Hypertrophic cardiomyopathy and dysregulation of cardiac energetics in a mouse model of biliary fibrosis

Abstract

Cardiac dysfunction is a major cause of morbidity and mortality in patients with end-stage liver disease, yet the mechanisms remain largely unknown. We hypothesized that the complex interrelated impairments in cardiac structure and function secondary to progression of liver diseases involve alterations in signaling pathways engaged in cardiac energy metabolism and hypertrophy, augmented by direct effects of high circulating levels of bile acids. Biliary fibrosis was induced in male C57BL/6J mice by feeding a 0.1% 3, 5-diethoxycarbonyl-1, 4-dihydroxychollidine (DDC) supplemented diet. After 3 weeks, mice underwent live imaging (DEXA scanning, 2DEcho, EKG, cardiac MRI), exercise treadmill testing and histological and biochemical analyses of livers and hearts. Compared to chow-fed mice, DDC-fed mice fatigued earlier on the treadmill, with reduced VO₂. Marked changes were identified electrophysiologically (bradycardia and prolonged QTc) and functionally (hyperdynamic left ventricular (LV) contractility along with increased LV thickness). Hearts of DDC-fed mice showed hypertrophic signaling (activation of AKT, inhibition of GSK3β and a 20-fold upregulation of β myosin heavy chain RNA) and elevated G_sα/G_iα ratio. Genes regulating cardiac fatty acid oxidation pathways were suppressed, along with a 3-fold increase in myocardial glycogen content. Treatment of mouse cardiomyocytes (which express the membrane bile acid receptor TGR5) with potent natural TGR5 agonists, taurochenodeoxycholic acid and lithocholic acid, activated AKT and inhibited GSK3β, similar to the changes seen in DDC-fed mouse hearts. This provides support for a novel mechanism whereby circulating natural bile acids can induce signaling pathways in heart associated with hypertrophy. In conclusion, 3 weeks of DDC feeding-induced biliary fibrosis leads to multiple functional, metabolic, electrophysiological and hypertrophic adaptations in the mouse heart recapitulating some of the features of human cirrhotic cardiomyopathy.

Cardiac dysfunction in cirrhosis (cirrhotic cardiomyopathy) is characterized by multiple electrical, physiological and structural responses in the heart including left ventricular (LV) hypertrophy with impaired relaxation, baseline hyperdynamic contractility, conduction abnormalities and an attenuated cardiovascular response to stress(1). The development of cirrhotic cardiomyopathy increases the risk of death in patients with cirrhosis and is associated with increased risks of hepato-renal syndrome, serious post-operative complications and dysrhythmias (2, 3). In addition to increased mortality, cardiomyopathy is a co-morbid contributor to exercise fatigue in patients with cirrhosis (4, 5).

Rodent models of liver injury (mainly common bile duct ligation [CBDL] and carbon tetrachloride treatment) have revealed critical insights into some of the molecular mechanisms underlying cirrhotic cardiomyopathy. Increased cholesterol/phospholipid ratio within the cardiomyocyte cell membrane, impaired β-adrenergic receptor (β-AR) function and abnormal gene expression of β- adrenergic signaling pathway members contributes to attenuated contractile response to inotropic stress in cirrhosis (6, 7). In addition, carbon monoxide, hemoxygenase, nitric oxide, endogenous cannabinoids, bile acids (BA) and endotoxins have been implicated in the pathophysiology of diminished cardiac function in cirrhotic rodents(8). Though these mediators help explain some of the attenuated cardiac contractility in cirrhosis, the causes of hyperdynamic contractility, hypertrophy and rhythm disturbances, key features of baseline cardiac phenotype in cirrhosis in adults and children, remain essentially unknown(9, 10).

Alteration in cardiac energy metabolism, especially substrate utilization, plays a major role in the physiology of cardiac function in health, disease and hypertrophy. Often metabolic changes precede structural and functional changes in the heart (11, 12). Dysregulation in cardiac energetics has been proposed as one of the potential mechanisms underlying cirrhotic cardiomyopathy.(13) Detailed investigations of whole body energetics, characterization of cardiac hypertrophy and altered metabolism in human and animal models have been limited and could provide a critical insight into the pathophysiology of cirrhotic cardiomyopathy. In addition, it is likely that several pathways are engaged prior to the development of cirrhosis, although the timeline and mediators remain to be determined.

