|Home | About | Journals | Submit | Contact Us | Français|
In the mammalian heart, cardiomyocytes withdraw from the cell cycle and initiate hypertrophic growth soon after birth, but the transcriptional regulatory mechanisms that control these neonatal transitions are not well-defined.
Forkhead family transcription factors have been implicated as positive (FoxM1) and negative (FoxO1 and FoxO3) regulators of cardiomyocyte proliferation prenatally, but their regulatory interactions and functions in neonatal cell cycle withdrawal have not been reported previously. Potential regulators of Fox activity, including the metabolic indicator AMP-activated protein kinase (AMPK), and Fox transcriptional targets (p21, p27, Insulin-like growth factor 1 (IGF1)) also were examined.
In cultured neonatal rat cardiomyocytes, AMPK activates FoxOs, and AMPK inhibition is sufficient to induce cell proliferation. In vivo, combined loss of FoxO1 and FoxO3 specifically in cardiomyocytes leads to delayed cell cycle withdrawal and increased expression of IGF1 and FoxM1. Conversely, cardiomyocyte-specific loss of FoxM1 results in decreased neonatal cardiomyocyte cell proliferation, decreased expression of IGF1, and increased expression of cell cycle inhibitors p21 and p27. IGF1 is a direct downstream target of cardiac Fox transcription factors, which is negatively regulated by FoxOs and positively regulated by FoxM1, dependent on AMPK activation status.
These data support a regulatory mechanism whereby the balance of FoxO and FoxM1 transcription factors integrates metabolic status, mediated by AMPK, and cell cycle regulation, through competitive regulation of target genes including IGF1, in neonatal cardiomyocytes.
After birth, cardiomyocytes withdraw from the cell cycle and the heart grows primarily by hypertrophy.1 In the neonatal heart, growth factor signaling is reduced, metabolism shifts from glycolytic to primarily oxidative, and the myocardium has the ability to regenerate.2 Approximately a week after birth, binucleated cardiomyocytes withdraw from the cell cycle, the myocardium loses its ability to regenerate, and the heart initiates hypertrophic growth.1, 2 Several transcription factors including GATA4, Nkx2.5, Tbx20, Tbx5, HAND1, Foxp1, and FoxM1 have been implicated in regulating cardiomyocyte cell proliferation in the developing heart.3–5 However, the transcriptional regulatory mechanisms that integrate growth factor signaling, metabolic status, and cell cycle regulation in the heart after birth are largely unknown.
Forkhead box (Fox) family transcription factors have been implicated in cardiomyocyte cell cycle control as well as adult cardiovascular disease. FoxM1 promotes cell cycle progression in a variety of cellular contexts and is required for cardiomyocyte proliferation in the developing heart.5, 6 FoxO factors have multifaceted functions in cell cycle regulation, metabolism, cell survival, autophagy, and aging.7 FoxOs are activated under conditions of growth factor deprivation, and FoxO1 and FoxO3 are expressed in the developing and diseased heart.8, 9 In the prenatal heart, forced expression of FoxOs inhibits cell proliferation and induces expression of cell cycle inhibitors p21 (p21Cip1) and p27 (p27Kip1).8 In the adult heart, FoxOs promote cardiomyocyte survival under conditions of oxidative stress and also induce autophagy after starvation.10, 11 In addition, FoxOs inhibit cardiac hypertrophy and are induced with diabetic cardiomyopathy in adults.9, 12 In cancer cells, FoxM1 and FoxOs have antagonistic transcriptional roles in cell cycle regulation.13, 14 However, the functions and potential interactions of these factors in neonatal cardiomyocyte cell cycle withdrawal have not been reported previously.
Immediately after birth and prior to feeding, neonatal mammals are subjected to a period of starvation.15 Growth factor deprivation via phosphoinositide 3-kinase (PI3K)/AKT inactivation leads to increased FoxO nuclear localization and transcriptional activity in a variety of cell types, including fetal cardiomyocytes.7, 8 Activation of AMP-activated protein kinase (AMPK) occurs as a result of metabolic insufficiency and promotes FoxO3 activity in longevity and tumor suppression16, but this interaction has not been demonstrated in the heart. Insulin-like Growth Factor 1 (IGF1) is among the growth factors that inactivate FoxOs and has been implicated in promoting prenatal cardiomyocyte proliferation.8 IGF1 expression also can promote cardiac hypertrophy, cell cycle reentry, and repair in adults17, 18, but its function and regulation in neonatal cell cycle withdrawal has not been fully characterized.
