Echocardiography phenotyping in murine genetic reference population of BXD strains reveals significant QTLs associated with cardiac function and morphology

Abstract

The genetic reference population of recombinant inbred BXD mice has been derived from crosses between C57BL/6J and DBA/2J strains. The DBA/2J parent exhibits cardiomyopathy phenotypes, whereas C57BL/6J has normal heart. BXD mice are sequenced for studying genetic interactions in cardiomyopathies. The study aimed to assess cardiomyopathy traits in BXDs and investigate the quantitative genetic architecture of those traits. Echocardiography, blood pressure, and cardiomyocyte size parameters obtained from 44 strains of BXD family (n > 5/sex) at 4–5 mo of age were associated with heart transcriptomes and expression quantitative trait loci (eQTL) mapping was performed. More than twofold variance in ejection fraction (EF%), fractional shortening (FS%), left ventricular volumes (LVVols), internal dimensions (LVIDs), mass (LVM), and posterior wall (LVPW) thickness was found among BXDs. In male BXDs, eQTL mapping identified Ndrg4 on chromosome 8 QTL to be positively correlated with LVVol and LVID and negatively associated with cardiomyocyte diameter. In female BXDs, significant QTLs were found on chromosomes 7 and 3 to be associated with LVPW and EF% and FS%, respectively, and Josd2, Dap3, and Tpm3 were predicted as strong candidate genes. Our study found variable cardiovascular traits among BXD strains and identified multiple associated QTLs, suggesting an influence of genetic background on expression of echocardiographic and cardiomyocyte diameter traits. Increased LVVol and reduced EF% and FS% represented dilated cardiomyopathy, whereas increased LV mass and wall thickness indicated hypertrophic cardiomyopathy traits. The BXD family is ideal for identifying candidate genes, causal and modifier, that influence cardiovascular phenotypes.

NEW & NOTEWORTHY This study aimed to establish a cardiac phenotype-genotype correlation in murine genetic reference population of BXD RI strains by phenotyping the echocardiography, blood pressure, and cardiomyocyte diameter traits and associating each collected phenotype with genetic background. Our study identified several QTLs and candidate genes that have significant association with cardiac hypertrophy, ventricular dilation, and function including systolic hyperfunction and dysfunction.

INTRODUCTION

Heart disease is the leading cause of death worldwide with global cardiovascular deaths increased by 41% since 1990, reaching 17.9 million deaths in 2019 and this curve is expected to rise to >23 million deaths annually by 2030 (1). One of the emerging practices to improve prevention and treatment of cardiac diseases is personalized or precision medicine, an integrative approach based on the use of genotypic (genetic variants and gene expression) or phenotypic (disease signs and severity) traits to predict patient-tailored risks for selecting optimal and effective disease management based on association between genes and disease signs (2). However, predictive precision medicine modeling in human populations and diseases is challenging due to complex interactions among genes, variants, and proteins, combined with lifestyles and environments. Therefore, to model some of the genetic and phenotypic complexity, experimental precision medicine involving large families of isogenic lines of recombinant inbred (RI) BXD mice representing a murine genetic reference population (GRP) have been developed. The BXD RI GRP panel, descended from crosses between C57BL/6J (B6) and DBA/2J (D2) strains, was first generated at the Jackson Laboratory in the mid-1970s (3). Parental D2 strain exhibited hallmarks of cardiac hypertrophy and fibrosis and thus has been recognized as a natural mouse model for cardiomyopathy (4–6). Currently, more than 150 lines of isogenic BXD strains have been produced to experimentally test causal modeling and effectively validate genotype-phenotype associations using a systems genetics approach, since heritable traits can be replicated in each BXD strain and mapped with high power and precision (7). This murine GRP panel segregates ∼6 million single nucleotide polymorphisms (SNPs) or variations (SNVs), thousands of insertions or deletions, and copy number variants that segregate among BXD strains (8, 9).

Numerous causal genes have been identified to cause cardiomyopathies, devastating progressive diseases of heart muscle with no effective treatment worldwide (10). Five types of cardiomyopathies are distinguished clinically: dilated cardiomyopathy (DCM), hypertrophic cardiomyopathy (HCM), restrictive cardiomyopathy (RCM), arrhythmogenic cardiomyopathy (ACM), and left ventricular noncompaction cardiomyopathy (LVNC) (11). DCM is distinguished by left ventricular (LV) dilation and systolic or contractile dysfunction (12); HCM is characterized by LV hypertrophy with diastolic or relaxation dysfunction, and RCM is associated with increased stiffness of the myocardium and dilated atria due to diastolic dysfunction without significant myocardial hypertrophy (13). ACM is characterized by frequent super- and ventricular tachyarrhythmias (VTAs) and sudden death, featuring progressive heart failure and fibro-fatty infiltration in the myocardium (14), whereas LVNC is distinguished by prominent LV trabeculae with intertrabecular recesses and a thin compacted myocardial layer (15). All of these morphofunctional features can be revealed by transthoracic echocardiography assessment.

According to the “final common pathway” hypothesis, disease-causal variants or mutations in sarcomeric contractile protein encoding genes are commonly associated with HCM and RCM (disease of the sarcomere) (16), whereas the genes that function as a link between cardiomyocyte membrane and sarcomere are the most familial in DCM cases (17, 18). ACM commonly develops as a result of mutations in genes expressed in cell-cell connections and desmosomes (19), whereas an “embryonic” hypothesis is more accepted for the development of LVNC (20). Overlap of the phenotypes and genetic etiologies is common between cardiomyopathy types, and phenotypes may vary considerably among patients with cardiomyopathy, even among family members who carry identical mutation (21, 22). This variability in cardiomyopathies is not explained by current knowledge, suggesting more complexity, potentially involving currently unknown genetic and epistatic factors. It is likely that cardiac phenotypes depend on epistatic interactions between multiple genes, causal and modifier, an individual’s genetic background, and environmental factors that differentially affect the manifestation of the cardiomyopathy phenotypes (23). Despite numerous animal studies employed to understand the mechanisms of cardiomyopathies and heart failure, the lack of their widespread use for human-based research persists (24). The reason is that most animal studies are carried out on a single genotype engineered on a fixed genetic background and therefore results are often challenging to translate into and explain human phenotype diversity (25).

In this study, we aimed to establish echocardiographic panels for cardiac function and morphology in male and female BXD strains of murine GRP and perform quantitative trait loci (QTL) mapping to understand the genetic basis of cardiac function and morphological diversity and translate the results into human cardiomyopathy phenotypes (26, 27). For this, we have carried out two-dimensional transthoracic echocardiography and blood pressure assessment in parental B6 and D2 strains and 42 BXD lines, evaluated the variation of all cardiac traits, and calculated heritability for each collected trait.