The goal of this study was to characterize the live mouse cardiac phenotype and explore the extrahepatic molecular consequences in the heart of a recently described mouse model of biliary fibrosis (14). We found multiple impairments in cardiac function and structure in mice with biliary fibrosis, along with distinct changes in cell signaling pathways leading to hypertrophy and altered metabolism. Finally, evidence is provided that supports a contributing role for elevated circulating levels of bile acids to altered cardiac signaling pathways in biliary fibrosis.

MATERIALS AND METHODS

Animals and Diet

Biliary fibrosis was induced by feeding 6-8 week old male C57BL6J mice (Jackson Labs; Bar Harbor, ME.) with 0.1% 3, 5-diethoxycarbonyl-1, 4-dihydroxychollidine (DDC) (Sigma-Aldrich. St. Louis, MO.) supplemented chow (Harlan-Teklad Inc, Madison, WI) for 3 weeks. Age matched male C57BL6J mice fed isocaloric chow were used as controls for all experiments. Mice were weighed twice a week. Feeds and bedding were weighed and changed twice a week to provide an estimate of food intake. Mice were fed ad libitum and had free access to water. Food was withdrawn 4 hours prior to experiments and all experiments were done in accordance with IACUC approved protocols at Baylor College of Medicine (BCM).

Anthropometric measurements

After 3 weeks DDC and control mice underwent dual energy x-ray absorptiometry (DEXA) scanning (GE-Lunar PIXImus, Madison, WI.) to evaluate lean mass, fat mass, bone mineral density, bone mineral content and percent fat content (Mouse Phenotype Core facility (MPC), BCM, Houston).

Exercise Tolerance and Energy Expenditure

After 3 weeks mice were exposed to a standard exercise protocol on a treadmill integrated with metabolic chamber (Oxymax Deluxe VO₂/VCO₂ System Columbus Instruments, Columbus, OH) (See Supplement). Indirect calorimetry was performed at baseline, during exercise and for 5 minutes after treadmill was stopped. Oxygen consumption (VO₂), carbon dioxide production (VCO₂) and Respiratory Exchange Ratio (RER) were analyzed from the gas sampled (MPC at BCM, Houston).

Cardiac Parameters

Continuous electrocardiograms (EC Genie (Mouse Systems Inc, Quincy MA) and tail-cuff blood pressures ( 6 channel NIBP system with BPMONWIN software [IITC Life Science Inc. CA, USA]) were recorded noninvasively in unsedated mice. Two-dimensional echocardiography (2DE) was performed in MPC (BCM) on sedated mice (Vevo 770 Digital RF, VisualSonic Inc. Toronto, CN). Cardiac MRI (Bruker Biospin, Billerica, MA) was performed in sedated mice and analyzed by a blinded investigator as described previously (15) (See Supplement).

Serum analyses

Sera and anti-coagulated blood were collected from the inferior vena cava (IVC) and analyzed for alanine aminotransferase (ALT), aspartate aminotransferase (AST), Lactate Dehydrogenase (LDH), total and conjugated bilirubin levels and Complete Blood Count (CBC) respectively (Cobas Integra 400+; Roche) at Center of Comparative Medicine (BCM). Serum bile acid levels and non-esterified fatty acid levels (NEFA) were evaluated by colorimetric methods (BioQuant Inc, San Diego, CA. & Roche). Serum leptin and TNFα levels were measured using commercial ELISA kits (RayBiotech Inc. Norcross, GA).

Cardiac and liver histology

Hearts and liver were rapidly isolated and flash frozen in liquid nitrogen, or used for subsequent histological analyses - routine H/E and Mason–Trichrome (for collagen). OCT embedded frozen sections were stained with Oil Red-O (for lipids) and Periodic Acid Schiff (PAS) for glycogen. Immunohistochemistry for TGR5 was performed on frozen hearts using rabbit polyclonal antibody (Sigma-Aldrich; St. Louis, MO.) at 1:200 dilution. Liver sections were stained with routine H/E. All histological studies were performed by the Texas Gulf Coast Digestive Disease Center. Myocardial glycogen content was compared enzymatically as described previously (16) and histologically using image analysis (see Supple1ment) .

Quantitative real time PCR (qRTPCR)

RNA levels were analyzed by qRTPCR (primer sequences in supplement Table 1.s), as described before (17) using SYBR® Green™ (Applied Biosystems, Foster city, CA.). Relative RNA expression was calculated by delta Ct method. Expression of target genes were normalized to internal standard (GAPDH) and reported as fold change compared to chow.

