CITED4 Protects Against Adverse Remodeling in Response to Physiological and Pathological Stress

Abstract

Rationale:

Cardiac CITED4 is induced by exercise and is sufficient to cause physiological hypertrophy and mitigate adverse ventricular remodeling after ischemic injury. However, the role of endogenous CITED4 in response to physiological or pathological stress is unknown.

Objective:

To investigate the role of CITED4 in murine models of exercise and pressure overload.

Methods and Results:

We generated cardiomyocyte-specific CITED4 knockout mice (C4KO) and subjected them to an intensive swim exercise protocol as well as transverse aortic constriction (TAC). Echocardiography, western blotting, qPCR, immunohistochemistry, immunofluorescence, and transcriptional profiling for mRNA and miRNA expression were performed. Cellular crosstalk was investigated in vitro.

CITED4 deletion in cardiomyocytes did not affect baseline cardiac size or function in young adult mice. C4KO mice developed modest cardiac dysfunction and dilation in response to exercise. After TAC, C4KOs developed severe heart failure with left ventricular dilation, impaired cardiomyocyte growth accompanied by reduced mammalian target of rapamycin (mTOR) activity and maladaptive cardiac remodeling with increased apoptosis, autophagy, and impaired mitochondrial signaling. Interstitial fibrosis was markedly increased in C4KO hearts after TAC. RNAseq revealed induction of a pro-fibrotic miRNA network. miR30d was decreased in C4KO hearts after TAC and mediated crosstalk between cardiomyocytes and fibroblasts to modulate fibrosis. miR30d inhibition was sufficient to increase cardiac dysfunction and fibrosis after TAC.

Conclusions:

CITED4 protects against pathological cardiac remodeling by regulating mTOR activity and a network of miRNAs mediating cardiomyocyte to fibroblast crosstalk. Our findings highlight the importance of CITED4 in response to both physiological and pathological stimuli.

Graphical Abstract

INTRODUCTION

Exercise promotes cardiovascular health and mitigates established disease.¹ Exercise training leads to an increase in cardiac mass, called physiological hypertrophy, which appears to mediate cardiac benefits.² In contrast, pathological stimuli, e.g. hypertension and aortic stenosis, lead to pathological growth, which often precedes heart failure and is associated with adverse outcomes. While physiological and pathological hypertrophy appear similar macro- and microscopically at early stages, the underlying mechanisms are often distinct.

Genome-wide analysis of all transcriptional components altered in exercised hearts found that CITED4 expression increased in hearts from swum mice.³ CITED (CBP/p300-Interacting Transactivators with E (glutamic acid)/D (aspartic acid)-rich-carboxylterminal domain⁴) proteins are 22- to 27 kDa proteins, which do not have DNA-binding motifs or DNA affinity and are thought to work through binding other transcription factors (e.g. CBP/p300).⁵ In the heart, CITED4 is expressed predominantly in cardiomyocytes and only at very low levels in fibroblasts or other non-cardiomyocytes.³

CITED4 overexpression induced hypertrophy in neonatal cardiomyocytes in vitro, and a high-throughput phenotypic screen of hypertrophic agonists found that neuregulin-1 (NRG1), a physiological stimulus increased in exercise, was linked to enhanced CITED4 expression.⁶ Interestingly, CITED4 also promoted a physiological cardiomyocyte growth pattern in vitro and prevented elongation in response to pathologic hypertrophic stimuli.⁶ Subsequently, we generated a transgenic mouse model with regulated, cardiomyocyte-specific CITED4 expression.⁷ Increased cardiac CITED4 expression in adult mice was sufficient to induce physiological cardiac hypertrophy with increased heart weight and cardiomyocyte size as well as normal cardiac function without fibrosis.⁷ CITED4 expression also enhanced functional recovery after ischemia/reperfusion injury with reduced ventricular dilation and scar size.⁷

However, overexpression models can lead to exaggerated or even nonphysiological effects, and may not reflect the role of the endogenous pathway. To understand the role of endogenous cardiomyocyte CITED4, we generated cardiomyocyte-specific CITED4 knockout mice (C4KO). C4KO mice were phenotyped at baseline and subjected to a swimming protocol as a physiological stimulus and transverse aortic constriction- (TAC-) induced pressure overload as a pathological stimulus. We found that CITED4 increases not only in response to exercise⁸ but also in response to TAC. Interestingly, while cardiac morphology and function were normal at baseline in young adult C4KO mice, they manifested a maladaptive response to exercise, characterized by ventricular dilation, reduced systolic function and increased expression of fibrosis-related and pathological genes. After TAC, C4KO mice had impaired cardiac hypertrophy corresponding with decreased activation of mTOR (mammalian target of rapamycin) and mitochondrial signaling as well as increased autophagy, apoptosis, and fibrosis. RNA sequencing revealed that CITED4 regulates a network of pro-fibrotic miRNAs. One of these, miR30d, correlates with outcomes in human heart failure patients and is released in exosomes from cardiomyocytes in response to stress.⁹ miR30d appeared to mediate cardiomyocyte-fibroblast cross talk and silencing of miR30d was itself sufficient to increase fibrosis and adverse cardiac remodeling after TAC similar to that seen in C4KOs. Taken together, these data demonstrate that CITED4 is upregulated in both pathological and physiological remodeling, and protects against maladaptive remodeling in both settings.

METHODS

Detailed methods and materials are in the Online Supplement. Microarray and Sequencing data have been made publicly available at the Gene Expression Omnibus (can be accessed at GSE149584, GSE150293) and the Sequence Read Archive (can be accessed at SRA PRJNA529610).

Further data and materials that support the findings of this study are available from the authors upon reasonable request. Please also see the Major Resources Table in the Online Supplement.

RESULTS

Generation of cardiomyocyte-specific CITED4 knockout mice.

To examine the role of cardiomyocyte CITED4 in vivo, we generated a cardiomyocyte-specific CITED4 knockout (C4KO) mouse by breeding αMHC-Cre transgenic to CITED4 floxed (F/F Ctr) mice. C4KO mice were born in the expected Mendelian ratio and survived into adulthood. Immunoblotting for CITED4 protein was unreliable due to insufficient and inconsistent antibody quality, but qPCR revealed ~80% reduction of CITED4 mRNA expression in adult whole heart lysates (Online Figure IA). To determine whether the residual CITED4 mRNA expression in whole heart lysates reflected expression in other cell types, we isolated cardiomyocytes from C4KO and littermate controls (Online Figure IB). Successful fractionation of cells was confirmed by immunoblotting for Troponin T (Online Figure IC). CITED4 mRNA expression was reduced >98% in cardiomyocytes isolated from C4KO mice in comparison to control mice. Of note, no reduction was seen in CITED4 expression in skeletal muscle in C4KO mice confirming the specificity of deletion (Online Figure ID).

