Author + information
- Received March 10, 2014
- Revision received April 25, 2014
- Accepted April 30, 2014
- Published online August 19, 2014.
- Firdos Ahmad, PhD∗,†,
- Hind Lal, PhD∗,†,
- Jibin Zhou, PhD∗,
- Ronald J. Vagnozzi, PhD∗,
- Justine E. Yu, BS∗,
- Xiying Shang, PhD∗,
- James R. Woodgett, PhD‡,
- Erhe Gao, PhD∗ and
- Thomas Force, MD∗,†,§∗ ()
- ∗Center for Translational Medicine, Temple University School of Medicine, Philadelphia, Pennsylvania
- †Division of Cardiovascular Medicine, Vanderbilt University, Nashville, Tennessee
- ‡Lunenfeld-Tanenbaum Research Institute, Mount Sinai Hospital, Toronto, Ontario, Canada
- §Cardiology Division, Temple University School of Medicine, Philadelphia, Pennsylvania
- ↵∗Reprint requests and correspondence:
Dr. Thomas Force, Temple University School of Medicine, Center for Translational Medicine and Cardiology Division, 3500 N. Broad Street, Philadelphia, Pennsylvania 19140.
Background Injury due to myocardial infarction (MI) is largely irreversible. Once an infarct has occurred, the clinical goal becomes limiting remodeling, preserving left ventricular function, and preventing heart failure. Although traditional approaches (e.g., β-blockers) partially preserve left ventricular function, novel strategies are needed to limit ventricular remodeling post-MI.
Objectives The aim of this study was to determine the role of glycogen synthase kinase–3α (GSK-3α) in post-MI remodeling.
Methods Mice with cardiomyocyte-specific conditional deletion of Gsk3α and littermate controls underwent sham or MI surgery. Heart function was assessed using serial M-mode echocardiography.
Results Gsk3α deletion in the heart markedly limits remodeling and preserves left ventricular function post-MI. This is due at least in part to dramatic thinning and expansion of the scar in the control hearts, which was less in the heart of knockout (KO) mice. In contrast, the border zone in the KO mice demonstrated a much thicker scar, and there were more viable cardiomyocytes within the scar/border zone. This was associated with less apoptosis and more proliferation of cardiomyocytes in the KO mice. Mechanistically, reduced apoptosis was due, at least in part, to a marked decrease in the Bax/Bcl-2 ratio, and increased cardiomyocyte proliferation was mediated through cyclin E1 and E2F-1 in the hearts of the KO mice.
Conclusions Taken together, these findings show that reducing GSK-3α expression in cardiomyocytes limits ventricular remodeling and preserves cardiac function post-MI. Specifically targeting GSK-3α could be a novel strategy to limit adverse remodeling and heart failure.
Myocardial infarction (MI) and its sequelae, in particular heart failure, are the leading causes of death in the developed world and rapidly are becoming major killers in the developing world (1,2). The neonatal heart has the capability of regeneration through cardiomyocyte proliferation (3), but it appears to lose much of this capability early post-birth (4). Various stressors, most importantly ischemic injury, can lead to progressive cardiomyocyte loss. Apoptosis is uncommon in the normal heart but is frequently seen in the ischemic heart (5). The endogenous regenerative capacity of the heart appears insufficient to replenish lost myocytes, so even relatively low levels of apoptosis can have profound effects on cardiac function (5). This has necessitated the use of strategies such as the injection of stem or progenitor cells into the infarcted region (6). Although these strategies hold promise, the activation of endogenous repair mechanisms within the heart could, in theory, represent a superior strategy.
Glycogen synthase kinase (GSK)–3β has been well studied in various pathophysiological conditions (7–11), but the role of GSK-3α in various pathologic settings, including ischemic injury and heart failure, is still poorly understood. This bias toward GSK-3β arose from early reports using Drosophila that concluded that GSK-3β was the dominant isoform because it was better able to rescue Wnt/wingless pathway defects (12,13). This bias appears to remain, with many more publications focusing on GSK-3β than on GSK-3α. To summarize, although the roles of GSK-3α remain unclear in many settings, it is clear that there are unique, overlapping functions of the 2 isoforms. In short, despite the 97% sequence similarity, these kinase have many nonoverlapping roles.