MATERIALS AND METHOD

BXD Lines

All animal studies were approved by the Institutional Animal Care and Use Committee (IACUC) of the University of Tennessee Health Science Center (UTHSC). BXD mice were maintained in micro-isolator cages at 25°C with free access to water and food at the 12:12-h light-dark cycle. For cardiac phenotyping, we used male and female B6 (control parental), D2 (natural model of cardiomyopathy), and 42 BXD RI mouse strains (n > 5) at 4–5 mo of age.

Evaluation of Heart Morphology and Function by Two-Dimensional Transthoracic Echocardiography

Transthoracic echocardiography was performed to evaluate heart function and morphology in mice by obtaining short- and long-axis cross-sectional, two-dimensional, and Doppler parameters using a Vevo2100 Micro-Imaging System (VisualSonics Inc., Toronto, Canada) with a 30-MHz transducer. A day before echocardiography, the chest of the mice was treated with a chemical hair remover to reduce ultrasound attenuation. Mice were anesthetized by oxygenated 1%–2% isoflurane and core temperature was maintained using a heated platform set at 37°C and monitored using rectal thermometer. Heart rate (HR) was recorded using metal probes connected to the limbs for electrocardiogram gating. LV posterior wall (LVPW) and interventricular septum (IVS) thickness and chamber dimensions of LV and RV at end-systole (s) and end-diastole (d) were acquired in parasternal long axis, M-mode, or short-axis view. Images were stored in the ultrasound system hard drive and transferred to an image server for off-line analysis. Two independent researchers blinded to genotype performed measurements of ventricular internal dimensions and wall thickness and chamber volumes and ejection fraction (EF) and fractional shortening (FS) of LV were calculated for evaluation of systolic function. Statistical analysis of echocardiograms was performed with one-way ANOVA with Bonferroni post tests using GraphPad Prism6 software. Cardiac parameters are reported as mean ± SD and probability (P) values ≤ 0.05 are considered significant (Supplemental Table S1).

Blood Pressure Assessment

The CODA tail-cuff noninvasive blood pressure (BP) system (Kent Scientific Corporation) was used to measure systolic and diastolic blood pressure (SBP and DBP, respectively) by determining the tail blood volume. Mice (n > 5 mice/line/sex) were acclimated to the animal holder for 15 min a day for 3 days. Before measuring BP, mice were allowed to acclimate in the holder for 5 min. Each mouse underwent 15 measuring cycles, with the first five being acclimation cycles. Only true cycles were used in data analysis.

Measurement of Cardiomyocyte Diameter

The cardiomyocyte diameter values among BXD RI panel have been generated previously (4, 28). In this study, we collected cardiomyocyte diameter in parental B6 and D2 strains. Cryostat cardiac sections (6 μm) were stained with hematoxylin/eosin and examined under a light microscope. Myocytes with nuclei in the middle or near the middle were selected for the analysis. Single cardiomyocyte diameter was measured with images captured from the septum and ventricular walls (three sections/heart). Diameter of 50 to 70 cardiomyocytes was traced in each section. Quantification of myocyte diameter was done using NIH ImageJ software as described previously (28).

Cardiac Gene Expression and Data Availability

The heart gene expression data used in this study were generated through our collaborative effort (29). Briefly, the ventricle tissues were harvested from 40 strains of the BXD family and both parental strains (B6 and D2) at around 6–7 mo of age. RNA was extracted using QIAGEN RNA extraction kits (https://www.qiagen.com) per the manufacturer’s instructions. RNA samples from ∼5 mice of the same strain were pooled equally (by microgram of RNA) into a single RNA sample. The pooled RNA samples were then purified using RNEasy (Qiagen, Hilden, Germany). Samples that passed quality control (RIN > 8.0) were run on Affymetrix Mouse Gene 2.0 ST in a single batch at the University of Tennessee Health Science Center (UTHSC). Raw microarray data were first normalized using the Robust Multichip Array (RMA) method; then, the data were logged and Z-normalized. To remove negative values from the tables, we shifted the mean to 8 units and increased the standard deviation of each array dataset to two units such that a twofold difference in expression corresponded to an ∼1 unit increase in expression as judged by the spike-in controls. This cardiac gene expression data “EPFL/LISP BXD CD Heart Affy Mouse Gene 2.0 ST Gene Level” have been deposited at our GeneNetwork.org under the accession ID GN485. More detailed information can be found at http://gn1.genenetwork.org/webqtl/main.py?FormID=sharinginfo&GN_AccessionId=485.

Heritability Calculation

The heritability (h2) was estimated by the broad sense heritability method (30), in which variances among strain means (VA) were compared with total variance. The equation used was

$$\text{h2}=\frac{\text{0.5\hspace{0.17em}VA}}{\text{0.5\hspace{0.17em}VA}+\text{VE}}$$

in which VA is the variance among strain means and VE is the variance within strains.

Quantitative Trait Loci Mapping

Quantitative trait loci (QTL) mapping allows the identification of chromosomal regions that specifically modulate variability in levels of cardiac-related traits (echocardiography, BP, and cardiomyocyte diameter). The analysis was done using the WebQTL tool in GeneNetwork (www.genenetwork.org) (3, 31). Likelihood ratio statistics (LRS) scores were computed initially using Haley-Knott equations (32). This score was used for evaluating the relations between the traits and particular genotype markers across the genome. Genome-wide significance (P < 0.05) and suggestive levels were estimated using results of 2,000 permutations. The 2-LOD confidence interval was used to filter the candidate genes.

Expression Quantitative Trait Loci Analysis

For the genomic loci associated with the cardiac-related traits, expression QTL (eQTL) mapping was conducted for all the genes located within the 2-LOD QTL confidence interval with the above QTL mapping methods. Here, we specifically focused on cis-regulated genes defined within 10 Mb of the genetic marker.

Correlation between Cardiac-Related Phenotypes and Cardiac Gene Expression

Correlation between cardiac gene expression level and cardiac-related traits (phenotypes) was estimated by computing Pearson product-moment correlation on GeneNetwork. Pearson correlation of P < 0.05 was used for determining the significance.

Genomic Sequence Variations between B6 and D2

Variations between B6 and D2 were obtained from our previously generated whole genome resequencing data (7) and Mouse Genome Project (https://www.sanger.ac.uk/sanger/Mouse_SnpViewer/rel-1505) (33). Here, we focused on the genetic variants located in the coding sequence regions (nonsense, missense, and frameshift) and potential regulatory regions such as gene up/downstream 3/5′ UTRs.

Online Public Resources

To identify the known functional-related genes in the QTL intervals, we retrieved the genes implicated in the cardiac-related traits from the following two online resources: the rat genome database (RGD) (34) and the International Mouse Phenotyping Consortium (IMPC) database (35). RGD database (https://www.rgd.mcw.edu) provides updated genetic, genomic, phenotype, and disease data generated from mouse, rat, and human. A total of 450 genes were downloaded from RGD using the keywords “cardiomyocyte,” “myocyte,” and “cardiomyopathy.” The IMPC database (https://www.mousephenotype.org/) provides detailed phenotype data for the knockout mouse. A total of 636 genes were downloaded from IPMC using the keywords “cardiomyocyte,” “myocyte,” and “cardiomyopathy.”

