Comparative Phenotypic Assessment of Cardiac Pathology,Physiology,and Gene Expression in C3H/HeJ,C57BL/6J,and B6C3F1/J Mice

Abstract

Human cardiomyopathies often lead to heart failure, a major cause of morbidity and mortality in industrialized nations. Described here is a phenotypic characterization of cardiac function and genome-wide expression from C3H/HeJ, C57BL/6J, and B6C3F1/J male mice. Histopathologic analysis identified a low-grade background cardiomyopathy (murine progressive cardiomyopathy) in eight of nine male C3H/HeJ mice (age nine to ten weeks), but not in male C57BL/6J and in only of ten male B6C3F1/J mice. The C3H/HeJ mouse had an increased heart rate and a shorter RR interval compared to the B6C3F1/J and C57BL/6J mice. Cardiac genomic studies indicated the B6C3F1/J mice exhibited an intermediate gene expression phenotype relative to the 2 parental strains. Disease-centric enrichment analysis indicated a number of cardiomyopathy-associated genes were induced in B6C3F1/J and C3H/HeJ mice, including Myh7, My14, and Lmna and also indicated differential expression of genes associated with metabolic (e.g., Pdk2) and hypoxic stress (e.g. Hif1a). A novel coexpression and integrated pathway network analysis indicated Prkaa2, Pdk2, Rhoj, and Sgcb are likely to play a central role in the pathophysiology of murine progressive cardiomyopathy in C3H/HeJ mice. Our studies indicate that genetically determined baseline differences in cardiac phenotype have the potential to influence the results of cardiotoxicity studies.

Keywords

heart genetic susceptibly genomics cardiomyopathy electrophysiology

Introduction

Heart failure is the leading cause of mortality (National Center for Health Statistics 2005a) and health care expenditures in the United States (National Center for Health Statistics 2005b). Human cardiomyopathy (hCM) may lead to heart failure, characterized by the inability of the heart to adequately pump blood, resulting in inadequate perfusion and oxygenation of tissues. The most common cause of hCM is myocardial infarction or coronary artery disease, both of which have environmental (http://www.mayoclinic.com/health/heart-disease/DS01120/DSECTION=causes) and genetic determinants (Kathiresan et al. 2009; Samani et al. 2007). In addition, thyroid disease and uncontrolled diabetes can promote hCM (Boudina and Abel 2007; Dahl et al. 2008). Specific exogenous causes of hCM include alcohol; cocaine or amphetamine abuse; chemotherapeutics, such as doxorubicin and anthracycline; as well as viral and bacterial infections (Jessup and Brozena 2003). A number of genetic risk factors for intrinsic hCM have been identified (Fowler et al. 2009). Most of the identified risk alleles compromise the structure or function of the sarcomere or interfere with ion homeostasis. Despite the mechanistically informative nature of these genetic determinants with respect to the pathogenesis of hCM, most are monogenic in origin and account for a small fraction of genetic risk factors for hCM-related disease in the general population (Morita et al. 2005). It is likely that a significant proportion of the hCM observed in the human population derives from a combination of independent and interacting environmental and genetic risk factors. Independent of the cause, CM and eventual heart failure is manifest because of the inability of myocardial cells to produce enough ATP to sustain high-energy processes such as maintenance of the ion gradient and calcium homeostasis, which are essential for excitation-coupled contraction and cardiomyocyte survival (Neubauer 2007; Rajabi et al. 2007; Taha and Lopaschuk 2007).

Because of the documented cardiotoxic effects of certain pharmacological and environmental agents, National Toxicology Program (NTP) scientists are working to develop more sensitive models to identify potential cardiac-toxic environmental chemicals (Dunnick et al. 2006; Dunnick, Johnson et al. 2004; Dunnick, Kissling et al. 2007; Dunnick, Lieuallen et al. 2004; Dunnick, Thayer et al. 2007; Jokinen et al. 2005; Nyska et al. 2005a, 2005b). The B6C3F1/J mouse (an F1 hybrid of the C57BL/6J and C3H/HeJ mice) is the mouse model most commonly used for environmental hazard identification and for regulatory decisions in the United States (Rao and Crockett 2003) in regard to human exposure and risk. The rates of spontaneous mCM (commonly referred to as murine progressive cardiomyopathy) vary considerably in two-year-old B6C3F1 mice (2% to 80%) (Maronpot et al. 1999), but this cardiomyopathy is not of clear functional consequence (i.e., early death). Cardiac physiology studies are not usually carried out on two-year-old mice to assess cardiac function such as cardiac output; hence, the possibility remains that more subtle cardiac deficits may exist in older mice with mCM. The NTP has shown treatment-related increases in mCM in B6C3F1 mice at the end of two years with several environmental agents (diethanolamine, urethane, methyl bromide, and bis[2-chloroethoxy]methane [CEM]) (Dunnick, Johnson et al. 2004; Jokinen et al. 2005). In these studies, the mCM was composed of cytoplasmic vacuolization, cellular infiltrate, necrosis, and/or fibrosis, but the mechanism for the cardiotoxicity was not always identified and may vary with the chemical (Dunnick, Johnson et al. 2004; Jokinen et al. 2005).

In contrast to mice from two-year studies, mCM is rarely observed in B6C3F1 mice from subchronic toxicity studies. The potential reasons for the relatively low rates of mCM include: (1) rodents in general may be less susceptible to cardiotoxicity; (2) the chemicals evaluated by the NTP may not include many acute cardiotoxins; (3) adaptation and resolution of injury in young rodents that occur following initial cardiac toxic events (Dunnick, Johnson et al. 2004); and (4) the B6C3F1 mice may not be sensitive to acute chemical-induced mCM owing to the presence of protective alleles. The genetic susceptibility hypothesis for low subchronic mCM rates is not unreasonable, considering the noteworthy differences in cardiac response between mouse strains following isoproterenol challenge (Berthonneche et al. 2009). Genetic susceptibility to cardiotoxicity can be conferred through different mechanisms. In some cases, it may be a result of toxicokinetic differences between strains, and other cases it may be derive from genetically determined subclinical disease in the test animal prior to chemical challenge (e.g., low-grade cardiomyopathy or elevated heart rate). Mouse strains that exhibit genetically determined subclinical cardiac disease may provide more sensitive models for detecting chemical-induced cardiotoxicity. It is important to point out the nuances of such a concept, specifically that strain susceptibility is likely a function of the type of cardiotoxic challenge and it may be necessary to test in multiple strains with different types of subclinical cardiac disease to identify the susceptible strain for a given test article. Bearing this latter issue in mind, the idea of testing in genetically susceptible populations has been employed as the justification for the use of transgenic and knockout mice in carcinogenicity hazard characterization (Pritchard et al. 2003), as well as when comparing the particulate matter-induced cardiopulmonary effects between normotensive Wistar Kyoto rats, spontaneously hypertensive rats, and stroke-prone spontaneously hypertensive rats (Wallenborn et al. 2007). In addition, the biological plausibility and relevance of this approach in relation to cardiotoxicity risk in humans is illustrated by the increased risk of drug-induced long-QT syndrome in individuals with subclinical long-QT that is a byproduct of mutations in SCN5A (Makita et al. 2002).

A convenient starting point for identifying more sensitive models for cardiotoxicity testing (i.e., those that exhibit subclinical/occult cardiac disease) is the parental strains of the B6C3F1 mouse (C3H/HeJ and C57BL/6J), as the cardiac phenotype of these strains has been studied in some detail. C3H/HeJ mice are susceptible to dystrophic cardiac calcinosis (DCC) and focal myocardial cell necrosis (Eaton et al. 1978). Quantitative trait loci that control the susceptibility to DCC in C3H/HeJ mice have been mapped to chromosomes 4, 6, 8, and 12 (Ivandic et al. 1996). Bacterial artificial chromosome transfer studies have since identified Abcc6 on chromosome 6 as a major determinant of DCC susceptibility in C3H/HeJ mice (Meng et al. 2007). The biochemical basis for DCC is related to altered calcium homeostasis, which also plays a critical role in hCM (Yano et al. 2005). C57BL/6 (ApoE−/−) mice are susceptible to the development of atherosclerosis when fed a high-fat diet, whereas C3H/HeJ (ApoE−/−) mice are particularly resistant (Shi et al. 2000). A number of QTL have been identified that confer risk of atherosclerosis in C57BL/6 (Wang, Schadt et al. 2007; Wang, Shi et al. 2007), however the underlying causal cardiac disease genes have not been explicitly identified. Despite these findings, it is uncertain whether a similar cardiac phenotype (specifically the mCM in C3H/HeJ mice) would spontaneously manifest under standard NTP study conditions. Furthermore, the underlying molecular-level processes that differ between the strains remain largely unexplored. An understanding of these molecular processes may serve to explain susceptibility to particular cardiotoxins.

One goal of the studies described here was to determine whether there is a difference in background cardiac disease phenotype (mCM) in C3H/HeJ, C57BL/6J, B6C3F1/J mice under NTP testing conditions. A second goal was to gain insight into the association between the histological and functional genomic phenotypes and to glean hypothesis-generating associations between the cardiomyopathy phenotype and genetic variation in the evaluated strains.

Materials and Methods

Animals

Ten C3H/HeJ, ten B6C3F1/J, and ten C57BL/6J male mice from The Jackson Laboratories, Bar Harbor, Maine were used for these studies. Surgery to implant a Data Sciences radiotelemetry transmitter (PhysioTelDSI TA10ETA-F20, DSI, Arden Hills, MN) in the abdominal cavity was performed on mice anesthetized with tribromoethanol (Avertine) 250 mg/kg IP at seven weeks of age. The ECG lead placement was a modified lead 1 (X axis). Prior to surgery, all mice received buprenorphine 0.05 mg/kg subcutaneously. Following surgery, all mice received 2 mL warmed saline given subcutaneously. All mice were housed individually with a twelve-hour light/dark cycle (6:00 AM to 6:00 PM). NTP 2000 diet (Ziegler Bros, Inc., Gardner, PA) was available ad libitum. The experiment was conducted at an AAALAC-accredited facility, and animal handling and husbandry met all NIH guidelines (Institute of Laboratory Animal Resources 1996).

Phenotype Measurements

Measurements of phenotype were performed two to two and a half weeks after surgery. Mean body weights for C3H/HeJ mice were 26.8 g, for B6C3F1/J mice 29.5 g, and for C57BL/6J mice 25.6 g at day one of the phenotype measurement. All data were recorded on conscious, unrestrained mice continuously for three days on a DSIK data exchange matrix through a receiver. Echocardiography (ECG) recordings were made using Datquest A.R.T., data acquisition and analysis software, components of the PhysioTel Telemetry System (DSI, Arden Hills, MI) on a schedule interval of a fifteen-second collection every ninety seconds for seventy-two hours beginning at 8:00 AM on study day one. The ECGs were evaluated for morphology, RR interval (msec) Bazett’s (Bazett 1920) corrected QT interval (millseconds), R_amp (MV)_, core body temperature (°C), and heart rate bpm). ECG waveforms were generated using ECGAUTO software (EMKA Technologies, Paris, France).

Necropsy

After collection of phenotype measurements (two to two and a half weeks post-surgery; nine and a half weeks of age), animals were euthanized with CO₂; the heart was collected for histopathologic analysis, and a piece of the heart apex was collected for RNA analysis. The average apex section of the heart taken for RNA extraction was 13.1 to 14.4 mg. No more than four minutes elapsed between euthanasia and the placement of cardiac tissue in RNA preservation solution.

Histopathology

Sections of the heart from all animals were placed in a prelabeled mega cassette, fixed in 10% neutral buffered formalin, then trimmed, dehydrated, cleared, and infiltrated with paraffin in a tissue processor. Tissues were embedded using the Tissue-Tek system, sectioned as a longitudinal section containing both atria and ventricles on a microtome at approximately 5 µm, and mounted on glass slides. After staining with hematoxylin and eosin, cover slips and labels were attached to the slides. Heart tissues were then evaluated microscopically at different magnifications by a board-certified veterinary pathologist, and they were then peer reviewed by another board-certified veterinary pathologist, an expert in cardiovascular pathology.

Statistical Analysis of Electrophysiology and Histopathology Data

Statistical analyses were performed independently for each factor. Means and standard errors of the average data were calculated for each strain. An analysis of variance (ANOVA) model was fit to the average data using the MIXED procedure in SAS, version 9.1 (SAS Institute Inc., Cary, NC, 2004). Next, t tests were used to compare each pair of strains. Each comparison was done at a .05/3 = .0167 significance level, to control the overall error rate across all three pairwise comparisons to no more than 0.05.

