The requirement for atrial function in developing heart is unknown. To address this question, we have generated mice deficient in atrial myosin light chain 2 (MLC2a), a major structural component of the atrial myofibrillar apparatus. Inactivation of the Mlc2a gene resulted in severely diminished atrial contraction and consequent embryonic lethality at ED10.5-11.5, demonstrating that atrial function is essential for embryogenesis. Our data also address two longstanding questions in cardiovascular development: the connection between function and form during cardiac morphogenesis, and the requirement for cardiac function during vascular development. Diminished atrial function in MLC2a-null embryos resulted in a number of consistent secondary abnormalities in both cardiac morphogenesis and angiogenesis. Our results unequivocally demonstrate that normal cardiac function is directly linked to normal morphogenic development of heart and vasculature. These data have important implications for the etiology of congenital heart disease.
During mammalian development, the heart is the first organ to develop and is essential for embryogenesis. Atrial function increases in parallel with morphogenesis during early heart development (Campbell et al., 1992). However, how the atrium contributes to embryonic cardiovascular function has not been evaluated.
Cardiac muscle expresses two major isoforms of myosin light chain 2, MLC2v (MYL2 – Mouse Genome Informatics) and MLC2a (MYLC2A – Mouse Genome Informatics). During cardiogenesis, MLC2v is expressed exclusively in ventricular and atrio-ventricular junction myocardium. Ablation of MLC2v results in disruption of ventricular function at embryonic day (ED) 11.5 and embryonic lethality at ED12.5 (Chen et al., 1998). MLC2a is initially expressed throughout the heart at ED7.5 and becomes restricted to the atria after ED12.5 (Kubalak et al., 1994). Despite its early expression throughout the heart until ED12.5, MLC2a protein is only incorporated into myofibrillar structures in atria, not in ventricle (Chen et al., 1998). As MLC2a is the only isoform expressed in atria, we reasoned that disruption of MLC2a would interrupt embryonic atrial function, and would therefore allow us to investigate the role of atrial function in embryogenesis. Our results have demonstrated that ablation of MLC2a disrupts the earliest functioning of the heart, demonstrating an early requirement for atrial function. Additionally, these results have allowed us to address longstanding questions in development concerning the role of embryonic heart function in cardiac morphogenesis and angiogenesis.
Recently, it has been shown that the embryonic heart in early chick embryos begins to beat substantially prior to a requirement for convective bulk transport to deliver oxygen and nutrients for growth (Burggren et al., 2000). Similar observations have been made for zebrafish, clawed frog and salamander embryos (Mellish et al., 1994; Pelster and Burggren, 1996; Territo and Burggren, 1998). These findings are consistent with the hypothesis that early blood flow may play a role not only as a transportation fluid, but also as a physical factor in development, perhaps influencing normal morphogenesis of the heart and angiogenesis.
Hemodynamic parameters are known to be a determinant of myocardial growth, structure and function in the adult heart, and may be important in the etiology of heart failure (Hutchins et al., 1978; Rockman et al., 1994; Zak, 1974). It has been postulated that normal flow and hemodynamics are also crucial determinants of normal cardiac morphogenesis, and may play a causative role in congenital heart disease (Sedmera et al., 1998). Previously, this issue has been addressed by ligation experiments to perturb blood flow in chick embryos (Hogers et al., 1997; Icardo, 1989; Sedmera et al., 1999). By technical necessity, these experiments perturbed blood flow in relatively late stage embryos, between HH stages 21-24, a time when considerable cardiac morphogenesis has already taken place. Because of the technically demanding nature of these experiments, small numbers of experimental embryos can be obtained, and from these variable phenotypes are observed (Broekhuizen et al., 1999). Additionally, mechanical intervention to perturb blood flow may also restrict cell migration, or may inflict non-specific tissue damage, complicating the interpretation of any effects on growth or morphogenesis. Genetic ablation of MLC2a has allowed us to selectively inactivate atrial function, altering hemodynamics from the earliest stages of heart development. Prior to any growth retardation, MLC2a mutants exhibit consistent phenotypic abnormalities in embryonic cardiac morphology. Our results unequivocally demonstrate that normal growth and morphogenesis of the heart are dependent on normal heart function from the earliest stages, and strongly suggest that altered hemodynamics during embryonic development can have significant morphologic consequences for heart development. Embryonic atrial insufficiency in MLC2a mutants has severe consequences for multiple aspects of chamber and looping morphogenesis, suggesting that a less severe perturbation could similarly affect these morphogenetic processes and give rise to congenital anomalies of the heart in humans.