Candidate Gene Prioritization

Our candidate gene prioritization included the following criteria: whether the QTL genes 1) harbor protein-altering variants or are cis-regulated in cardiac tissue; 2) are correlated with the measured cardiac-related traits including cardiomyocyte size; and 3) have been implicated in the cardiac function based on prior knowledge. We considered the strongest candidates with more evidence accumulated.

RESULTS

Variability in Left Ventricular Wall and Interventricular Septum Thickness among BXD Strains

Echocardiography used for assessing heart function and morphology revealed remarkably varied parameters and traits between the BXD lines as well as between male (M) and female (F) mice within the BXD strains. Cardiac hypertrophy is defined by increased thickness of LVPW and IVS at end-systole (s) and end-diastole (d) measured in M-mode. In male BXDs (Fig. 1A), the highest values for LVPW;s were found in BXD73M (1.45 ± 0.3 mm) and for LVPW;d in BXD71M (1.09 ± 0.2 mm). Eleven BXD strains in addition to parental D2M mouse showed significantly thicker LVPW;s than B6M parental strain and eight strains including B6 parental mice had significantly lower LVPW;s than that of D2 mouse, demonstrating important segregation of this trait among BXDs. Moreover, BXD71M, 73M, 70M, 90M, and 60M showed significantly thicker LVPW;d at end-diastole than B6 controls, suggestive of hypertrophy of myocardial walls. Strains D2M, BXD9M, 15M, 44M, and 74M had significantly thicker IVS at end-systole and 19 BXDs had significantly (P < 0.005) thicker IVS at end-diastole than that of B6M controls. The thickest IVS was found in BXD9M and BXD44M with average of 1.7 mm (end-systole) and 1.2 mm (end-diastole). Moreover, nine strains (D2M, BXD44M, 60M, 71M, 73M, 74M, 90M, and 171M) had hypertrophy of both LV walls and IVS. The thinnest LVPW and IVS were found in BXD78M (LVPW;d = 0.63 ± 0.02 mm, LVPW;s = 0.91 ± 0.1 mm, IVS;s = 1.1 ± 0.4 mm) and B6 (IVS;d = 0.75 ± 0.1 mm).

Figure 1.

Results of echocardiography demonstrating variability in left ventricular wall and interventricular septum thickness among BXD strains. Values of left ventricular posterior wall thickness (LVPW, mm, top) and interventricular septum thickness (IVS, mm, bottom) at end-systole (s) and end-diastole (d) in male (A) and female (B) BXDs. The x-axis represents the strains, whereas the y-axis represents the echocardiography value indicated as means ± SD (n > 5 mice/strain/sex). *Significant difference from C57B6J (B6) strains; #significant difference from DBA/2J (D2) strains (P < 0.05).

In the female group (Fig. 1B), BXD69F had the thickest LV walls (LVPW;s = 1.4 ± 0.3 mm and LVPW;d = 1.1 ± 0.3 mm, respectively) and BXD60F had the thickest IVS (IVS;s = 1.3 ± 0.2 mm, IVS;d = 0.93 ± 0.1). Eleven strains had significantly thicker LVPW;s than that of B6 controls. Among those, BXD60F, 69F, 70F, 74F, 77F, 89F, 90F, and 113F strains also had significantly higher LVPW;d than that of B6F mice. The end-systolic IVS thickness increased significantly in D2F, BXD60F, 77F, and 90F, whereas BXD40F, 60F, 73F, 77F, and 78F had significantly higher IVS at end-diastole compared with respective values of B6F controls. The lowest LVPW and IVS thickness was found in BXD65F (IVS;s = 1.0 ± 0.3 mm), BXD111F (IVS;d = 0.69 ± 0.06 mm), BXD78F (LVPW;s = 0.75 ± 0.1 mm), and BXD2F (LVPW;d = 0.62 ± 0.06 mm). Taken together, BXD60, 70, 71, 73, 74, 90, and 113 (male and female) strains demonstrated signs of significant LV hypertrophy inherited from parental D2 mouse.

Variability in Ventricular Chamber Volumes among BXD Strains

Increases in LV volumes are indicative for chamber dilation associated with DCM and can be assessed by volumes of blood in the ventricles during contraction at end-systole (Vol;s) and relaxation at end-diastole (Vol;d). We observed more than twofold variance in LV volumes among the tested BXD strains (Fig. 2). Among male mice, BXD34M (52 ± 13.0 μL) and BXD15M (48 ± 10 μL) had significantly increased LV Vol;s, and BXD15M, 34M, 78M, and 155M lines had significantly increased LV Vol;d compared with respective values recorded in B6 controls (Fig. 2A). In contrast, BXD24M, 61M, 69M, 71M, and 74M lines had the smallest LVVol;s and LVVol;d compared with B6 controls. Increases in internal dimension (ID) indicate LV chamber enlargement and significantly higher values were recorded in BXD34M (LVID;s) and BXD15M, 34M, and 78M (LVID;d) compared with respective parameters in B6 controls. Significantly smaller LV size was found in BXD24M, 61M, 70M, and 74M.

Figure 2.

Results of echocardiography demonstrating variability in left ventricular volumes and internal dimensions among BXD strains. Values of left ventricular volume (LVVol, µL, top) and internal dimension (LVID, mm, bottom) at end-systole (s) and end-diastole (d) in male (A) and female (B) BXDs. The x-axis represents the strains, whereas the y-axis represents the echocardiography value indicated as means ± SD (n > 5 mice/strain/sex). *Significant difference from C57B6J (B6) strains; #significant difference from DBA/2J (D2) strains (P < 0.05).

In the female group (Fig. 2B), the largest LV chamber was found in BXD32F (LVVol;s = 42 ± 13 μL) and six strains (BXD2F, 32F, 34F, 29F, 66F, and 78F) had significantly increased LV volumes and size compared with that in B6F controls. The lowest LV volumes and smallest LV were found in BXD70F (LVVol;s = 13 ± 7 μL; LVVol;d = 42 ± 11 μL; LVID;s = 2 ± 0.4 mm) with P < 0.05 compared with that in B6F mice. Taken together, male and female mice from BXD34 and 78 strains demonstrated echocardiography signs of LV dilation.

Abnormalities of RV with significantly increased RVID were found in 101M (1.8 ± 0.4, P < 0.05) compared with that in sex-matched B6 controls, suggestive of RV dilation in this strain (Fig. 3A).

Figure 3.

