Chromatin regulation by Brg1 underlies heart muscle development and disease.

Cardiac hypertrophy and failure are characterized by transcriptional reprogramming of gene expression. Adult cardiomyocytes in mice primarily express alpha-myosin heavy chain (alpha-MHC, also known as Myh6), whereas embryonic cardiomyocytes express beta-MHC (also known as Myh7). Cardiac stress triggers adult hearts to undergo hypertrophy and a shift from alpha-MHC to fetal beta-MHC expression. Here we show that Brg1, a chromatin-remodelling protein, has a critical role in regulating cardiac growth, differentiation and gene expression. In embryos, Brg1 promotes myocyte proliferation by maintaining Bmp10 and suppressing p57(kip2) expression. It preserves fetal cardiac differentiation by interacting with histone deacetylase (HDAC) and poly (ADP ribose) polymerase (PARP) to repress alpha-MHC and activate beta-MHC. In adults, Brg1 (also known as Smarca4) is turned off in cardiomyocytes. It is reactivated by cardiac stresses and forms a complex with its embryonic partners, HDAC and PARP, to induce a pathological alpha-MHC to beta-MHC shift. Preventing Brg1 re-expression decreases hypertrophy and reverses this MHC switch. BRG1 is activated in certain patients with hypertrophic cardiomyopathy, its level correlating with disease severity and MHC changes. Our studies show that Brg1 maintains cardiomyocytes in an embryonic state, and demonstrate an epigenetic mechanism by which three classes of chromatin-modifying factors-Brg1, HDAC and PARP-cooperate to control developmental and pathological gene expression.

Myosin heavy chain (MHC) is the molecular motor in muscle cells, and two isoforms, α- and β-MHC, are expressed specifically in mammalian hearts. The α-MHC has higher ATPase activity than β-MHC, and their relative amount changes under different pathophysiological conditions1 (Supplementary Fig. 1a). The α- and β-MHC ratio correlates directly with the overall cardiac performance in animals2-4 as well as in patients with cardiomyopathy and heart failure5-8. MHC protein mutations also cause cardiac dysfunction in mice and humans9, 10. Pathologic hypertrophy of adult hearts is associated with α-MHC downregulation and β-MHC induction11, returning to a fetal state of MHC expression. However, hearts expressing α-MHC have better outcome under stress conditions than those expressing mainly β-MHC2-4. Thus, strategies to control MHC expression represent attractive approaches for heart failure therapy1, 12.

Chromatin remodeling offers one such control to modulate gene expression. The Brg1/Brm-associated-factor (BAF) complex, consisting of 12 protein subunits, is a major type of ATP-dependent chromatin-remodeling complexes in vertebrates13. Our studies show that Brg1, the essential ATPase subunit of the BAF complex14, interacts with two other classes of chromatin-modifying enzymes, histone deacetylase (HDAC)15 and poly (ADP-ribose) polymerase (PARP)16, to regulate gene expression during cardiac growth, differentiation and hypertrophy in mice (Supplementary Fig. 1b, c). Both HDACs and PARP1 are therapeutic targets for cardiac hypertrophy since pharmacologic inhibition of their activities or genetic mutations of class I HDACs and PARP1 in mice reduce hypertrophy17-22. Brg1's interaction with HDACs and PARP1, and its activation in certain patients with hypertrophic cardiomyopathy suggest that Brg1/BAF may be a target for treating cardiac hypertrophy and failure.

Growth of Brg1-null myocardium

We used the Sm22αCre transgene23 to remove floxed alleles of Brg1 (Brg1)24 in the mouse myocardium (cardiomyocytes) by embryonic day 9.5 (E9.5)(Supplementary Fig. 2a-c). Sm22αCre;Brg1 embryos were grossly normal at E11.5, but died thereafter (Supplementary Fig. 2d-f). At E10.5 the heart had thin compact myocardium and failed to form interventricular septum (Fig. 1a, b; Supplementary Fig. 3a-c) while the trabeculation was normal23. At E11.5 trabeculation, cardiac jelly, and vasculature remained normal despite the thin myocardium and absent septum (Supplementary Fig. 3d-f; 4a-f; Supplementary text). Loss of compact myocardium could reduce cardiac output, causing embryonic lethality.