Cardiomyocyte specific CITED4 deletion does not affect baseline cardiac mass or function in young adult mice.

We first evaluated baseline cardiac function and morphology by echocardiography in 8-12 week old C4KO mice and littermate controls. No statistically significant differences were seen in fractional shortening (%FS) (Figure 1A), left ventricular mass index (LVMI, Figure 1B), or ventricular dimensions (Online Figure IE). Consistent with this, heart weight/tibia length (HW/TL, Figure 1C) and lung weight/tibia length (LW/TL, Figure 1D) ratios were not altered in C4KO mice confirming the absence of adverse cardiac remodeling or pulmonary congestion at baseline after CITED4 deletion. Moreover, there was no statistical difference in cardiomyocyte cross-sectional area in C4KO mice compared to controls (Figure 1E, 1F). There was also no statistical difference in gene expression of markers for pathological or physiological cardiac remodeling (e.g. ANP, BNP, αMHC, βMHC, C/EBPβ) or fibrosis markers (e.g. Col3a1, Col5a1, Col8a1, CTGF) in hearts from C4KO mice at baseline (Online Figure IIA, IIB). Furthermore, expression of relevant transcription factors/co-factors (e.g. Mef2a, Mef2c, p300, GATA4), as well as hypoxia (Hif1, VEGFα), autophagy (Atg5, Beclin1), and apoptosis (Bcl2, Birc2) genes was unchanged in C4KO hearts at baseline (Online Figure IIC). Collectively these data indicate that CITED4 is not necessary for normal cardiac development or maintenance of normal cardiac structure and function in the young adult heart.

Figure 1: Cardiomyocyte specific CITED4 knockout does not affect baseline cardiac mass or function in young adult mice.

A, Baseline Fractional Shortening (%FS) in C4KO compared to F/F Ctr mice in 8-12 week old mice (n=4 F/F Ctr, 3 C4KO, p=0.54). B, Left ventricular mass index (LVMI) for C4KO and control animals (n=4 F/F Ctr, 3 C4KO, p=0.64). C, Heart weight to tibia length ratios (HW/TL) in C4KO and F/F Ctr animals (n= 3 mice per group, p=0.56). D, Lung weight to tibia length ratios (LW/TL) in C4KO and F/F Ctr animals (n=3 mice per group), p=0.25). E, Cross sectional area (CSA) of C4KO relative to F/F Ctr cardiomyocytes at baseline (n=4 F/F Ctr, 3 C4KO, p=0.057). F, Micrographs of cardiac sections stained with Masson Trichrome Staining used to measure cross sectional area in E, scale bar = 75μm, images were chosen to represent average of both groups. For all graphs significance was determined by Student’s t-test, *p<0.05.

Cardiomyocyte specific deletion of CITED4 in mice causes maladaptive remodeling and a functional deficit in response to endurance exercise.

To evaluate whether CITED4, which is upregulated in response to exercise training,⁸ was also necessary for an adaptive response to endurance exercise, we carried out a ramp swimming exercise protocol in both C4KO and littermate controls. Interestingly, C4KO mice developed mildly reduced left ventricular systolic function (% Fractional Shortening) (Figure 2A) in response to exercise training. While the overall increase in heart weight normalized to tibia length was not different between C4KO and control animals (Figure 2B), cardiomyocyte cross sectional area was reduced (Figure 2C) and left ventricular dimensions increased at the end of the swimming period (Figure 2D). There was also an increase of fibrosis-related gene expression, including Col3a1, Col5a1, and CTGF (Figure 2E), as well as induction of markers of pathological remodeling such as BNP, and markers of mTOR suppression through upregulation of DDIT4 (Figure 2F). While relatively subtle, these results stand in contrast to the more typical physiological hypertrophy, gene expression patterns, and trend toward improved function seen with exercise. Taken together, these data indicate that CITED4 is necessary for a full cardiac adaptation to endurance exercise.

Figure 2: Cardiomyocyte specific deletion of CITED4 causes maladaptive remodeling and a functional deficit in response to endurance exercise.

A, Fractional Shortening (%FS) in C4KO compared to F/F Ctr mice after swimming exercise (n=5 F/F Ctr, 6 C4KO, p=0.005). B, Heart weight to tibia length ratios (HW/TL) in C4KO and F/F Ctr animals (n=5 F/F Ctr, 6 C4KO, p=0.29) after swimming exercise. C, Cross sectional area (CSA) of C4KO relative to F/F Ctr cardiomyocytes at baseline (n=3 F/F Ctr, 4 C4KO, p=0.015). D, Left ventricular internal dimension in end-diastole (LVIDd, p=0.043) and end-systole (LVIDs, p=0.049) as well as relative wall thickness (RWT, p=0.097) assessed by echocardiography in C4KO and control animals after swimming exercise (n=5 F/F Ctr, 6 C4KO). E, qPCR analysis of fibrosis-related genes in C4KO relative to F/F Ctr hearts after swimming exercise (Col3a1 p<0.001, Col5a1 p=0.008, CTGF p=0.017. F, qPCR analysis related to physiological and pathological remodeling (bMHC, ANP, BNP p=0.002), autophagy (Beclin1 p=0.023) and mitochondrial pathways (PGC1a, NDUFs1, NDUFv2) and mTOR regulators (DDIT4 p=0.027, DDIT4L) in C4KO relative to F/F Ctr hearts after swimming exercise (n=4 F/F Ctr, 5 C4KO). For all graphs significance was determined by Student’s t-test, *p<0.05.

Cardiomyocyte-specific deletion of CITED4 accelerates dilation and heart failure after TAC.

While CITED4 was identified in a genome-wide screen as a transcriptional co-activator increased by exercise but not by 2 weeks of TAC,³ we found that CITED4 expression transiently increased in hearts from control animals during the initial growth response one week after TAC, suggesting that it may play a role during that critical time. (Figure 3A). To investigate the functional role of CITED4 in pressure overload-induced cardiac remodeling, we subjected C4KO and littermate controls to TAC. First, mice were followed up to eight weeks post-surgery or until they developed severe heart failure with clinical symptoms that required euthanasia (endpoint).