Echocardiographic assessment of right ventricular internal dimension and heart function among BXD strains. Values of right ventricular internal diameter (RVID, mm; A), left ventricular ejection fraction (EF%; B), fractional shortening (FS%; C), and cardiac output (CO, mL/min; D) in male (left) and female (right) BXDs. The x-axis represents the strains and the y-axis represents the echocardiography value indicated as means ± SD (n > 5 mice/strain/sex). *Significant difference from C57B6J (B6) strains; #significant difference from DBA/2J (D2) strains (P < 0.05).

Variability in Cardiac Function Phenotypes among BXD Strains

Ejection fraction percentage (EF%) is a measurement of the fraction of blood pumped out each time the heart contracts, and fractional shortening (FS%) measures the changes in IDs during systole and diastole. Both EF% and FS% are indicators of pumping efficiency of blood into hemodynamic circulation reflecting a cardiac function. The echocardiography results demonstrated that both EF% and FS% varied greatly among BXD strains (Fig. 3, B and C). In male mice, BXD74M, 70M, 73M, 171M, and D2M strains had significantly increased EF% and FS%, indicative of hypercontractile LV as a result of myocardial hypertrophy in these strains. Furthermore, the highest EF% (76 ± 9%) and FS% (45 ± 9%) recorded in the BXD74M strain were significantly higher than that in D2 mouse. The lowest EF% (45 ± 3%) and FS% (22 ± 2%) were found in BXD51M, but these parameters failed to reach statistical significance compared with B6M controls (EF = 53 ± 3%, FS = 27 ± 2%).

In female mice (Fig. 3, B and C, right), strains D2F, BXD70F, and 74F displayed hypercontractile function evidenced by significantly increased EF% or FS% versus B6F strains (P < 0.05). Most female strains (31 strains) had significantly lower EF% or FS% compared with D2F parental strain, whereas BXD32F, 78F, 75F, and 66F had significantly reduced EF% or FS% compared with that in B6F (EF = 59 ± 6%, FS = 31 ± 4%) controls, indicating systolic dysfunction in these female BXDs.

Cardiac output (CO) is the amount of blood pumped by the heart per minute (mL/min) that distributes around the body providing the delivery of oxygen and nutrients to organs as well as the removal of waste products. We found nearly twofold differences in CO among BXD strains (Fig. 3D). CO was significantly increased in BXD15M, 49M, 155M, and 90M compared with B6 and D2 parental strains and no male strains were found to have significantly reduced CO. In contrast, 15 female strains had significantly reduced CO compared with B6F controls. Among those strains, BXD73F, 65F, and 44F had significantly reduced CO compared with both parental B6 and D2F strains.

LV Mass to Body Weight Ratio Varies among BXD Strains

The important parameter demonstrating cardiac hypertrophy is the ratio between heart weight (HW) and body weight (BW). In this study, we maintained the same approach by adjusting LV mass (LVM) to BW that has been used in previously reported similar association studies using murine GRPs (36). The parental D2 mice have significantly higher HW, averaging 185.1 mg relative to 129.8 mg in parental B6 mice and an average of HW among BXD cohort was 144.7 mg (4). We found the highest BW in the BXD60M (38.8 ± 5.5 g), and BXD2M, 49M, and 60M strains had significantly higher BW (P < 0.05) compared with that in B6M (27.5 ± 2.3 g) and D2M parental strains (24.2 ± 3.3 g) (Fig. 4A, left). Among females (Fig. 4A right), no difference in BW was noted in female parental strains and the BXD154F strain had the lowest BW (17 ± 0.7 g). To precisely ascertain the LV enlargement among BXDs, we correlated LVM (Fig. 4B) measured by echocardiography with BW (LVM/BW, Fig. 4C) and found 39 male and 41 female strains, including D2, had LVM that was higher than sex-matched B6 controls. The BXD9M strain had the largest LVM (173 ± 40 mg, P < 0.018 vs. 92 ± 10 mg in B6M) among males and BXD60F had the largest LVM (142 ± 58 mg, P < 0.0001 vs. 78 ± 10 mg in B6F) among females. Furthermore, 15 male and 15 female BXD strains had significantly increased LVM/BW ratio compared with that in B6 controls, suggesting LV enlargement. Among those strains, LVM/BW ratio was increased in D2, BXD34, 73, 87, 111, 113, and 155 mice of both sexes.

Figure 4.

Body weight, left ventricular mass, their ratios, and heritability of echocardiography traits among BXD strains. Values of body weight (BW; A), echocardiography corrected left ventricular mass (LVM, mg; B), and ratios between LVM and BW (LVM/BW, mg/g; C) in male (left) and female (right) BXDs. The x-axis represents the strains and the y-axis represents the trait value indicated as means ± SD (n > 5 mice/strain/sex). *Significant difference from C57B6J (B6) strains; #significant difference from DBA/2J (D2) strains (P < 0.05). D: bar plots of the heritability (y-axis, %) of echocardiographic traits (x-axis) in male (red) and female (green) BXD strains. The heritability was defined as the ratio of the within-strain variance to total sample variance. PV PP, pulmonary valve peak pressure; PV PVel, pulmonary valve peak velocity.

Taken together, our results indicate that all echocardiography traits were differentially segregated among BXDs and some traits presented sex-specific expression. Specifically, BXD32F, BXD66F, BXD75F, and BXD78F strains demonstrated systolic dysfunction (reduced EF% and FS%) and dilation of LV with increased Vols and IDs, consistent with DCM phenotype with impaired systolic function. Dilated LV phenotypes with preserved systolic function were found in BXD34M-F, 15M-F, 2M-F, 29M-F, 78M, 155M, 9M, 51M, and 87F strains. The strains D2M-F, BXD70M-F, 74M-F, and 73F had hypercontractile function with increased LVPW and IVS thickness consistent with HCM phenotype. Furthermore, BXD155M-F, BXD44M, BXD65M, BXD113F, BXD16F, and BXD77F strains had significantly greater LVM, LVPW, and IVS thickness compared with B6 controls, suggestive of physiological hypertrophy traits with normal cardiac function.