Global deletion of Gsk3α is detrimental in the setting of stress, leading to deterioration in cardiac function and increased hypertrophy, cardiac fibrosis, and ventricular remodeling (14,15). Surprisingly, a similar phenotype of heart failure and hypertrophy has been reported with a constitutively active (S21A mutant) Gsk3α global knock-in mouse when challenged with pressure overload (16). Furthermore, cardiac-specific overexpression of GSK-3α via transgenesis attenuates cardiac hypertrophy and increases fibrosis and apoptosis post transverse aortic constriction (17,18). These observations, from 3 different models, have led to uncertainty as to the roles of GSK-3α in the pathophysiology of heart failure. Moreover, global knockout (KO) or knock-in models can lead to compensatory effects that complicate the interpretation of phenotypes (19) and rarely, if ever, recapitulate the conditional gene deletion, for a variety of reasons.
Herein, for the first time, we use a cardiomyocyte-specific conditional KO mouse to characterize the rather surprising roles of GSK-3α in regulating the consequences of MI. We report that conditional deletion of GSK-3α leads to marked protection post-MI that is due to less cardiomyocyte death, less scar expansion and thinning, and increased cardiomyocyte proliferation in the KO hearts, the latter being driven by alterations in the activity of cell-cycle regulators. We believe that these studies will bring more clarity to the true roles played by GSK-3α in the infarcted heart.
A detailed method is supplied in the Online Appendix.
The Gsk3αflox/flox (fl/fl) mouse was generated as previously described (8). Mice expressing α-myosin heavy chain promoter-driven, tamoxifen (Tam)–inducible heterozygous Mer-Cre-Mer (a gift from Dr. J. Molkentin, Cincinnati Children’s Hospital, Cincinnati, Ohio) were crossed for 2 generations with Gsk3αfl/fl mice to generate Gsk3αfl/fl Cre mice. Both mouse strains were on the C57BL/6 background. At 12 weeks of age, when physiological development is largely complete, all male mice were placed on a Tam chow diet (400 mg/kg) for 15 days, followed by regular chow for an additional 15 days (to allow the clearance of Tam from the mice). Gsk3αfl/fl/ Cre+/− /Tam mice were conditional KO, whereas littermates Gsk3αfl/fl/Tam represented controls (wild-type [WT]). The Institutional Animal Care and Use Committee of Temple University approved all animal procedures and treatments.
A detailed antibody list and application is supplied in the Online Appendix.
After baseline echocardiography, permanent occlusion of the proximal left anterior descending coronary artery, or sham surgery, was performed in WT versus KO male littermates, as described previously (20). After the 8-week echocardiographic examination, the mice were euthanized for different studies.
Echocardiography was performed as described previously (15). In brief, transthoracic 2-dimensional motion-mode echocardiography was performed at 0, 1, 2, 4, and 8 weeks post-MI with a 12-MHz probe (VisualSonics, Toronto, ON, Canada) on mice anesthetized by inhalation of isoflurane (1.0% to 1.5%). Left ventricular (LV) end-systolic interior dimension, end-diastolic interior dimension, ejection fraction, and fractional shortening values were analyzed using the Vevo770 program (VisualSonics).
Histochemistry and immunohistochemistry
A detailed protocol is provided in the Online Appendix. In brief, 3 and 8 weeks post-MI, the whole heart was excised from anesthetized mice and fixed in 4% paraformaldehyde, dehydrated through increasing concentrations of ethanol, and then embedded in paraffin (9). Heart sections (5 μm) were stained using Masson’s trichrome (Sigma-Aldrich, St. Louis, Missouri) or immunofluorescence antibodies. A Nikon Eclipse 80i fluorescent microscope and NIS Elements software (Nikon Corporation, Tokyo, Japan) were used to capture the images and analysis.
Differences between data groups were evaluated for significance using unpaired Student t test (GraphPad Prism, GraphPad Software, San Diego, California). Data are expressed as mean ± SEM. For all tests, p values <0.05 were considered to indicate statistical significance.