In the current study, we examine AMPK function and the requirements for FoxOs and FoxM1 in neonatal cell cycle withdrawal. In addition we identify IGF1 as a common target of FoxOs and FoxM1, which is activated by FoxM1 in proliferating cardiomyocytes and repressed by FoxOs during neonatal cell cycle withdrawal, in response to AMPK activation status. These results provide evidence that FoxM1 inactivation and FoxO activation, subject to metabolic regulation, together regulate neonatal cardiomyocyte cell cycle withdrawal.
Primary neonatal (1–2 day) rat cardiomyocytes were isolated, infected with FoxO adenoviruses and analyzed as described previously.11 Cardiomyocyte-specific conditional loss of FoxOs and FoxM1 was achieved with β-myosin heavy chain (βMHC)-Cre using published mouse lines.5, 11 Proliferative indices were calculated as described previously.8, 11 Quantitative RT-PCR (qRT-PCR), Chromatin immunoprecipitation (ChIP), and IGF-1 reporter assays were performed as previously described.8, 11, 19 All experimental procedures with animals were approved by the Institutional Animal Care and Use Committee of the Cincinnati Children's Hospital Medical Center. An expanded Methods section is available online at http://circres.ahajournals.org.
Expression levels of the proliferative factor IGF1 and the activation status of the downstream kinase AKT were determined by Western blot analysis of wild type mouse heart lysates at embryonic day 14.5 (E14.5), E17.5, postnatal day 1 (pd1), pd7 and 1 month. In addition, the activation status of AMPK, an indicator of metabolic insufficiency, was determined relative to the activation status of FoxOs and expression of FoxM1. After birth, IGF1 protein expression is decreased by 50% in pd7 and 1 month old hearts as compared to E14.5. Similarly, the activity of AKT is also decreased by 40% at pd7 and 1 month old hearts compared to E14.5, as indicated by decreased p-AKT/total AKT (Figure 1A–C). Conversely, AMPK activation is increased postnatally (by 1.9-fold in pd7 and 2.25-fold at 1 month, compared to E14.5), as indicated by increased p-AMPK/total AMPK protein levels (Figure 1A, D). The activity of both FoxO1 and FoxO3 is also increased postnatally in mouse hearts (Figure 1A asterisks) as indicated by decreased levels of inactive phosphorylated FoxO1 (p-FoxO1; Ser-256)/total FoxO1 (30%-reduction in pd1 and 60% at 1 month, compared to E14.5, Figure 1E) and inactive p-FoxO3(Ser-318/321)/total FoxO3 (40% reduction in pd1 to 80% at 1 month, compared to E14.5, Figure 1F). In contrast, FoxM1 protein expression is decreased by 80% in postnatal mouse hearts compared to E14.5 (Figure 1A, asterisks and and1G).1G). Thus, the activity of both AMPK and FoxOs increases, whereas the activity of AKT and expression of IGF1 and FoxM1 protein decline, during the first week after birth in mouse hearts in vivo.
