Phosphoinositide 3-kinase(p110α) plays a critical role for the induction of physiological, but not pathological, cardiac hypertrophy

Abstract

An unresolved question in cardiac biology is whether distinct signaling pathways are responsible for the development of pathological and physiological cardiac hypertrophy in the adult. Physiological hypertrophy is characterized by a normal organization of cardiac structure and normal or enhanced cardiac function, whereas pathological hypertrophy is associated with an altered pattern of cardiac gene expression, fibrosis, cardiac dysfunction, and increased morbidity and mortality. The elucidation of signaling cascades that play distinct roles in these two forms of hypertrophy will be critical for the development of more effective strategies to treat heart failure. We examined the role of the p110α isoform of phosphoinositide 3-kinase (PI3K) for the induction of pathological hypertrophy (pressure overload-induced) and physiological hypertrophy (exercise-induced) by using transgenic mice expressing a dominant negative (dn) PI3K(p110α) mutant specifically in the heart. dnPI3K transgenic mice displayed significant hypertrophy in response to pressure overload but not exercise training. dnPI3K transgenic mice also showed significant dilation and cardiac dysfunction in response to pressure overload. Thus, PI3K(p110α) appears to play a critical role for the induction of physiological cardiac growth but not pathological growth. PI3K(p110α) also appears essential for maintaining contractile function in response to pathological stimuli.

Materials and Methods

Animal Experiments. Animal care and experimentation were approved by the Institutional Animal Care and Use Committee of Beth Israel Deaconess Medical Center. Experiments were performed on dnPI3K(p110α) and Ntg littermate controls (FVB/N background). dnPI3K(p110α) transgenic mice were generated as described (13).

Aortic Banding. Ascending aortic constriction was performed in male Ntg or dnPI3K mice (at least 11 weeks of age) as described (18), except that mice were anaesthetized with pentobarbital sodium salt (60 mg/kg) and the aorta was ligated by using a 25-gauge needle. One week after surgery, echocardiography was performed and mice were killed.

Swimming Training. For chronic swimming training, groups of 14–16 8- to 10-week-old male Ntg or dnPI3K transgenic mice were swum in water tanks for 4 weeks as described (9), except that mice were trained 7 days a week instead of 5 days. Further details are provided in Supporting Materials and Methods, which is published as supporting information on the PNAS web site, www.pnas.org.

Treadmill Performance. To compare the ability of dnPI3K and Ntg mice to perform exercise, a group of mice that did not undergo swimming training were tested for treadmill performance. An exhaustion test was performed as described (19) with some modifications. Details are provided in Supporting Materials and Methods.

Hemodynamic Measurements. LV pressure was measured as described in Supporting Materials and Methods.

Echocardiography. One week after aortic banding or at the completion of swimming training, echocardiography was performed as described (13), except that 2,2,2-tribromoethanol (Aldrich, 0.4–0.6 mg/g) was used for anesthesia.

Histological Analysis. Heart sections were cut and stained with Masson trichrome as described (13).

Morphometric Analysis of Isolated Cardiac Myocytes. Cardiac myocytes were enzymatically dissociated from hearts of NTg swimming and nonswimming mice, and the long axis, short axis, and cell area were measured as described (13).

Citrate Synthase Activity. Citrate synthase activity was measured in mixed gastrocnemius muscle as described (20), except that the concentration of oxaloacetate was 10.0 mM.

Northern Blot Analysis. Extraction of total RNA and Northern blotting was performed as described (13). Membranes were probed with atrial natriuretic peptide (ANP), brain natriuretic peptide, β-myosin heavy chain (β-MHC), α-MHC, α-skeletal actin, sarcoplasmic reticulum Ca²⁺ ATPase 2a, transforming growth factor β3, and GAPDH-radiolabeled probes. Details of the probes are provided in Supporting Materials and Methods.

Western Blotting. Phosphorylation of Akt, S6, and extracellular responsive kinase (ERK) 1/2 were examined by Western blotting in heart lysates as described (13). To measure the phosphorylation of p38, blots were probed with an anti-phospho p38 antibody (Cell Signaling Technology, Beverly, MA, 1:500) followed by an anti-p38 antibody (Santa Cruz Biotechnology, 1:200).

Statistics. Results are expressed as mean ± SEM. When comparing groups statistical significance was determined by using one-way ANOVA. If the ANOVA showed significance (P < 0.05), it was followed by the Fisher's projected least significant difference post hoc test. Significance level was P < 0.05.