RNA Preparation and Microarray Hybridization

A section of the heart apex (~ 10–18 mg) was placed into RNAlater (Ambion, Inc., Austin, TX) at necropsy. RNA was extracted from the hearts of five control C3H/HeJ, C57BL/6J, and B6C3F1/J mice. Total cardiac RNA was isolated from hearts using the QIAGEN RNeasy kit (QIAGEN, Valencia, CA). The RNA was analyzed for quantity and purity by ultraviolet analysis using the NanoDrop ND-1000 spetrophotometer (NanoDropTechnologies, Wilmington, DE). Samples were concentrated using Microcon filters (Millipore, Billerica, MA). All samples were evaluated for RNA integrity by gel electrophoresis using the Flash Gel RNA cassette system (Lonza, Rockland, ME). The average heart weights used for RNA extraction from the C3H/HeJ, B6C3F1/J, and C57BL/6J was 13.1, 14.4, and 13.9 mg, respectively. The total amount of RNA extracted for the B6C3F1/J and C57BL/6J mice was greater (average of ~ 1875 ng/14.4 mg and 1140 ng/13.9 mg) than for the C3H/HeJ mouse (375 ng/13.1 mg). Total RNA (50 ng) was used to synthesize double-stranded cDNA for each sample using Affymetrix GeneChipExpression 3′amplication two-cycle target labeling and control reagents (Affymetrix Inc. Santa Clara, CA). The cDNA served as a template to synthesize biotin-labeled antisense cRNA using an in vitro transcription labeling kit. Labeled cRNA was fragmented and hybridized to the Affymetrix Mouse Genome 430 2.0 Array. Array hybridization, washing, and staining were performed according to the Affymetrix recommended protocol EuKGE_Ws2v5. The chips were scanned using an Affymetrix GeneChipScanner 3000. Normalization was performed using the MAS5 algorithm and baseline transformed. Quality control measurements were evaluated to determine whether the data derived from the Affymetrix GeneChips were of sufficient quality prior to comparisons for differential expression. Affymetrix-recommended guidelines for evaluating quality were used to evaluate the output files for each GeneChip using the R/Bioconductor package, Simpleaffy (Wilson and Miller 2005). The quality control parameters evaluated for each chip were: average background; scale factor; number of genes scored as present; 3′ to 5′ ratios for the internal control genes, β-actin and glyceraldehydes-3-phosphate dehydrogenase; values for spike-in control transcripts; and value for poly (A) controls.

Microarray Data Analysis

All statistical analysis was performed on “present” probe sets (100% of the samples from at least one strain contained a “present” probe set) from the normalized data set. A complete compendium of probe sets exhibiting significant differential expression can be found in supplement 1, and the .cel files have been submitted to the GEO database. Using GeneSpring GX11 (Agilent Technologies, Santa Clara, CA), one-way ANOVA and Tukey honestly significant difference test were applied to distinguish differentially expressed sets of genes across the three experimental groups (C3H/HeJ vs. B6C3F1/J, C3H/HeJ vs. C57BL/6J, B6C3F1/J vs.C57BL/6J). A supplemental appendix to this article is published electronically only at http://tpx.sagepub.com/supplemental. P values listed in Supplementary Table 1 (Column Q) were adjusted using the Benjamini-Hochberg FDR multiple-testing correction. Self-organizing maps were created using the default setting in GeneSpring GX11. Literature-based, disease-centric enrichment analysis of the differentially expressed probe sets was done using CoPub (http://services.nbic.nl/cgi-bin/copub/CoPub.pl) (Frijters et al. 2008). All probe sets that exhibited significant change by ANOVA (all genes listed in Supplementary Table 1) were considered in the analysis. CoPub settings for the “disease” category analysis were an R score of >35 and a cross-species gene orthology selection. Four hundred forty-eight “disease” keyword categories were evaluated for enrichment in the analysis. The version of the CoPub database used for the analysis was based on Medline abstracts dated February 2008 and earlier. The “cardiomyopathy gene list” (used for selection of genes for cluster analysis) was created by searching the “disease” category of the CoPub “Bioconcept Search” with the keyword “cardiomyopathy” in (settings = all forms of cardiomyopathy selected, R score >20, copublications ≥ 3, mouse genes only) and by searching the keyword “cardiomyopathy” in the GeneCards database (http://www.genecards.org/). A nonredundant list was created from these searches, and overlap with the ANOVA gene list was determined using GeneSpring GX11. An identical approach was used to identify the “heart rate gene list.” Pathway enrichment analysis was carried out using Ingenuity Pathways Analysis (http://www.ingenuity.com/).

Table 1.

Mean^c cardiac telemetry metrics.

Metric	C3H/HeJ	B6C3F1/J	C57BL/6J
Heart rate (bpm)	650.7 ± 6.4^a,b	596.9 ± 8.7	596.7 ± 6.3
RR interval (msec)	94.2 ± 2.0^a,b	100.0 ± 3.3	101.8 ± 1.9
QTc Bazett's interval (msec)^d	187.1 ± 1.2^a	175.4 ± 1.1	183.2 ± .1
R_amp (mV)	0.46 ± 0.01	0.40 ± 0.02	0.57 ± 0.02^a
Body temperature (°C)	36.3 ± 0.1^a,b	36.9 ± 0.1	36.7 ± 0.1

^a Difference is statistically significant at p < .05 compared to B6C3F1/J.

^b Difference is statistically significant at p < .05 compared to C57BL/6J.

^c Data are presented as mean ± standard error of the mean.

^d QT interval corrected for heart rate.

Generation and Analysis of hCM Signatures

Human cardiac gene expression (Affymetrix GeneChip HG-U133A) from normal organ donors and patients suffering from familial, hypertrophic, idiopathic dilated, ischemic, and viral hCM were downloaded in the form of .cel files from the Harvard Cardiogenomics Web site (http://cardiogenomics.med.harvard.edu/project-detail?project_id=229). Individual experiments (five total for each form of hCM) containing normal samples and samples from a specific form of hCM were created in GeneSpring 11. The Harvard Cardiogenomics data were normalized using MAS5, baseline-transformed, and filtered to retain only probes that were flagged as “present” in at least one sample. For each hCM experiment, the “present” probe sets that overlap the “present” mouse probe sets from the current experiment were selected for further analysis. The selected probe sets were then tested for association with the specific form of hCM represented in that experiment using a t test in which the p values were adjusted using the Benjamini-Hochberg FDR multiple-testing correction. The top 50 probe sets, based on corrected p value, were selected as signature genes. Log₂ expression ratios (hCM/normal) were then obtained for each signature gene. Log₂ expression ratios (C3H/HeJ/C57BL/6J) for probe sets corresponding to those found in each hCM signature were then obtained and compared using a correlation metric in Microsoft Excel, 2007 (Microsoft Corp., Bellevue, WA).

Coexpression Network Analysis

The gene expression correlation data for genes involved in mouse heart development were downloaded from the StarNet2 Web site (Jupiter and VanBuren 2008) (http://vanburenlab.medicine.tamhsc.edu/CorrelationsDev.txt.gz) on January 4, 2010. The mouse heart developmental cohort was selected because disease processes such as CM often lead to the activation of developmental pathways in an attempt to repair the damaged tissue (Chien and Olson 2002). There were around 133 million pairs of gene expression correlations reported. A cutoff value of 0.90 for correlation with a corresponding significance cutoff 1e-10 was used to filter this list down to around 100 thousand pairs of genes and to create an associated network of correlated genes. This stringent cutoff was used because of lack of RAM to handle networks greater than this size. There were 3308 genes in the network. We denote this network by N_c. Let V_c denote the set of genes in this network.

The list of genes with significant differences in gene expressions across at least one of the three strains had 2074 genes. We denote this list by L_s. This list was mapped onto the 3308-gene network. There were 376 genes in common. We denote this number by N_common.

The node influence score of gene i in the network N_c is denoted by θ_i. Let d_ij denote the distance between genes i and j in the network N_c. The node influence score is defined as

θ_{i} = \sum_{j \in V_{c} \cap L_{s}} {0.5}^{d_{i j}}

The node influence score will be high for those genes in the correlation network that are in a neighborhood with a lot of genes from the list, L_s.

The significance of the node influence scores are computed using a gene permutation test. The null hypothesis of the test states that it is equally likely for every gene in the network N_c to belong to the list L_s.

The test is based on a 1000 simulations. During each simulation s, a random set of N_common genes are chosen from the set V_c. The node influence scores are denoted as Θ_i ^s. The p value for the influence score of node i is then computed as

p - {value}_{i} = \sum_{s = 1}^{1000} \frac{I (Θ_{i}^{s} \geq θ_{i})}{1000}

Where

\begin{array}{l} I (Θ_{i}^{s} \geq θ_{i}) = 1 if Θ_{i}^{s} \geq θ_{i} \\ ​ = 0 otherwise \end{array}

The smallest nonzero value p value that can be obtained by the above permutation-based procedure is .001. So a conservative Bonferroni correction for multiple tests for the 3308 genes in the network would be a p value cutoff of 0.

ConcensusPathDB Analysis

All probe sets that exhibited significant differential expression by one-way ANOVA (all genes listed in Supplementary Table 1) were subject to a neighborhood analysis using the ConcensusPathDB (http://cpdb.molgen.mpg.de/) (Kamburov et al. 2009). To carry out this analysis, human orthologues (identified using the Ensembl Biomart tool at http://uswest.ensembl.org/index.html) were used. The settings for the 1-next neighborhood analysis were as follows: minimal set size = 25, minimal connectivity, minimal number of contained candidates = 10, p value cutoff = .001, index = 0. The settings for the 2-next neighborhood analysis were as follows: minimal set size = 25, minimal connectivity, minimal number of contained candidates = 25, p value cutoff =.001, index = 0.

Identification of Genetic Variation

All nonsynonymous single-nucleotide polymorphisms that differ between the C57BL/6J and C3H/HeJ mice were identified using the Ensemble Biomart tool. All expression quantitative trait loci (eQTL) from the “BHF2 (ApoE-null) UCLA (June05)” brain, adipose, and liver database were downloaded from the GeneNetwork database (http://www.genenetwork.org/) (Wu et al. 2004). The BHF2 (ApoE-null) datasets are derived from multi-tissue, genome-wide mRNA expression studies that employed 334 BxH (C57BL/6J X C3H/HeJ) F2 mice that were homozygous null for ApoE (Yang et al. 2006). Genes with cis-eQTLs (peak QTL [quantified as an LRS value] found within ten megabases of that gene) in all three data sets were identified. For a detailed explanation of LRS values, the reader is referred to the GeneNetwork Glossary of Terms (http://www.genenetwork.org/glossary.html). Overlap between the genes with multi-tissue cis-eQTLs and significantly enriched gene nodes was then determined. We employed only cis-eQTLs in our analysis because they tend to be conserved across tissues (Meng et al. 2007; Yang et al. 2006), and therefore there is an increased likelihood that the cis-eQTLs that are identified in the brain, adipose, and liver datasets from GeneNetwork would also be found in cardiac muscle.

Results

Electrophysiology

Electrocardiogram evaluations showed quantitative differences among C3H/HeJ, B6C3F1/J, and C57BL/6J mice. As expected, heart rates covaried in the three strains, and heart rates were higher at night than during the day (data not shown). The C3H/HeJ mice had a 9% higher heart rate, and a 6%-8% shorter RR interval compared to the B6C3F1/J and C57BL/6J mice (Table 1). The heart rate corrected QT interval was longer in the C3H/HeJ mice compared to the B6C3F1 mice. R_amp was shorter in the C57BL/6J mice when compared to the B6C3F1/J mice. Core body temperature of the C3H/HeJ mice was lower than that of the B6C3F1/J and C57BL/6J mice.

Histopathology

Histopathologic analysis of the hearts showed that only the C3H/HeJ mouse had a background minimal focal mCM (Table 2 and Figure 1) involving 1%–10% of the heart section. Murine progressive cardiomyopathy is considered a spontaneous, progressive, age-related, degenerative spectrum of lesions consisting of focal to multifocal myocyte degeneration and/or necrosis, increased interstitial fibrous connective tissue, and minimal infiltrate of mononuclear inflammatory cells (Ruben et al. 2000). C3H/HeJ mice had one or two foci of degeneration between two longitudinal heart organ cross-sections of minimal severity. Foci were distributed throughout the apex, ventricle, and intraventricular septum, hence it is likely that differential gene expression found in apex samples reflects patterns observed in other regions of the heart. Each focus generally involved a single myocardial fiber with vacuolization, fragmentation, aggregation, and hyalinization of the cytoplasm. A few mononuclear cells often penetrated affected fibers. This was considered a very minimal mCM, without signs of clinical toxicity, or adverse effects and would not be anticipated to effect the survival of the animals in a two-year study.

Table 2.

Cardiac histopathology.

	C3H/HeJ	B6C3F1/J	C57BL/6J
Cardiomyopathy	8/9^a,b	1/10	0/10

^a Difference is statistically significant at p < .05 compared to B6C3F1/J.

^b Difference is statistically significant at p < .05 compared to C57BL/6J.

Figure 1.

Example histopathological focus identified in a C3H/HeJ mouse heart that is indicative of minimal CM. Note low magnification view (4X) of the heart. Arrow indicates the location of a single focus of cardiomyopathy. On the lower right side of the photo, there is a high magnification inset (40X), showing the minimal focal mononuclear inflammatory cells, composing the cardiomyopathy.