Many mouse mutants have been described that simultaneously affect early heart function and angiogenesis (Bi et al., 1999; Gerety and Anderson, 2002; Li et al., 1999; Lin et al., 1998; Lin et al., 1997; Lyons et al., 1995; Regan et al., 2002; Stanley et al., 2002; Tanaka et al., 1999; Yamagishi et al., 2000). However, each of these genes is expressed in both developing vasculature and heart, making it difficult to assess the independent effects of embryonic heart malfunction on the developing vasculature. The expression of MLC2a exclusively in developing myocardium allowed us to examine this question in our MLC2a knockouts, and has clearly demonstrated a requirement for cardiac function in both extra- and intraembryonic angiogenesis.
Together, our data imply that alterations in blood flow during early development, as a result of either genetic or environmental influences, may have an etiological role in congenital cardiovascular disease.
Materials and methods
A 15 kb mouse MLC2a genomic DNA clone was isolated from a 129SV library (Stratagene) and used to construct the MLC2a targeting vector by standard techniques. Briefly, a Cre-PGKNEOr cassette was inserted into pBluescript BKII (targeting vector) flanked by a 4.5 kb BamH1 fragment, containing the 5′ untranslated sequence of MLC2a and a 4.0 kb Kpn1/Acc1 fragment, containing the 3′ untranslated sequence of MLC2a, thereby replacing exon one through seven of MLC2a. The 1.3 kb Cre recombinase gene (Gu et al., 1994) and the 1.7 kb neomycin resistance (NEOr) gene were under the control of the MLC2a and phosphoglycerokinase (PGK) promoter, respectively. The targeting vector was subsequently linearized with XbaI and electroporated into R1 ES cells. G418-resistant ES clones were screened for homologous recombination by DNA blot analysis, as described below. One positive clone that underwent proper homologous recombination was microinjected into blastocysts from C57BL/6J mice at the Transgenic Core Facilities of the University of California, San Diego. Male chimeras were bred with female Black Swiss to test for germline transmission of the agouti coat phenotype from 129-derived ES cells.
DNA was extracted from G418-resistant ES cell clones, yolk sac and mouse tails, as previously described (Moens et al., 1993). ES cell DNA was digested with SstI, electrophoresed on a 0.8% (w/v) agarose gel, and subsequently blotted onto nitrocellulose. A 400 bp fragment, corresponding to the 5′ untranslated region of the MLC2a gene, was generated by polymerase chain reaction (PCR) using mouse genomic DNA and specific MLC2a primers (forward primer, 5′-GTGAGCCACTGAGAATGGTTGT-3′; reverse primer, 5′-CTTAGAACCCCACTTCATCCCT-3′), and subsequently radiolabeled using [32P]dATP by random priming (Invitrogen, San Diego, CA). DNA blots were hybridized to the radiolabeled probe at 68°C, washed twice for 15 minutes each time at 60°C in 0.1×SSC with 0.1% sodium dodecyl sulphate (SDS), and visualized by autoradiography. DNA from yolk sac and mouse tails was also subjected to PCR, using both Cre (forward primer, 5′-GTTCGCAAGAACCTGATGGACA-3′; reverse primer, 5′-CTAGAGCCTGTTTTGCACGTTC-3′) and MLC2a (forward primer, 5′-GGAACAGAGACCAGCCACA-3′; reverse primer, 5′-GGTCTGATTTGCAGATGATC-3′) specific primers, and products were visualized by ethidium bromide staining.