Neonatal cardiomyocytes normally exit the cell cycle, and post-natal proliferative rates are extremely low.1, 20 In order to determine the effects of altered activity of AMPK on cardiomyocyte cell cycle withdrawal, rat neonatal cardiomyocytes were treated with either AICAR (AMPK activator) or Compound C (AMPK inhibitor). AMPK inhibition with Compound C increases the cell cycle activity by 2.6-fold compared to vehicle (DMSO) treated cells, as determined by immunofluorescence and cell counts (Figure 2C,C'; Ki67+/α-actinin+ cardiomyocytes, indicated by white arrows).21 Activation of AMPK by AICAR treatment does not inhibit the already low rates of proliferation of neonatal cardiomyocytes compared to vehicle treated cells. The expression of cell cycle regulatory genes also was examined in rat neonatal cardiomyocytes treated with either AICAR or Compound C. Gene expression of cell cycle activators, IGF1 and FoxM1, and also of cell cycle inhibitors, p21 and p27 which are known targets of FoxOs, was determined by qRT-PCR. The activation of AMPK by AICAR results in decreased expression of FoxM1 and IGF1 with increased expression of p21 (Figure 2E). In contrast, inhibition of AMPK by Compound C results in significantly increased expression of FoxM1 and IGF1, with decreased expression of p21 and p27, in accordance with increased cell cycle activation. Thus inhibition of AMPK promotes cell cycle activity, with corresponding effects on cell cycle regulatory gene expression, in cultured rat neonatal cardiomyocytes.
The ability of AMPK to alter subcellular localization of endogenous FoxOs was determined in rat neonatal cardiomyocytes treated with either AICAR (Figure 3B,B' and E, E') or Compound C (Figure 3C,C' and F,F'). Nuclear versus cytoplasmic localization of FoxO1 and FoxO3 in treated and control cardiomyocytes was determined by immunofluorescence (Figure 3A,A' and D,D'). Activation of AMPK promotes nuclear localization of endogenous FoxO1 (Figure 3B,B', green as indicated by white arrow) and FoxO3 (Figure 3E,E', green as indicated by white arrow) in rat neonatal cardiomyocytes (α-actinin, red). Conversely, inhibition of AMPK leads to exclusion of FoxO1 (Figure 3C,C') and FoxO3 (Figure 3F,F') from the nucleus in rat neonatal cardiomyocytes. Quantitation of these results shows a significant increase in nuclear FoxO1 (75% versus 63%) and FoxO3 (45% versus 25%) in AICAR treated cardiomyocytes. In contrast, nuclear localization of FoxO1 (43% versus 63%) and FoxO3 (10% versus 25%) in cardiomyocytes is significantly decreased with AMPK inhibition, as compared to controls (Figure 3G). Increased nuclear localization of endogenous FoxO1 and FoxO3 by activation of AMPK is consistent with increased transcriptional function and suggests a critical role for AMPK and FoxOs in regulating neonatal cell cycle withdrawal.
In order to determine a direct link between AMPK and FoxO activation in rat neonatal cardiomyocytes, we examined the activation status of AKT, FoxOs, and the expression levels of FoxM1 in rat neonatal cardiomyocyte cultures treated with either AMPK activator AICAR or AMPK inhibitor Comp C. Activation of AMPK by AICAR, decreases AKT activity, as indicated by decreased p-AKT/total AKT, with concomitant increased FoxO1 and FoxO3 activity, as indicated by decreased AKT-specific phosphorylation of FoxO1(Ser-256) and FoxO3(Ser-318/321) (Figure 3H,I,J and K). Conversely, inactivation of AMPK by Comp C, increases AKT activity, as indicated by increased p-AKT/total AKT, with concomitant decreased FoxO1 and FoxO3 activity, as indicated by increased AKT-specific phosphorylation of FoxO1 and FoxO3 (Figure 3H,I,J and K). In addition, FoxM1 protein expression increases with Compound C treatment consistent with the decreased activation of FoxO1 and FoxO3 (Figure 3H,L and Figure 2C–D). Thus AMPK activation inhibits AKT activity and subsequent phosphorylation, nuclear localization, and downstream target gene expression of FoxOs in rat neonatal cardiomyocytes.
The function of FoxO transcription factors in regulation of neonatal cardiomyocyte cell cycle withdrawal was examined in cardiomyocytes infected with recombinant adenovirus expressing FoxO1 or FoxO3.11, 22, 23 FoxO gain of function was achieved using adenovirus expressing either constitutively active-FoxO1 (ADA) or constitutively active-FoxO3 (TmO3). FoxO loss of function was achieved by infection with adenovirus expressing Δ256-FoxO (dominant negative) that inhibits transcriptional activity of FoxOs. Cardiomyocytes infected in parallel with β-galactosidase (β-gal) adenovirus serve as a control. Inhibition of FoxO activity (Figure 4A, Δ256) results in increased cardiomyocyte cell cycle activation (Ki67+/α-actinin+ indicated by white arrows, 35% versus 11%, Figure 4Ad,d'). However, no changes were observed in cells infected with adenovirus expressing either ADA or TmO3 as compared to Adβ-gal infected controls (Figure 4A,B). Thus, inhibition of FoxO transcriptional activity promotes cardiomyocyte cell cycle activation.