Results

Response of dnPI3K Mice to Pressure Overload and Exercise Training. In a pilot study, we found that aortic banding was associated with a marked increase in peak LV systolic pressure (≈70 mmHg), whereas the LV pressure of swimming mice at rest was normal (Fig. 4, which is published as supporting information on the PNAS web site). As we reported (13), control dnPI3K transgenic mice (sham or nonswim) had smaller hearts compared with that of Ntg mice (Fig. 1). In response to aortic banding or exercise training the heart weights of Ntg mice increased significantly, and the degree of hypertrophy attained in Ntg mice with both models (i.e., swimming and aortic banding) was similar (Fig. 1 and Table 2, which is published as supporting information on the PNAS web site). The normalized HW/BW of banded dnPI3K mice was not different to that of banded Ntg (Fig. 1). These data suggest that PI3K(p110α) signaling is not essential for the development of pressure overload-induced hypertrophy. By contrast, the normalized HW/BW of dnPI3K mice subjected to swimming training was significantly lower than that of Ntg mice (Fig. 1). Thus, the dnPI3K mutant was able to significantly blunt the hypertrophic response observed in Ntg mice induced by exercise but not by pressure overload. The small degree of hypertrophy observed in dnPI3K transgenic mice in response to swimming may be caused by residual PI3K(p110α) activity (≈23%; ref. 13) or the activation of other signaling pathways.

PI3K(p110α) Is Critical for the Adaptation of the Heart to Exercise. Pathological and physiological stimuli are known to cause an increase in heart mass by different alterations in the volume of the cardiac chambers and in the thickness of ventricular walls (8, 10). In humans, isotonic exercise (swimming and running) has been associated with an increase in chamber dimensions and a proportional increase or no change in wall thickness (8, 10). By contrast, patients with chronic pressure overload have thick walls with relatively small ventricular chambers (8, 10). To investigate this in the current study, echocardiographic studies were performed. In response to swimming, the chamber dimensions of Ntg mice significantly increased (Table 1 and Fig. 5A, which is published as supporting information on the PNAS web site). In contrast, there was no significant change in the chamber dimensions or wall thickness of hearts from dnPI3K transgenic mice that underwent exercise training. Aortic banding resulted in a significant increase in the LV wall thickness of hearts from Ntg mice, and a comparable increase was observed in banded dnPI3K transgenic mice (Table 1). The degree of aortic stenosis, measured by the aortic pressure gradients (AoPg) across the bands of Ntg and dnPI3K transgenic mice, was similar (28.2 ± 4.1 and 24.8 ± 1.9 mmHg, respectively; Table 1). These data further suggest that PI3K(p110α) plays a critical role in the adaptation of the heart to exercise but not to pressure overload.

dnPI3K Mice Display Contractile Dysfunction in Response to Pressure Overload. In addition to the increased LV wall thickness observed in banded dnPI3K transgenic mice, hearts from these mice displayed marked dilation, as shown by a significant increase in LV end-diastolic dimension and LV end-systolic dimension (Table 1). Thus, LV dimensions can increase in both physiological and pathological hypertrophy as observed in Ntg swimmers and banded dnPI3K transgenics. However, the relative changes in these dimensions determine whether contractile function is depressed. Because a defining character of the pathological transition from hypertrophy to heart failure is decreased cardiac contractility, we measured fractional shortening, an echocardiographic index of LV systolic function. Even though the aortic gradients across the bands of Ntg and dnPI3K mice were similar (Table 1), fractional shortening measured 1 week after banding was significantly reduced in dnPI3K banded mice compared with Ntg banded mice (Table 1 and Fig. 5B). Further, the lung weight/body weight ratio, an index of pulmonary congestion and LV dysfunction, was greater in dnPI3K banded mice than in Ntg banded mice (dnPI3K: 9.5 ± 1.4, n = 8; Ntg: 7.2 ± 0.6, n = 13; P < 0.05). In contrast, cardiac contractility was not compromised in Ntg or dnPI3K mice subjected to swimming (fractional shortening normal in both; Table 1). Even though PI3K(p110α) does not play a critical role for the induction of hypertrophy in response to pressure overload (defined as an increase in heart size), it does play an essential role in maintaining cardiac function in response to pathological stimuli.