Immunoblotting

Proteins from homogenized whole hearts and membranes were extracted using standard procedures(17). After gel electrophoresis and immunoblotting, the gels were analyzed for expression of various proteins with specific antibodies (Supplement). Equal protein loading was confirmed by α-tubulin (Sigma-Aldrich, St. Louis, MO) for whole hearts or Sodium-Potassium ATPase plasma membrane antibody (Abcam, Cambridge, MA) and Ponceau staining for membrane extracts. Results were analyzed by densitometry (Kodak software, Scientific Imaging Systems; New Haven, CT.) and reported as fold change compared to chow.

Isolation of neonatal cardiomyocytes and bile acid treatments

Cardiomyocytes were isolated from 2-3 day old neonatal C57BL6J mouse pups (Cellutron Life Sciences, Baltimore, MD) as described previously(18). (Supplement) Isolated cardiomyocytes (0.5×10⁶cells/well) were incubated with 100μM sodium taurochenodeoxycholic acid (TCDCA), 10μM sodium lithocholic acid (LCA) (Sigma-Aldrich, St. Louis, MO) or vehicle (DMSO). After 4 hours, experiments were terminated with ice-cold PBS, cells were lysed and proteins extracted and analyzed by immunoblotting. Each experiment was performed in duplicate and repeated three times (n=3). For each set of experiment, cardiomyocytes were isolated and pooled from 15-20 neonatal hearts.

Statistical Analysis

All data are presented as Means ± SD (unless specified). Data was analyzed using non-parametric Mann-Whitney test (unless specified). All statistical calculations were done using the PRISM 5.0 software program (Graph-Pad Prism, San Diego, CA, USA). P < 0.05 was considered significant.

RESULTS

DDC-fed mice exhibit early fatigue, decreased whole body oxygen consumption (VO₂) and increased Respiratory Exchange Ratio (RER)

Three weeks of DDC-feeding led to biliary fibrosis and changes in serum indices of liver disease consistent with those noted by other investigators(14) (Supplement Fig.2s). DDC-fed C57BL/6J mice lost weight and demonstrated changes in body composition mainly ~ 30% decrease in lean and fat mass, without a change in percent fat or bone mineral content (Supplement Fig 1s). On the metabolic treadmill, DDC-fed mice were exhausted earlier than their chow-fed counterparts (Fig.1A). In addition, indirect calorimetry performed continuously during exercise showed that DDC-fed mice have decreased VO₂, decreased peak VO₂, and an increased RER (RER = VCO₂/VO₂) (Fig.1B and D). Thus, for whole body composition and exercise parameters, DDC-fed mice have several key similarities to those seen in patients with biliary cirrhosis (4, 5).

Fig.1. DDC-fed mice have limited tolerance for exercise and altered oxygen utilization.

Mice were challenged with acute maximal exercise on a treadmill. Time to exhaustion, (VO₂), (VCO₂) and (RER) were measured by indirect calorimetry as reported in Methods. DDC fed mice demonstrated early fatigue (A), lower VO₂ (B) and high RER (D) when compared to chow fed mice at each time point. VCO₂ did not did not differ between the groups. (*p<0.05; Results: Mean±SEM; n=6; Mann-Whitney for each time point). Note: Chow (□) and DDC (■)

The hearts of DDC-fed mice are hypertrophic and exhibit altered electrical and functional parameters

Unsedated DDC-fed mice exhibited bradycardia, prolonged QT intervals (QTc) and higher systolic BP (Fig.2B). Two-dimensional echocardiography (2DEcho) revealed hyperdynamic contractility evidenced by ~ 25% increase in both ejection fractions (%EF) and shortening fractions (%FS) (Fig.2A and B), similar to those seen in patients with cirrhosis or in infants with biliary atresia and portal hypertension (5, 9, 10). Cardiac MRI of DDC-fed mice showed ~ 25% increase in both LV and RV ejection fractions compared to chow-fed hearts (Table 1). Cardiac mass index, LV free wall and inter-ventricular septal thickness were significantly increased in DDC-fed mouse hearts (Fig. 3A and Table 1). LV stroke volume index tended to be higher in the DDC fed mice (Table.1).

Fig.2. DDC-fed mouse hearts have altered electrocardiographic, echocardiographic and hemodynamic parameters.