The ability of FoxO activation to block induction of neonatal cardiomyocyte cell cycle activation by AMPK inhibition was examined. Rat neonatal cardiomyocytes were infected with either ADA or TmO3 adenovirus, compared to Adβ-gal infected controls, for 24 hours and then treated with Compound C. The induction of cardiomyocyte cell cycle activation due to inhibition of AMPK activity (Figure 4Ca,a'), as detected by Ki-67 antibody reactivity, is attenuated in presence of constitutively active FoxO1 (Figure 4Cb,b', 20% versus 37%) or FoxO3 (Figure 4Cc,c', 12% versus 37%). Quantitative representation of Figure 4C is shown in Figure 4D. Thus FoxO activation is sufficient to inhibit cardiomyocyte cell cycle activity under conditions of AMPK inhibition.
The necessity of FoxO function in AMPK-mediated cell cycle withdrawal in neonatal cardiomyocytes was examined. Rat neonatal cardiomyocytes were infected with dominant negative Δ256-FoxO adenovirus or control Adβ-gal for 24 hours and then treated with the AMPK activator, AICAR. Inhibition of FoxO activity in Δ256-FoxO adenovirus-infected cardiomyocytes results in increased cardiomyocyte cell cycle activation (Ki67+/α-actinin+ indicated by white arrows, Figure 4Ad,d' and 4Ec,c') as compared to control cells (Figure 4Aa,a' and 4Ea,a'). Normally, AMPK activation leads to decreased proliferation, but cardiomyocyte cell cycle activation is increased with dominant negative Δ256-FoxO in the presence of AICAR (Figure 4Eb,b' (14%) versus 4Ed,d' (32%) and Figure 4F). Thus AMPK-mediated cell cycle withdrawal is dependent on FoxO activity in neonatal cardiomyocytes.
The requirements for FoxO function in neonatal cardiomyocyte cell cycle withdrawal in vivo was determined in mice with combined deficiency of FoxO1 and FoxO3 specifically in cardiomyocytes. Because both FoxO1 and FoxO3 are expressed in the heart and are functionally redundant in many cell types, both were deleted together in order to determine the effect of loss of FoxOs on cardiomyocyte cell cycle withdrawal. βMHC-Cre was used to specifically delete FoxO1 and FoxO3 from cardiomyocytes beginning at late fetal stages, and the mice are viable.11 Here we report the effects of cardiomyocyte-specific loss of FoxOs on neonatal cardiomyocyte cell cycle withdrawal. To determine cardiomyocyte mitotic activity in vivo, immunostaining was performed to colocalize Phospho-histone H3 (PHH3+) (proliferation marker) and MF20+ (cardiomyocyte-specific marker) cells in pd1, 3, 7 and 1 month old βMHC-Cre;FoxO1fl/fl/FoxO3fl/fl mouse hearts compared with littermate FoxO1fl/fl/FoxO3fl/fl controls.
Cardiomyocyte proliferation is increased in the βMHC-Cre;FoxO1fl/fl/FoxO3fl/fl mouse hearts compared to FoxO1fl/fl/FoxO3fl/fl control hearts at pd1 (4.3% versus 2.8%) and pd3 (1.8% versus 0.7%), as determined by PHH3 immunofluorescence (Figure 5A–C). However, by pd7 and in adults there were no significant differences in cardiomyocyte mitotic activity between the two genotypes. Thus, cardiomyocyte-specific loss of FoxO1 and FoxO3 results in increased cell proliferation in the first week after birth, but the cardiomyocytes exhibit delayed cell-cycle withdrawal apparent at pd7.