Figure 1

Brg1 promotes myocardial proliferation

a, b, H&E sections of E10.5 compact myocardium, whose thickness is denoted by arrowheads.

c, d, BrdU immunostaining of E10.5 compact myocardium.

e, BrdU incorporation quantitation. A: number of areas examined. p-value: Student's t-test. Error bar: standard deviation.

f, g, BMP10 in situ hybridization of E10.5 hearts.

h, i, p57kip2 immunostaining of E10.5 hearts.

j, p57kip2 quantitation. p-value: Student's t-test. Error bar: standard deviation.

These embryos had almost no myocardial apoptosis (Supplementary Fig. 3g-l). The compact and septal primordial myocardium had dramatic decrease of cell proliferation at E10.5, while other heart layers were normal (Fig. 1c-e; Supplementary Fig. 3m-o). Therefore, Brg1 is required for cell proliferation to form the compact and septal myocardium.

To identify genes responsible for these defects, we used RNA in situ hybridization to survey crucial myocardial transcripts in E10-E11 Sm22αCre;Brg1 hearts, including Nkx2.5, Gata4, MEF2C, Tbx3, Tbx5, CX43, Irx1, Irx2, NPPA, and BMP10. We found no changes in these transcripts (Supplementary Fig. 5) except for BMP10, a key factor required for myocardial proliferation25. BMP10 expression was nearly abolished in the compact myocardium of E10.5 Sm22αCre;Brg1 (Fig. 1f, g). We next examined p57, a cyclin-dependent kinase inhibitor, whose expression is normally suppressed by BMP1025. We found p57kip2 level correlated inversely with normal cardiac cell proliferation (Supplementary Fig. 6a). Also, p57kip2 appeared ectopically in E10.5 Sm22αCre;Brg1 myocardium (Fig. 1h-j; Supplementary Fig. 6b), correlating with BMP10 reduction and termination of myocardial cell proliferation. The BMP10/p57kip2-mediated proliferation was confirmed when we rescued myocardial proliferation in Sm22αCre;Brg1 with recombinant BMP10 by whole embryo cultures26 (Supplementary Fig. 6c-e).

We further used the Mef2cCre line27 to delete Brg1 in the right ventricular myocardium, leaving Brg1 intact in the endocardium and left ventricle (Supplementary Fig. 7a, b). Mef2cCre;Brg1 embryos had hypoplastic outflow tract and right ventricle (Supplementary Fig. 7c-f; 7i, j). The Brg1-null right ventricle phenocopied the defects in Sm22αCre;Brg1 ventricles, namely BMP10 downregulation, ectopic p57kip2 expression, and proliferation reduction (Supplementary Fig. 7c-h) while the Brg1-positive left ventricle was normal, demonstrating a primary and myocardially-specific regulation of BMP10-p57kip2 by Brg1.

Differentiation of Brg1-null myocardium

We examined if the early proliferation termination in Sm22αCre;Brg1 myocardium was coupled with premature differentiation. We analyzed the myofibril formation of E10.5 cardiomyocytes by α-actinin immunostaining and electron microscopy (EM). While control compact myocardial cells showed diffuse distribution of α-actinin, a component of z-lines that demarcate sarcomeres, those of Sm22αCre;Brg1 began to show striated patterns (Supplementary Fig. 9a, b). EM verified that Sm22αCre;Brg1 displayed consecutive sarcomeres while controls only had short myofibrils (Fig. 2a, b).

Figure 2

Brg1 suppresses myocardial differentiation

a, b, EM of the compact myocardium of E10.5 embryos.

c, Quantitative RT-PCR of ventricular α- and β-MHC at E10.5 and E11.5. Ctrl: control. Mut: Sm22αCre;Brg1. p-value: Student's t-test. Error bar: standard deviation.

d, Sequence alignment of the α-MHC locus from mouse, human, and rat. Peak heights indicate degree of sequence homology. Black boxes (a1-a7) are regions of high sequence homology and further analyzed by ChIP. Red: promoter elements. Salmon: introns. Yellow: untranslated regions.

e, PCR of Brg1-immunoprecipitated chromatin from E11.5 hearts. α-HRP: anti-horse radish peroxidase antibody.

f, Luciferase reporter assay of the proximal α-MHC promoter (-462 to +192) in SW13 cells. p-value: Student's t-test. Error bar: standard deviation.

g, Sequence alignment of the β-MHC locus from mouse, human, and rat. Black boxes (b1-b5) are regions of high sequence homology and further analyzed by ChIP. Green: transposons/simple repeats.