Whole-mount in situ hybridization analysis
In situ hybridization was performed on MLC2a-null and somite-matched embryos for Mlc2a, Mlc2v, Nkx2.5, dHand, eHand and Tbx5, using digoxigenin-labeled antisense riboprobes, essentially as previously described (Wilkinson, 1992) with modifications. Briefly, plasmids containing Mlc2v, Nkx2.5 and Tbx5 cDNAs were linearized with NotI, HindIII and SpeI, respectively. T7 RNA polymerase (Invitrogen) was used to synthesize RNA ribprobes by in vitro transcription in the presence of digoxigenin-11-dUTP (Roche). Fixed embryos were subsequently pre-hybridized in 50% formamide, 5×SSC (pH 5), 50 ug/ml yeast RNA, 1% SDS and 50 μg/ml heparin at 68°C overnight, and subsequently hybridized with riboprobes at 68°C for 18 hours. After vigorous washing, embryos were incubated with alkaline-phosphatase-conjugated anti-digoxigenin antibodies at 4°C overnight (Roche), vigorously washed again, and incubated in a NTMT buffer (Promega), including freshly added nitroblue tetrazolium (NBT; 4.5 μl) and 5-bromo-4-chloro-3-indolyl phosphate (BCIP; 3.5 μl) per ml NTMT. Embryos were subsequently fixed in 4% paraformaldehyde for 1-2 hours, until color development had occurred to the desired extent, and stored at 4°C in phosphate buffered saline (PBS) to terminate the color reaction. Whole embryos were analyzed and photographed on a ZEISS SV-5 dissecting microscope with a Nikon C-mount 35 mm camera.
Protein blot analysis
Total protein extracts were prepared from hearts of MLC2a-null, as well as wild-type, embryos, and protein analysis was performed as previously described (Chen et al., 1998). Immunodetection of MLC2a (28.2 kDa) and MLC2v (34.7 kDa) was performed in cardiac atria and ventricular samples by using a rabbit polyclonal antibody to MLC2a (1:500), as previously described (Chen et al., 1998). The blot was subsequently incubated with horseradish peroxidase-conjugated anti-rabbit Ig (1:1000; Sigma). Results were visualized by using enhanced chemiluminescence (Amersham).
Embryos and yolk sacs from MLC2a-null and wild-type, somite-matched embryos were fixed, dehydrated and embedded in paraffin wax. Serial transverse sections were obtained at 10 μm intervals. Sections were stained with Hematoxylin and Eosin, and photographed as previously described (Zhou et al., 2001).
Whole embryos and yolk sacs were fixed in 4% paraformaldehyde overnight at 4°C, and then washed in a series of methanol washes and bleached in 5% H2O2/100% methanol for 4 hours at room temperature (RT). Subsequently, embryos were rinsed in 100% methanol, rehydrated and incubated in PBSMT (3% milk and 0.1% Triton X-100 in PBS) for 2 hours at RT. Embryos were then incubated with (1) platelet endothelial cell adhesion molecule (PECAM; 1:75; Sigma) or (2) fetal liver kinase 1 (FLK1; 1:400; Sigma) overnight at 4°C, washed in PBSMT and incubated with anti-mouse horseradish peroxidase (1:1000) overnight at 4°C. Subsequently, embryos were washed in PBSMT and briefly rinsed in PBT (0.2% BSA and 0.1% Triton X-100 in PBS). For color development, embryos were incubated in 0.2 mg/ml DAB (Sigma) and 0.35% NiCl2 in PBT at RT, until the desired intensity was achieved. The reaction was stopped with the addition of 0.03% H2O2. The embryos were rinsed several times in PBT and PBS, and fixed in a 2% paraformaldehyde/0.1% gluteraldehyde solution overnight at 4°C. The next day, embryos were rinsed several times in PBS, and equilibrated at RT in 50% glycerol and then 70% glycerol for one hour each. Samples were analyzed and photographed on a ZEISS SV-5 dissecting microscope with a Nikon C-mount 35 mm camera.
Transmission electron microscopy
Embryos were fixed in 0.1 M cacodylate, containing 2% paraformaldehyde and 2% gluteradehyde, overnight at RT, and subsequently processed for transmission electron microscopy as previously described (Chen et al., 1998).
Assessment of heart rate
ED9.5 embryos (with deciduas and fetal blood vessels left attached) were dissected from anesthetized mothers (n=3) in 37°C DMEM containing 5% fetal bovine serum (Invitrogen), and placed within a closed 35 mm dish on the stage of a dissecting microscope connected to an image processor. Light was transmitted through a region of the atrioventricular junction and was detected by the image processor. Using the computer monitor, both atrial and ventricular beating rates were determined three times, simultaneously by two people within a three minute period.