During development, FoxM1 is required for cardiomyocyte proliferation.5, 24 FoxM1 expression is down regulated at birth in the heart, with little or no expression in the adult.25,5 In order to determine FoxM1 function in the regulation of the timing of neonatal cardiomyocyte cell cycle withdrawal in vivo, FoxM1fl/fl mice were bred with βMHC-Cre mice to delete FoxM1 specifically from cardiomyocytes beginning at late fetal stages in resulting βMHC-Cre;FoxM1fl/fl animals. In contrast to βMHC-Cre;FoxO1fl/fl/FoxO3fl/fl hearts, βMHC-Cre;FoxM1fl/fl neonatal hearts have decreased cardiomyocyte proliferation compared to FOXM1fl/fl control hearts at pd1 (2.4% vs 0.7%) and pd3 (1.5% vs 0.8%), as determined by immunofluorescence (Figure 5D–F). Similar to βMHC-Cre;FoxO1fl/fl/FoxO3fl/fl hearts, no significant differences in cardiomyocyte proliferation were observed between βMHC-Cre;FoxM1fl/fl and FoxM1fl/fl animals at pd7 and 1 month (Figure 5F). During the neonatal period, cardiomyocyte cell size and indicators of cardiac hypertrophy (βMHC, ANF, BNP) are not altered in cardiomyocyte-specific FoxO and FoxM1 deficient, mice as compared to littermate controls (Online Figure I). In addition, cell proliferation of noncardiomyocyte lineages was apparently unaffected with cardiomyocyte-specific loss of FoxOs or FoxM1 (data not shown); thus further studies are necessary to determine if FoxOs and FoxM1 regulate cell proliferation in coronary fibroblasts and smooth muscle cells. These results demonstrate that FoxOs and FoxM1 have opposing effects in neonatal cardiomyocytes, supporting a mechanism whereby FoxOs promote and FoxM1 inhibits neonatal cell cycle withdrawal.
During the neonatal period in mice, genes that promote cell proliferation are down-regulated and cell-cycle inhibitory genes are induced. IGF1 and FoxM1 promote prenatal cardiomyocyte proliferation, and p21 and p27 are direct downstream targets of FoxOs that inhibit cardiomyocyte proliferation.8, 26 Therefore, we examined the expression levels of these cell-cycle regulatory genes in βMHC-Cre;FoxO1fl/fl/FoxO3fl/fl neonatal hearts with delayed cell cycle withdrawal, compared to βMHC-Cre;FoxM1fl/fl neonatal hearts with accelerated cell cycle withdrawal (Figure 6). Expression of IGF1 (~2 fold, Figure 6A) and FoxM1 (~ 4 fold, Figure 6B) is increased in βMHC-Cre;FoxO1fl/fl/FoxO3fl/fl neonatal hearts. In contrast, cardiomyocyte-specific loss of FoxM1 results in decreased IGF1 gene expression (~80% reduced, Figure 6A) with increased gene expression of both FoxO1 and FoxO3 (~3–4 fold, Figure 6E; ~10 fold, Figure 6F). As expected, expression of both FoxO1 (Figure 6E) and FoxO3 (Figure 6F) is decreased in mice with conditional targeting of FoxOs in cardiomyocytes with βMHC-Cre, and there is negligible expression of FoxM1 in mice with conditional targeting of FoxM1 in cardiomyocytes with βMHC-Cre (Figure 6B). p27 is a direct transcriptional target of FoxOs, and p27 expression is decreased by 60–70% with cardiomyocyte-specific loss of FoxO1 and FoxO3 in neonatal (pd1 and pd3) hearts as determined by qRT-PCR (Figure 6D). However, gene expression of p21 remains unchanged, probably due to the very low level of expression in the control hearts. In contrast, expression of p21 (Figure 6C) and p27 (Figure 6D) is increased 2–4.5-fold in βMHC-Cre;FoxM1fl/fl hearts at pd1–3, consistent with observed premature cell cycle withdrawal in these animals. These results are in accordance with increased mitotic activity observed with cardiomyocyte-specific loss of FoxO1 and FoxO3, in contrast to decreased neonatal cardiomyocyte mitotic activity with loss of FoxM1. In addition, these data provide initial evidence for competitive regulation of shared target genes in the determination of the timing of cardiomyocyte cell cycle withdrawal.