h, PCR analysis of Brg1-immunoprecipitated chromatin from E11.5 hearts.

i, Luciferase reporter assays of the β-MHC proximal promoter (-835 to +222) in SW13 cells. p-value: Student's t-test. Error bar: standard deviation.

j, Immunostaining of HDAC1, 2, 3, 5, 6 and 9 (brown) in E11.5 hearts.

k, Co-immunoprecipitation of Brg1 with HDAC1, 2 and 9 in E11.5 hearts.

l, Quantitative RT-PCR of α- and β-MHC of cultured embryos treated with DMSO or TSA. p-value: Student's t-test. Error bar: standard deviation.

We then quantified mRNA expression of the two MHC isoforms, α-MHC, which is mainly expressed by adult hearts, and β-MHC, which is expressed primarily by embryonic hearts. E10.5 and E11.5 Sm22αCre;Brg1 ventricles highly expressed α-MHC, and down-regulated β-MHC, thereby increasing α-MHC/β-MHC ratio by 7-12 folds (Fig. 2c). Together with myofibril analysis, these data indicate that Brg1-null myocardial cells are highly differentiated and thus support a role of Brg1 in maintaining myocardial cells in an embryonic state of differentiation.

To test if Brg1 directly regulates MHC expression, we examined Brg1 binding to MHC promoters. With sequence alignment (www.dcode.org), we identified 7 evolutionarily conserved regions (a1-a7) in the mouse intergenic ~4 kb α-MHC promoter28 among mouse, rat, and human (Fig. 2d). Chromatin immunoprecipitation (ChIP) assay using E11.5 hearts with J1 anti-Brg1 antibody23 showed that of the 7 regions, Brg1 was strongly associated with the proximal promoter (a1) of α-MHC (Fig. 2e)(Supplementary text). In contrast to α-MHC, none of the 4 conserved regions in the 5kb upstream promoter of BMP10 was associated with Brg1 (Supplementary Fig. 8a, b). To test Brg1/BAF transcriptional activity, we cloned different regions of the α-MHC promoter into chromatinized episomal reporter pREP429 and then transfected the constructs into SW13 cells23, which lack Brg1 and Brm30 (Supplementary Fig. 8d). We found that restoring Brg1 expression caused approximately 65-75% reduction in the α-MHC reporter activity, and the proximal promoter (a1) was critical for α-MHC repression (Fig. 2f; Supplementary Fig. 8d). These observations support a direct repression of α-MHC by Brg1.

Since HDACs are chromatin modifiers that mediate transcriptional repression15, we tested if Brg1 requires HDACs to repress α-MHC. Indeed, Brg1 failed to repress α-MHC reporter in SW13 cells treated with trichostatin A (TSA), an HDAC inhibitor (Fig. 2f). Neither could HDAC repress the α-MHC reporter without Brg1 (Fig. 2f). HDAC proteins, including Class I HDAC1, 2, 3 and Class II HDAC5, 6, 9 were present in myocardial nuclei (Fig. 2j), and Brg1 co-immunoprecipitated with HDAC1, 2, 3 and 9 in E11.5 ventricles (Fig. 2k; data not shown). These findings indicate Brg1 and HDACs co-repress α-MHC in the embryonic myocardium.

We also analyzed the 5.5 Kb β-MHC promoter28, where we identified 5 highly conserved regions (b1 to b5 in Fig. 2g). Brg1 was widely associated with four of the five regions by ChIP experiments (Fig. 2h). Restoring Brg1 expression in SW13 cells activated β-MHC reporters (Fig. 2i). Deletional analysis of β-MHC promoter showed the proximal promoter (b1) was necessary for the Brg1-mediated β-MHC activation (Fig. 2i; Supplementary Fig. 8e). This activation of β-MHC did not require HDAC activity (Fig. 2i). However, HDAC is necessary for the basal activity of β-MHC since HDAC inhibition resulted in significant reduction of β-MHC promoter activity (Fig. 2i).

We then asked if HDAC inhibition in embryos causes premature α/β-MHC switches as observed in Brg1-null myocardium. Indeed, TSA-treated cultured embryos significantly up-regulated α-MHC while down-regulated β-MHC (Fig. 2l; Supplementary Fig. 9c, d). Overall, the biochemical studies, reporter assays, and embryo culture experiments support Brg1 and HDACs co-repress α-MHC, but independently activate β-MHC.