Targeted disruption of the MLC2a gene
The MLC2a gene replacement targeting construct is depicted in Fig. 1A (middle panel). The targeting construct was designed to replace the entire coding sequence (exons one to seven) of the MLC2a gene, with Cre cDNA and the neomycin cytidine deaminase gene cassette (NEO, Fig. 1A, middle panel), thereby inactivating the gene. Targeted ES cells were identified by Southern blot hybridization analysis, characteristic of the targeted allele, giving the expected 7.5 kb mutant fragment compared with the wild-type 9.5 kb fragment (Fig. 1B). This strategy resulted in a lack of MLC2a protein expression within homozygous MLC2a hearts, as detected by protein blot analysis using specific MLC2 antibodies (Fig. 1C), when compared with wild-type littermates.
The MLC2a-null mutation results in embryonic lethality
Heterozygous MLC2a-mutant offspring were viable, fertile and appeared normal in all respects. However, no viable MLC2a-null offspring were obtained in litters from MLC2a heterozygote intercrosses (Table 1), indicating that the MLC2a-null mutation was embryonic lethal. To assess the timing of lethality, Mendelian ratios of viable embryos from MLC2a heterozygote intercrosses were assessed at early embryonic developmental stages (ED7.5-ED11.5; Table 1). All MLC2a-null embryos were viable up to ED10.5 (Table 1). However, at ED10.5, null embryos were observed to have severe chest edema, indicative of heart failure, and growth arrest at approximately somite pair (sp) 24-28 (Fig. 2A,G). Assessment of MLC2a-null embryos at earlier stages of development revealed that growth retardation became evident at approximately ED 9.0-9.25 (sp 15-17). No viable MLC2a-null embryos were observed at ED11.5 (Table 1), suggesting that embryonic lethality occurred between ED10.5 and 11.5. As no embryos of normal appearance and growth were homozygous null for MLC2a, the observed phenotype of growth retardation and embryonic lethality was fully penetrant.
MLC2a-null embryos display aberrant cardiac morphogenesis from the linear heart tube stage
A comparison of MLC2a-null mutants and their wild-type or heterozygous littermates revealed no morphologically evident differences at the early cardiac crescent stage ED7.75 (5-7sp) (data not shown). However, aberrant cardiac morphogenesis in MLC2a mutants was apparent at the early linear heart tube stage ED8.5 (8-10 sp) (Fig. 3A-F). Relative to the linear heart tubes of wild-type or heterozygous littermates, mutant heart tubes appeared to be enlarged and amorphous, with no clear distinction between the bulbus arteriosus and the future left ventricle of the heart. With progression through looping morphogenesis, mutant hearts exhibited aberrant morphologies in each cardiac chamber, accompanied by overall abnormalities in looping architecture (Fig. 3G-R). By sp 22, enlarged atria were present in the mutant embryos relative to littermate controls (Fig. 3S-X). By sp 24-28, both enlarged atria and enlarged outflow tracts were apparent (Fig. 2). Morphological abnormalities in hearts from mutant embryos were consistent over a large number of mutants analyzed.
Histological analysis of MLC2a mutant embryos and their control littermates revealed a number of striking differences between them in myocardium and endocardium. Mutant ventricular myocardium was relatively thin walled, with underdeveloped trabeculae and left ventricular dilation when compared with controls, at ED10.0-10.5. At early stages of atrioventricular valve formation, the atrioventricular cushion becomes seeded by mesenchymal cells that have undergone an epithelial to mesenchymal transformation (Markwald et al., 1999). No seeding by mesenchymal cells was apparent in MLC2a mutant embryos at sp 24 (Fig. 2). Similar results were also obtained at sp 22 (data not shown).
We also performed whole-mount in situ hybridization analysis of MLC2a-null embryos and control littermates, using digoxigenin-labeled antisense riboprobes for a number of myocardial markers, including TBX5 (left ventricle and atria), MLC2v (ventricle) and NKX2.5 (entire myocardium). No differences were observed between MLC2a mutants and their control littermates (data not shown).
MLC2a-null embryos display lack of atrial myofibrillar organization and function
Ultrastructural analysis revealed that MLC2a-null atrial cardiomyocytes have a complete absence of myofibrillar organization, a lack of normal parallel alignment of thick and thin filaments when compared with somite-matched wild-type controls (Fig. 4A,C). By contrast, cardiomyocytes from the left ventricle of MLC2a-null embryos exhibited normal parallel alignment of thick and thin filaments, and Z-line formation, similar to somite-matched wild-type controls (Fig. 4B,D). In terms of function, severely diminished atrial beating was observed in MLC2a-null embryos (Fig. 5). By contrast, no significant difference in the rate of ventricular beating was observed between MLC2a-null and somite-matched wild-type controls (Fig. 5).