Cardiomyocyte-specific loss of FoxOs and FoxM1 has opposing effects on IGF1 gene expression in neonatal cardiomyocytes consistent with altered timing of cell cycle withdrawal (Figure 6A). Therefore, IGF1 genomic sequences were examined for conserved FOX binding consensus sequences and for transcriptional regulation by FoxO1 and FoxM1. The mouse IGF1 gene sequence (NC_000076.6) contains a conserved Forkhead (FOX) DNA binding sequence TAAACA located at −550 relative to the transcriptional start site (Figure 7A–B).19, 27, 28 Cotransfection experiments performed in C2C12 cells demonstrate that murine IGF1 (−1600) sequences linked to a pGL3-basic reporter are trans-activated by FoxM1 in a concentration-dependent manner (1.5–4.5 fold, Figure 7C). Mutagenesis of the conserved FOX binding site prevents trans-activation of IGF1 regulatory sequences by FoxM1 (Figure 7C). Conversely, cotransfection assays performed in HEK293 cells demonstrate that FoxO1 suppresses IGF1-luciferase reporter activity (Figure 7D). Together, these results show that FoxM1 and FoxO1 have opposite effects on transactivation of the IGF1 promoter in transfected cells.
In order to determine if FoxO1 and FoxM1 bind to the identified IGF1 promoter sequences, ChIP assays were performed in rat neonatal cardiomyocytes under different proliferative conditions (Figure 7E). Binding of FoxO1 and FoxM1 to IGF1 promoter sequences was examined in neonatal rat cardiomyocytes with altered AMPK activity resulting from treatment with either AICAR or Compound C. Immunoprecipitation with FoxO1 or FoxM1 antibodies was used to determine the fold enrichment of binding to IGF1 promoter sequences relative to the IgG control. In cardiomyocytes proliferating due to AMPK inhibition, FoxM1 binding to the IGF1 promoter region is increased by ~9 fold. In contrast, AMPK activation results in a ~6 fold enrichment of FoxO1 bound to the IGF1 promoter. Thus, FoxM1 and FoxO1 bind directly to the IGF1 promoter, and the specific FOX transcription factor bound to these sequences is subject to cardiomyocyte metabolic status. Together these data are consistent with a mechanism whereby differential binding of the IGF1 promoter by FoxM1 during development and by FoxO1 during the neonatal period regulates IGF1 gene expression and contributes to neonatal cell cycle withdrawal.
Here, we demonstrate that FoxOs and FoxM1 regulate cardiomyocyte proliferation and cell cycle withdrawal in the neonatal period. In the days after birth, FoxM1 is downregulated and FoxOs are activated concomitant with increased AMPK activation, decreased AKT activity, decreased IGF1 expression, and induction of cell cycle inhibitors. Studies in cardiomyocyte-specific FoxM1- and FoxO-deficient mice, together with experiments in cultured neonatal cardiomyocytes, support a model in which the balance of growth factor signaling and activity of FoxM1 versus FoxOs regulates cardiomyocyte proliferation in the fetal and neonatal periods (Figure 8). In fetal cardiomyocytes, growth factor signaling is high, FoxM1 is expressed, IGF1 is expressed, AKT is active, and cardiomyocytes are highly proliferative. During the neonatal period, growth factor signaling is reduced, AMPK is activated, AKT is inactivated, FoxOs are activated, FoxM1 is downregulated, IGF1 is downregulated, cell cycle inhibitors are upregulated, and cardiomyocytes withdraw from the cell cycle. Increased FoxO activity inhibits FoxM1 and IGF1 gene expression, while activating cell cycle inhibitor genes p21 and p27. ChIP and transfection studies demonstrate that FoxM1 binds and is a transcriptional activator of IGF1 regulatory sequences, whereas FoxO1 binds and represses the same IGF1 regulatory sequences, subject to regulation by AMPK. Together, these data support a regulatory mechanism whereby FoxO and FoxM1 transcription factors integrate metabolic status and cell cycle withdrawal in neonatal cardiomyocytes.