Differential Gene Expression

Analysis of variance (p < .05) analysis identified 2263 probe sets that exhibited significant differential expression between the three strains. A self-organizing map of the 2263 probe sets revealed the B6C3F1/J mice exhibited an expression pattern that was largely intermediate to the two parental strains with slight bias toward the C57BL/6J mice (Figure 2a). A complete list of probe sets exhibiting significant differential expression can be found in Supplementary Table 1. A Tukey HSB posttest identified 1183, 1350, and 2052 probe sets that exhibited significant differential expression between B6C3F1/J and C57BL/6J, C3H/HeJ and B6C3F1/J, and C3H/HeJ and C57BL/6J, respectively (Figure 2b and annotated in Supplementary Table 1, columns R, S and T). Genes exhibiting the largest differential expression between the strains are shown in Table 3. Of particular note is Tnni3k (higher expression in C57BL/6J), which is structurally polymorphic (missense mutations) between the strains and is a genetic modifier in the calsequestrin transgenic model of cardiomyopathy (Wheeler et al. 2009). Importantly, a comparison of global gene expression changes between C3H/HeJ mice with and without telemetry implants indicated there was no effect of the implant on cardiac gene expression (data not shown).

Figure 2.

Overview of microarray results. (A) Self organizing map based on probe sets that were differentially expressed between the 2 strains and the F1 hybrid. (B) Venn diagram illustrating the overlap of differentially expressed probe sets as determined by a Tukey honestly significant difference post-test.

Table 3.

Genes exhibiting the largest differential expression.

B6C3F1/J > C57BL/6J		C3H/HeJ > B6C3F1/J		C3H/HeJ > C57BL/6J
Gene symbol	Fold change	Gene symbol	Fold change	Gene symbol	Fold change
Baiap2l1	10.8	Dbp	4.9	Baiap2l1	26.6
Abhd1	10.5	Fam122b	4.3	Abhd1	13.8
Hddc3	5.0	Tfpi	4.0	Slc25a29	11.9
Itga9	4.6	Tmem164	3.8	Hddc3	8.8
Cuedc1	4.3	9530006C21Rik	3.1	Wdfy1	8.3
B6C3F1/J < C57BL/6J		C3H/HeJ < B6C3F1/J		C3H/HeJ < C57BL/6J
Gene symbol	Fold change	Gene symbol	Fold change	Gene symbol	Fold change
H2-K1	3.0	Rps9	24.3	H2-K1	37.4
Atp1a2	2.7	H2-K1	12.5	Rps9	27.2
Bcam	2.6	Sord	10.1	Tnni3k	22.0
2700089E24Rik	2.5	Tnni3k	9.9	Sord	16.7
Kif1c	2.5	Hsd17b7	8.3	EG633640	12.7

Disease-centric Enrichment Analysis

Considering that low-grade mCM was consistently observed in the C3H/HeJ mice, but not in the other groups of mice, we wanted to determine whether the differentially expressed genes are disproportionately associated with cardiac or muscle pathology. Using the CoPub microarray analysis tool, we identified diseases whose associated genes were overrepresented in genes identified in the ANOVA analysis, including heart hypertrophy and mCM (Table 4; Supplementary Table 2). Many of these disease gene sets identified, although not specifically associated with heart disease, are plausibly related to processes that are altered in cardiac disease, for example, hypoxia and starvation (related to metabolic adaptations following cardiac challenge). Differentially expressed genes that are specifically associated with CM (pan-species) in the CoPub and GeneCards databases were used to create a self-organizing map (Figure 3; Supplementary Table 1, column U). Using this gene set, the mice cluster into three groups; the B6C3F1/J mice cluster more closely to the C57BL/6J mice. The CM-associated genes that are expressed at relatively higher levels in C3H/HeJ mice are involved in PAK (Raf1, Pdgfa, Mapk8, Mapk10, My14), PDGF (Raf1, Pdgfa, Mapk8, Stat3), and ILK (Myh6, Mapk8, Mapk10, My14, Myh7) signaling, whereas genes that are expressed at relatively higher levels in C57BL/6J mice populate pathways associated with metabolic disease (Prkaa2, Cd36, Prkce, Tnfrsf1b) and mitochondrial dysfunction (Sdha, Xdh, Htra2, Ndufs4). Also notable is the differential expression of genes associated with the sarcoglycan complex (Higher in C3H/HeJ: Sgcb and Sgca; Higher in C57BL/6J: Sgcg).

Figure 3.

Cluster diagram of CM associated genes. Blue indicates lower and yellow indicates higher expression relative to the mean expression of a gene across the dataset. Duplicate genes indicate multiple probes for the same gene were differentially expressed.

Table 4.

Copub disease-centric enrichment analysis of all differentially expressed genes.

Keyword	p value	Number of genes^a	Top 5 genes^b
Starvation	2.80E-09	91	Akt, Pik3c3, Eif2s2, Eif2s3x, Eif2s2
Heart hypertrophy, enlarged heart, cardiomegaly, cardiac hypertrophy	1.90E-05	67	Myh6, Myh7, Ephb1, Akt1, Actb
Dilated cardiomyopathy, congestive cardiomyopathy, cardiomyopathy AND dilated, cardiomyopathy AND congestive, asymmetric septal hypertrophy	3.19E-05	33	Lmna, Myh6, Myh7, Speg, Pdlim5
Oxygen deficiency, hypoxia, anoxia	3.19E-05	64	Hif1a, Sod1, Akt1, Xdh, Egln2
Myopathies, muscular disease, muscle disorder	2.77E-04	31	Speg, Myh6, Myh7, Gaa, Lmna
Insulin resistance	6.03E-04	61	Akt1, Pik3c3, Plat, Ptk2b, Mapk8
Shock, hypovolemic shock, circulatory collapse	8.61E-04	55	Hspd1, Hspa8, Ahsa1, Dnaja2, Tra1
Myotonic dystrophy, muscular dystrophy	3.04E-03	40	Fcmd, Sgca, Lmna, Sgcg, Sgcb
Anemia	4.91E-03	27	Hba-a1, Kitl, Cp, Pafah1b1, Hfe
Sharp syndrome, mixed connective tissue disease, connective tissue disease AND mixed	4.91E-03	12	Trim21, Hnrpc, Calr, Top1, Ube3a
Giant platelet syndrome, bernard-soulier syndrome	5.34E-03	10	F2r, Prnp, Pola2, Cant1, Cd36
Seizure disorder, epileptic seizure, epilepsy	8.27E-03	34	Mal, Cbx5, Abat, Scn1b, Bre
Acute intermittent porphyria	9.55E-03	6	Hmbs, Alad, Uros, Cpox, Alas1

^a Number of genes found coexpressed with the keyword in PubMed.

^b Top genes based on the number of times the keyword co-occurred with a gene in a PubMed abstract.

A correlation analysis using the C3H/HeJ/C57BL/6J expression ratio and signatures derived from different forms of hCM indicated the mCM observed in the C3H/HeJ mice most closely resembles hCM that is a byproduct of an ischemia (Figure 4).

Figure 4.

Correlation with genomic signatures of human CM. The Log₂ (C3H/HeJ/C57BL/6J) expression ratios were compared using a correlation metric to 50 gene signatures derived from data reflecting 5 forms of human CM. Human CM signatures are the 50 probe sets exhibiting the lowest corrected p-value from a t-test that compared genome-wide expression from diseased and non-disease cardiac tissue. The signatures are Log₂ (Disease/Non-diseased) expression ratios for the top 50 probe sets (as define by p-value) for each comparison. The complete human CM signatures are in Supplementary 3.

The C3H/HeJ mice had higher heart rates than the B6C3F1/J and C57BL/6J mice. A search for differentially expressed genes that have been associated with heart rate identified genes that are involved in calcium homeostasis (Casq2, Calr, and Asph) and sodium/potassium transport (Abcc9, Atp1a2, Kcnh2, and Gja1) (Supplementary Table 1, column V).

Ingenuity Pathways Enrichment Analysis

The most enriched pathways from each comparison are shown in Tables 5, 6, and 7, and the complete list can be found in Supplementary Table 4. Genes differentially expressed between C57BL/6J (no mCM) and B6C3F1/J (low incidence mCM) mice (Table 5) populated pathways associated with translation (generally higher expression in C57BL/6J), amino acid/fatty acid metabolism (generally higher expression in C57BL/6J), intermediate metabolism (generally higher expression in C57BL/6J), actin cytoskeleton signaling (generally higher expression in B6C3F1/J), Akt signaling (generally higher expression in B6C3F1/J), and PKA signaling (generally higher expression in B6C3F1/J). Genes differentially expressed between B6C3F1/J (low-incidence mCM) and C3H/HeJ (high-incidence mCM) (Table 6) mice populated pathways associated with mTor signaling (generally higher expression in B6C3F1/J), amino acid/fatty acid metabolism (generally higher expression in B6C3F1/J), actin cytoskeleton signaling (generally higher expression in C57BL/6J), and PKA/Adrenergic signaling (generally higher expression in C3H/HeJ). Pathways enriched by genes differentially expressed between C57BL/6J (no mCM) and C3H/HeJ (high-incidence mCM) (Table 7) are largely a composite of the other two comparisons. Genes that were expressed at higher levels in C57BL/6J mice populated pathways that control adaptation to energy deficiencies (mTor and AMPK), whereas genes that were expressed at higher levels in C3H/HeJ mice populated pathways associated with cAMP/PKA/Adrenergic signaling.

Table 5.

Ingenuity Canonical Pathway enrichment analysis: C57BL/6J vs. B6C3F1/J.^c

C57BL/6J vs. B6C3F1/J^a		C57BL/6J > B6C3F1/J^b		C57BL/6J < B6C3F1/J^b
Ingenuity Canonical Pathways	-log (p value)	Ingenuity Canonical Pathways	-log (p value)	Ingenuity Canonical Pathways	-log (p value)
Aminoacyl-tRNA Biosynthesis	5.12	Aminoacyl-tRNA Biosynthesis	5.57	Actin Cytoskeleton Signaling	3.07
WARS, FARS2, CARS2, GARS, KARS, TARS, AARS, QARS, NARS2, AARSD1, FARSB		FARS2, CARS2, GARS, KARS, TARS, AARS, QARS, AARSD1, FARSB		ABI2, F2R, ACTB, PPP1CB, CRK, MYH7, PIP5K1B, SSH1, MYL1, MYL12B, ARHGEF6, ARPC4, MYL4, PIK3R2
Alanine and Aspartate Metabolism	4.69	Valine, Leucine and Isoleucine Degradation	5.12	Regulation of Actin-based Motility by Rho	2.83
PDHA1 (includes EG:5160), ADSL, ADSSL1, DDO, ABAT, ASRGL1, KARS, AARS, PDHB, NARS2, AARSD1		ACAA1B, ACAD11, ACAA1, MCCC1, MUT, ACAD9, IVD, HMGCS1, ACAA2, ALDH7A1, MCCC2		MYL12B, ACTB, ARPC4, MYL4, PPP1CB, RHOJ, PIP5K1B, MYL1
Valine, Leucine and Isoleucine Degradation	4.04	Alanine and Aspartate Metabolism	3.44	Calcium Signaling	2.82
ACAA1, ABAT, MUT, ACAD9, ACAA2, ACAA1B, ACAD11, BCAT2, MCCC1, IVD, HMGCS1, ALDH7A1, MCCC2		PDHA1 (includes EG:5160), ADSL, ASRGL1, KARS, AARS, PDHB, AARSD1		CREB3, PRKAR1B, MEF2A (includes EG:4205), RCAN3, MYL4, MYH7, ASPH, SLC8A1, MYL1, GRINA, PRKAR1A, CAMK2G
Purine Metabolism	3.35	Antigen Presentation Pathway	3.13	Tight Junction Signaling	2.65
ADSSL1, DDX39, KIF1B, NUDT9, XDH, IDE, POLB, ADSL, POLD3, POLR3A, POLR2A, DDX3Y (includes EG:26900), DLG1, PDE2A, RP2, POLR1A, POLR1B, POLG, PAICS, MYH7, HSPD1, PAPSS2, GART, NME3, INO80, ITPA, MPP5, ATF7IP, POLA2, AFG3L2, AK2, CANT1		HLA-E, HLA-DQA1, HLA-DRB1, CANX, HLA-G, HLA-C		ACTB, VAPA, PRKAR1B, MYL4, CPSF3, MYH7, ARHGEF2, SPTAN1, INADL, MYL1, PRKAR1A
Insulin Receptor Signaling	3.23	Butanoate Metabolism	2.54	Insulin Receptor Signaling	2.3
RPS6KB1, RAF1, GRB2, PPP1CB, CRK, INPPL1, ACLY, VAMP2, PTPRF, PTEN, EIF4EBP1, CBL, EIF2B5, PRKAR1B, PIK3R2, STX4, PRKAR1A		SDHA, PDHA1 (includes EG:5160), ACAA1, SDHC, HMGCS1, PDHB, ALDH7A1		CBL, PRKAR1B, PPP1CB, CRK, PIK3R2, VAMP2, PTPRF, EIF4EBP1, PRKAR1A
Actin Cytoskeleton Signaling	2.96	Purine Metabolism	2.5	ERK/MAPK Signaling	2.27
ACTR2, RAF1, ABI2, F2R, GRB2, ACTB, RDX, MYLK3, PPP1CB, MYH7, CRK, PIP5K1B, SSH1, MYL1, FGF1, CYFIP2 (includes EG:26999), MYL12B, ARHGEF6, ARPC4, MYL4, ARPC1A, PIK3R2, PIP4K2C		DLG1, PDE2A, POLR1B, POLR1A, KIF1B, NUDT9, XDH, PAICS, HSPD1, IDE, PAPSS2, GART, ADSL, POLD3, INO80, ITPA, MPP5, ATF7IP, AK2		YWHAQ (includes EG:10971), DUSP6, CREB3, PRKAR1B, PPP1CB, CRK, HSPB7, STAT3, PIK3R2, EIF4EBP1, PRKAR1A
RhoA Signaling	2.59	Glycolysis/Gluconeogenesis	2.05	Clathrin-mediated Endocytosis Signaling	2.2
ACTR2, RAF1, ABI2, F2R, GRB2, ACTB, RDX, MYLK3, PPP1CB, MYH7, CRK, PIP5K1B, SSH1, MYL1, FGF1, CYFIP2 (includes EG:26999), MYL12B, ARHGEF6, ARPC4, MYL4, ARPC1A, PIK3R2, PIP4K2C		PDHA1 (includes EG:5160), GPI, PFKP, ADHFE1, PDHB, ACYP2, BPGM, ALDH7A1		CD2AP, EPN1, ARF6, CBL, F2R, ACTB, ARPC4, CSNK2A1, LDLRAP1, PIK3R2
Butanoate Metabolism	2.16	Pyruvate Metabolism	1.99	RhoA Signaling	2.19
SDHA, PDHA1 (includes EG:5160), BDH1, ACAA1, ABAT, SDHC, HMGCS1, PDHB, ALDH7A1		PDHA1 (includes EG:5160), ACAA1, ME1, PDHB, HAGH, ACYP2, ALDH7A1		ARHGAP9, MYL12B, ACTB, ARPC4, SEPT4, MYL4, PIP5K1B, MYL1
Regulation of Actin-based Motility by Rho	2.14	Protein Ubiquitination Pathway	1.98	Amyloid Processing	1.92
ACTR2, MYL12B, ACTB, ARPC4, MYL4, PPP1CB, RHOJ, ARPC1A, PIP5K1B, PIP4K2C, MYL1		UBE2G2, UBE2M, STUB1, USP36, PSMD2, ANAPC5, MDM2, HSPA5, NEDD4L, UBE2J1, HLA-C, ANAPC1		CSNK2A1, PRKAR1B, CSNK1A1, BACE1, PRKAR1A
Protein Kinase A Signaling	2.02	Bile Acid Biosynthesis	1.88	ILK Signaling	1.88
RAF1, AKAP8, PDIA3, MYLK3, PPP1CB, YWHAQ (includes EG:10971), GNB4, PRKAR1B, SMAD4, MYL4, PRKCE, YWHAQ (includes EG:22630), CTNNB1, PDE2A, CREB3, CREBBP, PYGL, MYL1, MYL12B, ANAPC5, GNB2, KDELR3, KDELR1, ANAPC1, PRKAR1A, CAMK2G		ACAA1B, ACAA1, ADHFE1, ACAA2, ALDH7A1		PARVB, PPAP2B, ACTB, ARHGEF6, CREB3, MYL4, RHOJ, MYH7, PIK3R2, MYL1