MLC2a-null embryos display defects in both extraembryonic and intraembryonic vasculature
Macroscopic analysis of yolk sacs from MLC2a-null embryos revealed that they lack both large and small vessels, as compared with wild-type or heterozygous littermate controls (Fig. 6A-J). Histological analyses demonstrated that the endodermal and mesodermal layers within the yolk sac of MLC2a-null embryos were completely dissociated, which was in contrast to findings with control littermates (Fig. 6E,J). Lack of vessel formation in yolk sacs of MLC2a-null embryos was also confirmed by whole-mount immunostaining using the endothelial cell specific marker PECAM (Fig. 6K-P).
MLC2a-null mutants also exhibited defects in formation of intraembryonic vasculature, evident by sp 24, as revealed by PECAM staining (Fig. 6Q-T). In the mutants, cranial vessels were less complex, and intersegmental vessels were disorganized relative to those observed in control littermates.
In the present study, we have investigated the role of atrial function during embryogenesis by generating a mouse model bearing a null mutation of MLC2a, a structural protein present in cardiac atria. The requirement for atrial function during embryogenesis has remained unexplored (Campbell et al., 1992).
MLC2a-null mice died at approximately ED10.5-11.5 of cardiovascular insufficiency, as indicated by pericardial edema. The proximal cause of death was atrial malfunction. Consistent with the ablation of the atrial specific isoform of myosin light chain 2, the atrial chamber of null embryos contracted infrequently and sporadically, and displayed a lack of myofibrillar assembly. By contrast, the left ventricle contracted normally, and displayed normal myofilament assembly. These results are consistent with previous studies from our laboratory, which demonstrated that although MLC2a is initially expressed throughout the linear heart tube before becoming restricted to the atria at ED12.5 (Kubalak et al., 1994), MLC2a protein is not incorporated into the sarcomeric structure of the ventricles (Chen et al., 1998).
We have utilized the fact that atrial function is disrupted from the beginning of cardiogenesis to investigate longstanding questions as to the role of early heart function in cardiac morphogenesis, and in both extra- and intraembryonic vasculogenesis. Our studies clearly demonstrate that disruption of embryonic heart function has severe consequences on crucial aspects of morphological development of the heart, and on angiogenesis. These results have important implications for the etiology of congenital heart disease.
Despite our demonstration that myofibrillogenesis in extra-atrial myocardium was intact in MLC2a-null mutants, defects in chamber morphogenesis were observed throughout the heart, beginning at the early linear heart tube stage. Just prior to this stage, cardiac contraction has initiated, and embryonic and extraembryonic circulations have amalgamated (Kaufman, 1998). Linear heart tubes of MLC2a mutants appeared relatively large and amorphous relative to control littermates. At subsequent stages of heart development, looping morphogenesis occurred in mutant hearts, but was aberrant in a number of respects, including the length, shape and size of each cardiac segment, and their geometrical relationship to each other. Histological analysis revealed multiple abnormalities within myocardial and endocardial lineages of MLC2a homozygous-knockout embryos. The ventricular myocardium of MLC2a mutants was relatively thin, with fewer trabeculae, and left ventricular (LV) dilation may have occurred secondary to diminished cardiac function.
Similar secondary effects on ventricular morphogenesis consequent to atrial dysfunction have been made in a zebrafish line carrying a mutation in an atrial myosin heavy chain, weak atrium (wea) (see Berdougo et al., 2003). As with the left ventricle in MLC2a mouse mutants, ventricles in wea mutants exhibit a decreased ventricular lumen.
A crucial aspect of cardiac morphogenesis is the formation of septa, which divide the mature heart into four chambers. As an initial stage in the division of the atrial and ventricular chambers, endothelial cells in the region of the atrioventricular canal become activated to invade the cardiac jelly (Markwald et al., 1999). The seeding of cardiac jelly by endocardial cells occurs at approximately ED10 in the mouse. In MLC2a homozygous-knockout embryos, this seeding was not observed. This observation suggests that septation and valve formation, among the most common manifestations of congenital heart disease (Epstein and Buck, 2000), can be critically affected by aberrations of early myocardial function.