Variability in Blood Pressure among BXDs and Association between Echocardiography and Blood Pressure Parameters

Variabilities in SBP and DBP are major risk factors for cardiovascular diseases including stroke, coronary heart disease, and heart failure (37), while the role of arterial hypertension (AH) or hypotension influencing cardiomyopathy phenotypes is not known. Among BXDs, we found significant variabilities in SBP and DBP (Supplemental Fig. S1). In male BXDs, SBP was significantly increased in BXD86M (205.4 ± 23 mmHg, highest), D2M, BXD73M, 34M, 90M, 69M, 87M, and 66M compared with B6M controls (140.8 ± 6.9 mmHg), whereas significantly higher DBP was found in BXD86M, D2M, and BXD73M. The BXD171M strain had the lowest SBP (95.2 ± 18.7 mmHg) and DBP (51.7 ± 17.1 mmHg). Strains BXD73aM, 89M, 75M, 84M, and 60M had significantly lower DBP when compared with control B6M (122.8 ± 5.9 mmHg). In female BXDs, BXD68F (194.9 ± 5.7 mmHg, highest), D2F, BXD65F, and BXD113F displayed significantly higher SBP, whereas BXD170F (92.1 ± 19.1 mmHg, lowest), 74F, 34F, 73F, and 171F strains had significantly decreased SBP when compared with that in control B6F (143.0 ± 4.9 mmHg). The significantly higher DBP was found in BXD68F (160.3 ± 3.0 mmHg), 155F, D2F, and 65F, whereas the lowest DBP was recorded in BXD170F (59.4 ± 14.1 mmHg), 34F, 74F, 73F, 171F, 69F, 172F, and 24F when compared with control B6F (120.3 ± 7.2 mmHg). We then performed correlation analysis to determine whether echocardiography and BP parameters are correlated and found no correlation between SBP or DBP and any of the echocardiography traits presented in BXDs. These results suggested that echocardiography traits inherited from parental strains are expressed independently from variabilities in blood pressure among BXD strains.

Heritability of Cardiac Phenotypes among BXD Strains

We estimated the heritability (h2) as the ratio of the within-strain variance to total sample variance (Fig. 4D). Most echocardiography parameters have moderate to high heritability, averaging 0.40 in male and 0.36 in female, suggesting that genetic factors contribute to the phenotypic variation. Overall, the heritability in males was higher than that in females with the exception of RVID, heart weight, CO (LV trace), IVS;d, and IVS;s. In males, the LVPW;s showed the highest heritability (h2 = 0.52), followed by RVID (h2 = 0.48) and weight (h2 = 0.45). In females, only RVID (h2 = 0.59) and weight (h2 = 0.46) showed heritability greater than 0.4. The parameters IVS;s and IVS;d showed the lowest heritability both in males and females. Notably, LVPW;s showed the highest heritability in males, however, moderate in females (h2 = 0.35); this was inverse for RVID.

QTL Analysis Identifies Locus Associated with LVID and LVVol in Male Mice

Due to clear variability in echocardiography measurements detected between male and female BXD strains, we performed QTL mapping analysis for all echocardiography-related traits in male and female cohorts separately. QTL mapping was conducted in the male BXD panel of ∼7,200 genetic markers using fast linear regression equations of Haley and Knott (38) within GeneNetwork. Then, 2,000 permutations were performed to assess the genome-wide threshold for significant and suggestive associations. We found a locus on chromosome 8 to be significantly associated with LVID;d, LVID;s, LVVol;d, and LVVol;s, with the maximum LRS score of 19.44, 16.93, 18.36, and 15.62, respectively (Fig. 5A). As those traits mapped to the same position and are highly correlated (average absolute Pearson correlation coefficient r > 0.93), we conducted principal component analysis (PCA) for those traits (Fig. 5B). The PC1 explained ∼80% overall variance, hence we used PC1 to perform interval mapping and a significant QTL (peak LRS =18.50) was identified at the same position (chromosome 8 at 95.75 Mb) encompassing 6.4 Mb from 91.0 to 97.4 Mb (Fig. 5C).

Figure 5.

QTL mapping for cardiac-related traits in male mice. A: QTL heatmap identified significant overlapping loci on chromosome 8 related to left ventricular internal diameter (LVID) and LV volumes (LVVol) at end-systole (s) and end-diastole (d). B: distribution of PC1 values generated from the four traits (LVID;s, LVID;d, LVVol;s and LVVol;d) across the BXD strains. The x-axis represents the strains and the y-axis represents the value of PC1. C: interval mapping for PC1 identified a significant QTL on chromosome 8 between 91.0 and 97.4 Mb. The x-axis denotes a position on the mouse genome in megabases (Mb) and the y-axis gives the likelihood ratio statistic scores (LRSs). Significant LRS = 18.22 (pink line) and suggestive LRS = 11.25 (gray line) threshold was determined with 2,000 permutation tests. LVID, left ventricular internal dimension; LVVol, left ventricular volume; QTL, quantitative trait loci.

The chromosome 8 QTL harbors 131 genes within the 2-LOD QTL interval from 91.0 to 97.4 Mb. We first explored whether the observed variabilities of the LVID and LVVol could be mediated via the functional coding sequence variants of these 131 QTL genes. By looking up and comparing the whole genome sequences of the BXD parental strain B6 and D2, 14 genes (Chd9, Rbl2, Rpgrip1l, Fto, Irx3, Irx6, Lpcat2, Capns2, Ces1a, Ces1b, Ces1c, Kifc3, Cngb1, and Zfp319) were identified to harbor nonsynonymous SNPs. However, none was correlated with the investigated echocardiography traits (Supplemental Table S2). Next, we explored whether observed phenotype differences across the BXD mice could be mediated via transcriptional mechanisms. Using heart gene expression data generated for the BXD RI panel, we conducted eQTL mapping for each of the QTL genes and identified six genes (Coq9, Ndrg4, Crnde, Irx3, Rpgrip1l, and Rbl2) that are cis-regulated (Supplemental Table S3). It should be worth noting that the eQTL peak signal (chromosome 8 at 95.7 Mb) of Ndrg4 was almost identical with the QTL locus (Fig. 6A). Moreover, Ndrg4 was also found to be positively correlated with the locus-associated traits such as LVID;d, LVID;s, LVVol;d, and LVVol;s (Fig. 6, B–E). Although no coding sequence variant was found for Ndrg4, two 5′ UTR (rs33223753 and rs225352854) and several gene upstream variants were identified. Expression of another gene at this locus, Slc6a2, was positively correlated (P < 0.05) with the above four traits, LVID;d (r = 0.493), LVID;s (r = 0.531), LVVol;d (r = 0.474), and LVVol;s (r = 0.525). Taken together, our approach identified Ndrg4 and Slc6a2 as strong candidate genes at the chromosome 8 locus that was involved in the regulation of LV dilation in males.

Figure 6.

Ndrg4 is a candidate gene for chromosome 8 QTL in male mice. A: interval mapping of Ndrg4 mRNA level identified a significant cis-eQTL on chromosome 8 at 95.7 megabases (Mb). The x-axis denotes a position on the mouse genome in Mb and the y-axis gives the likelihood ratio statistic scores (LRSs). Significant LRS = 17.42 (pink line) and suggestive LRS = 11.08 (gray line) threshold were determined with 1,000 permutation tests. B and C: Ndrg4 mRNA level showed correlation with left ventricular internal diameter at end-systole and end-diastole (LVID;s and LVID;d, respectively). Each dot represents the mean value of n ≥ 5 mice/group. D and E: Ndrg4 mRNA level showed correlation with left ventricular volumes at end-systole and end-diastole (LVVol;s and LVVol;d, respectively). LVID, left ventricular internal dimension; LVVol, left ventricular volume; QTL, quantitative trait loci.