Multiple regulatory pathways contribute to neonatal cardiomyocyte cell cycle withdrawal and control cardiomyocyte cell cycle activity after birth.29, 30 Loss of the cell cycle inhibitors p27 or p130, or alternatively forced expression of the cell cycle activator cyclinA2, leads to delayed cardiomyocyte cell cycle withdrawal after birth, similar to that observed with cardiomyocyte-specific loss of FoxOs in the current study.20, 30 However, the mechanism by which cell cycle withdrawal is delayed, but not prevented, in these models is not known. Increased expression of cyclinD2 or Myc can promote neonatal cardiomyocyte cell cycle activity, but also leads to multinucleation of cardiomyocytes later in life.29 Additional pathways, including the Hippo/Yap pathway, have been implicated in cardiomyocyte cell cycle inhibition, and overexpression of activated Yap leads to prolonged cell cycle activity and cardiomyocyte hyperplasia after birth.31 FoxOs inhibit cell cycle progression through a variety of mechanisms, including activation of p21 and p27, as well as inhibition of cyclinD1/2, gene expression.32 While the regulatory hierarchies that control neonatal cell cycle withdrawal are not fully defined, FoxOs appear to be critical mediators of this process through inhibition of cell cycle activators in addition to activation of cell cycle inhibitors.
There is increasing evidence that the balance of FoxOs and FoxM1 regulates cell cycle activity in a variety of developmental and disease processes. In cancer, FoxM1 promotes cell proliferation, while FoxOs can act as tumor suppressors.6, 32 In developing cardiomyocytes, loss of FoxM1 in mice results in myocardial hypoplasia and fetal death, whereas FoxO inhibition leads to increased cardiomyocyte proliferation at E17.5.5, 8 FoxOs and FoxM1 have opposing effects on cell cycle inhibitors in that FoxM1 promotes the protein degradation of p21 and p27, whereas FoxOs are transcriptional activators of p21 and p27 gene expression.6, 32 In breast cancer cells, FoxO3 represses FoxM1 expression while VEGF and estrogen receptor α genes are transcriptionally inhibited by FoxO3 and activated by FoxM1.14, 33 Here we provide evidence for a similar antagonistic relationship in neonatal cardiomyocytes in which FoxOs repress FoxM1 gene expression in addition to competing for binding of IGF1 gene regulatory sequences.
IGF1 gene expression is dynamically regulated in heart development and disease. Before birth, IGF1 activates PI3K/AKT signaling leading to phosphorylation, nuclear exclusion, and inactivation of FoxOs in cardiomyocytes.8 FoxM1 activation of IGF1 gene expression in prenatal cardiomyocytes has not been examined directly, but is a possible mechanism by which FoxM1 promotes cell proliferation in the developing heart.5 During the neonatal period, IGF1 gene expression is decreased, potentially due to decreased expression of FoxM1, supporting a feedback mechanism of IGF1 downregulation and FoxO activation. Forced expression of IGF1 can prolong neonatal cell cycling30, but it is not known if this is mediated through FoxO inactivation. Additional factors also contribute to IGF1 gene regulation. An NFAT-responsive regulatory element was previously identified in IGF1 promoter sequences19, and Yap, a transcriptional regulator of the Hippo pathway, also has been implicated as a positive regulator of IGF1 signaling in cardiomyocytes.31 In the adult heart, there are conflicting reports as to whether IGF1 signaling is damaging or cardioprotective during injury or oxidative stress, and it is likely that secreted and locally activated isoforms have distinct functions.18 IGF1 also has been implicated in cardiac myocyte regeneration and repair by acting on cardiac stem cells. However, the potential interactions of FoxOs and IGF1 signaling in cardiac disease and repair have not yet been explored.