Although TSA causes MHC switches in embryos, HDAC inhibition did not produce reduction in myocardial proliferation (Supplementary Fig. 9e-g). Conversely, BMP10 rescued myocardial proliferation of Sm22αCre;Brg1, but did not influence α/β-MHC expression (Supplementary Fig. 9h, i). Therefore, Brg1 governs two parallel pathways to independently control myocardial growth and differentiation in embryos.

Cardiac hypertrophy and MHC changes in adult

We asked if Brg1 is also critical for cardiac growth and differentiation in stressed adult hearts. To bypass embryonic lethality, we used the doxycycline-inducible Tnnt2-rtTA;Tre-Cre mouse line31 to effect adult myocardial gene deletion. A 5 day-doxycycline treatment was sufficient to activate a β-galactosidase reporter (Supplementary Fig. 10a). We surgically constricted the transverse aorta (TAC) to pressure-overload the heart and induce cardiac hypertrophy in control and Tnnt2-rtTA;Tre-Cre;Brg1 littermates. The transgene and doxycycline alone did not cause hypertrophy (Fig. 3a; Supplementary Fig. 10f). Four weeks after surgery, the control and Tnnt2-rtta;Tre-Cre;Brg1 mice fed with normal diet developed severe cardiac hypertrophy with increased cardiomyocyte size (Fig. 3a; Supplementary Fig. 10b, c, f), ventricular/body weight ratio (Supplementary Fig. 10f), and cardiac fibrosis (Supplementary Fig. 10g, i). In contrast, doxycycline-treated Tnnt2-rtTA;Tre-Cre;Brg1 mice exhibited only mild cardiac hypertrophy with slight increase in cardiomyocytes size (Fig. 3a; Supplementary Fig. 10d, e) and ventricular/body weight ratio (Supplementary Fig. 10f), and without fibrosis (Supplementary Fig. 10h, j). Overall, Brg1-null myocardium had a 63-73% reduction of cardiac hypertrophy. Thus, Brg1 is essential for the pressure-induced cardiac hypertrophy.

Figure 3

Brg1 is required for cardiac hypertrophy

a, Cardiomyocyte size quantitation. Ctrl: control. Mut: Tnnt2-rtTA;Tre-Cre;Brg1. p-value: Student's t-test. Error bar: standard deviation.

b, c, Quantitative RT-PCR of α- and β-MHC in cardiac ventricles of doxycycline-treated control and Tnnt2-rtTA;Tre-Cre;Brg1 mice 4 weeks after sham/TAC operation. p-value: Student's t-test. Error bar: standard deviation.

We next investigated whether Brg1 regulates MHC expression in hypertrophic hearts. Control hypertrophic hearts underwent canonical MHC changes, namelyα-MHC downregulation and β-MHC upregulation (Fig. 3b). In contrast, doxycycline-treated Tnnt2-rtTA;Tre-Cre;Brg1 mice showed a 2.1-fold increase of α-MHC and a 51% reduction of β-MHC (Fig. 3b). Consequently, the pressure-stressed Brg1-null myocardium expressed 4.4-fold α-MHC and 0.13 fold β-MHC as the control myocardium (Fig. 3c). This reversal of MHC was not caused by reduced hypertrophy, which could only lessen, but not reverse, canonical MHC changes. Therefore, Brg1 is critical for α/β-MHC switch in hypertrophic hearts.

While Brg1 is highly expressed in embryonic hearts23, it is turned off in adult myocardium with some expression in endothelial or interstitial cells (Fig. 4a). Brg1 became detectable in cardiomyocytes within 7 days after TAC by immunostaining (Fig. 4b), confirmed by western blot (Fig. 4d); while it remained absent in Tnnt2-rtTA;Tre-Cre;Brg1 myocardium (Fig. 4c). Also, Brg1 mRNA increased by 1.8 fold within 2 weeks after TAC (Fig. 4e), indicating Brg1 reactivation by stress signals is essential for the hypertrophic process.