^a All genes exhibiting significant differential expression in the comparison were used for pathway analysis.

^b Only genes that exhibited significantly higher levels in the comparison were used for pathway analysis.

^c Genes list below enriched pathways are those that exhibited significant differential expression in the indicated comparison.

Table 6.

Ingenuity canonical pathway enrichment analysis: B6C3F1/J vs. C3H/HeJ.

B6C3F1/J vs. C3H/HeJ^a		B6C3F1/J > C3H/HeJ^b		B6C3F1/J < C3H/HeJ^b
Ingenuity Canonical Pathways	-log (p value)	Ingenuity Canonical Pathways	-log (p value)	Ingenuity Canonical Pathways	-log (p value)
Aminophosphonate metabolism	4.86	mTOR signaling	5.8	Cardiac β-adrenergic signaling	4.75
PIGO, PIGF, METTL2B, METTL1, AS3MT, ALAS1, COQ5, PHKA1, CCBL1, BCKDHB		PRKAB2, DDIT4, PPP2R5C, PPP2R2A, RPS6, EIF3J, RHOJ, HIF1A, EIF4E, PPP2CB, PPP2R1A, RRAS2, AKT1, PIK3C3, PRKAA2, PRKCE, RPS6KA2, EIF4B		ADCY9, PDE7A, AKAP8, ADRBK2, PRKAR2A, AKAP6, ADCY5, PPM1L, GNB2, PRKAR1B, SLC8A1, AKAP9, PKIG
mTOR Signaling	4.07	Valine, Leucine and Isoleucine Degradation	3.81	α-Adrenergic Signaling	4.71
NAPEPLD, PPP2R5C, PRKAB2, DDIT4, PPP2R2A, RPS6, EIF3J, RHOJ, HIF1A, EIF3E, EIF4E, PPP2CB, PPP2R1A, AKT1, RRAS2, PIK3C3, PPM1L, PRKAA2, PRKCE, RPS6KA2, EIF4B		ACAA1B, ACAA1, MCEE, MCCC1, MUT, IVD, ALDH9A1, ACAA2, ALDH7A1, MCCC2		ADCY9, RAF1, ITPR2, ADCY5, CALM2, GNB2, PRKAR1B, PRKAR2A, PYGL, SLC8A1, PHKG1
Cardiac β-adrenergic Signaling	3.94	Regulation of eIF4 and p70S6K Signaling	3.45	Purine Metabolism	3.63
ADCY9, PPP1R14C, PDE7A, AKAP8, PPP2R5C, PPP2R2A, ADRBK2, PRKAR2A, AKAP6, PPP2CB, PPP2R1A, PPP1R10, ADCY5, PPM1L, GNB2, PRKAR1B, SLC8A1, AKAP9, PKIG		EIF2S3, PPP2CB, PPP2R1A, AKT1, RRAS2, PPP2R5C, PPP2R2A, PIK3C3, EIF3J, RPS6, EIF2A, EIF4E		ADCY9, ATP1B1, PDE7A, DDX39, KIF1B, PAICS, MYH7, REV3L, POLB, GART, BCKDHB, ADSL, PRPS2, ITPA, DDX3Y (includes EG:26900), ADCY5, ABCC1, AK3L1, AFG3L2, PPAT
α-Adrenergic Signaling	3.6	Aminophosphonate Metabolism	3.09	Protein Kinase A Signaling	3.57
ADCY9, RAF1, ITPR2, CALM2, PRKAR2A, PYGL, GNAI2, RRAS2, ADCY5, GNB2, PRKAR1B, PRKCE, SLC8A1, PHKG1, CALM1		PIGO, METTL2B, METTL1, AS3MT, ALAS1, CCBL1		ADCY9, RAF1, PDE7A, AKAP8, ITPR2, PTCH1, PRKAR2A, CALM2, AKAP6, PYGL, NTN1, ROCK2, ADCY5, PRKAR1B, GNB2, SMAD4, MYL4, AKAP9, PHKG1
Valine, Leucine and Isoleucine Degradation	3.59	Glycosaminoglycan Degradation	2.56	Hepatic Cholestasis	3.11
ACAA1, ABAT, MUT, ALDH9A1, ACAA2, BCKDHB, ACAA1B, BCAT2, MCEE, MCCC1, IVD, ALDH7A1, MCCC2		MGEA5, IDUA, IDS, PNPT1, COASY, SULF2		ADCY9, ABCB4, CYP27A1, SREBF1, ADCY5, ABCC1, MAPK8, MAPK10, PRKAR1B, PRKAR2A, TIRAP
Purine Metabolism	3.5	Production of Nitric Oxide and Reactive Oxygen Species in Macrophages	2.51	cAMP-mediated Signaling	3.05
ENTPD4, NME1, DDX39, PDE7A, KIF1B, NME2, BCKDHB, POLB, ADSL, DDX3Y (includes EG:26900), ADCY5, ABCC1, ADCY9, DLG1, RP2, TJP2, ATP1B1, POLR1A, POLR1B, PAICS, MYH7, HSPD1, REV3L, PAPSS2, GART, PRPS2, INO80, GMPS, ITPA, ATF7IP, PNPT1, AK3L1, AFG3L2, AK2, PPAT		PPP1R14C, PPP2R5C, PPP2R2A, RHOJ, MAP3K4, PPP2CB, PPP2R1A, AKT1, NFKBIA, PPP1R10, PIK3C3, PRKCE, TNFRSF1B		ADCY9, RAF1, PDE7A, AKAP8, ADCY5, CALM2, PRKAR1B, PRKAR2A, AKAP6, STAT3, AKAP9, PKIG
Regulation of eIF4 and p70S6K Signaling	3.18	p70S6K Signaling	2.46	Pancreatic Adenocarcinoma Signaling	2.79
EIF2S3, RAF1, PPP2R5C, PPP2R2A, RPS6, EIF3J, EIF3E, EIF4E, EIF2A, PPP2CB, PPP2R1A, AKT1, RRAS2, PIK3C3, PPM1L, EIF2B5		GNAI2, PPP2CB, PPP2R1A, AKT1, RRAS2, PPP2R5C, PPP2R2A, PIK3C3, MAPT, PRKCE, RPS6		RAF1, NAPEPLD, TFDP1, MAPK8, MAPK10, SMAD4, HBEGF, MDM2, STAT3
CDK5 Signaling	3.09	fMLP Signaling in Neutrophils	2.37	Renin-Angiotensin Signaling	2.76
ADCY9, PPP1R14C, RAF1, PPP2R5C, PPP2R2A, MAPT, PRKAR2A, PPP2CB, PPP2R1A, RRAS2, PPP1R10, ADCY5, PPM1L, PRKAR1B		GNAI2, ACTR2, ACTR3, RRAS2, NFKBIA, ARPC2, PIK3C3, PRKCE, NFATC2, CALM1		ADCY9, RAF1, ITPR2, ADCY5, MAPK8, MAPK10, PRKAR1B, PRKAR2A, STAT3
Dopamine Receptor Signaling	2.75	Bile Acid Biosynthesis	2.25	Cellular Effects of Sildenafil (Viagra)	2.56
ADCY9, PPP2CB, PPP1R14C, PPP2R1A, PPP1R10, PPP2R5C, PPP2R2A, ADCY5, COMT, PPM1L, PRKAR1B, PRKAR2A		ACAA1B, ACAA1, ADHFE1, ALDH9A1, ACAA2, ALDH7A1		ADCY9, ITPR2, ADCY5, CALM2, PRKAR1B, PRKAR2A, MYL4, MYH7, KCNH2, PRKG1
fMLP Signaling in Neutrophils	2.72	Ceramide Signaling	2.19	Corticotropin Releasing Hormone Signaling	2.56
RAF1, ACTR2, ITPR2, CALM2, GNAI2, NFKBIA, ACTR3, RRAS2, ARPC2, PIK3C3, GNB2, CYBB, PRKCE, NFATC2, CALM1		PPP2CB, PPP2R1A, AKT1, RRAS2, PPP2R5C, PPP2R2A, PIK3C3, TNFRSF1B		ADCY9, RAF1, ITPR2, ADCY5, PTCH1, CALM2, PRKAR1B, PRKAR2A, JUND

^a All genes exhibiting significant differential expression in the comparison were used for pathway analysis.

^b Only genes that exhibited significantly higher levels in the comparison were used for pathway analysis.

^c Genes list below enriched pathways are those that exhibited significant differential expression in the indicated comparison.

Table 7.

Ingenuity Canonical Pathway enrichment analysis: C67BL/6J vs. C3H/HeJ.