To test our hypothesis that GSK-3α plays a critical role in MI-induced remodeling, we first determined the effect of ischemic injury on GSK-3α activity, as assessed by phosphorylation. WT mice were subjected to MI surgery, and 3 weeks later, LV lysates were analyzed for GSK-3α phosphorylation (on Ser21). Phosphorylation of GSK-3α was significantly increased, confirming that chronic MI leads to inhibition of GSK-3α (Figs. 1A and 1B, Online Figs. 1A and 1B). Therefore, we asked if GSK-3α might have a role in MI-induced ventricular remodeling. We used the Tam-inducible Mer-Cre-Mer system to delete Gsk3α specifically in cardiomyocytes. The progeny were viable, fertile, and showed no overt pathologic phenotype. After Tam treatment, expression levels of GSK-3α and GSK-3β were determined by immunoblotting. GSK-3α protein expression in the KO mice was reduced to 12% of baseline, and GSK-3β expression and activity were comparable (Figs. 1C to 1F).
Deletion of Gsk3α limits post-MI cardiac remodeling and preserves contractile function
Gsk3α KO and littermate control mice underwent sham or MI surgery, and cardiac function was determined using serial 2-dimensional motion-mode echocardiography at various time points up to 8 weeks. Although post-MI infarct size, hypertrophy, fibrosis, and inflammation (as assessed by nuclear factor κB activation) were comparable in WT and KO mice (Online Figs. 1 to 4), ventricular chamber dilation, both in systole and diastole, was significantly less in the KO mice, consistent with reduced remodeling (Figs. 2A and 2B). The preserved chamber dimensions in the KO mice were associated with significantly better LV function, as reflected by preserved ejection fraction and fractional shortening (Figs. 2C and 2D). The smaller LV internal dimension and the preserved cardiac function in the KO mice were consistent and remained significantly better until the end of the study (8 weeks post-MI). Although global Gsk3α deletion leads to detrimental phenotypes due to up-regulation of m-TORC1 (14,21), this was not the case in the cardiac-specific conditional KO mice (Online Figs. 5A to 5C). These observations suggest that the inhibition of GSK-3α could be a novel strategy to limit adverse ventricular remodeling and dysfunction post-MI.
Deletion of Gsk3α attenuates post-MI myocyte death and scar expansion
The most striking finding in our studies was a marked difference in scar expansion and content of the scar. In the WT mice, the scar was markedly thinned and comprised a much larger percent circumference of the left ventricle than in the KO mice (Figs. 3A to 3C and 3F). Furthermore, there were very few viable myocytes within the scar area in the WT mice. In contrast, the scar in the Gsk3α KO mice’s hearts was much thicker and, importantly, contained significantly greater numbers of viable cardiomyocytes (Figs. 3D and 3E). To determine the mechanism of the increased thickness and reduced scar circumference in the KO mice’s hearts, we performed terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) staining using 3-week post-MI heart sections. TUNEL-positive cardiomyocytes were significantly increased post-MI irrespective of genotype. However, the number of apoptotic cells was significantly less in the hearts of KO mice (Figs. 4A and 4B). The KO mice also showed increased expression of the antiapoptotic factor Bcl-2 and decreased expression of proapoptotic Bcl-2-associated X (Bax) protein expression post-MI (Figs. 4C and 4D). Thus, the ratio of Bax to Bcl-2, an important indicator of caspase-3 activation and apoptosis, was significantly lower in the KO mice’s hearts (Fig. 4D). These findings suggest that GSK-3α promotes MI-induced cardiomyocyte death, which leads to scar expansion and thinning and adverse remodeling.
