation of ICa,L by PI(3,4,5)P3 or p110α/p85α might be mediated by Akt. Intracellular infusion of α−/− myocytes with activated Akt1 also caused a rapid increase in ICa,L density (Figure 2F) and restored the mean ICa,L density at +10 mV to control levels (10.5 ± 0.6 pA/pF, N = 14 cells). Inactive Akt1 did not increase ICa,L density (Figure 2F). Consistent with the inability of insulin to activate Akt in the α−/− heart, insulin treatment of α−/− myocytes failed to increase ICa,L density (Table 1).
The decrease in ICa,L density caused by p110α ablation could be due to altered voltage dependence of the LTCC. We addressed this possibility by analyzing the steady-state activation and inactivation of ICa,L at different voltages. No significant differences between α−/− and α+/+ myocytes were found in these experiments (supplementary Figure 3). Therefore, differences in voltage gating cannot account for the difference in ICa,L density observed between the two groups. Treatment of α−/− or α+/+ cells with isoproterenol to stimulate β-adrenergic receptors caused a 1.8-fold increase in ICa,L density in both cases, although the absolute value for ICa,L density was higher in α+/+ myocytes (Table 1). These results suggested that p110α ablation does not affect the response of the LTCC to β-adrenergic activation, but it may affect the number of channels available for activation.
We next assessed the level of LTCC protein on the surface of cardiac myocytes. Isolated myocytes from α−/− and α+/+ hearts were subjected to protein biotinylation to label cell surface Cav1.2. The amount of biotinylated Cav1.2 was markedly reduced in α−/− myocytes as compared to control cells (Figure 3A). However, immunoblotting did not reveal a difference in the total expression of Cav1.2 in membrane fractions of α−/− and control hearts (Figure 3B). These results suggest that the ICa,L deficit in α−/− myocytes is due mainly to a decrease in the number of channels on the cell surface.
The large reduction in ICa,L density in α−/− myocytes led us to suspect that these cells would exhibit a contractility defect. Single myocytes isolated from hearts of α−/− and α+/+ littermates were stimulated electrically and the change in sarcomere length was recorded as shown in Figure 4A. Ablation of p110α decreased myocyte contractility by 23% (P = 0.02, t-test; α−/−, N = 28 cells; α+/+, N = 21 cells). Myocyte contraction is directly regulated by changes in intracellular Ca2+ concentration. Therefore, we measured Ca2+ transients in α−/− and α+/+ myocytes as shown in Figure 4B. Similar to the change in myocyte contraction, ablation of p110α decreased the amplitude of Ca2+ transients by 24% (P = 0.03, t-test; α−/−, N = 22 cells; α+/+, N = 26 cells). Next, echocardiographic analysis of heart function was performed to determine if the cellular contractility defect translates into a cardiac phenotype in vivo. Indeed, α−/− mice showed a 28% decrease in fractional shortening (a measure of cardiac contractility) as compared to controls (Figure 4C). The left ventricular chamber dimensions were also mildly but significantly enlarged and the posterior wall thickness was slightly decreased in the α−/− mice (Figure 4C).