C57BL/6J vs. C3H/HeJ^a		C57BL/6J > C3H/HeJ^b		C57BL/6J < C3H/HeJ^b
Ingenuity Canonical Pathways	-log (p value)	Ingenuity Canonical Pathways	-log (p value)	Ingenuity Canonical Pathways	-log (p value)
Aminoacyl-tRNA Biosynthesis	4.47	Aminoacyl-tRNA Biosynthesis	5.41	Cellular Effects of Sildenafil (Viagra)	3.45
CARS2, GARS, LARS2, HARS, TARS, QARS, NARS2, AARSD1, FARSB, WARS, FARS2, MTFMT, IARS		WARS, FARS2, CARS2, GARS, LARS2, TARS, QARS, MTFMT, IARS, AARSD1, FARSB		ADCY9, MYH6, ITPR2, CALM2, PRKAR2A, PPP1CB, MYH7, KCNH2, PRKG1, MYL1, MYL12B, ADCY5, PRKAR1B, MYL4, PRKAR1A
Aminophosphonate Metabolism	4.12	Valine, Leucine and Isoleucine Degradation	4.34	Protein Kinase A Signaling	3.39
PIGO, PIGF, METTL2B, METTL1, AS3MT, PRMT5, ALAS1, COQ5, PHKA1, CCBL1, BCKDHB		ACAA1, MUT, ACAD9, ALDH9A1, ACAA2, ACAA1B, ACAD11, MCEE, MCCC1, IVD, HMGCS1, ALDH7A1, MCCC2		PDE7A, CALM2, PPP1CB, NTN1, ROCK2, YWHAQ (includes EG:10971), GNB4, ADCY5, PRKAR1B, SMAD4, MYL4, YWHAQ (includes EG:22630), ADCY9, ITPR2, PTCH1, CREB3, PRKAR2A, PYGL, AKAP6, MYL1, MYL12B, KDELR3, AKAP9, PRKAR1A, CAMK2G
Valine, Leucine and Isoleucine Degradation	3.41	mTOR Signaling	3.64	Cardiac β-adrenergic Signaling	3.38
ACAA1, ABAT, MUT, ACAD9, ALDH9A1, ACAA2, BCKDHB, ACAA1B, ACAD11, BCAT2, MCEE, MCCC1, IVD, HMGCS1, MCCC2, ALDH7A1		RPS6KB1, PRKAB2, DDIT4, PPP2R5C, PPP2R2A, RPS6, EIF3J, HIF1A, EIF4E, PPP2CB, PPP2R1A, RRAS2, AKT1, PIK3C3, PRKAA2, PRKCE, RPS6KA2, EIF4B		ADCY9, PDE7A, ADRBK2, PRKAR2A, PPP1CB, AKAP6, GNB4, ADCY5, PPM1L, PRKAR1B, SLC8A1, AKAP9, PRKAR1A, PKIG
RhoA Signaling	3.16	Antigen Presentation Pathway	3.38	Calcium Signaling	3.11
ACTR2, PTK2B, SEPT4, MYLK3, RDX, CDC42EP3, PIP5K1B, DLC1, MYL1, ROCK2, ACTR3, ARHGAP9, MYL12B, ARPC2, CFL2, ARPC4, MYL4, ARPC1A, CDC42EP1, SEPT6, PIP4K2C		CALR, HLA-E, HLA-DQA1, HLA-DRB1, CANX, CD74, HLA-G, HLA-C		MYH6, ATP2C1, ITPR2, CREB3, MEF2A (includes EG:4205), CALM2, PRKAR2A, MYH7, MYL1, GRINA, PRKAR1B, RCAN3, MYL4, ASPH, SLC8A1, PRKAR1A, CAMK2G
Purine Metabolism	2.99	Regulation of eIF4 and p70S6K Signaling	2.99	Purine Metabolism	2.97
ADSSL1, ENTPD4, MYH6, NME1, PDE7A, DDX39, KIF1B, XDH, NUDT9, NME2, IDE, ATP11B, POLB, BCKDHB, POLD3, POLR3A, POLR2A, ADCY5, DDX3Y (includes EG:26900), ABCC1, ADCY9, DLG1, PDE2A, TJP2, ATP1B1, POLR1A, POLR1B, POLG, MYH7, REV3L, HSPD1, PAPSS2, GART, NME3, PRPS2, INO80, GMPS, ATF7IP, PNPT1, POLA2, AK3L1, AFG3L2, AK2, PPAT, CANT1		EIF2S3, RPS6KB1, PPP2R5C, GRB2, PPP2R2A, RPS6, EIF3J, EIF2A, EIF4E, PPP2CB, PPP2R1A, AKT1, RRAS2, PIK3C3		MYH6, ADSSL1, DDX39, PDE7A, KIF1B, POLB, BCKDHB, POLR3A, POLR2A, ADCY5, DDX3Y (includes EG:26900), ABCC1, ADCY9, ATP1B1, POLG, MYH7, REV3L, GART, NME3, PRPS2, POLA2, AK3L1, AFG3L2, PPAT, CANT1
p70S6K Signaling	2.98	Aminophosphonate Metabolism	2.98	Amyloid Processing	2.78
RPS6KB1, YWHAG, PPP2R5C, F2R, PDIA3, PPP2R2A, GRB2, MAPT, RPS6, GNAI2, YWHAQ (includes EG:10971), PPP2CB, PPP2R1A, AKT1, RRAS2, PIK3C3, PPM1L, PRKCE, PIK3R2, YWHAQ (includes EG:22630), EEF2K, BCAP31		PIGO, METTL2B, METTL1, AS3MT, PRMT5, ALAS1, CCBL1		CSNK1E, MARK1, CSNK2A1, PRKAR1B, PRKAR2A, CSNK1A1, BACE1, PRKAR1A
Protein Kinase A Signaling	2.82	EIF2 Signaling	2.95	α-Adrenergic Signaling	2.49
PDE7A, PTK2B, PDIA3, CALM2, MYLK3, PPP1CB, NTN1, ROCK2, YWHAQ (includes EG:10971), GNB4, NFKBIA, PPP1R10, ADCY5, PRKAR1B, PRKCE, MYL4, SMAD4, YWHAQ (includes EG:22630), CTNNB1, CALM1, ADCY9, PPP1R14C, PDE2A, YWHAG, ITPR2, PTCH1, CREB3, CREBBP, PRKAR2A, PYGL, AKAP6, MYL1, GNAI2, MYL12B, KDELR3, ANAPC5, NFATC2, KDELR1, AKAP9, PRKAR1A, CAMK2G, ANAPC1		EIF2AK1, EIF2S3, AKT1, RRAS2, GRB2, PIK3C3, EIF5, EIF3J, RPS6, PPP1CB, EIF2A, EIF4E		ADCY9, GNB4, ITPR2, ADCY5, CALM2, PRKAR1B, PRKAR2A, PYGL, SLC8A1, PRKAR1A
AMPK Signaling	2.77	p70S6K Signaling	2.91	Dopamine Receptor Signaling	2.46
RPS6KB1, PFKFB1, PPP2R5C, PRKAB2, PPP2R2A, PRKAR2A, PFKP, PFKFB2, EIF4EBP1, PPP2CB, PPP2R1A, AKT1, PPM1B, PIK3C3, PPM1L, PRKAA2, PRKAR1B, CPT2, AK3L1, PIK3R2, AK2, PPAT, PRKAR1A		RPS6KB1, YWHAG, PPP2R5C, GRB2, PPP2R2A, PDIA3, MAPT, RPS6, GNAI2, PPP2CB, PPP2R1A, AKT1, RRAS2, PIK3C3, PRKCE		ADCY9, GCH1, ADCY5, COMT, PPM1L, PRKAR1B, PRKAR2A, PPP1CB, PRKAR1A
TR/RXR Activation	2.75	PI3K/AKT Signaling	2.61	RAR Activation	2.43
UCP2, MED1, ME1, PFKP, HIF1A, THRSP, KLF9, UCP3, AKT1, SCARB1, ENO1, SREBF1, PIK3C3, NCOA1, NCOR1, PIK3R2, TBL1XR1		RPS6KB1, YWHAG, PPP2R5C, GRB2, PPP2R2A, INPPL1, EIF4E, PTEN, PPP2CB, PPP2R1A, AKT1, RRAS2, NFKBIA, CTNNB1		ADCY9, MED1, PRKAR2A, RBP1, ADCY5, CSNK2A1, NCOA1, MAPK10, PRKAR1B, SMAD4, NRIP1, PIK3R2, STAT5B, SCAND1, PRKAR1A
Melanocyte Development and Pigmentation Signaling	2.62	NRF2-mediated Oxidative Stress Response	2.47	cAMP-mediated Signaling	2.3
ADCY9, RPS6KB1, GRB2, CREBBP, CREB3, PRKAR2A, CRK, KITLG, RRAS2, ADCY5, PIK3C3, PRKAR1B, RPS6KA2, PIK3R2, DCT, PRKAR1A		MGST1, SOD1, CREBBP, GSTA5, HSPB8, DNAJC3, MAFK, GSTO1, RRAS2, AKT1, SCARB1, DNAJB11, PIK3C3, CCT7, DNAJC18, PRKCE, DNAJA2, DNAJC16		ADCY9, PDE7A, DUSP6, CREB3, CALM2, PRKAR2A, AKAP6, STAT3, ADCY5, PRKAR1B, AKAP9, CAMK2G, PKIG, PRKAR1A

^a All genes exhibiting significant differential expression in the comparison were used for pathway analysis.

^b Only genes that exhibitted significantly higher levels of in the comparison were used for pathway analysis.

^c Genes list below enriched pathways are those that exhibited significant differential expression in the indicated comparison

Identification of Highly Integrated Signaling Nodes Using Network Analysis

Critical signaling hubs are potential targets for toxicity and in some cases may be causally related to the phenotype of concern. To identify these critical regulators, we performed two network neighborhood enrichment analyses, one that is a novel analysis that is based on co-expression data from the mouse heart developmental cohort found in the Starnet2 database, and another that is based on a composite of protein interactions and canonical pathways (ConcensusPathDB). The complete results of the individual analyses can be found in Supplementary Tables 5 and 6. Seven hundred forty-eight significantly enriched gene nodes were identified in the coexpression network analysis, and 1007 were identified in the ConcensusPathDB analysis. A pathway analysis of the significantly enriched gene nodes from the coexpression network indicated a strong association with energy generation and mitochondrial dysfunction pathways (Supplementary Table 7). A similar analysis of the enriched gene nodes (centers) from the ConcensusPathDB analysis indicated a significant enrichment of signaling pathways associated with the control of energy homeostasis such as mTor, AMPK, and PI3K/AKT (Supplementary Table 8). To identify the gene nodes that are likely to have the strongest influence, we determined which of the significantly enriched gene nodes occurred in both analyses (Table 8). The genes listed in Table 8 are found in a number of pathways including actin cytoskeleton signaling (Rhoq, Rhot2, Rhoj, Gsn, Actc1), pyruvate metabolism (Pdha1, Dlat, Bckdha, Dld, Mdh1, Hadha), mitochondrial dysfunction (Ndufa4, Atp5a1, Ndufb8, Ndufb10), AMPK signaling (Prkaa2, Lipe, Pfkm), and sarcoglycan complex formation (Sgcb and Sgca). The Prkaa2 second neighborhood coexpression network diagram containing only differentially expressed genes is shown in Figure 5 . Notably, a number of the genes in the Prkaa2 network, including Abcc9, Sgcb, Nrap, Dsp, Sdha, Hspb8, Hopx, and Ldb3, are associated in the literature with cardiomyopathy (i.e., they are on the “cardiomyopathy gene list”). One potential reason for nodal enrichment is the gene that occupies the node varies at a functional level owing to genetic variation between C3H/HeJ and C57BL/6J. Hypothetically, the perturbation of the network created by the functional variation leads to local network adaptations that are manifest as differential gene expression. Three of the genes in Table 8, Prmt5, Anxa5, and Dst, encode proteins that differ in amino acid sequence between the two mice strains. Four genes in Table 8, Prmt5, Prkaa2, Mccc2, and Agl, exhibit genetically determined expression (cis-eQTL) as determined by genetic genomic mapping of F2 mice from a C3H/HeJ X C57BL/6 cross (Yang et al. 2006). All significantly enriched gene nodes that exhibit genetic variation are noted in Supplementary Table 5 (columns R–V).

Table 8.

Enriched signaling nodes from network analysis.

Differentially Expressed^a
1st and 2nd CPDB^C			1st CPDB			2nd CPDB
Entrez ID	Gene symbol	Nodal enrichment score^d	Entrez ID	Gene symbol	Nodal enrichment score	Entrez ID	Gene symbol	Nodal enrichment score
210789	Tbc1d4	22	24051	Sgcb	51	18597	Pdha1	29
76614	Immt	20	18604	Pdk2	49	14194	Fh1	26
			20928	Abcc9	46	108079	Prkaa2^f	24
			56357	Ivd	42	78038	Mccc2^f	21
			20391	Sgca	41	80837	Rhoj	21
			18810	Plec1	33	69574	Cmbl	19
			20742	Spnb2	29	77559	Agl^f	12
			27273	Pdk4	19	22318	Vamp2	7
			72198	Skiv2l2	4	69178	Snx5	6
			85305	Kars	4	229317	Eif2a	4
			27374	Prmt5^e,f	3	11747	Anxa5^e	4
						26754	Cops5	3
Not differentially expressed^b
1st and 2nd CPDB			1st CPDB			2nd CPDB
Entrez ID	Gene symbol	Nodal enrichment score	Entrez ID	Gene symbol	Nodal enrichment score	Entrez ID	Gene symbol	Nodal enrichment score
11938	Atp2a2	38	227753	Gsn	49	18642	Pfkm	52
76263	Gstk1	36	235339	Dlat	44	20193	S100a1	47
20528	Slc2a4	35	59010	Sqrdl	42	97212	Hadha	44
			71960	Myh14	41	68342	Ndufb10	43
			53857	Tuba8	39	104776	Aldh6a1	41
			13628	Eef1a2	34	93747	Echs1	41
			78920	Dlst	33	380698	Obscn	38
			13382	Dld	32	110119	Mpi	37
			225115	Svil	30	67264	Ndufb8	33
			12039	Bckdha	27	94245	Dtnbp1	31
			11464	Actc1	26	110821	Pcca	31
			11739	Slc25a4	26	17449	Mdh1	30
			108058	Camk2d	25	214952	Rhot2	29
			11946	Atp5a1	22	19039	Lgals3bp	29
			50708	Hist1h1c	21	106628	Trip10	27
			22145	Tuba4a	20	223697	Unc84b	26
			13518	Dst^e	16	66904	Pccb	26
			76707	Clasp1	15	16890	Lipe	25
			666794	Rbm24	13	19156	Psap	25
			18674	Slc25a3	11	104215	Rhoq	21
			68794	Flnc	8	17992	Ndufa4	19
			11749	Anxa6	4	432676	Gm5436	16
						12301	Cacybp	7

Abbreviation: CPDB, ConsensusPathDB.