Figure 3: Cardiomyocyte specific CITED4 knockout mice develop accelerated heart failure after transverse aortic constriction (TAC).

A, qPCR analysis of CITED4 mRNA in hearts of control mice at 1 week (n=3) and 8 weeks (n=5) after TAC and at baseline (n=4) (p<0.035 baseline vs. 1 week, p<0.028 1 week vs. 8 weeks, p=0.874 baseline vs. 8 weeks, one-way ANOVA with Sidak’s multiple comparisons test). B, Echocardiographic images from F/F Ctr and C4KO 8 weeks after TAC, images were chosen to represent average of both groups. C, Assessment of ventricular function by echocardiography in C4KO compared to F/F Ctr mice at indicated time points before and after TAC surgery (n=6 F/F Ctr, 6 C4KO, p=.999, 0.050, 0.064, 0.002, <0.001, respectively). D, Left ventricular dimensions in end-diastole (LVIDd, p=0.999, 0.245, 0.009, 0.025, 0.282, respectively) and E, end-systole (LVIDs) at indicated time points before and after TAC surgery (n=6 F/F Ctr, 6 C4KO, p=0.999, 0.426, 0.076, 0.017, <0.0001, respectively). F, Relative wall thickness assessed by echocardiography in C4KO and control animals at indicated time points before and after TAC surgery (n=6 F/F Ctr, 6 C4KO, p=0.998, 0.026, 0.391, 0.049, <0.607, respectively). For all graphs in C-F significance was determined by repeated-measures two-way ANOVA and Sidak’s multiple comparisons test, *p<0.05.

Echocardiography revealed an accelerated decline in cardiac function in C4KO mice after TAC compared to littermate controls (Figure 3B, 3C). C4KO mice also rapidly developed left ventricular chamber dilation (Figure 3D, 3E), and decreased relative wall thickness (Figure 3F). Of note, C4KO hearts did not undergo a hypertrophic phase in response to pressure overload suggesting that CITED4 is necessary for initial cardiac growth in response to TAC. Increased heart weight (Figure 4A) measurements confirmed altered cardiac remodeling in C4KO mice after surgery. Lung weights were also substantially increased in C4KO mice after TAC consistent with pulmonary congestion (Figure 4B). Interestingly, cardiomyocyte cross sectional area (Figure 4C) was decreased in C4KO hearts consistent with the absence of an early hypertrophic phase and the model that CITED4 is necessary for cardiomyocyte growth in response to pressure overload. Histological analyses even as early as one week after surgery revealed increased cardiac fibrosis (Figure 4D, 4E). Longitudinally sectioned hearts showed increased interstitial fibrotic tissue throughout the ventricular wall in C4KO hearts, while fibrosis was localized to the base region in control hearts (Figure 4E). Cardiomyocyte apoptosis was also increased in C4KO hearts one week after TAC compared to littermate controls (Figure 4F, 4G), without a statistically significant difference in capillary density (Online Figure IIIA). Since adverse remodeling was most dramatic after TAC, we focused further detailed analyses and mechanistic studies on this model.

Figure 4: Cardiomyocyte specific CITED4 knockout mice develop ventricular dilation and maladaptive remodeling in response to pressure overload.

A, Heart weight to tibia length ratios (HW/TL) in C4KO and F/F Ctr animals at endpoint after TAC surgery (n=6 F/F Ctr, 7 C4KO, p=0.046). B, Lung weight to tibia length ratios (LW/TL) in C4KO and F/F Ctr animals at endpoint after TAC surgery (n=6 F/F Ctr, 7 C4KO, p=0.049). C, Cross sectional area measurement from C4KO compared to F/F Ctr hearts at endpoint after TAC surgery (n=4 F/F Ctr, 7 C4KO, >200 cells per section, p<0.001). D, Quantification of fibrosis measured as percentage of total myocardial area in C4KO compared to F/F Ctr hearts at the endpoint after TAC surgery (n=5 F/F Ctr, 4 C4KO, p=0.017). E, Micrographs of cardiac sections stained with Masson Trichrome Staining to label fibrotic tissue one week after TAC surgery, scale bar = 1mm, images were chosen to represent differences of fibrosis pattern in both groups. F, Images of TUNEL staining in C4KO compared to F/F Ctr hearts one week after TAC surgery. Sections were stained with Wheat Germ Agglutinin (WGA) to mark cellular membranes (green), TUNEL (red), and DAPI (blue), scale bar=100μm, images were chosen to represent TUNEL positive areas in both groups. G, Quantification of TUNEL-positive cardiomyocytes in C4KO relative to F/F Ctr hearts one week after TAC surgery (n=5 F/F Ctr, 4 C4KO, p=0.044). For all graphs significance was determined by Student’s t-test, *p<0.05 unless indicated otherwise.

As CITED4 overexpression after ischemia-reperfusion injury induced protective activation of mTOR signaling,⁷ we analyzed mTOR downstream targets in C4KO hearts after TAC. We found decreased S6 protein and 4EBP1 phosphorylation and increased Beclin1 and Lc3II protein expression, indicating decreased cellular growth and translational mTOR activity and increased autophagy in C4KO hearts one week after TAC (Figure 5A, 5B). Reduced mTORC1 activation after TAC persisted in C4KO hearts as long as 8 weeks post-surgery (Online Figure IIIB, IIIC) while mTOR signaling appeared unchanged in unstressed hearts from C4KO mice (Online Figure IIID, IIIE). Growth factor signaling usually activates mTORC1 through AKT-mediated inhibitory phosphorylation of hamartin/tuberin (TSC1/2).¹⁰ Interestingly, phosphorylation of AKT and (AKT-dependent) TSC2 (T1462, S939) phosphorylation were upregulated in C4KO mice one week after TAC, potentially indicating compensatory feedback signaling involving mTORC2 in response to decreased mTORC1 activity (Figure 5A, 5B). Phosphorylation of S1387-TSC2 usually resulting in mTOR suppression remained unchanged in C4KO hearts after TAC when compared to controls. However, gene expression of the TSC-dependent mTORC1 suppressor REDD1 (DDIT4) and its homolog REDD2 (DDIT4L) were upregulated in C4KO hearts one week after TAC (Figure 5C). In addition, REDD1 (DDIT4) but not REDD2 (DDIT4L) was upregulated in isolated ventricular myocytes from C4KO mice at baseline and after phenylephrine (PE) stimulation (Figure 5D), providing a potential mechanism contributing to CITED4’s regulation of mTORC1 signaling. These results suggest that transient CITED4 upregulation after TAC is necessary for the early growth response of the heart, while deletion of CITED4 reduces mTORC1 activation after TAC, impairs early cardiac hypertrophy and leads to accelerated heart failure with enhanced fibrosis, apoptosis, and autophagy.