Our observations have demonstrated that absence of atrial contraction can impact multiple aspects of cardiac morphogenesis, and can affect both myocardial and endocardial lineages. These effects may be secondary to alterations in hemodyamic fluid forces, or secondary to oxygen or nutrient delivery. We favor the explanation that the first alterations in cardiac morphology observed in MLC2a mutants can be attributed to changes in hemodynamic forces for the following reasons. Initial aberrations in mutant heart morphologies are observed shortly after embryonic and extraembryonic circulations have merged. Oxygen-carrying red blood cells are not present within the bloodstream of the mouse embryo until the 20 to 25 somite stage (Palis et al., 1999), rendering it unlikely that insufficient circulation of blood affects tissue oxygenation prior to this stage. Growth retardation in MLC2a embryos is not apparent until approximately 16 somites, making it unlikely that nutrients are limiting at early heart tube stages. Consistent with these observations, experiments in a number of vertebrate species have demonstrated that heart beat and circulation considerably precede the demand for oxygen or nutrient delivery (Burggren et al., 2000). Alterations in fluid forces, distinct from effects on oxygen or nutrient delivery, have previously been shown to be crucial for kidney morphogenesis (Serluca et al., 2002). Our observations suggest that fluid forces also play a crucial role at the earliest stages of cardiac morphogenesis.
A number of previous studies have demonstrated that alterations in fluid forces experienced by endothelial cells, including those of the endocardium, can result in changes in their alignment and changes in gene expression (Icardo, 1989; Resnick et al., 2000; Shyy and Chien, 2002; Topper and Gimbrone, 1999). Genes that are regulated within endothelium in response to hemodynamic sheer stress include those involved in extracellular matrix remodeling and growth factor pathways. In this light, it is interesting to note the changes observed in both cardiac jelly and extra-atrial myocardium in the MLC2a mutant, where alterations in hemodyamic sheer stress experienced by the endocardium would be expected to occur owing to the lack of atrial function.
In addition to defects in cardiac architecture, MLC2a-null mice also displayed defects in both yolk sac and intraembryonic angiogenesis. In MLC2a mutants, the initial vascular plexus of the yolk sac forms, but does not remodel into a vascular network of larger vessels. Similar secondary effects of cardiac malfunction on yolk sac angiogenesis have been observed in a mouse knockout of the sodium calcium exchanger, and in a cardiac specific rescue of an N-cadherin-null mouse, where a previous defect in yolk sac angiogenesis was rescued by expression of either N- or E-cadherin in heart (Koushik et al., 2001; Luo et al., 2001; Wakimoto et al., 2000).
However, in addition to observed effects on yolk sac angiogenesis, MLC2a mutants also show defects in intraembryonic vasculogenesis. Initial stages of angiogenesis appear normal, but subsequent remodeling of vessels does not occur normally in the MLC2a homozygous-null embryos. These results demonstrate aberrations in intraembryonic angiogenesis in MLC2a mutants, which do not reflect an intrinsic defect in endothelial or smooth muscle cells, but rather are secondary to cardiac malfunction. Similar defects in vasculogenesis have been reported in a number of knockout mice that are mutant for genes that are expressed both in myocardium and in vascular cells, or in vascular endothelium and endocardium (Puri et al., 1999). Our results uniquely demonstrate that cardiac function is required for normal maturation of intraembryonic vessels. Our results also caution that an intrinsic requirement for gene function in vascular cell types cannot be inferred in cases where defects in vascular maturation are accompanied by defects in myocardial function. The latter could arise owing to defects within myocardium itself, as seen here with the MLC2a knockout, or within cell lineages that affect the myocardium, including the endocardium (Puri et al., 1999).
In summary, our analysis of the MLC2a mutant has demonstrated that ablation of atrial function can have drastic secondary consequences on cardiac morphology and vascular development. They also suggest that more minor perturbations of atrial or cardiac contractile function, including perturbations of flow during development, could have less drastic but serious consequences for normal cardiovascular development in the human fetus. In particular, our findings suggest that normal alignment of cardiac chambers and cushion formation, crucial aspects in septation of the normal heart that are perturbed in the majority of cases of congenital cardiac disease, can be perturbed consequent to aberrant cardiac function.
This work was supported by grants from NIH (J.C. and S.E.). F.S. is a recipient of the CIHR/Heart and Stroke Foundation of Canada partnership program postdoctoral fellowship. P.-H.C. is supported by grants NHRI-EX91-9108SC and NHRI-EX92-9108SC from the National Health Research Institute, Taiwan.
- Accepted August 27, 2003.
- © 2003.