^a Gene exhibited significant differential expression between the 3 strains by analysis of variance.

^b Gene did not exhibit significant differential expression between the 3 strains by analysis of variance.

^c Gene was signficantly enriched in a CPDB First and Second Neighborhood Analysis.

^d Enrichment score from coexpression network analysis. The higher the score, the greater the enrichment of the node (gene).

^e Gene exhibits a protein coding difference between C57BL/6J and C3H/HeJ mice.

^f Gene exhibits a cis-expression QTL in 3 tissues (adipose, brain, and liver), as determined by QTL mapping from a C57BL/6J X C3H/HeJ F2 cross.

Figure 5.

Prkaa2 2^nd neighborhood network of differentially expressed genes. In the co-expression network Prkaa2 is linked to 138 genes in the second neighborhood through the genes, Asb2 and Tns1 (1^st neighborhood genes). Of the 138 in the second neighborhood, 39 were differentially expressed. Prkaa2 2^nd neighborhood genes that exhibited significantly higher expression in C3H/HeJ mice are shown in the ovals, where as those that exhibited significantly lower expression in C3H/HeJ are shown in boxes (including Prkaa2). Genes in the diamonds were not differentially expressed.

Discussion

Distinct cardiac phenotypes were observed in the two mouse strains and the F1 hybrid. Electrophysiological measurements indicated the C3H/HeJ mice exhibited differences in heart rate and ECG parameters. A histopathology diagnosis consistent with a minimal mCM was significantly increased in the C3H/HeJ mouse. Although the C3H/HeJ mice had a minimal mCM, there were no clinical signs indicating any reduced cardiac function. However, analysis of the heart gene expression patterns suggested there are noteworthy differences at the functional genomic level between the two mouse strains and the F1 hybrid.

C3H/HeJ mice exhibited significantly different ECG metrics and heart rate when compared to B6C3F1/J and C57BL/6J. The increased heart rates in the C3H/HeJ mice may be attributable to higher levels of T4 in this strain (http://phenome.jax.org/pub-cgi/phenome/mpdgrcgi?rtn=views/measplot&brieflook=22820) (Dahl et al. 2008). Changes in ECG metrics that are indicative of cardiac disease are often manifested at the genomic level as complex changes in expression of ion channels (Le Bouter et al. 2003). In this study, we observed differential expression of a number of sodium, potassium, and calcium channels than confer depolarizing and repolarizing effects in the heart (Aiba and Tomaselli 2010). Scn4b (up-regulated in C3H/HeJ) and Scn1b (up-regulated in C3H/HeJ) encode subunits of the voltage-gated sodium channels. Both are auxiliary proteins of the pore forming the α subunits (e.g., Scn5a) that regulate channel gating and expression. Both subunits facilitate sodium current (Medeiros-Domingo et al. 2007; Watanabe et al. 2008), hence the up-regulation of these genes in C3H/HeJ mice is potentially an adaptation that sustains action potential propagation during chronotropic stress. Slc8a1 (NCX1; up-regulated in C3H/HeJ mice) is a sodium/calcium antiporter that under normal conditions removes calcium from the cytosol, leading to relaxation of the cardiac muscle. Slc8a1 is typically up-regulated in heart failure (Aiba and Tomaselli 2010). The higher level of Slc8a1 gene in C3H/HeJ mice likely reflects an adaptation to the shorter R-R interval (chronotropic stress), which requires a more rapid removal of calcium form the cytosol. Atp1a2 (down-regulated in C3H/HeJ) encodes an ATPase that is preferentially located in the T-tubules of cardiomyocytes, where its activity is coupled NCX (encoded by Slc8a1) (James et al. 1999). The down-regulation in C3H/HeJ mice likely represents an attempt by the cardiomyocyte to curtail energy expenditure during metabolic stress. The decreased expression of Atp1a2 has two consequences: (1) increased action potential duration, and (2) reversal of the NCX transporter leading to higher levels of intracellular calcium (Akar and Tomaselli 2005).The Kcnh2 (up-regulated in C3H/HeJ mice) encodes a delayed-rectifying potassium channel that underlies the I _Kr current of the cardiomyocyte (Aiba and Tomaselli 2010). The I _Kr current is critical in the late phase of membrane repolarization, and hence it is likely that the increased expression of Kcnh2 in C3H/HeJ mice is an adaptive response that allows for more rapid repolarization during chronotropic stress. Kcnj3 (Girk1; decreased in C3H/HeJ) is a G-protein–controlled potassium channel that contributes to heart rate decreases elicited by the parasympathetic nervous system (Bettahi et al. 2002). Knockout of Kcnj3 in mice leads to a mild resting tachycardia (Bettahi et al. 2002). The decreased expression of Kcnj3 in C3H/HeJ may reflect an adaptive mechanism that allows for maintenance of cardiac output (through increased heart rate) in the presence of low-grade mCM. Kcnt2 (up-regulated in C3H/HeJ) is an outward rectifying, ATP-sensitive potassium channel that is rapidly activated following sodium channel activation. Activity of this channel is increased under metabolic stress. This increase is predicted to have a protective effect by limiting excitability and by extension calcium-associated excitotoxicity (Bhattacharjee et al. 2003). Kcnj11 (down-regulated in C3H/HeJ) and Abcc9 (up-regulated in C3H/HeJ) encode subunits of the cardiac sarcolemmal ATP-sensitive potassium (K_ATP) channels. The K_ATP channels mediate shortening of the action potential and therefore decrease calcium influx and the related excitotoxicity (Kane et al. 2005). Differential expression of Kcnt, Kcnj11, and Abcc9 likely reflects a protective mechanism that attenuates further cardiac damage. Overall, the differential expression of ion channel genes paints a complicated picture of the electrophysiological remodeling that takes place at the genome level during subclinical mCM. One intriguing hypothesis is that the adrenergic signaling cascade, which is differentially active in the two mouse strains and the F1 hybrid, serves as a driving force behind the electrophysiological remodeling (Galindo et al. 2009).

Cardiac stress/toxicity causes activation of a fetal gene expression program in cardiac muscle that is considered by some to be an adaptive change. The two most noted changes are in the contractile and metabolic features of the myocardium (Rajabi et al. 2007). Specific hallmarks of the fetal gene expression program are an increase in expression of fetal isoforms of myosin and actin along with a metabolic switch from β-oxidation of fatty acids to anerobic glycolysis as a means of energy production (Rajabi et al. 2007).

The C3H/HeJ, and to a lesser extent B6C3F1/J, mice in our study exhibit many of the characteristic gene expression changes observed in fetal heart, both at the pathway and at the specific gene level. Increased expression of the fetal isoform of the myosin heavy chain Myh7 (Rajabi et al. 2007) and other genes associated with actin cytoskeletal/tight junction signaling are induced in B6C3F1/J mice (low incidence of cardiomyothy), indicating that changes in this pathway may be one of the earliest indicators of cardiac stress. The cytoskeleton of the cardiomyocyte functions as a bridge to transmit the contractile force between the sarcomere and the extracellular matrix. This connection is essential for transforming sarcomere contraction to cardiac output, hence the induction of this system when the myocardium is stressed likely reflects an adaptive response to maintain cardiac output. The mediators of this adaptive response include a number of transcription factors such as calcium-regulated NFATc, JNK-regulated ATF2, Rho/ROCK-regulated GATA4, and p38-regulated Mef2 (Rohini et al. 2009). Notably, all three of these signaling modules are differentially regulated between the two mouse strains and the F1 hybrid, either at the transcription factor or kinase level.

Pathway analysis also indicted that genes associated with fatty acid metabolism (branched-chain amino acid degradation) were expressed at lower levels in C3H/HeJ mice, indicating a shift toward a fetal heart metabolic profile. The fetal heart relies more heavily on anaerobic metabolism of glucose for energy production (Rajabi et al. 2007). Gene expression changes observed in C3H/HeJ mice that are particularly indicative of this metabolic shift include the up-regulation of Pfkfb1 (glycolysis regulation), Gapdh (glycolysis), and Eno2 (glycolysis) and the down-regulation of Pfkfb1 (glycolysis regulation, cardiac specific), Pdk2 (negative regulation of pyruvate metabolism), Pdk4 (negative regulation of pyruvate metabolism), Bpgm (counterproductive to glycolysis), Pdhb (pyruvate metabolism), Slc25a20 (β-oxidation), Acad9 (β-oxidation), Acot1 (β-oxidation), Acot2 (β-oxidation), Cpt2 (β-oxidation), Acaa1a (β-oxidation), Acaa1b (β-oxidation), Acaa2 (β-oxidation), Sdha (TCA cycle), Ndufaf1 (oxidative phosphorylation/electron transport), Ndufa12 (oxidative phosphorylation/electron transport), Coq7 (oxidative phosphorylation/electron transport), Coq3 (oxidative phosphorylation/electron transport), Atpaf2 (oxidative phosphorylation/electron transport), and Atpaf1 (oxidative phosphorylation/electron transport). A striking exception to these observations is the notably higher expression of Slc25a29 (β-oxidation) in C3H/HeJ mice. Anaerobic metabolism gives rise to acidosis because of the production of lactic acid. C3H/HeJ mice express relatively higher levels of Slc16a7, which is involved in the export of H+ ions and monocarboxylic acids such as lactate. Overall, these findings indicate the C3H/HeJ mice are experiencing mild cardiac stress and the cardiac muscle is attempting to adapt to the adverse environment by altering its contractile and metabolic characteristics in a fashion that is in line with previously documented genomic changes associated with mCM (Rajabi et al. 2007).

Using a gene network–based approach, we identified genes that occupy regulatory gene nodes that are likely to have a critical role in driving the molecular level changes observed in mCM. Of the genes that were identified using this approach, Prkaa2 is one of the most likely to play the role of central regulator of the observed genome level responses. Prkaa2 encodes a catalytic subunit of the AMP-activated protein kinase (AMPK) (Arad et al. 2007) that is expressed at relatively high levels in both cardiac and skeletal muscle (http://biogps.gnf.org/#goto=genereport&id=5563). AMPK is a heterotrimer composed of an α catalytic subunit, and noncatalytic β and γ subunits. AMPK monitors cellular energy status (Viollet et al. 2009). In response to metabolic stress that leads to an increase in the AMP/ATP ratio, AMPK is activated and in turn activates expression of genes that promote glycolysis and mitochondrial energy production and shuts down energy-consuming processes such as protein synthesis (Viollet et al. 2009). The evidence supporting a role for Prkaa2/AMPK, both in C3H/HeJ mouse mCM and cardiac function in general, are manifold. The heart has a very high demand for energy, consuming more energy (ATP) per day than any other organ (Neubauer 2007). Hence, slight alterations in the AMPK signaling pathway that is central to maintaining a proper AMP/ATP balance has a high likelihood of compromising cardiac function. Along these lines, knockout and transgenic studies indicate Prkaa2 is necessary for maintaining myocardial energy homeostasis during ischemia (Zarrinpashneh et al. 2006), attenuating pressure-overload–induced cardiac hypertrophy (Zhang et al. 2008), maintaining mitochondrial function (Athea et al. 2007), and controlling circulating levels of catecholamines (Viollet et al. 2003). The increase in circulating catechols in the Prkaa2 knockout mice may be of particular significance because the focal pathology and ECG results observed in the C3H/HeJ mice are consistent with excessive adrenergic tone (Jokinen et al. 2005).

A mutation in human PRKAG2 that produces a constitutive active form of this protein alters the levels of Prkaa2 activity and leads to a rare form of hypertrophic hCM (Blair et al. 2001; Gollob et al. 2001). In our study, compared to B6C3F1/J and C57BL/6J mice, C3H/HeJ mice express lowered levels of not only Prkaa2 mRNA, but also of Prkag2 and Prkab2. Pathway analysis demonstrated a number of processes regulated by AMPK signaling were altered, such as fatty acid and carbohydrate metabolism. In addition, ConsensusPathDB pathway neighborhood analysis revealed that there was a significant enrichment of differentially expressed genes that interact at a pathway level with four additional subunits of AMPK (Prkab1, Prkaa1, Prkag1, and Prkag3), not including Prkaa2. In combination, these data strongly suggest that there is differential activity of the AMPK pathway between the two mouse strains and the F1 hybrid. Hypothetically, the lower activity of this pathway in C3H/HeJ may render these mice less capable of maintaining energy homeostasis during metabolic stress. Compromised energy levels could potentially lead to an inability to maintain Ca⁺² homeostasis (Yano et al. 2005), leading to the activation of caspases, cardiomyocyte apoptosis (Foo et al. 2005), and mCM.