(A) denotes representative M-mode cardiac 2DEcho pictures of chow fed and DDC fed mice showing hyperdynamic LV in DDC fed mice. Bar graph in (B) denotes heart rate (HR), QT interval, corrected QT interval (QTc) as analyzed by rhythm strips, systolic (SBP) and mean (MBP) blood pressures as calculated in unsedated mice using tail-cuff, shortening fractions (%FS), ejection fractions (%EF) and calculated end diastolic volume (LVEDV) of the left ventricle by 2DEcho. (* p<0.05; n=6). Note: Chow (□) and DDC (■)

Table. 1.

Cardiac parameters from MRI: Table depicts values obtained for the respective cardiac parameters after quantitative analysis of cardiac MR. (Statistics: Students t test)

CARDIAC PARAMETERS	CHOW (n=5) (Results±SD)	DDC (n=5) (Results±SD)	P value
LV Ejection Fraction (%)	48.03±4.7	66.9±3.7	0.0002
RV Ejection Fraction (%)	49.7±1.35	62.12±4.5	0.007
Posterior Wall thickness (mm)	0.71±0.04	0.88.±0.035	0.0001
Septal thickness (mm)	0.92±0.06	1.168±0.036	0.0001
Cardiac Mass (Normalized to body wt)	0.0021±0.0003	0.0037±0.0007	0.008
LVEDV (ml)	0.054±0.004	0.043±0.009	0.0547
LVESV (ml)	0.025±0.006	0.014±0.004	0.009
LV Stroke Volume Index (ml/g)	1.44±0.1	1.75±0.3	0.07
RV Stroke Volume Index (ml/g)	1.10±0.14	1.4±0.34	0.13

Fig.3. Cardiac hypertrophy and hypertrophic signaling in the DDC-fed heart.

(A) Representative pictures of cardiac MR short axis mid-ventricular slices in diastole shows chow fed hearts (CH) on the left and DDC fed hearts on the right. Concentric LV hypertrophy in DDC hearts is evident on these images showing increased LV posterior wall and septal thickness. This was verified by quantitative analysis of the posterior wall, septal thickness and LV mass normalized to the body weight (see Table.1). (B) shows qRTPCR analysis of key genes involved in cardiac hypertrophy. Note upregulation of β-MyHC, BNP and eEF2. (C) shows representative immunoblots for AKT, Ser473-phospho-AKT, GSK3β and Ser9-phospho-GSK3β with α-tubulin (loading control). (D) shows densitometric analysis of the respective bands normalized to α-tubulin and analyses of the degree of phosphorylation (pAKT/AKT and pGSK3b/GSK3b). (*p<0.05; n=5-7) Chow (□) and DDC (■)

DDC-induced biliary cirrhosis leads to multiple alterations in cardiac hypertrophy signaling pathways and expression of relevant target genes

Central among the many changes associated with cardiac hypertrophy in a variety of mouse models are activation of v-akt murine thymoma viral oncogene/Protein kinase B (AKT) and inhibition of Glycogen Synthase Kinase- 3β (GSK3β) (19). DDC feeding resulted in a 1.5 fold increase in cardiac levels of phospho-Thr308-AKT levels (activation) and a 1.75 fold increase in phospho-Ser9-GSK3β (inhibition) (Fig.3C and D), both associated with increased expression of genes responsible for cardiac hypertrophy(20). Expression of several, but not all key genes involved in cardiac hypertrophy were upregulated: notably β myosin heavy chain (β MyHC; 20-fold), brain natriuretic peptide (BNP; 2-fold) and elongation factor eEF2 (1.8-fold) (Fig. 3B and Supplement Fig. 3s A).

β-adrenergic receptor signaling

Several researchers have identified altered β-adrenergic signaling in cirrhosis, linked to reductions in either β-adrenergic receptor (βAR) expression (7) or impairments in post-receptor signaling pathways (6). Myocardial contractility is governed in large part by signaling through βAR-1 and βAR-2 and their associated co-regulatory G_sα and G_iα subunits. DDC-feeding did not change RNA or protein expression of either β-ARs (Fig.4). However, a significant suppression of G_iα, with a modest activation of G_sα, led to a 2-fold increase in the G_sα/G_iα ratio.

Fig.4. Elevated G_sα/ G_iα ratios in the hearts of DDC-fed mice.

