Author + information
- Received June 17, 2014
- Revision received August 18, 2014
- Accepted September 4, 2014
- Published online December 30, 2014.
- Anne-Cecile Huby, PhD∗,
- Uzmee Mendsaikhan, MD∗,
- Ken Takagi, MD†,
- Ruben Martherus, PhD∗,
- Janaka Wansapura, PhD‡,
- Nan Gong, MD∗,
- Hanna Osinska, PhD∗,
- Jeanne F. James, MD, PhD∗,
- Kristen Kramer, BS∗,
- Kazuyoshi Saito, MD∗,
- Jeffrey Robbins, PhD∗,
- Zaza Khuchua, PhD∗,
- Jeffrey A. Towbin, MD∗ and
- Enkhsaikhan Purevjav, MD, PhD∗∗ ()
- ∗Heart Institute, Department of Pediatrics, Cincinnati Children’s Hospital Medical Center, Cincinnati, Ohio
- †Jikei University, Tokyo, Japan
- ‡Department of Radiology, Imaging Research Center, Cincinnati Children’s Hospital Medical Center, Cincinnati, Ohio
- ↵∗Reprint requests and correspondence:
Dr. Enkhsaikhan Purevjav, Cincinnati Children’s Hospital Medical Center, MLC7020, 3333 Burnet Avenue, Cincinnati, Ohio 45229.
Background Familial restrictive cardiomyopathy (FRCM) has a poor prognosis due to diastolic dysfunction and restrictive physiology (RP). Myocardial stiffness, with or without fibrosis, underlie RP, but the mechanism(s) of restrictive remodeling is unclear. Myopalladin (MYPN) is a messenger molecule that links structural and gene regulatory molecules via translocation from the Z-disk and I-bands to the nucleus in cardiomyocytes. Expression of N-terminal MYPN peptide results in severe disruption of the sarcomere.
Objectives The aim was to study a nonsense MYPN-Q529X mutation previously identified in the FRCM family in an animal model to explore the molecular and pathogenic mechanisms of FRCM.
Methods Functional (echocardiography, cardiac magnetic resonance [CMR] imaging, electrocardiography), morphohistological, gene expression, and molecular studies were performed in knock-in heterozygote (MypnWT/Q526X) and homozygote mice harboring the human MYPN-Q529X mutation.
Results Echocardiographic and CMR imaging signs of diastolic dysfunction with preserved systolic function were identified in 12-week-old MypnWT/Q526X mice. Histology revealed interstitial and perivascular fibrosis without overt hypertrophic remodeling. Truncated MypnQ526X protein was found to translocate to the nucleus. Levels of total and nuclear cardiac ankyrin repeat protein (Carp/Ankrd1) and phosphorylation of mitogen-activated protein kinase/extracellular signal–regulated kinase 1/2 (Erk1/2), Erk1/2, Smad2, and Akt were reduced. Up-regulation was evident for muscle LIM protein (Mlp), desmin, and heart failure (natriuretic peptide A [Nppa], Nppb, and myosin heavy chain 6) and fibrosis (transforming growth factor beta 1, alpha–smooth muscle actin, osteopontin, and periostin) markers.
Conclusions Heterozygote MypnWT/Q526X knock-in mice develop RCM due to persistence of mutant MypnQ526X protein in the nucleus. Down-regulation of Carp and up-regulation of Mlp and desmin appear to augment fibrotic restrictive remodeling, and reduced Erk1/2 levels blunt a hypertrophic response in MypnWT/Q526X hearts.