A number of pharmacological inhibitors of PI3K are being developed as drugs to treat cancer and other disorders14. The data above suggest that drugs targeting p110α could potentially have adverse side effects on cardiac contractility. We examined LTCC function in myocytes treated with PI-103, an inhibitor that is relatively selective for p110α vs. the other PI3K isoforms15 and that retarded tumor growth in a mouse model of glioma16. Canine myocytes were used in these experiments because their electrophysiological properties closely resemble those of human myocytes. Exposure of cells to PI-103 markedly inhibited ICa,L as measured by whole-cell patch clamping (Figure 5A) without affecting cell capacitance (supplementary Table 2). The I–V relationships show that PI-103 caused a reduction in ICa,L density at positive potentials (Figure 5B). The peak ICa,L density at +10 mV was reduced by 28% in the presence of the drug (Figure 5C). In contrast, pretreating myocytes with TGX-221, a relatively selective inhibitor of p110β17, had no effect on ICa,L density (Figure 5C). PI-103 induced a shift to the right in the steady-state inactivation curve of ICa,L, causing a significant increase in the membrane potential at which 50% of the channels are inactivated (supplementary Figure 4B), but this change cannot account for the large drop in ICa,L density. Infusion with PI(3,4,5)P3 had no effect on vehicle-treated canine cells, but it reversed the inhibition of ICa,L density by PI-103 (Figure 5A–C and supplementary Figure 5A), confirming that the effect of PI-103 on ICa,L is due to inhibition of PI3K. ICa,L in myocytes treated with PI-103 still responded to isoproterenol stimulation (Figure 5C and supplementary Figure 5B), and isoproterenol plus PI(3,4,5)P3 maximally stimulated ICa,L density in drug-treated cells (supplementary Figure 5C). Treatment with an Akt1/2 inhibitor also decreased peak ICa,L density at +10 mV to about the same level as that obtained with PI-103 (Figure 5C). This result is consistent with the conclusion that Akt acts downstream of p110α to maintain ICa,L at control levels. Finally, we confirmed that the reduction of ICa,L in the presence of PI-103 leads to a decrease in myocyte contractility. The change in sarcomere length was decreased by 23% in drug-treated as compared to vehicle-treated cells (Figure 5D; P = 0.02, t-test; Vehicle, N = 18; PI-103, N = 14).
Pharmacological and genetic studies have indicated that p110α plays an important role in insulin action, cell size control and tumorigenesis5, 12, 15. This study shows that p110α also plays a critical role in regulating LTCC function in the heart. Cardiac-specific ablation of p110α in adult mice caused significant reductions in ICa,L density, Ca2+ transients, and myocyte contraction that resulted in compromised cardiac contractility. Reduced cell surface expression of the LTCC in α−/− myocytes can account for the defects in Ca2+ handling and contractility. The decrease in cell surface LTCC in the knockout myocytes is greater than expected when compared to the more modest reduction in ICa,L density. This could be due to a preferential internalization of inactive channels following downregulation of p110α signaling. The rapid increase in ICa,L density in α−/− myocytes in response to infusion of PI(3,4,5)P3, recombinant p110α/p85α PI3K or activated Akt1 protein is probably due to redistribution of the LTCC from internal membrane compartments to the plasma membrane. In a previous study using myocytes from diabetic Ins2Akita mice, we showed that this response is microtubule-dependent2. Akt phosphorylation of an intracellular subunit of the LTCC also plays a role in increased trafficking to the cell surface18.
Earlier studies that sought to define a role for PI3Ks in cardiac physiology concluded that, in general, p110γ negatively regulates contractility and p110α positively controls heart size19, 20. Experiments that examined the function of p110α were done using transgenic mice that overexpress either constitutively active or dominant-negative (DN) p110α proteins in cardiac myocytes3, 19–21. Mice carrying DN p110α showed a 16% decrease in heart weight to body weight ratio, similar to the α−/− mice described herein. In contrast to our α−/− mice, expression of DN p110α did not cause a change in cardiac contractility or ICa,L density3, 19, 20. Aside from strain differences, other distinctions between the two models could account for these discrepancies. First, the DN p110α protein could potentially sequester p85 subunits and inhibit activation of other class IA PI3Ks, but ablation of p110α would not be expected to have such a global inhibitory effect. Second, the animals used in this study were subjected to gene ablation after reaching adulthood, whereas the DN p110α transgene was active in the heart during embryonic development and after birth20. Thus, any compensatory processes that might affect cardiac function might be distinct in the two models. For example, we observed no difference in Akt expression in the hearts of α+/+ and α−/− mice, but the amount of Akt protein was higher in hearts of DN p110α transgenic mice than in controls20. Increased Akt activity in the transgenic myocytes could explain why ICa,L density and cardiac contractility were not reduced. Since human heart disease usually occurs long after postnatal development, we believe that our strategy of modifying PI3K in adult animals is more relevant to the human condition than the use of standard transgenic or knockout animals. Moreover, our demonstration that a p110α inhibitor reduces ICa,L activation and contractility in canine myocytes suggests that this might be a general mechanism to regulate cardiac function.