(A) shows QPCR results of βAR-1 and βAR-2 RNA standardized to GAPDH and (B) depicts densitometric analysis of βAR-1and βAR-2 protein expression normalized to α-tubulin (C) Immunoblot pictures of whole heart incubated with antibodies for G_sα, G_iα and α-tubulin. (D) shows densitometric analysis of the respective bands normalized to α-tubulin. (*p<0.05; n=5-7) Note: Chow (□) and DDC (■)

Glycogen accumulation and altered fatty acid metabolism in the hearts of DDC-fed mice

Cardiac histology revealed increased staining with PAS (Fig.5A). Quantitative image analysis and biochemical analysis of cardiac tissue showed a 3-fold increase in myocardial glycogen content (Fig 5B and C) after DDC feeding. Both GLUT-1 and GLUT-4 membrane protein levels, which are responsible for the majority of cellular glucose uptake, were increased in DDC-fed mouse hearts. RNA levels of GLUT-1, but not GLUT-4 were elevated (Fig. 5D and E). Serum free fatty acids (a major energy substrate for a healthy resting heart), and leptin levels (a key regulator of cardiac fatty acid oxidation [FAO]) (21) were, respectively, 60% and 80% reduced in DDC-fed mice (Fig. 6A). Expression of heart-type fatty acid binding protein (h-FABP/FABP-3), mitochondrial-carnitine palmitoyl transferase-2 (mCPT-2), short chain acyl-coA-dehydrogenase (s-ACAD) and uncoupling protein-3 (UCP-3) - key genes responsible for cardiac FAO, were suppressed in DDC-fed mice (Fig.6B). CD36, mCPT-1, medium-ACAD, long-ACAD and very long-ACAD RNA levels did not significantly change with DDC feeding (data not shown). DDC-feeding induces activation of cardiac acetyl-CoA carboxylase (ACC, via a ~50% decrease in phospho-ACC, Fig 6D) which has been shown to contribute to suppression of FAO (22). AMPKα protein levels were unchanged by DDC feeding (Fig.6C). Finally, UCP-3 which is involved in cardiac FAO (23) and remodeling (24), was ~ 50% suppressed after DDC-feeding at both RNA and protein levels (Fig. 6B and D). This reduction in FAO was not linked to changes in myocardial lipid accumulation (Fig. 6E).

Fig.5. Increased glucose transporter expression and glycogen deposition in hearts of DDC-fed mice.

Representative frozen sections (A) of heart (n=5/group) after PAS staining shows increased glycogen deposition in DDC hearts compared with chow fed hearts. Myocardial glycogen content was quantified enzymatically (B) and by blinded colorimetric image analysis to quantify percent staining (C) by PAS. (D) denotes representative immunoblot results of equally loaded isolated membrane protein samples probed with antibodies for GLUT-1 and GLUT-4 proteins. (E) shows gene expressions of GLUT-1 and GLUT-4 standardized to GAPDH and (lower row) densitometric analysis of GLUT-1 and GLUT-4 bands. Equal loading confirmed by Sodium-Potassium-ATPase membrane protein and Ponceau staining. (*p<0.05; n=5; and # p<0.05; n=3, Students t test.) Note: Chow (□) and DDC (■)

Fig. 6. Reduced fatty acid oxidation in hearts of DDC-fed mice.

(A) shows decreased levels of serum NEFA levels and leptin levels in DDC fed mice. (B) shows QPCR results of key genes regulating fatty acid oxidation. (C) shows immunoblots of membranes incubated with antibodies for AMPKα and Thr172-phospho-AMPKα with bar graph below showing densitometric analysis of the respective bands standardized to α-tubulin and analysis of the degree of phosphorylation of AMPKα. (D) shows membranes incubated with antibodies for ACC, Ser79-phospho-ACC and UCP-3 and α-tubulin as loading control with bar graph showing densitometric analysis of the respective bands standardized to α-tubulin and analysis of the degree of phosphorylation of ACC. (*p<0.01; n=5) Note: Chow (□) and DDC (■). (E) shows representative pictures of Oil-Red-O stained frozen sections of the hearts showing no evidence of lipid accumulation in the DDC fed cardiac muscle.