Restrictive cardiomyopathy (RCM) accounts for approximately 5% of diagnosed cardiomyopathies and is characterized by diastolic dysfunction and restrictive physiology (RP), whereas systolic function typically remains normal or near normal (1). The volume and wall thickness of the ventricles are usually normal or small, whereas atrial or biatrial enlargement occurs due to impaired ventricular filling during diastole (2). Particularly in children, RCM has the poorest prognosis among all types of heart muscle diseases, with 2- and 5-year mortality rates of 50% and 70%, respectively, and the highest rate of sudden cardiac death (SCD) (3). Survivors ultimately develop heart failure (HF) due to RP, as well as pulmonary hypertension; however, the mechanistic basis of RP with diastolic dysfunction, myocardial fibrosis, and cardiac stiffness is unclear.
A history of familial RCM (FRCM) is reported in approximately 30% of RCM cases, with autosomal dominant inheritance most commonly noted (4). Several genes, typically encoding proteins of the sarcomere, Z-disk, cytoskeleton, or intermediate filament network, have been associated with autosomal dominant FRCM (4). The myopalladin (MYPN) gene, which is located at chromosome 10q21.3, encodes a 145-kDa protein that participates in linking regulatory molecules involved in sarcomeric I-band and Z-disk assembly and muscle gene expression (5). The N-terminal MYPN interacts with cardiac ankyrin repeat protein (CARP/ANKRD1), a transcriptional coinhibitor of genes involved in the development of HF and hypertrophy (6). MYPN has dual localization, sarcoplasm and nucleus, similar to that seen with CARP (5). At the Z-disk, MYPN interacts with alpha-actinin and with the SRC homology 3 domain of nebulette (7). Mutations in the MYPN gene cause diverse phenotypes in humans, including dilated cardiomyopathy, hypertrophic cardiomyopathy, and RCM (8,9). We previously reported a nonsense autosomal dominant mutation (MYPN-Q529X) that resulted in FRCM in siblings via disturbed myofibrillogenesis and sarcomeric Z-disk disruption (9).
In this study, knock-in mice carrying a heterozygous and homozygous Mypn-Q526X mutation in exon 10 of the murine Mypn gene, homologous to the human MYPN-Q529X mutation, were analyzed to determine the pathophysiology and molecular mechanism(s) of FRCM.
Generation of knock-in mice
The study conformed to the protocols approved by the Institutional Animal Care and Use Committee at Cincinnati Children’s Hospital Medical Center. To generate a murine Mypn-Q526X mutation, we targeted exon 10 in the Mypn gene (Online Figure 1A) using a homologous recombination method as described previously (10) and detailed in the Online Appendix.
Evaluation of heart function in mice
Serial echocardiography and electrocardiography were performed in mice at 6 and 12 weeks of age (12 animals/group). Cardiac magnetic resonance imaging was performed in 12-week-old animals when mice showed a markedly increased early (E) and late (A) diastolic velocities (E/A) ratio, signs of “restrictive filling,” or diastolic dysfunction by echocardiography. See the Online Appendix for experimental details.
Histopathology, immunohistochemistry, quantitative real-time polymerase chain reaction, and electron microscopy
Histopathology, including hematoxylin and eosin, Masson trichrome, immunohistochemical, transcriptional, and terminal deoxynucleotidyl transferase dUTP nick end-labeling (TUNEL) analysis, was performed to assess structural, fibrotic, hypertrophic, and/or apoptotic remodeling in the heart. Transmission electron microscopy was performed on glutaraldehyde-perfused hearts as previously described (11). After isolation of total RNA from ventricular tissues, quantitative real-time polymerase chain reaction (PCR) was performed as described in the Online Appendix. Six 12-week-old animals per group were used. See experimental details in the Online Appendix.
Protein expression, pull-down, and Western blotting
Human embryonic kidney (HEK293) cells were transfected with different chimeras of MYPN–green fluorescent protein (GFP) and CARP-V5 complementary DNA (cDNA) to evaluate MYPN and CARP interactions using immunoprecipitation and coimmunoprecipitation. Cellular fractionation was performed using the NE-PER kit (Pierce, Rockford, Illinois). Western blotting was used for protein evaluation, and levels of proteins were quantified in relative density units using ImageJ software as described in the Online Appendix.