QTL Analysis Identifies Loci Associated with LVPW, EF%, and FS% in Female Mice

QTL mapping for cardiac-related traits in female BXD mice was performed as described for the male BXD mice. We identified a significant genomic locus on chromosome 7 at 40.2 Mb that was associated with LVPW at diastole (Fig. 7A). The 2-LOD QTL interval encompassed 10 Mb from 36.6 to 46.4 Mb. In addition, a suggestive QTL located on chromosome 3 at 92.6 Mb was identified for EF% and FS% (Fig. 7, B and C, respectively), with QTL interval from 87.8 to 100.1 Mb. We did not find any statistically significant QTL for any other cardiac-related traits in females.

Figure 7.

QTL mapping for cardiac-related traits in female mice. A: one locus on chromosome 7 found to significantly associate with left ventricular posterior wall thickness (LVPW, mm). The x-axis denotes a position on the mouse genome in megabases (Mb) and the y-axis gives the likelihood ratio statistic scores (LRSs). Significant LRS (pink line) and suggestive LRS (gray line) threshold was determined with 2,000 permutation tests. Overlapping suggestive locus related to fractional shortening (FS%; B) and ejection fraction (EF%; C) on chromosome 3 has been identified. QTL, quantitative trait loci.

The QTL identified on chromosome 7 contains 363 genes within the 2-LOD region. Among these, 98 genes contained protein-altering variants between B6 and D2, of which expression of 9 genes (Klk15, Spib, Nup62, Fuz, Pnkp, Josd2, Rcn3, Ftl1, and Klk1b27) was significantly correlated (P < 0.05) with the LVPW at diastole (LVPW;d) as shown in Table 1. We next explored if the QTL genes were cis-regulated, and identified 17 significant cis-eQTL genes, with Josd2 expression being correlated with the LVPW;d. Among other genes at the same locus, Mybpc2, Pold1, Klk1, Gys1, Ccdc114, Nkg7, Grin2d, Emc10, and Vrk3 are known to be associated with cardiac-related functions, six of which (Mybpc2, Pold1, Nkg7, Grin2d, Ccdc114, and Zfp719) harbored coding sequence variants. However, no correlation between expression of these genes and LVPW;s was found.

Table 1.

List of candidate genes for chromosome 7 QTL

Gene Symbol	Entrez ID	Gene Name	Variant	Cis-eQTL Max LRS	LVPW;d
					R	P
Klk15	317652	Kallikrein related-peptidase 15	NS SNP		0.613	0.001
Spib	272382	Spi-B transcription factor (Spi-1/PU.1 related)	NS SNP		0.552	0.004
Nup62	18226	Nucleoporin 62	NS SNP		0.532	0.005
Fuz	70300	Fuzzy planar cell polarity protein	NS SNP		0.531	0.005
Pnkp	59047	Polynucleotide kinase 3′- phosphatase	NS SNP		0.516	0.007
Josd2	66124	Josephin domain containing 2	Splice site mutation	39.3	0.499	0.010
Rcn3	52377	Reticulocalbin 3, EF-hand calcium binding domain	NS SNP		0.476	0.015
Ftl1	14325	Ferritin light polypeptide 1	NSSNP		0.456	0.021
Klk1b27	16619	Kallikrein 1-related peptidase b27	NS SNP		-0.456	0.021

LRS, likelihood ratio statistics; LVPW;d, left ventricular posterior wall at diastole; NS, nonsynonymous; SNP, single nucleotide polymorphism.

For suggestive chromosome 3 QTL, we identified 35 genes that have coding sequence variants, with 5 genes (Gon4l, Tchhl1, Pdzk1, Prcc, and Lor) expression significantly correlated (P < 0.05) with EF% and FS% traits (Table 2). In addition, 15 genes were shown to be cis-regulated, with 7 genes (Gon4l, Snapin, Tchhl1, Prcc, Zfp697, Dap3, and Sprr2a1) expression significantly correlated (P < 0.05) with the QTL traits. In addition to the above potential candidates, we identified 15 other cardiac function-related genes (Mef2d, Lmna, Dpm3, Tpm3, Itga10, Prkab2, Pde4dip, Wars2, Snapin, Ash1l, Cgn, Paqr6, Ivl, Snx27, and Gja8) in this locus, with 3 genes (Snapin, Tpm3, and Wars2) significantly correlated (P < 0.05) with EF% and FS%.

Table 2.

List of candidate genes for chromosome 3 QTL

Gene Symbol	Gene Name	Entrez ID	Variant	Cis-eQTL Max LRS	EF (%)		FS (%)		Function
					R	P	R	P
Gon4l	Gon-4-like (C. elegans)	76022	NS SNP	82.8	−0.607	0.001	−0.619	0.001
Tchhl1	Trichohyalin-like 1	71325	NS SNP	23.2	0.569	0.002	0.577	0.002
Pdzk1	PDZ domain containing 1	59020	NS SNP		0.563	0.003	0.570	0.002
Prcc	Papillary renal cell carcinoma (translocation-associated)	94315	NS SNP	20	−0.542	0.004	−0.535	0.005
Lor	Loricrin	16939	NS SNP		−0.420	0.036	−0.409	0.042
Snapin	SNAP-associated protein	20615		23.9	−0.577	0.002	−0.563	0.003	IMPC
Zfp697	Zinc finger protein 697	242109		30.7	−0.529	0.006	−0.502	0.010
Dap3	Death-associated protein 3	65111		12.9	−0.514	0.008	−0.515	0.008
Sprr2a1	Small proline-rich protein 2A1	20755		22.1	−0.418	0.037	−0.411	0.040
Tpm3	Tropomyosin 3, gamma	59069			0.535	0.005	0.551	0.004	RGD
Wars2	Tryptophanyl tRNA synthetase 2 (mitochondrial)	70560			0.397	0.049	0.379	0.061	RGD

EF, ejection fraction; FS, fractional shortening; IMPC, International Mouse Phenotyping Consortium; LRS, likelihood ratio statistics; NS, nonsynonymous; RGD, rat genome database; SNP, single nucleotide polymorphism.