Sun et al. reported that PTEN ablation or IGF-1 treatment of mouse cardiac myocytes led to increased ICa,L density and their data suggested that increased p110α/Akt signaling mediates this response3. Although their results are complementary to our findings, there are significant differences between our studies. As shown here and elsewhere2, we did not observe an increase in ICa,L in response to insulin treatment or PI(3,4,5)P3 infusion of control myocytes. Furthermore, while Sun et al.3 reported that inactivation of p110α or Akt did not affect ICa,L density relative to controls, we found that downregulating p110α/Akt signaling from the basal level decreases ICa,L. Technical differences might explain some of these differences. In our system, most LTCCs appear to be already located at the plasma membrane in the basal state and upregulating PI3K signaling would not be expected to stimulate a large net movement of LTCCs to the cell surface. Our data suggest that basal p110α/Akt signaling maintains LTCCs at the cell surface and therefore manipulations that block this signaling result in a net movement of LTCCs to the intracellular membrane compartment, thus resulting in decreased ICa,L density.
Ablation of genes for the p85α and p85β PI3K regulatory subunits led to a loss of p110α protein in the heart with no deleterious effect on cardiac contractility22. Interestingly, like the DN p110α mice, the amount of Akt protein was higher in p85 knockout hearts than in controls. The relatively normal cardiac function in the p85 knockout mice is surprising. It would be interesting to determine if LTCC function is affected by loss of the p85 proteins.
To our knowledge, this is the first study to interrogate the cardiac function of p110β. Our results suggest that p110β does not play a major direct role in regulating cardiac LTCC function. Insulin activation of Akt was normal in β−/− hearts, in agreement with recent studies demonstrating that p110β is not a major effector of insulin signaling in other tissues4, 12, 13. Our data indicate that insulin activation of Akt in myocytes is mediated mainly by p110α. Together, our results suggest that decreased ICa,L and contractility defects in insulin-resistant or -deficient diabetic animals can be attributed at least in part to a reduction in p110α activation2, 8, 9.
Mutations in the p110α gene have been detected at a high frequency in human cancers. Since most of these mutations lead to constitutive activation of the enzyme, the therapeutic benefit of inhibitors that target p110α is potentially very high5, 6, 14. A number of drugs that target PI3K have either entered clinical trials or are about to enter clinical trials. Our results raise the possibility that systemic inhibition of p110α might cause adverse cardiac side effects in some patients. Although LTCC blockers are used for the treatment of hypertension, these drugs are often contraindicated in patients with underlying contractile dysfunction or heart failure. We suspect that p110α inhibitors may have a similar deleterious effect in susceptible patients. Our results also suggest that Akt inhibitors, also being investigated for cancer therapy23, might have similar negative effects on cardiac contractility. It is interesting to note that some cancer patients treated with imatinib or sunitinib develop left ventricular dysfunction24, 25. We speculate that these tyrosine kinase inhibitors might exert adverse cardiac side effects in part by inhibiting the activation of p110α. These results will have important implications for the clinical use of drugs that target this important signaling pathway.
We appreciate the technical assistance of Joan Zuckerman. We thank J. D. Molkentin (Univ. of Cincinnati) for the generous donation of the transgenic MerCreMer mouse.
This study was funded in part by grants from the American Heart Association, Juvenile Diabetes Research Foundation, the Department of Veterans Affairs Merit Review program (R.Z.L.) and the National Institutes of Health (DK62722, R.Z.L.; HL28958, HL67101, I.S.C.; HL-85221, R.T.M.).
Small molecule inhibitors of phosphatidyinositol 3-kinases are currently being tested in early clinical trials for treatment of cancer. These drugs inhibit all four phosphatidyinositol 3-kinase catalytic isoforms, including p110α. Results from this study show that inhibition of p110α in cardiac myocytes reduces the ion current through the L-type Ca2+ channel. Since the L-type Ca2+ current regulates myocyte contraction, treating cancer patients with underlying heart problems with non-selective phosphatidyinositol 3-kinase inhibitors might lead to deleterious cardiac side effects.
Subject codes:  Ion channels/membrane transport;  Cell signalling/signal transduction;  Contractile function.