Figure 4

MHC regulation by Brg1, PARP and HDAC

a, b, c, Brg1 immunostaining in ventricular myocardium of doxycycline-treated control and Tnnt2-rtTA;Tre-Cre;Brg1 mice 1 week after sham/TAC operation. Arrows: cardiomyocyte nuclei.

d, Brg1 immunoblot of cardiac nuclear extracts from wildtype mice 2 weeks after TAC.

e, Quantitative RT-PCR of Brg1 mRNA in wildtype mice 2 weeks after TAC. p-value: Student's t-test. Error bar: standard deviation.

f, PCR of Brg1- and PARP1- immunoprecipitated chromatin from thymus and adult hearts 2 weeks after TAC.

g, h, Luciferase reporter assays of α-MHC (g) and β-MHC (h) promoter in SW13 cells with PARP inhibition. p-value: Student's t-test. Error bar: standard deviation.

i, j, Co-immunoprecipitation of Brg1, PARP1, HDAC2 and 9 in TAC-treated adult hearts (i) and in E11.5 hearts (j).

k, Quantitative RT-PCR of α- and β-MHC of PJ34-treated cultured embryos. p-value: Student's t-test. Error bar: standard deviation.

l, PCR of PARP1- immunoprecipitated chromatin from E11.5 hearts.

m, PCR of HDAC2- immunoprecipitated chromatin from adult hearts 2 weeks after TAC.

We then determined if reactivated Brg1 controls MHC expression through direct binding of MHC promoters. ChIP analysis of TAC-treated hearts showed Brg1 was highly enriched in the proximal promoters of both α-MHC and β-MHC, but not BMP10 (Fig. 4f; Supplementary Fig. 8a, c). Furthermore, Brg1 binding to MHC promoters was detectable only in TAC-treated, but not sham-operated, hearts (Supplementary Fig. 10k), consistent with Brg1 reactivation by pressure overload. The binding pattern was similar to that in embryonic hearts (Fig. 2e, 2h), indicating a common mechanism underlying the Brg1-mediated MHC control in embryonic and hypertrophic hearts.

MHC regulation by Brg1, HDAC and PARP

Besides HDACs18-20, PARP1 is the only other chromatin-modifying enzyme16 known to regulate cardiac hypertrophy17, 22. However, it is unknown if PARP1 binds to MHC promoters. ChIP analysis of TAC-treated hearts showed PARP1 bound to the proximal promoters of α-MHC and β-MHC, but not BMP10 in a pattern similar to that of Brg1 (Fig. 4f; Supplementary Fig. 11a). Like Brg1, PARP1 binding occurred only in TAC-treated, but not sham-operated, hearts (data not shown). Furthermore, inhibiting PARP1 activity by PJ-3432 reduced both Brg1-mediated α-MHC repression and β-MHC activation in reporter assays in SW13 cells, indicating Brg1 and PARP1 cooperate to regulate MHC (Fig. 4g, h). Indeed, PARP1 and Brg1 co-immunoprecipitated in both TAC-treated hearts and E11.5 hearts (Fig. 4i, j). Embryos cultured with PJ-34 had normal myocardial proliferation, but exhibited α/β-MHC switches characteristic of Brg1-null myocardium (Supplementary Fig. 11b; Fig. 4k). Immunostaining and ChIP analyses of E11.5 hearts showed PARP1 was present in myocardial nuclei and bound to the proximal promoters of α- and β-MHC, but not BMP10, in a pattern similar to that of Brg1 (Fig. 4l; Supplementary Fig. 11c, d). These findings indicate Brg1 complexes with PARP1 to regulate MHC in embryonic and stressed adult hearts.

Brg1 and PARP1 co-immunoprecipitated with HDAC1, 2 or 9 in E11.5 and stressed adult hearts (Fig. 2k; 4i, j; Supplementary Fig. 11e). We asked if HDACs are present on MHC promoters. Using ChIP with two cross-linking steps33 in TAC-treated hearts, we found HDAC2 and HDAC9 were enriched in the α-MHC promoter, but bound minimally to the β-MHC promoter (Fig. 4m; Supplementary Fig. 11f), suggesting direct α-MHC and indirect β-MHC regulation by HDACs. Together with reporter assays (Fig. 2f, i; 4g, h), these biochemical studies suggest that Brg1, PARP, and HDAC physically form a chromatin-remodeling complex on the α-MHC promoter to repress α-MHC, while Brg1 complexes with PARP on the β-MHC promoter to activate β-MHC.