The differential activity between the two mouse strains and the F1 hybrid of the AMPK pathway and the closely related PI3K/PDK1/Akt/mTor/S6K signaling network suggests the orthogonal networks in humans are likely to exhibit polymorphic activity. The differential activity is likely to be a result of a combination of underlying disease/physiological state and genetic polymorphism. A number of FDA-approved and investigational drugs target these networks directly or indirectly, raising the possibility that differential activity of these networks may confer increased risk of toxicity in certain subpopulations. An example of such a drug is the tyrosine kinase inhibitor Sunitinib, which is used to treat renal cell carcinoma and gastrointestinal stromal tumors and has been associated with cardiotoxicity (Cheng and Force 2010). The cause of the sunitinib-associated cardiotoxicity is likely to be multifactorial, however, one facet of the toxicity appears to be related to inhibition of AMPK, which compromises the ability of the cardiomyocytes to adapt to energy stress (Cheng and Force 2010). Another example of a cardiotoxic drug that indirectly interacts with the AMPK pathway is Zidovudine, a nucleoside analog reverse transcriptase inhibitor used to treat human immunodeficiency virus (Walker et al. 2004). The cardiotoxicity of this drug appears to be related to its inhibition of mitochondrial thymidine phosphorylation (McKee et al. 2004), which leads to mitochondrial toxicity and energy stress (Chan et al. 2007). Because of its critical role in adaptation to energy stress, differential activity of the AMPK pathway has the potential to confer differential susceptibility to Zidovudine-induced cardiotoxicity. The PI3K/PDK1/Akt/mTor/S6K signaling network is central to adaptive/maladaptive hypertrophy and the survival of cardiomyocytes (Cheng and Force 2010). Among the drugs that target this pathway are the mTor inhibitors, temsirolimus, everolimus, and sirolimus. On their own, these drugs attenuate cardiac hypertrophy in experimental models (Shioi et al. 2003). However, there is concern that inhibition of mTor, which blocks the ability of the cardiomyocyte to adapt to energy stress, may confer increased risk of cardiotoxicity in the presence of drugs (currently under development) that target other kinases in the PI3K/PDK1/Akt/mTor/S6K network (Cheng and Force 2010). In individuals/strains that exhibit compromised mTor activity, the risk of drug-induced cardiotoxicity may be greater.

These studies characterize the distinct cardiac phenotypes of C3H/HeJ, C57BL/6J, and B6C3F1 mice. Although others have evaluated vascular gene expression differences between these strains (Miyoshi et al. 2006; Yuan et al. 2009), the studies presented here are the first to detail the functional genomic differences in cardiac muscle between the strains. Major phenotype differences among the mouse strains studied included differential expression of cardiac gene transcripts for ion channel regulation/expression, for cardiac fetal development gene patterns, and for regulation of energy generation. Based on the conservation of genes among mammalian species, the mouse is considered to be a good model to study the relationship between gene transcript expression and function. Importantly, a large number of gene expression changes identified here have counterparts in human cardiac disease.

A supplemental appendix to this article is published electronically only at http://tpx.sagepub.com/supplemental. The original microarray data are available on the CEBS database.

Supplemental Material

Supplementary_tables - Comparative Phenotypic Assessment of Cardiac Pathology, Physiology, and Gene Expression in C3H/HeJ, C57BL/6J, and B6C3F1/J Mice

Supplementary_tables for Comparative Phenotypic Assessment of Cardiac Pathology, Physiology, and Gene Expression in C3H/HeJ, C57BL/6J, and B6C3F1/J Mice by Scott S. Auerbach, Reuben Thomas, Ruchir Shah, Hong Xu, Molly K. Vallant, Abraham Nyska and June K. Dunnick in Toxicologic Pathology

Footnotes

Acknowledgments

We thank Dr. Jef French, NIEHS, and Dr. K. Shockley, NIEHS, for their excellent review of the manuscript. We would also like to thank Dr. Vincent VanBuren and Dr. Daniel C. Jupiter for generously providing the cardiac gene expression correlation matrix used in the creation of the coexpression network analysis. This work was supported by the intramural program of the National Institute of Environmental Health Sciences Intramural Program, Research Triangle Park, NC. The in-life portion of this study was conducted under NIEHS contract N01-ES-55536. Statistical support was provided under NIEHS contract ES 55547.

References

Aiba

Tomaselli

G. F.

(2010). Electrical remodeling in the failing heart. Curr Opin Cardiol 25, 29–36.

Akar

F. G.

Tomaselli

G. F.

(2005). Ion channels as novel therapeutic targets in heart failure. Ann Med 37, 44–54.

Arad

Seidman

C. E.

Seidman

J. G.

(2007). AMP-activated protein kinase in the heart: role during health and disease. Circ Res 100, 474–88.

Athea

Viollet

Mateo

Rousseau

Novotova

Garnier

Vaulont

Wilding

J. R.

Grynberg

Veksler

Hoerter

Ventura-Clapier

(2007). AMP-activated protein kinase alpha2 deficiency affects cardiac cardiolipin homeostasis and mitochondrial function. Diabetes 56, 786–94.

Bazett

(1920). An analysis of time relation of electrocardiograms. Heart 7, 353–67.

Berthonneche

Peter

Schupfer

Hayoz

Kutalik

Abriel

Pedrazzini

Beckmann

J. S.

Bergmann

Maurer

(2009). Cardiovascular response to beta-adrenergic blockade or activation in 23 inbred mouse strains. PloS One 4, e6610.

Bettahi

Marker

C. L.

Roman

M. I.

Wickman

(2002). Contribution of the Kir3.1 subunit to the muscarinic-gated atrial potassium channel IKACh. J Biol Chem 277, 48282–88.

Bhattacharjee

Joiner

W. J.

Yang

Sigworth

F. J.

Kaczmarek

L. K.

(2003). Slick (Sl02.1), a rapidly gating sodium-activated potassium channel inhibited by ATP. J Neurosci 23, 11681–91.

Blair

Redwood

Ashrafian

Oliveira

Broxholme

Kerr

Salmon

Ostman-Smith

Watkins

(2001). Mutations in the gamma(2) subunit of AMP-activated protein kinase cause familial hypertrophic cardiomyopathy: evidence for the central role of energy compromise in disease pathogenesis. Human Mol Genet 10, 1215–20.

10.

Boudina

Abel

E. D.

(2007). Diabetic cardiomyopathy revisited. Circulation 115, 3213–23.

11.

Chan

S. S.

Santos

J. H.

Meyer

J. N.

Mandavilli

B. S.

Cook

D. L.

Jr. McCash

C. L.

Kissling

G. E.

Nyska

Foley

J. F.

van Houten

Copeland

W. C.

Walker

V. E.

Witt

K. L.

Bishop

J. B.

(2007). Mitochondrial toxicity in hearts of CD-1 mice following perinatal exposure to AZT, 3TC, or AZT/3TC in combination. Environ Mol Mutagen 48, 190–200.

12.

Cheng

Force

(2010). Molecular mechanisms of cardiovascular toxicity of targeted cancer therapeutics. Circ Res 106, 21–34.

13.

Chien

K. R.

Olson

E. N.

(2002). Converging pathways and principles in heart development and disease: CV@CSH. Cell 110, 153–62.

14.

Dahl

Danzi

Klein

(2008). Thyrotoxic cardiac disease. Curr Heart Fail Rep 5, 170–76.

15.

Dunnick

Blackshear

Kissling

G. E.

Cunnningham

Parker

Nyska

(2006). Critical pathways in heart function: bis(2-chloroethoxy)methane-induced heart gene transcript change in F344 rats. Toxicol Pathol 34, 348–56.

16.

Dunnick

Johnson

Horton

Nyska

(2004). Bis(2-chloroethoxy)methane-induced mitochondrial and myofibrillar damage: short-term time-course study. Toxicol Sci 80, 243–52.

17.

Dunnick

J. K.

Kissling

Gerken

D. K.

Vallant

M. A.

Nyska

(2007). Carditoxicity of Ma Huang/caffeine or ephedrine/caffeine in a rodent model system. Toxicol Pathol 35, 657–664.

18.

Dunnick

J. K.

Lieuallen

Moyer

Orzech

Nyska

(2004). Cardiac damage in rodents after exposure to bis(2-chloroethoxy)methane. Toxicol Pathol 32, 309–317.

19.

Dunnick

J. K.

Thayer

K. A.

Travlos

G. S.

(2007). Inclusion of biomarkers for detecting perturbations in the heart and lung and lipid/carbohydrate metabolism in National Toxicology Program studies. Toxicol Sci 100, 29–35.

20.

Eaton

G. J.

Custer

R. P.

Johnson

F. N.

Stabenow

K. T.

(1978). Dystrophic cardiac calcinosis in mice: genetic, hormonal, and dietary influences. The American journal of pathology 90, 173–86.

21.

Foo

R. S.

Mani

Kitsis

R. N.

(2005). Death begets failure in the heart. The Journal of clinical investigation 115, 565–71.

22.

Fowler

S. J.

Napolitano

Priori

S. G.

(2009). The genetics of cardiomyopathy: genotyping and genetic counseling. Current treatment options in cardiovascular medicine 11, 433–46.

23.

Frijters

Heupers

van Beek

Bouwhuis

van Schaik

de Vlieg

Polman

Alkema

(2008). CoPub: a literature-based keyword enrichment tool for microarray data analysis. Nucleic Acids Res 36, W406–10.

24.

Galindo

C. L.

Skinner

M. A.

Errami

Olson

L. D.

Watson

D. A.

McCormick

J. F.

McIver

L. J.

Kumar

N. M.

Pham

T. Q.

Garner

H. R.

(2009). Transcriptional profile of isoproterenol-induced cardiomyopathy and comparison to exercise-induced cardiac hypertrophy and human cardiac failure. BMC physiology 9, 23.

25.

Gollob

M. H.

Green

M. S.

Tang

A. S.

Gollob

Karibe

Ali Hassan

A. S.

Ahmad

Lozado

Shah

Fananapazir

Bachinski

L. L.

Roberts

(2001). Identification of a gene responsible for familial Wolff-Parkinson-White syndrome. The New England journal of medicine 344, 1823–31.

26.

Institute of Laboratory Animal Resources (1996). Guide for the care and use of laboratory animals. National Academy Press, Washington, D. C.

27.

Ivandic

B. T.

Qiao

J. H.

Machleder

Liao

Drake

T. A.

Lusis

A. J.

(1996). A locus on chromosome 7 determines myocardial cell necrosis and calcification (dystrophic cardiac calcinosis) in mice. Proc Natl Acad Sci U S A. 1996 May 28;93(11):5483–8.

28.

James

P. F.

Grupp

I. L.

Grupp

Woo

A. L.

Askew

G. R.

Croyle

M. L.

Walsh

R. A.

Lingrel

J. B.

(1999). Identification of a specific role for the Na,K-ATPase alpha 2 isoform as a regulator of calcium in the heart. Molecular cell 3, 555–63.

29.

Jessup

Brozena

(2003). Heart failure. The New England journal of medicine 348, 2007–18.

30.

Jokinen

M. P.

Lieuallen

W. G.

Johnson

C. L.

Dunnick

Nyska

(2005). Characterization of spontaneous and chemically induced cardiac lesions in rodent model systems: the National Toxicology Program experience. Cardiovascular toxicology 5, 227–44.

31.

Jupiter

D. C.

VanBuren

(2008). A visual data mining tool that facilitates reconstruction of transcription regulatory networks. PloS one 3, e1717.

32.

Kamburov

Wierling

Lehrach

Herwig

(2009). ConsensusPathDB—a database for integrating human functional interaction networks. Nucleic Acids Res 37, D623–8.

33.

Kane

G. C.

Liu

X. K.

Yamada

Olson

T. M.

Terzic

(2005). Cardiac KATP channels in health and disease. Journal of molecular and cellular cardiology 38, 937–43.

34.