Figure 5: mTOR signaling is impaired in C4KO hearts after pressure overload.

A, Western Blot of protein isolated from heart lysates one week after TAC surgery comparing AKT, TSC2, S6- and 4EBP-protein-phosphorylation and Beclin1 and LC3 protein expression in C4KO hearts with F/F Ctr hearts. GAPDH served as protein loading control. B, Quantification of AKT (p=0.008), T1462-TSC2 (p=0.024), S939-TSC2 (p=0.047), S1387-TSC2 (p=0.063), pS6 (p=0.026) and p4EBP (p=0.012) protein-phosphorylation and Beclin1 (p=0.003) and LC3 A/B (p=0.087) protein expression in C4KO hearts relative to F/F Ctr hearts one week after TAC (n=6 F/F Ctr, 6 C4KO, *p<0.05, Student’s t-test). C, qPCR analysis of mTORC1 inhibitors REDD1 (DDIT4) and REDD2 (DDIT4L) in C4KO relative to F/F Ctr hearts one week after TAC surgery (n=4/group, *p=0.032 and 0.023, respectively, Student’s t-test). D, qPCR analysis of REDD1 (DDIT4) and REDD2 (DDIT4L) mRNA expression in isolated ventricular myocytes from F/F Ctr and C4KO hearts at baseline (n=5/group, p=0.029 REDD1 and 0.799 for REDD2) and after Phenylephrine (PE) stimulation (n=4 F/F Ctr, 3 C4KO, *p=0.002 for REDD1 −PE, 0.035 +PE, ns for REDD2, *p<0.05 one-way ANOVA and Sidak’s post test).

Gene expression array reveals impaired mitochondrial pathways in C4KO hearts after TAC.

To better understand the pathways contributing to C4KO phenotypes, we performed microarray analyses of hearts from C4KO and littermate controls one week after TAC when the phenotypes just began to diverge in order to avoid confounding by the heart failure phenotype that develops later in these mice. Differentially expressed mRNAs (p≤0.005; fold change≥1.5) were subjected to hierarchical cluster analysis and displayed in a heat map (Figure 6A) and a volcano plot to highlight the most significant differences (Online Figure IVA). We also assessed differentially regulated pathways using Gene Set Enrichment Analysis (GSEA) using Gene Ontology (GO) gene sets. Interestingly, all top ten downregulated pathways were linked to mitochondria or mitochondrial function (Figure 6B), while significantly upregulated pathways included those relevant to extracellular matrix, among others (Figure 6B). The most differentially expressed genes in the combined top ten downregulated pathways were mitochondrial complex I NADH:ubiquinone oxidoreductase subunits (NDUF) and genes encoding for mitochondrial inner membrane related proteins (TIMM9, IMMP1L) (Online Figure IVB). Similarly, genes necessary for mitochondrial function, such as subunits of ATP synthase (e.g. ATP5K, ATP8B3, data not shown), cytochrome c oxidase (e.g. COX7b, COX5a), and relevant Complex II genes (SDHA, SDHB) were reduced in hearts from C4KO mice after TAC (Online Figure IVB). Applying KEGG analysis, a similar gene set was significantly downregulated (Online Figure VA). KEGG-pathways enriched for these genes included oxidative phosphorylation, citrate cycle/TCA cycle, fatty acid metabolism, cardiac muscle contraction, and dilated cardiomyopathy, among others (Online Figure VB). In an independent cohort, we confirmed significant downregulation of NDUFs1 and NDUFv2, both known downstream targets of the transcriptional co-activator PGC1α (PPARGC1A), which was also decreased in the hearts of C4KO mice after TAC (Figure 6C). PGC1α regulates oxidative phosphorylation and fatty acid oxidation gene expression and plays a key role in mitochondrial biogenesis and metabolism in the heart.¹¹ Of note, PGC1α deletion was previously shown to accelerate cardiac dysfunction in response to pressure overload and its transcription is modulated by mTORC1 in other cellular systems.^{12, 13} To determine if CITED4 regulates PGC1α autonomously, we investigated gene expression in isolated adult ventricular myocytes from C4KO and control animals at baseline and after PE stimulation. While we found no statistically significant difference in PGC1α or its tested downstream targets in isolated adult C4KO cardiomyocytes, we did see a trend towards a more pronounced reduction in PGC1α signaling after PE stimulation and a significant reduction in its target NDUFs1 (Online Figure VIA). C4KO mice also showed a nonsignificant trend to decreased mitochondrial DNA content, indicative of either less mitochondrial biogenesis or increased mitophagy in response to pressure overload (Online Figure VIB). Electron microscopy of heart sections, however, did not reveal consistent morphological alterations in mitochondria between C4KO and control mice after TAC, but intracellular edema and dilation of sarcotubular structures was noted particularly in some cells in the C4KO group (Online Figure VIC). While impaired mitochondrial pathways are commonly seen in heart failure, our data indicate changes in these pathways as early as one week after TAC, when most phenotypes were not significantly different from controls and overt heart failure had not developed. Thus, the accelerated and exaggerated expression changes seen in C4KO mice likely contribute to rather than reflect the more severe ventricular dysfunction that develops later in these mice.

Figure 6: Gene expression array reveals impaired mitochondrial pathways in C4KO hearts after pressure overload.

A, Heatmap depicts differentially expressed mRNA (applying the combined criteria of p≤0.005 and fold change ≥1.5) in C4KO and F/F Ctr heart samples (n=4/group) one week after TAC surgery. The dendrogram was constructed using the Manhattan-Ward clustering algorithm. B, Bar-plot of the top 10 most significantly enriched pathways, Gene Set Enrichment Analysis (GSEA) using Gene Ontology database. Bars depict the normalized enrichment score of pathway up- or downregulation in C4KO hearts relative to F/F Ctr hearts one week after TAC surgery (n=4/group). C, qPCR for validation of reduced PGC1α, NDUFs1, and NDUFv2 mRNA expression (n=4 F/F Ctr, 4 C4KO, p=0.001, 0.039, 0.024, respectively), Student’s t-test, *p<0.05.