Implication in human cardiomyopathy

To investigate if Brg1 is activated in human hypertrophic hearts, we studied patients with hypertrophic cardiomyopathy (HCM) of unknown etiology, who required surgical myectomy34 to relieve cardiac obstruction caused by prominent ventricular or septal hypertrophy (Supplementary Fig. 12a). HCM severity was measured by the maximal thickness of interventricular septum during diastole (IVSd). IVSd in HCM patients was 2.02 folds of the control (Supplementary Fig. 12b, c). Quantitative RT-PCR analyses showed HCM hearts had 48-fold reduction of α-MHC, 5.5-fold increase of β-MHC and 2-fold increase of Brg1 expression (Fig. 5a). The loss of α-MHC, gain of β-MHC, and activation of Brg1 resembled the changes observed in mice with hypertrophy, suggesting a similar role of Brg1 in human disease. Consistent with this notion, the IVSd and β/α-MHC ratio correlated well with Brg1 level between control and HCM subjects, with sigmoidal regression curves inflecting at 1.50-fold and 1.45-fold of Brg1, respectively (Fig. 5b, c; Supplementary Fig. 12d, e). At the inflection point of 1.50-fold Brg1, IVSd equals 1.54 cm, coinciding with a clinical criterion (IVSd >1.50 cm) in HCM diagnosis35. We therefore speculate a 50% increase of Brg1 may be a threshold for disease development in certain patients.

Figure 5

Brg1 activation in human cardiomyopathy

a, Quantitative RT-PCR of α-MHC, β-MHC, and Brg1 expression in normal and HCM subjects. p-value: Student's t-test. Error bar: standard deviation.

b, IVSd (y) plotted against the Brg1 RNA level (x). Red: regression curve. e: the base of natural logarithm (~2.718). Arrow and dashed line: the inflection point.

c, The β /α-MHC RNA ratio (y) plotted against the Brg1 RNA level (x).

d, Model of developmentally-activated and stress-induced assembly of BAF/HDAC/PARP complexes on the α-MHC, and BAF/PARP complex on the β-MHC promoter.

Discussion

Brg1/BAF may have regenerative and therapeutic implications given its newly identified roles in both embryonic and adult cardiomyocytes. Similar mechanisms are directed by Brg1 to control cardiac growth, differentiation and gene expression under developmental and pathological conditions (Supplementary Fig. 1b, c). The stress-dependent assembly of a developmental complex to modify chromatins in adult hearts provides a molecular explanation for fetal gene activation in the diseased adult myocardium. HDAC, PARP and now Brg1/BAF are the only classes of chromatin-modifying factors known to regulate cardiac hypertrophy. Cardiac stresses activate Brg1, which then assembles a BAF/HDAC/PARP chromatin complex on MHC promoters (Fig. 5d), where they may interact with transcription factors such as TR, TEF1, MEF2, SRF, GATA4 and NFAT36, 37 to control MHC expression. The induction of Brg1 by hypertrophic stimuli suggests that chromatin may ultimately be where all the stress-response signals converge for the regulation of MHC genes, a critical step in the myopathic process.

Brg1/BAF, besides thyroid hormone receptors36, provides the only known direct mechanism that antithetically regulates α-MHC and β-MHC. This opposite MHC regulation may underlie the on-off, rather than graded, switching of MHC in individual cardiomyocytes of hypertrophic hearts28. Exactly how HDACs and PARPs contribute to this BAF-mediated process awaits further investigations. HDACs and PARPs may covalently modify histones and thereby help anchor BAF to certain sites of MHC promoters. BAF proteins may also be modified through acetylation/deacetylation or poly-ADP-ribosylation by HDACs and PARPs. Such modifications may decide the composition of BAF complex and how BAF interacts with chromatin as well as other MHC regulators such as miR-2081. Elucidating these issues will help determine how the chromatin and target specificities of BAF may be established by its possible combinatorial assembly with the large families of 17 PARP16 and 18 HDAC15 proteins.

METHODS

Sm22αCre, Brg1, Mef2cCre, R26R and Tnnt2-rtTA;Tre-Cre mice have been described23, 24, 27, 31, 38. Immunostaining, RNA in situ hybridization, quantitative RT-PCR, and whole embryo culture were performed as described23, 26. TAC was modified from previous descriptions20. The pressure load caused by TAC was verified by the pressure gradient across the aortic constriction measured by echocardiography. Only mice with a pressure gradient > 30 mmHg were analyzed for cardiac hypertrophy and gene expression. Curve modeling was performed with the Levenburg-Marquardt non-linear regression method and XLfit software. Detailed methods can be found in the Supplementary Information.