Kathiresan

Voight

B. F.

Purcell

Musunuru

Ardissino

Mannucci

P. M.

Anand

Engert

J. C.

Samani

N. J.

Schunkert

Erdmann

Reilly

M. P.

Rader

D. J.

Morgan

Spertus

J. A.

Stoll

Girelli

McKeown

P. P.

Patterson

C. C.

Siscovick

D. S.

O’Donnell

C. J.

Elosua

Peltonen

Salomaa

Schwartz

S. M.

Melander

Altshuler

Ardissino

Merlini

P. A.

Berzuini

Bernardinelli

Peyvandi

Tubaro

Celli

Ferrario

Fetiveau

Marziliano

Casari

Galli

Ribichini

Rossi

Bernardi

Zonzin

Piazza

Mannucci

P. M.

Schwartz

S. M.

Siscovick

D. S.

Yee

Friedlander

Elosua

Marrugat

Lucas

Subirana

Sala

Ramos

Kathiresan

Meigs

J. B.

Williams

Nathan

D. M.

MacRae

C. A.

O’Donnell

C. J.

Salomaa

Havulinna

A. S.

Peltonen

Melander

Berglund

Voight

B. F.

Kathiresan

Hirschhorn

J. N.

Asselta

Duga

Spreafico

Musunuru

Daly

M. J.

Purcell

Voight

B. F.

Purcell

Nemesh

Korn

J. M.

McCarroll

S. A.

Schwartz

S. M.

Yee

Kathiresan

Lucas

Subirana

Elosua

Surti

Guiducci

Gianniny

Mirel

Parkin

Burtt

Gabriel

S. B.

Samani

N. J.

Thompson

J. R.

Braund

P. S.

Wright

B. J.

Balmforth

A. J.

Ball

S. G.

Hall

A. S.

Schunkert

Erdmann

Linsel-Nitschke

Lieb

Ziegler

Konig

Hengstenberg

Fischer

Stark

Grosshennig

Preuss

Wichmann

H. E.

Schreiber

Schunkert

Samani

N. J.

Erdmann

Ouwehand

Hengstenberg

Deloukas

Scholz

Cambien

Reilly

M. P.

Chen

Wilensky

Matthai

Qasim

Hakonarson

H. H.

Devaney

Burnett

M. S.

Pichard

A. D.

Kent

K. M.

Satler

Lindsay

J. M.

Waksman

Epstein

S. E.

Rader

D. J.

Scheffold

Berger

Stoll

Huge

Girelli

Martinelli

Olivieri

Corrocher

Morgan

Spertus

J. A.

McKeown

Patterson

C. C.

Schunkert

Erdmann

Linsel-Nitschke

Lieb

Ziegler

Konig

I. R.

Hengstenberg

Fischer

Stark

Grosshennig

Preuss

Wichmann

H. E.

Schreiber

Holm

Thorleifsson

Thorsteinsdottir

Stefansson

Engert

J. C.

Xie

Anand

Kathiresan

Ardissino

Mannucci

P. M.

Siscovick

O’Donnell

C. J.

Samani

N. J.

Melander

Elosua

Peltonen

Salomaa

Schwartz

S. M.

Altshuler

(2009). Genome-wide association of early-onset myocardial infarction with single nucleotide polymorphisms and copy number variants. Nature genetics 41, 334–41.

35.

Le Bouter

Demolombe

Chambellan

Bellocq

Aimond

Toumaniantz

Lande

Siavoshian

Baro

Pond

A. L.

Nerbonne

J. M.

Leger

J. J.

Escande

Charpentier

(2003). Microarray analysis reveals complex remodeling of cardiac ion channel expression with altered thyroid status: relation to cellular and integrated electrophysiology. Circulation research 92, 234–42.

36.

Makita

Horie

Nakamura

Sasaki

Yokoi

Sakurai

Sakuma

Otani

Sawa

Kitabatake

(2002). Drug-induced long-QT syndrome associated with a subclinical SCN5A mutation. Circulation 106, 1269–74.

37.

Maronpot

R. R.

Boorman

G. A.

Gaul

B. W.

(1999). Pathology of the mouse: reference and atlas. Cache River Press, Vienna, IL.

38.

McKee

E. E.

Bentley

A. T.

Hatch

Gingerich

Susan-Resiga

(2004). Phosphorylation of thymidine and AZT in heart mitochondria: elucidation of a novel mechanism of AZT cardiotoxicity. Cardiovascular toxicology 4, 155–67.

39.

Medeiros-Domingo

Kaku

Tester

D. J.

Iturralde-Torres

Itty

Valdivia

Ueda

Canizales-Quinteros

Tusie-Luna

M. T.

Makielski

J. C.

Ackerman

M. J.

(2007). SCN4B-encoded sodium channel beta4 subunit in congenital long-QT syndrome. Circulation 116, 134–42.

40.

Meng

Vera

Che

Wang

S. S.

Ingram-Drake

Schadt

E. E.

Drake

T. A.

Lusis

A. J.

(2007). Identification of Abcc6 as the major causal gene for dystrophic cardiac calcification in mice through integrative genomics. Proc Natl Acad Sci U S A. 2007 Mar 13;104(11):4530–5. Epub 2007 Mar 6.

41.

Miyoshi

Tian

Matsumoto

A. H.

Shi

(2006). Differential response of vascular smooth muscle cells to oxidized LDL in mouse strains with different atherosclerosis susceptibility. Atherosclerosis 189, 99–105.

42.

Morita

Seidman

C. E.

(2005). Genetic causes of human heart failure. The Journal of clinical investigation 115, 518–26.

43.

National Center for Health Statistics (2005a). Mortality data from the national vital statistics system. http://www.cdc.gov/nchs/deaths.htm

44.

National Center for Health Statistics (2005b). Heart Disease Fact Sheet. http://www.cdc.gov/cvh/library/fs_heart_disease.htm

45.

Neubauer

(2007). The failing heart—an engine out of fuel. The New England journal of medicine 356, 1140–51.

46.

Nyska

Murphy

Foley

J. F.

Collins

B. J.

Petranka

Howden

Hanlon

Dunnick

J. K.

(2005a). Acut hemorrhagic myocardial necrosis and sudden death of rats exposed to a combination of ephedrine and caffeine. Tox Sci 83, 388–396.

47.

Nyska

Murphy

Foley

J. F.

Collins

B. J.

Petranka

Howden

Hanlon

Dunnick

J. K.

(2005b). Acute hemorrhagic myocardial necrosis and sudden death of rats exposed to a combination of ephedrine and caffeine. Toxicol Sci 83, 388–96.

48.

Pritchard

J. B.

French

J. E.

Davis

B. J.

Haseman

J. K.

(2003). The role of transgenic mouse models in carcinogen identification. Environmental health perspectives 111, 444–54.

49.

Rajabi

Kassiotis

Razeghi

Taegtmeyer

(2007). Return to the fetal gene program protects the stressed heart: a strong hypothesis. Heart Fail Rev 12, 331–43.

50.

Rao

G. N.

Crockett

P. W.

(2003). Effect of diet and housing on growth, body weight, survival and tumor incidences of B6C3F1 mice in chronic studies. Toxicol Pathol 31, 243–50.

51.

Rohini

Agrawal

Koyani

C. N.

Singh

(2009). Molecular targets and regulators of cardiac hypertrophy. Pharmacol Res.

52.

Ruben

Arceo

R. J.

Bishop

S. P.

Elwell

M. R.

Kerns

W. D.

Mesfin

G. M.

Sandusky

G. E.

Van Vleet

J. F.

(2000). Non-proliferative lesions of the heart and vasculature in rats. STP/ARP/AFIP, Washington DC.

53.

Samani

N. J.

Erdmann

Hall

A. S.

Hengstenberg

Mangino

Mayer

Dixon

R. J.

Meitinger

Braund

Wichmann

H. E.

Barrett

J. H.

Konig

I. R.

Stevens

S. E.

Szymczak

Tregouet

D. A.

Iles

M. M.

Pahlke

Pollard

Lieb

Cambien

Fischer

Ouwehand

Blankenberg

Balmforth

A. J.

Baessler

Ball

S. G.

Strom

T. M.

Braenne

Gieger

Deloukas

Tobin

M. D.

Ziegler

Thompson

J. R.

Schunkert

(2007). Genomewide association analysis of coronary artery disease. The New England journal of medicine 357, 443–53.

54.

Shi

Wang

N. J.

Shih

D. M.

Sun

V. Z.

Wang

Lusis

A. J.

(2000). Determinants of atherosclerosis susceptibility in the C3H and C57BL/6 mouse model: evidence for involvement of endothelial cells but not blood cells or cholesterol metabolism. Circulation research 86, 1078–84.

55.

Shioi

McMullen

J. R.

Tarnavski

Converso

Sherwood

M. C.

Manning

W. J.

Izumo

(2003). Rapamycin attenuates load-induced cardiac hypertrophy in mice. Circulation 107, 1664–70.

56.

Taha

Lopaschuk

G. D.

(2007). Alterations in energy metabolism in cardiomyopathies. Annals of medicine 39, 594–607.

57.

Viollet

Andreelli

Jorgensen

S. B.

Perrin

Geloen

Flamez

Lenzner

Baud

Bennoun

Gomas

Nicolas

Wojtaszewski

J. F.

Kahn

Carling

Schuit

F. C.

Birnbaum

M. J.

Richter

E. A.

Burcelin

Vaulont

(2003). The AMP-activated protein kinase alpha2 catalytic subunit controls whole-body insulin sensitivity. The Journal of clinical investigation 111, 91–8.

58.

Viollet

Athea

Mounier

Guigas

Zarrinpashneh

Horman

Lantier

Hebrard

Devin-Leclerc

Beauloye

Foretz

Andreelli

Ventura-Clapier

Bertrand

(2009). AMPK: Lessons from transgenic and knockout animals. Front Biosci 14, 19–44.

59.

Walker

D. M.

Poirier

M. C.

Campen

M. J.

Cook

D. L.

Jr. Divi

R. L.

Nagashima

Lund

A. K.

Cossey

P. Y.

Hahn

F. F.

Walker

V. E.

(2004). Persistence of mitochondrial toxicity in hearts of female B6C3F1 mice exposed in utero to 3’-azido-3’-deoxythymidine. Cardiovascular toxicology 4, 133–53.

60.

Wallenborn

J. G.

Schladweiler

M. C.

Nyska

Johnson

J. A.

Thomas

Jaskot

R. H.

Richards

J. H.

Ledbetter

A. D.

Kodavanti

U. P.

(2007). Cardiopulmonary responses of Wistar Kyoto, spontaneously hypertensive, and stroke-prone spontaneously hypertensive rats to particulate matter (PM) exposure. Journal of toxicology and environmental health 70, 1912–22.

61.

Wang

S. S.

Schadt

E. E.

Wang

Ingram-Drake

Shi

Drake

T. A.

Lusis

A. J.

(2007). Identification of pathways for atherosclerosis in mice: integration of quantitative trait locus analysis and global gene expression data. Circulation research 101, e11–30.

62.

Wang

S. S.

Shi

Wang

Velky

Greenlee

Wang

M. T.

Drake

T. A.

Lusis

A. J.

(2007). Mapping, genetic isolation, and characterization of genetic loci that determine resistance to atherosclerosis in C3H mice. Arteriosclerosis, thrombosis, and vascular biology 27, 2671–6.

63.

Watanabe

Koopmann

T. T.

Le Scouarnec

Yang

Ingram

C. R.

Schott

J. J.

Demolombe

Probst

Anselme

Escande

Wiesfeld

A. C.

Pfeufer

Kaab

Wichmann

H. E.

Hasdemir

Aizawa

Wilde

A. A.

Roden

D. M.

Bezzina

C. R.

(2008). Sodium channel beta1 subunit mutations associated with Brugada syndrome and cardiac conduction disease in humans. The Journal of clinical investigation 118, 2260–8.

64.

Wheeler

F. C.

Tang

Marks

O. A.

Hadnott

T. N.

Chu

P. L.

Mao

Rockman

H. A.

Marchuk

D. A.

(2009). Tnni3k modifies disease progression in murine models of cardiomyopathy. PLoS genetics 5, e1000647.

65.

Wilson

C. L.

Miller

C. J.

(2005). Simpleaffy: a BioConductor package for Affymetrix Quality Control and data analysis. Bioinformatics (Oxford, England) 21, 3683–5.

66.

C. C.

Huang

H. C.

Juan

H. F.

Chen

S. T.

(2004). GeneNetwork: an interactive tool for reconstruction of genetic networks using microarray data. Bioinformatics (Oxford, England) 20, 3691–3.

67.

Yang

Schadt

E. E.

Wang

Arnold

A. P.

Ingram-Drake

Drake

T. A.

Lusis

A. J.

(2006). Tissue-specific expression and regulation of sexually dimorphic genes in mice. Genome research 16, 995–1004.

68.

Yano

Ikeda

Matsuzaki

(2005). Altered intracellular Ca2+ handling in heart failure. The Journal of clinical investigation 115, 556–64.

69.

Yuan

Miyoshi

Bao

Sheehan

J. P.

Matsumoto

A. H.

Shi

(2009). Microarray analysis of gene expression in mouse aorta reveals role of the calcium signaling pathway in control of atherosclerosis susceptibility. American journal of physiology 296, H1336–43.

70.

Zarrinpashneh

Carjaval

Beauloye

Ginion

Mateo

Pouleur

A. C.

Horman

Vaulont

Hoerter

Viollet

Hue

Vanoverschelde

J. L.

Bertrand

(2006). Role of the alpha2-isoform of AMP-activated protein kinase in the metabolic response of the heart to no-flow ischemia. American journal of physiology 291, H2875–83.

71.

Zhang

Fassett

Zhu

Viollet

Wiczer

Bernlohr

D. A.

Bache

R. J.

Chen

(2008). AMP activated protein kinase-alpha2 deficiency exacerbates pressure-overload-induced left ventricular hypertrophy and dysfunction in mice. Hypertension 52, 918–24.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

9.42 MB