Statistical analysis, reported as mean ± standard error of the mean, was performed with the Student t test or 1-way analysis of variance using GraphPad5 software (GraphPad Software, Inc., La Jolla, California). A probability value of p ≤ 0.05 was considered significant.
Manifestation of diastolic dysfunction
Given that persistence of the 65-kDa MypnQ526X peptide in vivo may potentially cause a “poison peptide” effect, levels of the Mypn protein were confirmed by Western blotting; the MypnWT/Q526X line 3 with the highest 65-kDa MypnQ526X protein expression was selected for further breeding (Online Figure 1B). Heterozygous and homozygous recombination of the Mypn-Q526X mutation was confirmed by sequencing of Mypn exon 10 in genomic DNA obtained from mouse tails (Online Figure 1C). We verified Mypn mRNA transcription levels in the heart, skeletal muscle, liver, and kidneys by reverse transcriptase PCR. As expected, homozygotes had impaired Mypn mRNA transcription in heart and skeletal muscle compared with wild-type (WT) and heterozygote littermates (Online Figures 1D and 1E). Thus, homozygous MypnQ526X mice were also generated as a model of Mypn gene ablation.
Both MypnWT/Q526X and MypnQ526X mice were born and grew normally, with no skeletal myopathy observed up to 12 weeks of age. At 6 weeks, increased E/A ratios, features of RP in humans (12), were detected in MypnWT/Q526X mice compared with WT and homozygotes. The same cohort of mice underwent a serial echocardiography at 12 weeks of age, and the E/A ratio increased significantly (Figure 1A). Although systolic function and chamber dimensions were preserved, MypnWT/Q526X mice showed signs of impaired diastolic filling of the left ventricle (LV), including decreased end-diastolic volume (LVEDV) and internal dimensions compared with WT and homozygotes (Online Table 1).
Further, cardiac magnetic resonance imaging was used in the same cohort of 12-week-old mice to precisely evaluate cardiac anatomy and diastolic function. No significant differences between groups were found in circumferential strain, strain rate, torsion, or torsion rate (data not shown). The left atrial cross-sectional area of MypnWT/Q526X mice (3.8 ± 0.5 mm2) was larger compared with that of WT mice (3.5 ± 0.6 mm2) and homozygotes (3.0 ± 0.3 mm2; p = 0.02), as measured in the 4-chamber view (Figure 1B, Online Table 2). The LVEDV was significantly lower in the MypnWT/Q526X mutants compared with WT mice (p = 0.03). The sphericity index, the ratio of the short- to long-axis dimensions of the LV, was also lower in the MypnWT/Q526X mice (0.51 ± 0.03) compared with WT animals (0.56 ± 0.03; p = 0.02), consistent with LV remodeling in heterozygotes only.
Arrhythmias and conduction abnormalities are reportedly present in 15% to 30% of patients with RCM, especially in pediatric patients (13). Electrocardiography revealed that heart rate, pulse, pulse rate, and QRS durations were similar among mutant and WT mice, whereas T-wave duration was decreased in MypnWT/Q526X versus WT or homozygotes (Online Table 3). Arrhythmias, including premature atrial contractions, premature ventricular contractions, and type II second-degree atrioventricular block (Figure 1C), were observed in heterozygotes only.
Fibrosis in heterozygous MypnWT/Q526X hearts
Morphological evaluation of cardiac muscle revealed no hypertrophy in any animal (Figure 2A). Histologically, diffuse interstitial and perivascular fibrosis was found in MypnWT/Q526X ventricular myocardium only. Dystrophin staining revealed intact sarcolemmal integrity and normal cardiomyocyte size among the different groups, confirming absence of cardiomyocyte hypertrophy (Online Figure 2A). No positive TUNEL staining was seen in any group, excluding the involvement of apoptosis. Ultrastructurally, t-tubule enlargement and mild intercalated disk disruption were seen in heterozygous and homozygous animals (Figure 2B).