TGR5 and bile acid signaling in mouse heart

As a first step to identify the molecular mediators of altered cardiac myocyte function in biliary fibrosis, we focused upon exploring roles for bile acids (BA) and the recently discovered G protein coupled receptor (GPCR) for BAs -TGR5(25). DDC-fed mice have ~50-fold elevated circulating levels of serum bile acids compared to chow-fed mice (Fig 7E), at concentrations capable of activating TGR5(26). When activated by extracellular BAs, including taurine-conjugated primary BAs, this GPCR initiates G protein containing signaling pathways in several metabolically-active tissues(27). However, little is known for a role for cardiac TGR5. Immunoblot and qRTPCR analysis of RNA indicates that TGR5 is present in mouse heart and isolated cardiomyocytes (Fig.7A and B). Immunohistochemistry indicates its presence in both endothelium and cardiomyocytes (Fig 7C). Cardiac TGR5 RNA and protein expression did not change in DDC fed mouse hearts (Fig.7 D). To study if there are direct effects of BAs (natural TGR5 agonists) on cardiomyocytes, we exposed neonatal mouse cardiomyocyte cultures to the TCDCA (EC₅₀ for TGR5 of 1.9 μM) and LCA (EC₅₀ for TGR5 of 0.58μM)(26). After incubation for 4 hours, TCDCA (100μM) as well as LCA (10μM) activated AKT and inhibited GSK3β, similar to the changes in-vivo after DDC feeding (compare Fig 7G to Fig 3).

Fig. 7. TGR5 and role for bile acids.

(A) shows immunoblot of 4 whole heart tissues [H] showing a band detected at 37kDa using TGR5 antibody with spleen (S) as a positive control. (B) shows PCR analysis of flash frozen whole heart (H), isolated neonatal cardiomyocytes (CM), spleen [S] as positive control and water [W] used as a negative control. Note TGR5 band at 126 base pair besides molecular weight marker (M). (C) shows presence of TGR5 (red staining) in the OCT fixed flash frozen hearts by immunohistochemistry using rabbit polyclonal antibody for TGR5 (Bar=20 μm thick). No staining seen in the isotype control stained heart. (D) shows bar graph comparing QPCR results of TGR5 RNA (standardized to GAPDH) and TGR5 protein expression (normalized to α-tubulin) between chow (□) and DDC fed (■) groups and expressed as fold change compared to chow group. (E) depicts circulating serum bile acids in DDC fed mice. (*p<0.05; Results: Mean±SD; (n=5) Note: Chow (□) and DDC (■)) (F) shows representative immunoblot results from n=3 experiments of equally loaded protein samples from isolated neonatal cardiomyocytes (0.5×10⁶ cells/well) incubated with DMSO as vehicle (Con), TCDCA (100μM) and LCA (10μM) for 4 hours. Membranes were incubated with antibodies for AKT, Ser473-phospho-AKT, GSK3β and Ser9-phospho-GSK3β with α-tubulin as the loading control. (G) depicts bar graphs showing densitometric analysis of the degree of activation (phosphorylation) of AKT (p-AKT/AKT) and inhibition (phosphorylation) of GSK3β (p-GSK3β/GSK3β). (*p<0.05; Statistics:Mann-Whitney as compared with DMSO control; n=3).

DISCUSSION

Extra-hepatic consequences of cirrhosis such as cirrhotic cardiomyopathy strongly impact upon morbidity and mortality(8). In this study, we report that the DDC-fed model of biliary fibrosis, cholestasis and liver injury, recapitulates many of the cardiac abnormalities seen in human cirrhotic cardiomyopathy. In addition, changes were identified in multiple signaling pathways which could potentially provide insight into the development of cirrhotic cardiomyopathy. DDC-fed mice have multiple alterations in exercise physiology (early fatigability, decreased VO₂ and increased RER), cardiac rhythm (bradycardia and prolonged QT interval), cardiac morphology (LV and septal hypertrophy) and cardiac function (increased LV ejection and shortening fractions) which model most of the essential elements of overt cardiac dysfunction in cirrhotic patients(5, 8, 9, 28, 29). At the molecular level, these hearts exhibit activation of AKT and inhibition of GSK3β pathways: both of which contribute to hypertrophic and hyperdynamic changes in the heart(30-35). Though cardiac expression of β-ARs did not change, G_iα was significantly reduced with a 2-fold increase in G_sα/G_iα ratio; altering βAR signal transduction. (36)

DDC-feeding leads to suppression of key regulators of cardiac FAO, enhanced expression of membrane glucose transporters and a 3 fold increase in glycogen content suggesting a novel dysregulation in cardiac energetics and metabolism. Finally, a direct activation of AKT and inhibition of GSK3β is seen in isolated cardiomyocytes cultures by TCDCA and LCA (potent natural TGR5 ligands), supporting a potential direct role for high levels of circulating BAs.