Correlation between Cardiomyocyte Size and Candidate Genes among BXDs

An increase in cardiomyocyte size indicative of cellular hypertrophy is one of the decisive components of remodeling in many types of cardiomyopathies and the magnitude of increase in cardiomyocyte diameter was found to be associated with DCM with systolic dysfunction and advanced heart failure (39). Previous independent measurements of cardiomyocyte diameter found more than 1.5-fold variance among BXDs (4, 28). This study found that B6 parental mice have the smallest cardiomyocytes (≤25 µm), whereas the D2 parental strain is one of the strains with large cardiomyocytes (>35 µm) and there is a significant difference in cardiomyocyte diameter between B6 and D2 (P < 0.01) (Fig. 8A). Pearson correlation analysis of BXD strains used for echocardiography found 14 BXDs that had an overlapped data between the two datasets, cardiomyocyte diameter trait, and BXD heart transcriptomes (Fig. 8B). When we performed correlation of cardiomyocyte diameter values with the genes found within QTLs on chromosomes 8, 7 and 3, we identified Ndrg4 (r = −0.588, P = 0.025), Dap3 (r = 0.539, P = 0.046), and Tpm3 (r = 0.576, P = 0.030) to be positively correlated with the cardiomyocyte diameter, respectively (Fig. 8, C–E). Furthermore, the Bayesian network modeling linking the chromosome 8 QTL (peak SNP: rs32133186) with LVID and LVVol and their PCA value through Ndrg4 demonstrated that the phenotype variation is the consequence of the variation in expression of Ndrg4 (Fig. 8F). These results further support that Ndrg4 is a strong candidate gene associated with both, cardiac function and cardiomyocyte hypertrophy.

Figure 8.

Cardiomyocyte diameter and genetic correlation in BXD strains. A: values of cardiomyocyte diameter (µm) in B6 and D2 parental strains (n ≥ 5 mice/group). B: among 42 BXD strains utilized for echocardiography, 14 BXD strains had overlapped data between the two datasets: cardiomyocyte diameter trait (blue circle) and heart transcriptome (red circle). Correlation of mRNA levels of Ndrg4 (C), Dap3 (D), and Tpm3 (E) expression in heart tissues of BXDs (y-axis) with cardiomyocyte diameter (µm) indicated by x-axis. Each dot represents the mean values of n ≥ 5 mice/group. F: Bayesian network modeling linking the chromosome 8 QTL (peak SNP: rs32133186) with LVID, LVVol, and their PCA value through Ndrg4, demonstrating that the phenotype variation is the consequence of the variation of Ndrg4 expression. LVID, left ventricular internal dimension; LVVol, left ventricular volume.

DISCUSSION

Due to the complex polygenic and epistatic nature of cardiac traits, deep association studies between phenotypes and interactome (genome, transcriptome, proteome, and metabolome) networks is important for advancing precision medicine as a strategy to improve cardiovascular health and disease prevention. Furthermore, discovering the underlying genetic elements of complex phenotypes is critical in predicting the vulnerability of individuals “at risk” and developing personalized diagnostics, early care, and management before the disease onset or deterioration. Despite tremendous progress in genetic and molecular research, the underlying basis remains unclear or imprecise in 60%–70% of cardiomyopathy cases (11). In this study, to translate the genetic variants and phenotypes associated with heterogeneity of human cardiomyopathies, we used a systems genetics approach to determine genetic and epistatic contributions to the variation in echocardiography phenotypes in murine GRP of BXD strains. We chose the BXD family of RI mice for number of reasons. First, the BXD family is the largest and comprehensively genotyped murine GRP (3, 7), representing a reproducible high-resolution mapping panel with ∼6 million SNPs, ∼500,000 insertion-deletions, and thousands of copy number variants that segregate among BXD strains similar to the human population (26). Second, the parental D2 strain has been shown to have some signs of cardiac pathology by several laboratories, such as increased heart weight, cardiomyocyte size, calcifications, elevated markers of hypertrophy, and fibrosis including β-MHC, ANP, BNP, and SMA-α1 (4–6, 36). Comparative cardiac magnetic resonance imaging (cMRI) revealed cardiac hypercontraction with increased EF% (75.86 ± 1.08%) with smaller ventricular systolic volumes for the D2 mouse compared with EF% (55.2 ± 6.0%) and systolic volumes for the B6 strain (40). Based on whole genome sequencing results, the D2 strain carries missense variants in Mybpc3, Prkag2, Myocd, Tnnt2, and Ankrd17, the candidate GWAS genes associated with cardiomyopathy in humans. Finally, BXDs have been bred in the same controlled laboratory conditions to minimize the environmental effects on cardiac morphology and physiology. In this condition, we believe that all phenotypic variations are caused by the genetic background of different BXD strains that carry different alleles of cardiomyopathy candidate genes (causal and modifier) inherent from the D2 parental strain. Previous association analyses of cardiac remodeling traits performed in Hybrid Mouse Diversity Panel (HMDP) of mouse GRPs, which included several male BXD strains, demonstrated a remarkable diversity in echocardiographic values among BXDs studied (36). In this study, we hypothesized that natural phenotypic and genetic features in D2 mice will be segregated among male and female offspring derived from B6 and D2 crosses, allowing QTL and eQTL mapping to explain variance in phenotype and expression traits based on sex. To test this, we evaluated echocardiography and blood pressure parameters and cardiomyocyte diameter traits across 42 BXD strains and parental B6 and D2 mice while sedentary followed by systematic correlation between cardiac traits and heart transcriptomes, and identification of genetic loci that regulate those collected traits. Elevated BP has a direct impact on the development of abnormal geometry of the heart as a result of LV hypertrophy and chamber dilatation in response to hemodynamic and pressure overload. The prevalence of isolated LV hypertrophy (nondilated) ranges from 20% to 100% in patients with AH depending on severity and duration of an increase in BP (41). Dilatation of LV chamber overlapped with hypertrophy of LV myocardial walls in contrast conferred a significant risk of unfavorable cardiovascular events in patients with severe AH (42). Thus, we correlated BP values and echocardiography parameters in BXD strains and validated that echocardiography traits were expressed independently from BP variabilities among BXDs.

Echocardiography phenotypes associated with HCM (hypercontraction with increased EF% and FS%, thickened IVS, and LVPW) were found in parental D2M-F, BXD70M-F, 74M-F, 73M, and 171M, while several BXD strains had increased ventricular wall thickness and normal cardiac function consistent with physiological hypertrophy. Traits consistent with DCM (systolic dysfunction with reduced EF% and FS% and increased LVVol and ID) have been found in BXD78F, 32F, 75F, and 66F strains. Some strains (BXD62M-F, 65bM-F, and 102M-F) demonstrated dilated RV, predicting ACM phenotype with predominant RV pathologies. Based on our echocardiography results, we found most these traits are highly variable among male and female BXD strains. Their heritability ranges were from 0.24 to 0.59, indicating that genetic background has great effect on those traits and demonstrating that the BXD family is an exceptional mouse model to investigate effects of individual genetic background on cardiomyopathy phenotypes and explore the mechanisms of how individual’s genetic variants modify cardiac morphology and function phenotypes.