Fibrosis-related microRNA pathways are dysregulated in C4KO hearts after TAC.

We also analyzed microRNAs (miRNAs) that were differentially expressed in C4KO hearts after TAC by RNASeq. We found that over 70% (32 out of 43) of differentially regulated miRNAs (p≤0.05; fold change≥1.25) were linked to fibrosis (either through Ingenuity Pathway analysis (IPA, Qiagen) and/or prior publications) (Figure 7A). Select miRNAs either up- or downregulated in C4KO hearts after TAC were validated in an independent cohort (Online Figure VIIA). These results are consistent with the most significantly upregulated pathways detected in the gene expression analysis, which included Serpine1, TGFB2, TIMP3, CTGF, Col5a1, Col6a1, Col8a1, TIMP1, for example, and were enriched in extracellular matrix, extracellular structure organization, biological adhesion, and extracellular matrix component pathways, among others (data not shown). We analyzed expression of fibrosis-related genes targeted by miRNAs that were differentially expressed in C4KO hearts after TAC and found inversely regulated in the transcriptome of C4KO hearts after TAC, including Col3a1, Col8a1, Col5a1, TGFb2, CTGF and PDGFc, which were validated in an independent cohort of C4KO and littermate controls (Figure 7B).

Figure 7: Fibrosis-related pathways are dysregulated in C4KO hearts after pressure overload.

A, Differentially expressed, fibrosis-related (annotated and/or published) miRNAs in C4KO compared to F/F Ctr hearts one week after TAC surgery assessed by miRNASeq profiling (n= 5/group). B, qPCR validation of fibrosis-related genes (mRNA) inversely regulated by differentially expressed miRNAs (A) in C4KO relative to F/F Ctr hearts after TAC surgery (n=4/group, p=0.041, 0.033, 0.021, 0.045, 0.048, 0.036, respectively, *p<0.05, Student’s t-test). C, qPCR validation of miR-30d expression in C4KO relative to F/F Ctr hearts one week after TAC surgery (n=8/group, p=0.045) and in hearts from inducible CITED4 mice relative to tTA Ctr hearts (n=4/group, p=0.029), *p<0.05, Student’s t-test. D, Neonatal rat ventricular myocytes (NRVM) were treated with negative control (Ctr) and CITED4 (C4) siRNA. C4 siRNA treatment lead to a reproducible decrease in CITED4 mRNA expression (p=0.002) and caused a decrease in miR30d expression (p=0.006) (n=4 individual experiments, *p<0.05, Student’s t-test). E, CTGF (*p<0.001, ^#p<0.001, Ctr miR vs. C4 siRNA p=0.002), Col1a1 (*p<0.001, Ctr miR vs. C4 siRNA p<0.001, p=0.039 Ctr miR vs. C4+miR30d, miR30d vs. C4 siRNA p=0.037), Acta2 (*p<0.001, ^#p<0.001, miR30d vs. C4 siRNA p<0.001) and Postn (*p=0.010, ^#p=0.033, miR30d vs. C4 siRNA p=0.014) mRNA expression evaluated in mouse embryonic fibroblasts (MEF) that were either treated with conditioned media from Ctr siRNA, Ctr miRNA mimic or C4 siRNA, miR30d mimic, or C4 siRNA in addition to miR30d mimic NRVMs (n=3 individual experiments, *p<0.05 vs. Ctr siRNA and Ctr miR, ^#p<0.05 vs. C4 siRNA, one-way ANOVA and Sidak’s post test).

miR30 mediates crosstalk between cardiomyocytes and fibroblasts to control CITED4-dependent fibrosis.

Among the network of miRNAs differentially regulated in C4KO hearts after TAC, four members of the miR30 family were significantly downregulated when compared to littermate controls (Figure 7A, Online Figure VIIA). miR30 family members have been shown to not only provide a useful biomarker predicting the response to therapy in heart failure patients^{9, 14} but also to directly regulate cardiac fibrosis by targeting CTGF and PDGFc in cardiomyocytes and fibroblasts.^{15, 16} We were particularly intrigued by miR30d, as miR30d is highly expressed in cardiomyocytes and packaged in extracellular vesicles,⁹ suggesting it could mediate cross-talk between cardiomyocytes and fibroblasts in our cardiomyocyte-specific genetic model. Therefore, we first confirmed significant downregulation of miR30d in an independent sample of C4KO hearts after TAC (Figure 7C). Additionally, we showed that transgenic cardiomyocyte CITED4 expression increases cardiac miR30d in vivo⁷ (Figure 7C) while miR30d expression was unchanged in hearts from C4KO at baseline (Online Figure IID). To investigate the connection between miR30d and CITED4, we turned to an in vitro model. siRNA-mediated knockdown of CITED4 in neonatal rat ventricular myocytes (NRVM) resulted in downregulation of miR30d expression (Figure 7D), suggesting CITED4 regulates miR30d in a cell autonomous way. We next conducted conditioned media experiments to determine if CITED4 expression regulates crosstalk between cardiomyocytes and fibroblasts. Media conditioned from NRVM in which CITED4 had been knocked down contained reduced exosomal miR30d (Online Figure VIIB) and provoked an increase in extracellular matrix genes CTGF, Col3a1, Col1a1, as well as fibroblast-to-myofibroblast differentiation marker alpha smooth muscle actin (ACTA2) and proliferation marker Periostin (POSTN) expression in neonatal rat ventricular fibroblasts (NRVF) compared to conditioned media from NRVM treated with a scrambled siRNA (Online Figure VIIC). Results were similar in mouse embryonic fibroblasts (MEF) treated with conditioned media from NRVM in which CITED4 had been knocked down in presence of the potent fibroblast activator Transforming growth factor–β1 (TGF-β1) (Online Figure VIID). On the other hand, we observed a trend towards lower pro-fibrotic gene expression in NRVF and MEFs with conditioned media from NRVMs alone when compared to untreated fibroblasts (Online Figure VIIC and D), which was further amplified by transfer of extracellular vesicle media (Online Figure VIIIA). To confirm the role of decreased miR30d for fibroblast activation in C4 knockdown media, we overexpressed miR30d in NRVMs through a miR-mimic and found that it blocked the MEF activation seen with conditioned media from NRVMs in which CITED4 was knocked down (Figure 7E). Exploratory bioinformatic analyses revealed additional potential downstream effectors. Comparing global transcriptional changes in C4KO hearts after TAC to transcriptional changes in response to miR30d knockdown in cardiomyocytes (NRVM), we found Serpine1 and ITGA5 as commonly regulated pro-fibrotic genes, for example (Online Figure VIIIB), while other gene expression changes were unique in either setting (e.g. mitochondrial genes GLUL and UQCRB were only upregulated in C4KO) underscoring the complexity. Differences in the models may also account for pathways that are not common between C4KO hearts after TAC and LNA-miR30d treated NRVMs and present a limitation of this approach.