2DE and cardiac MRI of DDC-fed mice show increased thickness of the LV free wall and interventricular septum, consistent with findings in hearts of adults (37), as well as infants (9, 10) with biliary cirrhosis. Several animal models of liver injury (carbon tetrachloride(38), CBDL (39), and now the DDC feeding) lead to LV hypertrophy, which, in turn, can lead to diastolic dysfunction and the inability to compensate for increased demands (e.g., systolic dysfunction when stressed). From a cell signaling perspective, several critical mediators have emerged as central to cardiac hypertrophy--AKT activation and GSK3β inhibition.(40). Chronic activation of AKT leads to myocardial hypertrophy in mice via mechanisms including either inhibition of GSK3β(19, 20) or activation of p70S6K via mTOR(35). Transgenic mice overexpressing cardiac AKT demonstrate hypertrophy, hypercontractility and an attenuated contractile response to dobutamine (30) which mirror the cardiac impairments seen in cirrhotic patients. In the DDC-fed mouse model, the heart displays several phenotypic and molecular features of physiologic hypertrophy such as: hypercontractile LV, AKT activation and upregulation of eEF2(41) as well as characteristics of pathologic hypertrophy such as “re-expression” of fetal growth genes - β-MyHC and BNP(42). Whether these changes reflect adaptive or maladaptive responses to liver injury/biliary fibrosis, or response to vascular changes remain to be seen. Importantly, whether or not the remodeling is reversible in the DDC model as seen in patients with cirrhotic cardiomyopathy after liver transplantation, remains to be determined.

Cardiopulmonary exercise testing of cirrhotic patients revealed early fatigue, lower VO₂, and lower maximal VO₂ (4, 5). DDC-fed mice showed similar physiological changes to humans throughout the course of exercise treadmill testing. Though not statistically significant, these mice demonstrated higher VCO₂ which also contributed for higher RER. High RER is often seen when carbohydrates are oxidized preferentially over lipids(43), suggesting that substrate utilization by metabolically active tissues, including the heart, is impaired in these mice.

A resting heart utilizes circulating fatty acids for 60-70% of its energy needs while glucose and lactate supply the remainder. The heart switches its substrate preference from fatty acids to glucose as a compensatory mechanism under stressors like congestive heart failure, pathologic hypertrophy, and cardiomyopathy(11, 12). Histologic and biochemical evidence suggest a switch in cardiac substrate utilization favoring glycogen accumulation and reduced fatty acid oxidation in DDC fed mice(Fig. 5 and 6). The presence of glycogen in the hearts of this model is surprising and intriguing, and may be a contributor to the structural and functional alterations, rather than an innocent bystander. It is possible that in the DDC model, cardiomyocytes store glycogen as endogenous fuel in the face of shortage of exogenous fuel (fatty acids) to maintain vital functions as a survival mechanism. Energy efficiency (ATP/mole O₂) improves when heart “switches” from fats to glycogen as preferred fuel for respiration(44). This could explain the hypercontractility in an “un-stressed state”. During catecholamine stress as in exercise, this endogenous glycogen is rapidly depleted leading to decompensation--a hallmark of cirrhotic cardiomyopathy. Accumulation of minimal amounts of glycogen as seen in mice over-expressing PRKAG2 leads to hypertrophy and increased contractility, while excessive glycogen deposition leads to severe cardiac dysfunction(45-47). At a molecular level, accumulation of glycogen can be explained by activation of AKT with resultant inactivation of GSK3β(44). At a physiologic level, myocardial glycogen content is increased in fasting or as an adaptive response to stress “hibernation”(44). . Overall, detailed mechanisms and consequences underlying the marked alterations in cardiac metabolism, along with changes in cell signaling pathways and their physiologic relevance in the development of cardiomyopathy in this liver injury model await further study.