Our further genetic association analysis identified two significant QTL loci on chromosome 8 that regulate phenotype variations of LVID and LV volumes at end-diastole and end-systole in the males. In female BXDs, one locus on chromosome 7 regulating thickness of LVPW at end-diastole and one suggestive QTL on chromosome 3 that regulates EF% and FS% phenotype variations have been identified. Among genes at chromosome 8 locus, Ndrg4 is a strong candidate for LV dilation as it is not only cis-regulated but also positively correlated with LVVol and LVID QTLs. Pathological cardiomyocyte hypertrophy is the feature of both, HCM characterized by concentric hypertrophy where thickness of cardiomyocytes increases more than in length, resulting in increased LV wall thickness, as well as DCM characterized by eccentric hypertrophy due to LV chamber dilation where lengthening of individual cardiomyocytes excesses its thickening (43). Therefore, a negative correlation of Ndrg4 with cardiomyocyte diameter supports echocardiography data consistent with LV dilation. The NDRG (N-myc downstream regulatory gene) family of hydrolase enzymes includes NDRG1, NDRG2, NDRG3, and NDRG4 encoding proteins that play a crucial role in growth and development, cell proliferation, tumor, stress response, and endocrine and apoptosis regulation (44). Among NDRGs, the only gene expressed in cardiomyocytes is NDRG4, however its role in the heart is unclear beyond its involvement in myocyte proliferation and epicardial cell migration (45). Recently, NDRG4 p.T256M variant has been identified in patients with pulmonary atresia and ventricular septal defect (46).

In addition, Slc6a2 at chromosome 8 locus has been previously implicated in cardiovascular function according to the IMPC databases. The Slc6a2 (solute carrier family 6 member 2) encodes norepinephrine transporter (NET) protein, which is responsible for the sodium-chloride (Na⁺/Cl⁻)-dependent reuptake of extracellular norepinephrine and dopamine and epigenetic modification of NET has been linked to postural tachycardia syndrome (47). Moreover, the c.A3081T (rs28386840) SNV has been shown to modify blood pressure and heart rate during subanesthetic ketamine administration in a sex-dependent manner (48), suggesting that Slc6a2 may be a candidate modifier gene that regulates LV dilation in males.

In female BXDs, phenotype variations, such as LVPW thickness and cardiac systolic function (EF%, and FS%) are affected by major genetic loci with two QTLs identified on chromosome 7 (significant correlation) and 3 (suggestive correlation), respectively. The locus on chromosome 7 has been previously shown to be associated with LVM in isoproterenol-induced heart failure male HMDP of mouse GRPs, but not in sedentary conditions (36). Moreover, Myh14 encoding a nonmuscle myosin involved in mechanotransduction has been determined in chromosome 7 QTL as a negative regulator of isoproterenol-induced LV hypertrophy. In our study, we identified the supporting evidence for strong candidate genes within chromosome 7 QTL for LVPW thickness and suggest that multiple genes, each of which has a small effect, may be involved in myocardial hypertrophy. Particularly, we emphasize Josd2 encoding Josephin domain-containing protein2, a deubiquitinating (DUB) enzyme, out of 17 cis-eQTLs associated with LV wall thickness. Another DUB enzyme in this QTL is kallikrein, encoded by Klk1 that cleaves kininogen into the active kinin, which is involved in broad biological processes, such as vasodilation, smooth muscle contraction, and inflammation (49). Tissue kallikrein mediates the release of active epidermal growth factor EGF, an important positive regulator of cardiac hypertrophy (50), suggesting mechanistic studies on DUB enzymes may explain the roles of DUB enzymes in cardiac function and hypertrophy.

Among genes identified in the chromosome 3 QTL, Dap3 and Tpm3 genes earned our attention as candidate regulators for cardiac systolic function and cardiomyocyte hypertrophy. The Tpm3 gene encodes tropomyosin 3, a protein that is involved in regulating the stability of actin filaments and muscle contraction via interactions with sarcomeric proteins, such as troponins. Genetic abnormalities in TPM3 are associated with congenital nemaline and cap myopathies, while TPM3 deletions cause hypercontractile congenital muscle stiffness (51). Consistent with our results of positive correlation between Tpm3 and cardiomyocyte diameter in BXDs, overexpression of TPM3 in mouse heart has resulted in hyperdynamic (hypercontractile) systolic and diastolic function via decreasing Ca²⁺ sensitivity of the sarcomere (52). Another candidate, Dap3 encodes death-associated protein 3, a 28S subunit protein expressed in the mitochondrial ribosome and loss of DAP3 resulted in mitochondrial fragmentation sensitizing cells to the intrinsic mitochondria-mediated death (53), one of the apoptotic mechanisms involved in the pathogenesis of cardiomyopathies and failing heart.

Conclusions

In summary, this study aimed to establish a cardiac phenotype-genotype correlation in murine genetic reference population of male and female BXD RI strains by phenotyping the echocardiography, blood pressure, and cardiomyocyte diameter traits and associating each collected phenotype with genetic background. Our study identified several QTLs and candidate genes that have a significant association with cardiac hypertrophy, ventricular dilation, and function including systolic hyperfunction and dysfunction. We conclude that the BXD family is an ideal mammalian model to map loci and gene variants affecting cardiomyopathy phenotypes and provide important data for future studies with potential of identifying novel key genetic components and mechanisms of cardiac diseases in human population.

Limitations of the Study

We acknowledge that the use of expression arrays does not provide information about alternative isoform usage, which can be a cis-directed phenomenon and may contribute to trait distribution. In addition, we used strain mean for our mapping analysis of echo phenotypes and cardiac gene expression, which discounts nongenetic sources of individual differences within strain. We are developing linear mixed models that could map BXD cases at the individual level. We will test this approach for our future study.

DATA AVAILABILITY

Cardiac gene expression data “EPFL/LISP BXD CD Heart Affy Mouse Gene 2.0 ST Gene Level” have been deposited at our GeneNetwork.org under accession ID GN485. Detailed information can be found at http://gn1.genenetwork.org/webqtl/main.py?FormID=sharinginfo&GN_AccessionId=485.

SUPPLEMENTAL DATA

Supplemental Table S1: https://doi.org/10.6084/m9.figshare.20368935.

Supplemental Table S2: https://doi.org/10.6084/m9.figshare.21397215.

Supplemental Table S3: https://doi.org/10.6084/m9.figshare.21397380.

Supplemental Fig. S1: https://doi.org/10.6084/m9.figshare.21583509.

GRANTS

This study was supported by NIH Grants R01 HL128350 (to L.L. and Y.S.) and R01 HL151438 (to J.A.T., L.L., and E.P.) and by UTHSC, Center for Integrative and Translational Genomics (CITG) Research Grant (to E.P.).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

E.P. conceived and designed research; B.-O.O., U.M., and N.R.A. performed experiments; B.-O.O., F.X., U.M., N.R.A., A.K.B., J.N.J., Y.S., L.L., and E.P. analyzed data; B.-O.O., F.X., J.N.J., J.A.T., and E.P. interpreted results of experiments; B.-O.O., F.X., U.M., N.R.A., A.K.B., Y.S., L.L., and E.P. prepared figures; B.-O.O. and U.M. drafted manuscript; B.-O.O., F.X., J.A.T., L.L., and E.P. edited and revised manuscript; E.P. approved final version of manuscript.