We considered that the attenuated cardiac hypertrophic response of dnPI3K transgenic mice to swimming training might be caused by the inability of these mice to exercise to the same degree as Ntg mice. This idea was considered unlikely for a number of reasons. First, blinded observers were unable to distinguish dnPI3K mice from Ntg mice while they underwent swimming training; second, the dnPI3K mutant was expressed only in the heart, hence PI3K in skeletal muscle was unaffected; and finally, dnPI3K mice showed no signs of heart failure when followed for up to 18 months (13). However, to more quantitatively address this question we performed a number of experiments. Citrate synthase activity, an index of muscle oxidative capacity and hence physical training, was measured in mixed gastrocnemius muscle of mice that underwent swimming training. After 4 weeks of exercise, citrate synthase activity was elevated to a similar degree in skeletal muscle of Ntg and dnPI3K transgenic mice compared with nonswimming mice (Fig. 6A, which is published as supporting information on the PNAS web site). In addition, groups of mice that did not undergo swimming training were tested for treadmill performance (exhaustion and endurance tests). During the exhaustion test, the distance traveled and work capacity of Ntg and dnPI3K transgenic mice were not different (Fig. 6 B and C). Further, both dnPI3K and Ntg were able to sustain aerobic exercise (horizontal running at 14 m/min) for 60 min without displaying any signs of exhaustion.

Pressure Overload- and Exercise-Induced Hypertrophy Are both Associated with an Increase in Myocyte Size but with Distinct Molecular and Histological Phenotypes. We previously reported that aortic banding was associated with an increase in myocyte area (21). In the current study, the long axis, short axis, and mean cell area of isolated cardiac myocytes from hearts of swimming mice were greater than that of myocytes from nonswimming mice (Fig. 2A Upper). Next, we examined whether swimming-induced hypertrophy and pressure overload-induced hypertrophy were associated with distinct histological and molecular phenotypes, as described in physiological and pathological models of cardiac hypertrophy (5, 6, 8, 22–24). In the present study, aortic banding was associated with interstitial fibrosis in hearts from Ntg and dnPI3K mice, whereas there was no fibrosis in the ventricular walls of swimming mice in either group (Fig. 2 A Lower). In addition, pressure overload was associated with marked changes in cardiac gene expression (Fig. 2B). ANP, brain natriuretic peptide, β-MHC, α-skeletal actin, and transforming growth factor β3 mRNAs were elevated in hearts of banded Ntg mice compared with sham (Fig. 2B). By contrast, expression of these genes was not altered in hearts of swimming Ntg mice. There were no significant differences between levels of α-MHC and sarcoplasmic reticulum Ca²⁺ ATPase 2a (SERCA2a), although there was a trend for depressed levels of SERCA2a in Ntg band compared with Ntg sham (P < 0.09). Interestingly, ANP, brain natriuretic peptide, β-MHC and α-skeletal actin mRNAs were elevated in hearts from control dnPI3K mice (sham or nonswim) compared with Ntg mice from the same groups (Fig. 2B). Swimming, in part, reversed the changes in ANP, β-MHC, and α-skeletal actin mRNAs found in dnPI3K nonswimmers. This finding is reminiscent of early reports in which swimming training was shown to prevent or reverse molecular abnormalities associated with pathological hypertrophy (7). The functional significance of the discordant regulation of these genes in dnPI3K transgenic mice at baseline and in response to pathological or physiological stimuli requires further investigation.

Exercise Training Activates Akt, a Downstream Target of PI3K(p110α). Signaling pathways activated in response to pressure overload have been relatively well characterized (23, 25), whereas those activated in response to exercise are less clear. To identify potential downstream signaling pathways involved in the hypertrophic response induced by swimming, we measured the phosphorylation of Akt, S6, ERK1/2, and p38 at four time points during the 4-week training period in Ntg mice. Akt is the best-characterized target of PI3K(p110α) (26), and Gq-dependent activation of ERK1/2 and p38 has been reported in response to pressure overload (25). Phosphorylation of Akt (pAkt) and S6 (pS6) were elevated at some time points during the development of swimming induced-hypertrophy (Fig. 2C), whereas phosphorylation of ERK (pERK1/2) or p38 did not significantly change (Fig. 7, which is published as supporting information on the PNAS web site). These results are reminiscent of the findings in PI3K(p110α) transgenic mice. In hearts of dnPI3K transgenic mice pAkt and pS6 were depressed, whereas they were increased in caPI3K mice; pERK1/2 was unaffected in both transgenics (13).

Phosphorylation of Akt was not different in hearts from banded mice compared with sham operated mice at any time point (Fig. 8, which is published as supporting information on the PNAS web site). By contrast, pERK1/2 was elevated at 48 h and phospho-p38 was elevated at 1 week in banded mice (Fig. 2D). Of note, the p110γ isoform of PI3K but not the p110α isoform was shown to be activated during pressure overload (27). Thus, PI3K coupled to receptor tyrosine kinases (p110α) appears to be important for developmental- and exercise-induced cardiac growth, whereas pressure overload activates PI3K coupled to Gq-coupled receptors (p110γ) (27).