High circulating levels of BAs, in the order of fifty to hundred-fold higher than normal (48) are one of the central components of cirrhosis, and biliary cirrhosis in particular. The primary BAs such as TCDCA and LCA, whose levels are elevated in cirrhosis, activate AKT and inhibits GSK3β in isolated cardiomyocyte cultures at concentrations seen in cirrhosis (Fig 7), supporting a potential novel role in mediating hypertrophy, hypercontractility and metabolic changes. BAs are multifunctional molecules, that have recently been identified as true hormones(49), with intracellular (cell signaling cascades, nuclear receptors) and extracellular functions targeting cell surface receptors (TGR5) (25). BA nuclear receptors and BA transporters are poorly expressed in the myocardium, but TGR5 is present (fig 7)(50, 51). Thus, there is likely a pathway for circulating BAs to signal to cardiomyocytes extracellularly, through TGR5 (52). These data along with others (53, 54) are among the first to identify cardiomyocytes as a direct target for BA signaling, suggesting a strong role for BAs in the pathophysiology of cirrhotic cardiomyopathy. It is likely that the hypertrophy, hypercontractility, electrocardiographic disturbance and altered cardiac metabolism are interlinked and the “cross-talk” needs further analysis. It is also likely that BAs do not act alone—rather they act in concert with other circulating molecules (cytokines, endotoxin, chemokines, lipids, etc) to contribute to the ultimate changes seen in the cirrhotic heart.

Though head to head comparison between cardiac findings in the mouse DDC model and rat CBDL and rat carbon tetrachloride (CCl₄) models are difficult secondary to distinct modes of injury, timelines and species differences, there are some similarities of note. All three models demonstrate cardiac hypertrophy(38, 55). Attenuated contractility to inotropic stress in CBDL and CCl₄ models has been shown to be secondary to downstream effects of decreased β-adrenergic signaling and upregulated cannabinoid receptor signaling likely mediated by endotoxemia, high circulating TNFα, NFκB, high levels of HO-1 and increased NO production. In contrast, cardiac physiology in the DDC mouse model is mostly hyperdynamic, with a robust up-regulation of cardiac HO-1 RNA, similar to CBDL rat hearts ((56) and Supplemental Fig.4s). There are other differences as well, and direct comparisons of the energetic, metabolic, and molecular phenotypes between these models in mice are needed to properly compare the consequences of these distinct models.

Among the more relevant issues is whether or not the mouse DDC model develops portal hypertension and systemic hemodynamic changes seen in some of the other models. There is some indirect evidence for this, via an increase in spleen weight index (Supp. Fig.2) but direct determination of portal pressures and hemodynamics will be needed to be certain. In fact, it may be even more significant that such changes in the hearts of these mice can develop with biliary fibrosis alone. Further studies are needed to properly define these important issues.

In conclusion, multiple inter-related impairments in cardiac cell biology and function are evident in the DDC-induced model of biliary fibrosis which involve multiple metabolites and signaling pathways, including those engaged by high circulating levels of BAs. Investigating the mechanisms of the multiple metabolic, functional, and molecular signaling alterations induced by liver failure on cardiac function, along with analyzing in detail the contributions of systemic bile acids on cardiomyocyte biology, will help design rational targets to counteract the serious clinical cardiac consequences of cirrhosis.

Abbreviations

2DEcho

Two-dimensional echocardiography

ALT

alanine aminotransferase

AST

aspartate aminotransferase

ACC

acetyl-CoA carboxylase

AKT

v-akt murine thymoma viral oncogene/Protein kinase B

βAR

β-adrenergic receptor

BA

bile acids

β MyHC

β myosin heavy chain

BNP

brain natriuretic peptide

CBDL

common bile duct ligation

DDC

0.1% 3, 5-diethoxycarbonyl-1, 4-dihydroxychollidine

DEXA

dual energy x-ray absorptiometry

%EF

ejection fractions

eEF2

eukaryotic elongation factor-2

%FS

shortening fractions

FAO

fatty acid oxidation

GLUT

Glucose transporter proteins

GPCR

G protein coupled receptor

GSK3β

Glycogen Synthase Kinase- 3β

h-FABP/FABP-3

heart-type fatty acid binding protein

LDH

Lactate Dehydrogenase

mCPT-2

mitochondrial-carnitine palmitoyl transferase-2

NEFA

non-esterified fatty acid levels

PAS

Periodic Acid Schiff

RER

Respiratory Exchange Ratio

s-ACAD

short chain acyl-coA-dehydrogenase

TCDCA

taurochenodeoxycholic acid

LCA

lithocholic acid

UCP-3

uncoupling protein-3

iNOS

inducible nitric oxide synthase

eNOS

endothelial nitric oxide synthase

HO-1

hemoxygenase-1

VO₂

Oxygen consumption

VCO₂

Carbon dioxide production