However, we found some of the pro-fibrotic miRNA network was recapitulated in C4KO mice after swimming. For example, miR-30d and miR-133b (both anti-fibrotic) were mildly decreased, while miR-376c (pro-fibrotic miRNA) was increased (Online Figure VIIIC). These data are consistent with the fibrotic mRNA expression and mildly reduced systolic function seen in C4KO mice after swimming (Figure 2A, 2E).

To further investigate how CITED4 regulates miR30d, we performed proteomic analyses to evaluate the CITED4 interactome in cardiomyocytes. Tandem mass spectrometry after isobaric tagging (iTRAQ)¹⁷ of overexpressed CITED4 in NRVMs revealed CBP/p300 as the most highly enriched CITED4 binding partner relative to control transduced NRVMs (Online Figure IXA). Of note, an enhancer (GH08J134830) proximal to miR30d and known to regulate miR30d expression contains a binding site for p300.¹⁸ Further miRNA promoter sequence analysis using the PROmiRNA tool¹⁹ revealed transcription factor binding sites for p300 and known CITED4-binding partners TFAP2A and p300²⁰ in many of the other differentially expressed miRNAs in C4KO hearts after TAC (summarized in Online Figure IXB). Interestingly, CBP/p300 binding to CITED4 was detected exclusively in the cytoplasm of NRVMs. This could reflect the developmental stage or the quiescent state of the cells studied. Although preliminary attempts failed to identify stimuli that induced nuclear translocation in vitro, such pharmacological treatments obviously do not recapitulate the biology of in vivo pressure overload. Moreover, in collaborative studies, we have previously demonstrated that p300 can acetylate cyclophilin D, thereby altering the mitochondrial permeability transition pore (mPTP) and apoptosis²¹, suggesting a relevant role for cytoplasmic p300. Taken together, these data suggest that CITED4-p300 binding in cardiomyocytes could play a significant role in the phenotypes observed through either regulation of miR30d expression and/or mitochondrial function and apoptosis.

We then asked whether reduced miR30d could contribute the phenotypes observed in C4KO after TAC in vivo by examining the effects of miR30d knockdown using a locked nucleic acid (LNA) antisense inhibitor specific for miR30d. LNA-antimiR30d effectively reduced cardiac miR30d expression (Figure 8A). There was no effect of LNA-antimiR30d on baseline cardiac structure and function (Figure 8B, 8C and data not shown). However, as with CITED4 deletion, depletion of miR30d dramatically accelerated the development of cardiac dysfunction after TAC (Figure 8B) as well as ventricular dilatation (Figure 8C, 8D) and pulmonary congestion (Figure 8E). Also similar to C4KO mice, interstitial fibrosis was increased in miR30d-inhibited hearts compared to control LNA-antimiR-treated animals after TAC (Figure 8F, 8G).

Figure 8: LNA mediated miR30d depletion leads to rapid cardiac failure and fibrosis in response to pressure overload.

A, Confirmation of successful locked nucleic acid (LNA) mediated miR30d depletion, relative miR30d expression in heart lysates from LNA-anti-miR30d and scramble (Scr) control LNA treated mice (n=4 Scr-LNA, 7 LNA-anti-miR30d, Student’s t-test, p<0.001, *p<0.05). B, Assessment of ventricular function by echocardiography in LNA-anti-miR30d compared to Scr-LNA treated mice at indicated time points before and after TAC surgery (n=6 Scr-LNA, 6 LNA-anti-miR30d, p=0.999, 0.999, 0.035, 0.033, respectively, repeated-measures two-way ANOVA with Sidak’s post test for multiple comparisons, *p<0.05). C, Left ventricular dimensions in end-diastole (LVIDd) at indicated time points before and after TAC surgery (n=6 Scr-LNA, 6 LNA-anti-miR30d, p=0.999, 0.922, 0.037, 0.034, respectively, repeated-measures two-way ANOVA with Sidak’s post test for multiple comparisons, *p<0.05). D, Heart weight to tibia length ratios (HW/TL) in LNA-anti-miR30d and Scr-LNA treated animals 4 weeks after TAC surgery (n=6 Scr-LNA, 9 LNA-anti-miR30d, Student’s t-test, p=0.010, *p<0.05). E, Lung weight to tibia length ratios (LW/TL) in LNA-anti-miR30d and Scr-LNA treated animals 4 weeks after TAC surgery (n=6 Scr-LNA, 9 LNA-anti-miR30d, Student’s t-test, p=0.036, *p<0.05). F, Quantification of fibrosis measured as percentage of total myocardial area in LNA-anti-miR30d compared to Scr-LNA treated hearts 4 weeks after TAC surgery (n=6 Scr-LNA, 5 LNA-anti-miR30d, Student’s t-test, p=0.047, *p<0.05) G, Micrographs of cardiac sections stained with Masson Trichrome Staining to label fibrotic tissue 4 weeks after TAC surgery, scale bar=100μm, images were chosen to represent differences of fibrosis pattern in both groups.

DISCUSSION

In this study, we provide evidence that CITED4, previously identified as a member of the gene program activated by exercise,^{3, 6, 7} is also upregulated in response to pressure overload and is necessary for cardiomyocyte growth and to prevent adverse remodeling in both settings.^{3, 6, 7} This suggests that CITED4 is a notable exception to the general rule that the pathways and transcriptional mechanisms involved in physiological and pathological growth of the heart are largely distinct,³ since it appears essential in both settings. Moreover, in the absence of this growth and other CITED4-dependent effects, signs of maladaptive remodeling are seen with both contexts.