Deletion of Gsk3α promotes post-MI cardiomyocyte proliferation
The increased number of viable cardiomyocytes in the hearts of KO mice also raised the possibility that deletion of Gsk3α might drive cardiomyocyte proliferation. The role of GSK-3α in cardiomyocyte proliferation post-MI is unknown. To determine if GSK-3α regulates this process, WT and KO animals were subjected to MI surgery, and then 5-bromo-2′-deoxyuridine (BrdU) was injected starting from the third week and continuing for 5 days. Immunostaining for BrdU and the cardiomyocyte-specific marker troponin I was performed using 3-week post-MI LV sections. Only cells that were both BrdU and troponin I positive, with BrdU colocalizing with 4,6-diamidino-2-phenylindole, were counted as proliferating cardiomyocytes. Indeed, the number of proliferating cardiomyocytes was significantly increased in the KO mice’s hearts (Figs. 5A and 5B). Immunostaining for another deoxyribonucleic acid synthesis marker, Ki67, also showed increased numbers of Ki67-positive cardiomyocytes in the KO hearts (Figs. 5C and 5D). To further investigate the role of GSK-3α on MI-induced cardiomyocyte proliferation, the mitosis marker phospho-histone H3 (Ser10) was examined by immunostaining. Consistent with the aforementioned findings, an increased number of cardiomyocytes positive for phospho-histone H3 were observed in the hearts of KO mice (Figs. 5E and 5F). These results confirm robust post-MI cardiomyocyte proliferation in the KO mice’s hearts and suggest that GSK-3α is a potent regulator of cardiomyocyte proliferation in the adult ischemic heart.
Gsk3α deletion promotes post-MI cyclin E1 and E2F-1 recruitment
To determine how GSK-3α might regulate cardiomyocyte proliferation, we examined pro-proliferative factors that might be responsible, including the Wnt/β-catenin pathway, cyclin-dependent kinases, cyclins, and the E2F-1 transcription factor (Online Figs. 6A to 6C). From this list, we identified 2 candidates, cyclin E1 and E2F-1, in post-MI KO mice’s hearts. The protein level of the transcription factor E2F-1 was significantly increased in the hearts of KO mice post-MI (Figs. 6A and 6B). This was not regulated through retinoblastoma protein (Rb), because phosphorylation (Ser780) of Rb was comparable between WT and KO mice (Figs. 6A and 6B). To further dissect molecular mechanisms responsible for GSK-3α-mediated E2F-1 signaling, we examined the possibility of direct interaction between these proteins by immunoprecipitation assay using neonatal rat ventricular myocyte lysates. We observed a strong interaction between GSK-3α and E2F-1 (Fig. 6C), suggesting that GSK-3α directly regulates E2F-1 levels in cardiomyocytes independent of Rb phosphorylation.
Cyclin E1 is an important G1-to-S transition-phase marker and plays an important role in cell proliferation. A significant up-regulation of cyclin E1 was observed in the KO mice’s hearts post-MI (Figs. 7A and 7B). To examine cyclin E1 specifically in cardiomyocytes, immunostaining was performed on 3-week post-MI heart sections. Only double-positive cells with cyclin E1 and α-actinin (a cardiomyocyte marker), with cyclin E1 colocalizing with 4,6-diamidino-2-phenylindole, were counted as proliferating cardiomyocytes. The number of cyclin E1–positive cardiomyocytes was significantly increased in the KO mice (Figs. 7C and 7D). To study the mechanism responsible for GSK-3α-mediated cyclin E1 regulation, neonatal rat ventricular myocytes were treated with the GSK-3 inhibitor SB216763 for 1.5 h, and cyclin E1 phosphorylation was determined. Phosphorylation of cyclin E1 (Thr395) was significantly decreased by GSK-3 inhibition (Fig. 7E). To determine if GSK-3α and cyclin E1 directly interact, we performed co-immunoprecipitation assays using cyclin E1 antibody, which suggested a direct interaction between GSK-3α and cyclin E1. To examine if this interaction is dependent on GSK-3α kinase activity, co-immunoprecipitation assay was performed in the presence of GSK-3 inhibitor (SB216763). Indeed, the interaction between cyclin E1 and GSK-3α was reduced in the presence of GSK-3 inhibitor, as reflected by a lower level of pull-down of GSK-3α (Fig. 7F). These data suggest that GSK-3α directly interacts with and phosphorylates cyclin E1 in cardiomyocytes. To our knowledge, this is the first report of direct regulation of cyclin E1 by GSK-3α. Taken together, these findings suggest that deletion of GSK-3α promotes post-MI E2F-1 and cyclin E1 recruitment and induces the reentry of adult cardiomyocytes into the cell cycle.