Discussion

To develop more effective strategies to treat cardiac hypertrophy and failure it is critical that we have a better understanding of the signaling pathways involved in the development of pathological versus physiological cardiac hypertrophy. Although previous studies have shown that physiologic and pathologic hypertrophy have distinct functional characteristics, it was not clear whether these two forms of hypertrophy were mediated by common or distinct signaling pathways. This work illustrates that PI3K(p110α) is critical for the induction of physiological (swimming induced) but not pathological hypertrophy (pressure overload induced).

As previously stated, the dnPI3K mutant did not completely block the response to exercise training. The increase in the HW/BW of ≈20% in dnPI3K swimmers compared with dnPI3K nonswimmers is most likely caused by residual PI3K(p110α) activity in dnPI3K transgenics (≈23%; ref. 13) or the activation of other signaling pathways. Further, it is worth noting that any increase in the heart weight of dnPI3K transgenics appears relatively large when expressed as a percent of dnPI3K hearts, because of their smaller size at baseline compared with Ntg. To put this in perspective, the HW/BW of Ntg subjected to aortic banding or swimming increased by ≈40% in each model, whereas the HW/BW of dnPI3K transgenics increased by only 20% in response to swimming but >65% in response to aortic banding. We believe this is compelling evidence to support the claim that PI3K(p110α) plays a more critical role for the induction of physiological than pathological cardiac hypertrophy.

Of clinical relevance with regard to the transition from hypertrophy to heart failure, we also show that PI3K(p110α) is essential for maintaining cardiac function in response to a pathological stimuli. dnPI3K mice displayed marked cardiac dysfunction in response to pressure overload. In just 1 week, fractional shortening was reduced by ≈35%, and hearts of dnPI3K transgenic mice displayed significant dilation. In Ntg mice contractile dysfunction in response to aortic banding does not usually occur until at least 4 weeks (28, 29). This study shows that the p110α isoform of PI3K plays an important role with respect to contractile function under a pathological stress. Further studies are required to elucidate the mechanisms responsible for the depressed function in dnPI3K transgenics in response to aortic banding (e.g., increased fibrosis, necrosis, depressed excitation-contraction coupling).

Previous studies have suggested that the p110γ isoform is critical for contractile function rather than the p110α isoform (17, 27). However, an important distinction between these isoforms may exist. Under basal conditions dnPI3K(p110α) transgenics display no signs of cardiac dysfunction and there is no evidence of increased necrosis, apoptosis, or fibrosis compared with Ntg mice (13). In contrast, the study by Crackower et al. (17), which used PI3K(p110γ) null mutant mice, suggests that the p110γ isoform is critical for the regulation of contractility under basal conditions. Thus, it appears that the p110γ isoform is critical for contractile function under basal conditions, whereas p110α is critical for the regulation of contractility in response to a pathological stress.

In the current study, Akt was activated in response to swimming whereas two mitogen-activated protein kinase pathways (i.e., ERK1/2 and p38) were not. It was previously reported that p38 and ERK1/2 were activated in response to treadmill exercise (30). However, in that study the treadmill protocol involved running mice only twice a week for 10 weeks at a speed and inclination that induced acute stress. Further, in that study there was no significant increase in the HW/BW of wild-type mice. Our model was designed to mimic daily aerobic exercise and clearly resulted in an increase in the HW/BW. We believe these differences can account for this discrepancy.

Signaling Pathways Involved in the Development of Pathological and Physiological Cardiac Hypertrophy. The current study suggests that PI3K(p110α) plays an important role for the induction of physiological cardiac hypertrophy but not pathological hypertrophy. The idea that different signaling cascades may be important for the induction of these two forms of hypertrophy is supported by other reports in the literature. Based on these findings, we present a model illustrating pathways that may be involved for the development of physiological and pathological cardiac hypertrophy (Fig. 3). Cardiac-specific overexpression of Gαq, which is activated by a number of ligands, including angiotensin II (Ang II) and endothelin-1, induced cardiac hypertrophy akin to pathological hypertrophy (14). In addition, transgenic mice expressing a GqI peptide (specific for inhibiting Gq-coupled receptor signaling) in cardiac myocytes (31), and mice lacking the trimeric G proteins Gα_q and Gα₁₁ in cardiomyocytes (23), did not develop cardiac hypertrophy in response to pressure overload, suggesting the G_q/11 pathway is important for the induction of pathological hypertrophy. Furthermore, pressure overload-induced hypertrophy but not swimming-induced hypertrophy was inhibited by Ang II receptor blockade (24, 32). Downstream of Gq, p38, c-Jun NH(2)-terminal kinase (JNK), and PKCβ all have been implicated in mediating pathological hypertrophy (15, 33–35). Recent studies suggest that p38 plays a critical role in the development of fibrosis in response to pathological stimuli but not to cardiac growth itself (36, 37). Further, using mice deficient in mitogen-activated protein kinase kinase kinase (MEKK1, preferentially activates the JNK pathway) it was shown that MEKK1–JNK does not mediate cardiac hypertrophy (38). Thus, the model presented is a working model that will require modification once the role of these signaling molecules are more extensively studied.