Constitutive cardiomyocyte-specific CITED4 deletion did not affect baseline structure or function of the heart in young adult mice. However, in a ramp swimming exercise protocol, C4KO mice developed reduced systolic function and increased ventricular dilation with decreased cardiomyocyte cross sectional area. Maladaptive ventricular remodeling in C4KO after exercise was further supported by increased gene expression markers of cardiac fibrosis (e.g. CTGF) and pathological stress (e.g. BNP). Even more dramatically, C4KO mice developed accelerated heart failure in response to pressure overload induced by transverse aortic constriction (TAC), with ventricular dilatation, increased autophagy and apoptosis. The usual initial hypertrophic response to TAC was absent in C4KO mice. While C4KO heart weights were increased after TAC, this reflected an increase in left ventricular dilation rather than hypertrophy. Given Laplace’s law, failure of left ventricular wall thickening as a compensatory mechanism to reduce wall stress, could contribute to ventricular dilation and ultimately failure.² Interestingly, prior collaborative studies demonstrated that CITED4 inhibits cardiomyocyte elongation in response to hypertrophic agonists, driving the cell morphology towards a pathological pattern in vitro.⁶ Our current studies demonstrate that CITED4 is essential for adaptive cardiomyocyte growth in vivo both in response to exercise and pressure overload.

One likely contributor to impaired cardiac growth after TAC is the observed reduction in mTORC1 activation. CITED4 deletion in cardiomyocytes led to increased expression of the mTORC1 inhibitor REDD1 (DDIT4) in both the TAC and swimming models. mTORC1 is a key regulator of cardiomyocyte hypertrophy, apoptosis, autophagy, and mitochondrial structure.²² Processes potentially underlying accelerated maladaptive ventricular remodeling in C4KO mice were similarly multifaceted. Increased apoptosis and autophagy were also present in C4KO hearts after TAC. In line with these findings, our previous work showed that CITED4 overexpression promoted adaptive ventricular remodeling after myocardial injury, increased mTORC1 activity, and mitigated autophagy and apoptosis.⁷ Of note, gene products of the autophagic pathway were increased in CITED4 depleted hearts after TAC. It has been suggested that apoptosis itself is a causal component of the pathogenesis of heart failure. As few as 23 apoptotic myocytes in 10⁵ were sufficient to cause lethal dilated cardiomyopathy in mice.²³ Expression profiling in NRVMs revealed upregulation of the anti-apoptotic gene Bcl2 in response to forced CITED4 expression.⁷ A direct anti-apoptotic effect has further been confirmed in these cells by showing substantially reduced cardiomyocyte apoptosis after hypoxia-reoxygenation in vitro.⁷ As noted above, the cytoplasmic binding of CBP/p300 by CITED4 could also regulate apoptosis as we have previously shown that p300 can acetylate cyclophilin D, thereby altering the mPTP and apoptosis.²¹

Transcript profiling at an early time-point after TAC further revealed impaired mitochondrial pathways. Many of the genes downregulated in our dataset are downstream targets of PGC1α, which was also repressed in C4KO hearts after TAC when compared to controls. The expression of PGC1α and its target genes is reduced in many mouse models of heart failure.¹² However, CITED4 depletion led to a further decrease of PGC1α target genes when compared to control animals after TAC and in vitro in response to PE.¹² PGC1α is also regulated by mTORC1 signaling in other cellular systems and mediates mTOR’s control of mitochondrial function.¹³ In line with these data, PGC1α knockout mice are able to develop initial hypertrophy in response to TAC, but subsequently develop profound cardiac dysfunction and dilated cardiomyopathy characterized by compromised mitochondrial function and energy deficit.¹² However, we cannot entirely exclude the possibility that the observed downregulation of mitochondrial pathway genes downstream of PGC1α is due to a secondary effect of heart failure rather than a direct effect of CITED4 depletion. Downreguation of these pathways can occur early during pathological remodeling before overt heart failure presents.²⁴ Collectively, these data suggest that CITED4 affects cell growth, survival, and energy metabolism upstream of mTORC1 in cardiomyocytes in response to pressure overload. Interestingly, however, phosphorylation of upstream AKT and TSC2 were upregulated in C4KO mice after TAC, which should lead to increased mTORC1 activity. Previous studies investigating the consequences of decreased mTOR activity in an inducible mTOR knockout mouse model after pressure overload found that even a modest reduction in mTOR activity in the adult myocardium results in dilated cardiomyopathy characterized by apoptosis, autophagy, and altered mitochondrial structure, as seen in C4KO mice after TAC.²² Even though both mTOR complexes were affected in this model, AKT phosphorylation (an mTORC2 substrate) was also increased in mTOR knockout mice.²² In C4KO mice, the increase in AKT phosphorylation may reflect disinhibition of mTORC2 as a result of reduced mTORC1 activity, but could also indicate the presence of another activated kinase, as speculated with mTOR knockout.²⁵ In any event, increased AKT and TSC2 phosphorylation suggest CITED4 regulates mTORC1 downstream of AKT and TSC2. Two potential mediators of this inhibition emerged from our expression profile: REDD1 and REDD2. Both REDD1 (DDIT4)¹⁰ and REDD2 (DDIT4L)²⁶ function as mTORC1 suppressors while simultaneously increasing upstream mTORC2 signaling. REDD1 releases TSC2 from its growth-factor-induced association with inhibitory 14-3-3 proteins irrespective of TSC2’s AKT-dependent phosphorylation. REDD1 is necessary for hypoxia-stimulated mTORC1 inhibition¹⁰ and increased in C4KO mice after TAC. REDD2 is activated by pathological stress in the heart, leading to inhibition of mTORC1 signaling and increased autophagy while upstream signaling is upregulated.²⁶ REDD2 is increased ~2.2fold in C4KO hearts after TAC. Of note, overexpression of REDD2 in turn leads to decreased CITED4 expression.²⁶ Further in vitro experimentation showed REDD1 to be upregulated in ventricular myocytes at baseline and more pronounced in response to phenylephrine stimulation, indicating an effect of CITED4 on REDD1 independent of established cardiac injury in vivo. REDD2 was predominantly upregulated in vivo in C4KO mice and not in vitro. One limitation of our study is that our data do not prove a causal role for changes in mTORC1 and mitochondrial pathways. However, mTORC1-inhibition and a reduction in PGC1α signaling occur earlier and are exaggerated in C4KO mice, suggesting they likely contribute to the accelerated ventricular dysfunction seen in C4KO mice after TAC. While the phenotype observed in C4KO mice after TAC shows similarity to mTOR knockout mice as noted above,²² we acknowledge that mTOR manipulation can causes different phenotypes depending on important technical differences in the model investigated, as well as the specific molecular manipulation, timing, and degree of activation/inhibition.²⁵ Finally, in C4KO, these alterations in mTOR signaling also occur in the context of other important gene network changes that likely act in combination to produce the phenotypes observed.