AMPK activation is an indicator of energy depletion that induces oxidative metabolism in addition to regulating mitochondrial biogenesis, autophagy, cell growth, and proliferation.34 In the heart, AMPK is activated during the neonatal period coincident with energy depletion and induction of oxidative metabolism. In addition, AMPK inhibition in neonatal cardiomyocytes is sufficient to promote cell cycle activation, consistent with a role for AMPK activation in neonatal cardiomyocyte cell cycle withdrawal. Here, we show that AMPK inhibition leads to increased activation of AKT with increased phosphorylation of FoxO1 at Ser-256 and of FoxO3 at Ser-318/321 AKT target sites, concomitant with nuclear exclusion and inactivation. In contrast, AMPK directly phosphorylates FoxO3 at alternative sites leading to increased transcriptional activity in response to nutrient deprivation16, and AMPK activation causes cell cycle arrest in a variety of cell types.35, 36 AMPK activation of FoxOs also promotes oxidative metabolism, but this has not yet been demonstrated directly in neonatal cardiomyocytes. In the adult heart, AMPK37 and FoxOs11 have been implicated in cardioprotection under conditions of oxidative stress or cardiac injury, but intersecting regulatory mechanisms have not yet been defined. Interestingly, increased activity of AMPK and FoxOs, along with decreased IGF1, are associated with prolonged lifespan due to caloric restriction and also with cardiovascular aging.18 Thus, there is accumulating evidence that the AMPK/FoxO/IGF1 regulatory network has multiple roles in cardiac development and disease from birth to old age.
The neonatal heart is undergoing multiple critical transitions, including cell cycle withdrawal, metabolic substrate shifts, and loss of regenerative potential, apparent one week after birth. Here, we show that the AMPK activation of FoxOs promotes cell cycle withdrawal and inhibits IGF1 gene expression in neonatal cardiomyocytes. It would be interesting to determine if the activation status of this pathway affects the ability of the neonatal heart to regenerate or the loss of regenerative potential approximately a week after birth. The balance of AMPK and FoxO activation versus IGF1 signaling has multiple functions during cardiomyocyte development and disease processes that must be taken into account when considering therapeutic approaches to heart failure and enhancement of repair. In addition, AMPK activation by metaformin has been used to treat diabetes and metabolic syndrome but also has likely implications for cardiac disease.34 Likewise, IGF1 treatment has been proposed as a mechanism promote cardiac repair but could also lead to increased oxidative injury due to inhibition of AMPK and FoxOs.18 Thus, therapeutic approaches that affect AMPK/FoxO/IGF1 signaling pathways could have varied and complex effects on cardiac disease and repair processes.
Here, we report that FoxO and FoxM1 transcription factors regulate cardiomyocyte proliferation and cell cycle withdrawal in the neonatal period. Immediately after birth and prior to feeding, the mammalian heart is subjected to a period of starvation in which growth factor signaling is reduced and cardiomyocytes are under metabolic stress. At the same time, expression of FoxM1, which promotes cardiomyocyte proliferation, is downregulated and FoxOs are activated, concomitant with induction of cell cycle inhibitors. Mice lacking FoxM1 in cardiomyocytes exhibit decreased cell proliferation after birth, whereas cardiomyocyte-specific loss of FoxO1 and FoxO3 delays neonatal cell cycle withdrawal. In cultured neonatal cardiomyocytes, FoxO activity is subject to AMP-kinase activation, an indicator of metabolic stress. Insulin-like growth factor-1 (IGF1) can promote cardiac hypertrophy, cell cycle reentry, and repair in adult hearts. Our results suggest that IGF1 is a direct downstream target of cardiac Fox transcription factors, which is negatively regulated by FoxOs and positively regulated by FoxM1, dependent on AMPK activation status. Together, these data support a regulatory mechanism in which FoxO and FoxM1 transcription factors integrate metabolic status and cell cycle withdrawal in neonatal cardiomyocytes.
We thank Michelle Sargent for cardiomyocyte isolation, Christina Alfieri and Jonathan Cheek for technical support, and Craig Bolte for assistance with the FoxM1 mutant mice.
SOURCES OF FUNDING This work was supported by NIH/NHLBI grant P01 HL069779 to K.E.Y., R01 HL84151 to V.V.K. and an AHA-Great Rivers Affiliate Post-doctoral Fellowship 11POST7210026 to A.S.
DISCLOSURES No relationships to disclose
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.