Perhaps the most likely candidate responsible for mediating exercise-induced cardiac hypertrophy is IGF-I, which activates PI3K(p110α) coupled to receptor tyrosine kinases. Serum levels of IGF-I were increased in competitive swimmers (39) and rodents that underwent chronic exercise training (40, 41). In addition, cardiac-specific overexpression of the IGF-I receptor in transgenic mice resulted in cardiac hypertrophy that was characteristic of physiological hypertrophy, and the increase in heart weight was suppressed when these mice were crossed with dnPI3K transgenic mice.¶ Finally, cardiac formation of IGF-I, but not endothelin-1, or Ang II, was higher in professional athletes than in control subjects (42). By contrast, cardiac formation of Ang II was increased in hypertrophied hearts of patients with heart failure (43). In agreement with our finding that PI3K(p110α) does not play a critical role for pressure overload-induced hypertrophy, mice with severe IGF-I deficiency developed significant cardiac hypertrophy in response to pressure overload (44).

Even though PKCε and ERK1/2 are activated by Gq in response to pathological stimuli, transgenic models have suggested that both these signaling molecules may result in physiological hypertrophy, which is a beneficial compensatory event (Fig. 3 and refs. 45–47). Calcineurin signaling, another important hypertrophic mediator, has been implicated for the induction of pathological and physiological cardiac hypertrophy (48). Calcineurin inhibitors were shown to inhibit cardiac hypertrophy in response to pressure overload or in transgenic mouse models of cardiomyopathy (49). More recently, cardiac-specific overexpression of the calcineurin inhibitory protein, myocyte-enriched calcineurin-interacting protein (MCIP) 1, in transgenic mice was shown to inhibit the hypertrophic response in a genetic model of cardiomyopathy (ca form of calcineurin), to β-adrenergic receptor stimulation or exercise training (50). It has been suggested that glycogen synthase kinase-3β (GSK-3β) provides a mechanism for cross talk between the Akt and calcineurin pathways (51). Akt directly phosphorylates GSK-3β, resulting in its inactivation. It was reported that mice expressing a ca form of GSK-3β did not develop exercise-induced hypertrophy, thereby, indicating the importance of the PI3K/Akt/GSK-3β pathway in physiological hypertrophy.∥

We previously reported that PI3K(p110α) was essential for developmental heart growth; dnPI3K transgenics had smaller hearts, whereas caPI3K transgenics had larger hearts, which resembled physiological hypertrophy (13). Taken together with the current study, these findings suggest that developmental cardiac growth and exercise-induced cardiac growth may be mediated by the same signaling pathways. In support of this hypothesis, developmental heart growth was not affected by inhibition of the Gq “pathological” signaling pathway (23, 31).

Clinical Significance. The current study suggests that PI3K(p110α) could be a potential tool for augmenting physiological growth and improving function of the diseased heart, while used in conjunction with other agents that inhibit pathological hypertrophy. Growth hormone and IGF-I were considered as potential therapeutic agents for circumstances in which hypertrophy of cardiomyocytes would be desirable, such as postmyocardial infarction or dilated cardiomyopathies (52). However, the results have been conflicting, probably because of growth hormone and IGF-I acting via other signaling pathways in addition to PI3K (52). Based on the work presented here, it may be of benefit to selectively target PI3K(p110α) as a potential strategy for the treatment of heart failure.

Supplementary Material

Acknowledgments

We thank M. Rivera for assistance with swimming the mice, P. Jay and O. Rozhitskaya for Northern probes, S. Ngoy for performing some of the mouse operations, and W. T. Pu, P. Jay, and M. Schinke for discussions. This work was supported by National Institutes of Health Grants RO1 HL65742 and U01HL66582 (to S.I.).

Footnotes

References

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.