Of particular interest was the profound fibrosis that developed in the ventricular wall in C4KO hearts after TAC. While this could be a consequence of increased mechanical stress or increased tissue loss through apoptosis and reduced cell size requiring tissue replacement,²⁷ results from small RNAseq profiling led us to further investigate fibrosis related pathways in C4KO mice after TAC. Over 70% of differentially expressed miRNAs were linked to organ fibrosis. Many of those miRNAs had predicted or validated target genes that were inversely altered in the transcriptome of C4KO hearts, including Col3a1, Col8a1, Col5a1, TGFb2, CTGF and PDGFc. Interestingly, among the most differential miRNAs were members of the miR133 and miR30 family, both consistently downregulated in several models of pathological hypertrophy and heart failure and previously shown to regulate fibrosis, including the expression of CTGF in fibroblasts.¹⁵ miR30d is highly expressed in cardiomyocytes and packaged in extracellular vesicles,⁹ suggesting it could mediate crosstalk between cardiomyocytes and fibroblasts. We turned to an in vitro model and investigated whether CITED4-dependent miR30 reduction in NRVMs and consequently reduced exosomal miR30d would influence fibroblast extracellular matrix gene expression levels. Indeed, cardiomyocyte CITED4 silencing led to reduced miR30d expression and further experiments revealed that conditioned media from these cells provoked an increase in fibroblast CTGF and collagen levels. Alpha smooth muscle actin (acta2) as a marker for fibroblast to myofibroblast transformation and periostin (postn) as a marker for fibroblast proliferation were also increased in fibroblasts after treatment with conditioned media from CITED4 depleted cardiomyocytes. Inducing miR30d in the context of CITED4 knockdown in NRVMs antagonizes fibroblast activation induced by CITED4 knockdown media alone. These results suggest that miR30d mediates crosstalk between cardiomyocytes and fibroblasts dependent on CITED4, and could contribute to the increased fibrosis seen in vivo though likely in concert with other miR-30 family members and pro-fibrotic miRNA changes observed. In support of the relevance of our findings, miR30d depletion in vivo, led to rapid heart failure and cardiac fibrosis in a murine TAC model similar to that seen in C4KO mice.

As noted, while we have here demonstrated the functional relevance of miR30d, our in vivo results show that cardiac loss of CITED4 disturbs a whole panel of miRNAs involved in the regulation of fibrosis and previously linked to fibrotic diseases. For example, a role in cardiac fibrosis has been established for miR21, miR133, miR708, miR101, and miR29,^{15, 28–35} miRNAs that were also differentially regulated in C4KO hearts after TAC. Therefore, it is unlikely that any one miRNA accounts fully for the fibrosis seen in C4KO hearts after TAC. Interestingly, some significantly downregulated miRNAs, including miR30 family members, and miR101a, and miR181, are predicted to target REDD1 and could thereby also influence the cell autonomous effect in C4KO cardiomyocytes as described above. Moreover, some miRNAs in our dataset share promoter sequences with regulatory elements and gene enhancers that bind CITED4 (e.g. EP300, TFAP2). Further investigation is warranted to understand how CITED4 regulates the expression of miRNAs.

We conclude that CITED4 is upregulated in response to both physiological and pathological stimuli, and is necessary for adaptive cardiomyocyte growth as well as to prevent adverse remodeling in both settings. These findings reveal an important role of CITED4 as a central regulator of cardiac homeostasis in the adult heart and a common node controlling cardiac growth, adaptation, and fibrosis in vivo.

NOVELTY AND SIGNIFICANCE.

What Is Known?

• Exercise promotes cardiovascular health and mitigates established disease.

• Regular exercise leads to physiological cardiac growth, while diseases like hypertension and aortic stenosis, for example, lead to pathological growth.

• CITED4 is part of a gene program induced by exercise and is sufficient to cause physiological cardiac growth, mimicking the exercise response of the heart, and can mitigate adverse cardiac remodeling after ischemic injury.

What New Information Does This Article Contribute?

• CITED4 is upregulated in both physiological and pathological cardiac growth and is necessary to prevent adverse remodeling in both settings.

• Cardiomyocyte CITED4 knockout (C4KO) leads to rapid heart failure characterized by ventricular dilation, apoptosis, autophagy, and cardiac fibrosis in a murine model of pathological cardiac growth.

• CITED4 protects against pathological cardiac remodeling by regulating mTOR (mammalian target of rapamycin) activity and a network of miRNAs mediating cardiomyocyte to fibroblast crosstalk

Although the beneficial preventive and therapeutic effects of exercise are well established, the underlying molecular mechanisms are not well understood. Cardiac CITED4 was previously shown to be induced by exercise and sufficient to cause physiological hypertrophy and mitigate adverse remodeling after ischemic injury. However, the role of endogenous CITED4 in response to physiological or pathological stress was unknown. We investigated the role of CITED4 using a cardiomyocyte-specific knockout model. We found that CITED4 increased not only in response to exercise but also in response to pathological stress, making it a notable exception to the general rule that the pathways involved in physiological and pathological growth of the heart are largely distinct. While cardiac morphology and function were normal at baseline in young adult C4KO mice, they manifested a maladaptive response to exercise. After pressure overload, C4KO mice showed impaired cardiac hypertrophy, decreased activation of mTOR and mitochondrial signaling, increased autophagy, apoptosis, and fibrosis. CITED4 regulates a network of pro-fibrotic miRNAs, and one of these, miR30d, appeared to mediate cardiomyocyte-fibroblast cross talk dependent on CITED4. These data demonstrate that CITED4 is upregulated in both pathological and physiological remodeling and in both settings protects against maladaptive remodeling.

Nonstandard Abbreviations and Acronyms:

C4KO

CITED4 Knockout

GO

Gene Ontology

GSEA

Gene Set Enrichment Analysis

KEGG

Kyoto Encyclopedia of Genes and Genomes

MEF

Mouse embryonic fibroblasts

miR/miRNA

Micro RNA

NRVF

Neonatal rat ventricular fibroblasts

NRVM

Neonatal rat ventricular myocytes

PE

Phenylephrine

siRNA

Small interfering RNA

TAC

Transverse aortic constriction