After an MI, the heart undergoes the process of remodeling, which consists of LV dilation and contractile dysfunction, often leading to heart failure. Strategies to prevent remodeling largely center around neurohormonal blockade (e.g., β-blockers and angiotensin-converting enzyme inhibitors) and, although beneficial to some degree, often cannot prevent heart failure in many patients, especially those with larger infarcts. Herein, we identify the α isoform of GSK-3 as a novel and potent promoter of remodeling post-MI, in part driven by profound scar expansion and leading to marked LV dysfunction. In stark contrast, the deletion of GSK-3α leads to much less dilative remodeling and better preserved LV function. The diminished remodeling post-MI in the KO mice is accompanied by a thicker scar and markedly increased numbers of viable cardiomyocytes within the scar area of KO mice’s hearts compared with those of WT mice. The profound reduction in ventricular dilation and scar expansion in the KO mice was supported by both less apoptosis and new myocyte formation post-MI (Central Illustration).
The traditional strategy (e.g., total somatic [global] gene deletion) can lead to secondary and compensatory effects that greatly complicate the interpretation of phenotypes (19). Global gene deletion can result in embryonic lethality if the gene is required for development. Moreover, global deletion induces compensatory changes in other genes as a means of preventing lethality. One of the best examples is from the GSK-3 family (e.g., global deletion of Gsk3β is lethal because of hyperproliferation of cardiomyoblast and congenital heart defects) (9). However, cardiac-specific conditional deletion has no overt effect and is protective post-MI in adults (11). Conversely, germline deletion of Gsk3α does not affect the embryonic development and pathophysiology of adult mice up to 4 months post-birth but leads to functional deterioration with aging (15,21). Given these complications of global gene targeting, tissue-specific approaches have become more widely adopted to bypass the confounding phenotypes observe in global gene-targeting strategies.
Both global deletion and constitutive activation of GSK-3α, under a variety of stresses, have been shown to induce detrimental phenotypes of cardiac hypertrophy and remote fibrosis (15,16). Moreover, germline global deletion of Gsk3α has been shown to induce progressive cardiac hypertrophy and contractile dysfunction, even in the absence of a cardiac-specific stress (14,21). Importantly, our findings, which show no hypertrophy or fibrosis post-MI, provide a stark contrast to previous reports and suggest that GSK-3α likely plays little or no direct role in stress-induced cardiac hypertrophy and remote fibrosis. In summary, the previously reported role of GSK-3α in various stress settings is likely due to secondary effects of global gene deletion and/or transgenesis.
Cardiomyocyte death, even at very low rates, can lead to significant collective loss of cardiomyocytes over time and is reported to cause post-MI cardiac dysfunction and heart failure (22). Our findings show that GSK-3α is a potent promoter of cardiomyocyte death and apoptosis in the ischemic heart by inducing Bcl-2 and inhibiting Bax activity. Bcl-2 is a known antiapoptotic factor (23) that protects against oxidative stress-induced apoptosis and preserves cardiomyocyte viability and LV function in the ischemic heart when overexpressed (24). Our findings suggest that cardiomyocyte-specific deletion of Gsk3α leads to increased expression of Bcl-2, and this limits cardiomyocyte death in chronic MI.
We also report robust cardiomyocyte proliferation in the Gsk3α-KO mice’s hearts, as evidenced by increased Ki67- and BrdU-positive cardiomyocytes. Mechanistically, we found that deletion of Gsk3α increases E2F-1 recruitment in the KO mice’s hearts post-MI. Although a previous study showed that GSK-3α regulates E2F through cyclin D1 post transverse aortic constriction (16), we observed that cyclin D1 played no role in the cardiomyocyte-specific KO mice post-MI (Online Figs. 6A to 6C). These observations suggest that GSK-3α might regulate E2F-1 in a stress-dependent manner, as cyclin D1 was involved in transverse aortic constriction but not in MI. Several studies have shown that activation of Rb regulates E2F-1 (25,26), but this was not a mechanism in the present study, as phosphorylation of Rb was comparable between WT and KO mice. Furthermore, we found that GSK-3α directly interacts with E2F-1 in cardiomyocytes. These findings suggest that GSK-3α is a direct regulator of E2F-1, independent of cyclin D1 and Rb. Similarly, an increased level of the G1/S transition-phase marker cyclin E1 was observed in the hearts of KO mice. Our data suggest that GSK-3α phosphorylates and regulates cyclin E1 levels and emphasize the fact that GSK-3α is a key and novel regulator of cyclin E1 in the cardiomyocyte. The elevated levels of E2F-1 and cyclin E1 in the KO mice appear to be the central mechanism for cardiomyocyte proliferation.
Herein we propose that selective inhibition of GSK-3α could significantly improve outcomes in patients with ischemic injury. Selective inhibition of GSK-3α has not, to our knowledge, been achieved with available drugs. In fact, inhibition of GSK-3β can have adverse effects on the heart. The ideal scenario would be to have a compound that selectively inhibits GSK-3α (and not the other isoform, GSK-3β). Eventually this hurdle will be conquered and small molecules will be developed that selectively target GSK-3α. These agents could be very beneficial in patients with ischemic disease.
We present findings that reveal the potential translational value of GSK-3α inhibition in the setting of chronic MI. Cardiomyocyte-specific conditional deletion of Gsk3α preserves cardiac function, limits ventricular remodeling, attenuates cardiomyocyte death, restricts scar expansion and thinning, and promotes cardiomyocyte proliferation post-MI. Taken together, these observations suggest that selective inhibition of GSK-3α could be a therapeutic strategy to limit MI-induced cardiac remodeling and heart failure.
COMPETENCY IN MEDICAL KNOWLEDGE: In mouse models of left ventricular function after MI, genetic expression of GSK-3α is associated with more severe ventricular dilatation, adverse remodeling, and heart failure.
TRANSLATIONAL OUTLOOK: All currently available GSK-3 inhibitors target both the α and β isoforms, but the development of new drugs that specifically inhibit GSK-3α could be promising for study in large animal models of MI.
This work was supported by grants HL061688 and HL091799 to Dr. Force from the National Heart, Lung, and Blood Institutehttp://dx.doi.org/10.13039/100000050 and grant MOP74711 to Dr. Woodgett from the Canadian Institutes of Health Researchhttp://dx.doi.org/10.13039/501100000024. The authors have reported that they have no relationships relevant to the contents of this paper to disclose. Drs. Ahmad and Lal contributed equally to this work.
- Abbreviations and Acronyms
- glycogen synthase kinase
- homozygous Gsk3α floxed mouse
- homozygous Gsk3α floxed mouse with heterozygous Mer-Cre-Mer
- left ventricular
- myocardial infarction
- retinoblastoma protein
- terminal deoxynucleotidyl transferase dUTP nick end labeling
- Received March 10, 2014.
- Revision received April 25, 2014.
- Accepted April 30, 2014.
- American College of Cardiology Foundation
- Velagaleti R.S.,
- Pencina M.J.,
- Murabito J.M.,
- et al.
- Kenneth D.J.X.,
- Sherry L.M.,
- Arialdi M.M.,
- Hsiang-Ching K.
- Porrello E.R.,
- Mahmoud A.I.,
- Simpson E.,
- et al.
- Pasumarthi K.B.,
- Field L.J.
- Haq S.,
- Michael A.,
- Andreucci M.,
- et al.
- Woulfe K.C.,
- Gao E.,
- Lal H.,
- et al.
- Lal H.,
- Zhou J.,
- Ahmad F.,
- et al.
- Matsuda T.,
- Zhai P.,
- Maejima Y.,
- et al.
- Zhai P.,
- Gao S.,
- Holle E.,
- et al.
- Maejima Y.,
- Galeotti J.,
- Molkentin J.D.,
- et al.
- Davis J.,
- Maillet M.,
- Miano J.M.,
- Molkentin J.D.
- Gao E.,
- Lei Y.H.,
- Shang X.,
- et al.
- Kung G.,
- Konstantinidis K.,
- Kitsis R.N.
- Harbour J.W.,
- Dean D.C.
- Nevins J.R.