Taken together, these results showed that heterozygote MypnWT/Q526X mutants demonstrated fibrotic remodeling without an overt hypertrophic response, accurately recapitulating the human RCM phenotype. Homozygous MypnQ526X mice had no or a minimal phenotype as result of MYPNQ526X ablation. Therefore, further comparative molecular studies were carried out in MypnWT/Q526X and WT littermates.
Putative Z-disk stress-sensor dysregulation in MypnWT/Q526X hearts
Immunohistochemical analysis revealed no differences in Mypn at the Z-disks. Notably, coincident delocalization of Carp and desmin (Des) to the periphery was found in MypnWT/Q526X cardiomyocytes only (Figure 3, arrows). On a protein level, although no MypnWT was seen, the 65-kDa MypnQ526X was detected in the hearts of heterozygous mice as expected (Figure 4A). A significant increase in the expression of the Mypn homolog, palladin (90 kDa), was evident in MypnWT/Q526X hearts and confirmed by monoclonal antibody and immunohistochemical staining (Online Figure 2B). Nebulette and alpha-actinin were mildly up-regulated, whereas Carp was considerably down-regulated in MypnWT/Q526X (p = 0.03). Levels of calpain 3 were similar in mutants and WT, excluding the likelihood of Carp degradation due to increased protease activation. Des and muscle LIM protein (Mlp/cysteine-rich protein 3 [Csrp3]), the putative Z-disk stress-sensors, were significantly up-regulated in heterozygotes versus WT (p < 0.05). Mild down-regulation of alpha-tubulin, caveolin 3, and vinculin occurred in MypnWT/Q526X compared with WT, suggesting early recruitment of t-tubules and adherens junctions, respectively (Online Figure 2B). Intercalated disk proteins such as desmoplakin 2, connexin 43, or N-cadherin were not affected. Levels of cardiac troponin I (cTnI) and phospho-cTnI were normal, consistent with those of intact sarcomeres.
Analysis of downstream signaling pathways
Because cardiac stretch is regulated mainly at the Z-disks by extracellular signal–regulated kinase 2 (ERK2)–dependent phosphorylation of titin at the Zis1/Z-repeats (14), we sought to determine whether the ERK cascade is altered. Reduced levels of phosphorylated–mitogen-activated protein kinase (MAPK)/ERK1/2 (pMek1/2) and pErk1/2 were seen in MypnWT/Q526X hearts compared with those of WT (Figure 4B), whereas other potential MAPK-associated targets, including p53, c-Jun N-terminal kinase, p38, focal adhesion kinase, nuclear factor kappa–light chain enhancer of activated B cells, signal transducer and activator of transcription 3, ERBB4, and transforming growth factor (TGF)-β1, were not affected (Online Figure 2C). However, there was less phosphorylation of Smad2 and Akt in MypnWT/Q526X animals. Further, CARP-dependent genes, including natriuretic peptide A (Nppa), Nppb, and myosin heavy chain 6 mRNA (Online Table 4), were more highly expressed in MypnWT/Q526X hearts, supporting the idea of diminished inhibitory effects of CARP due to reduced levels of CARP protein (Figure 4C). Significant increases in the expression of fibrotic (α-smooth muscle actin, Tgfβ1, periostin, and osteopontin), inflammatory (vascular cell adhesion protein, intercellular adhesion molecule, and interleukin-1β), and antiapoptotic (B-cell lymphoma 2) genes were identified in MypnWT/Q526X hearts.
Persistence of nuclear 65-kDa Mypn peptide associated with RCM phenotype
Mouse hearts from all groups (WT and heterozygote and homozygote mutants) were fractionated and levels of total, nuclear, and cytoplasmic Mypn, palladin, and Carp proteins were quantitated to determine whether the location and expression levels of the proteins were associated with the RCM phenotype in vivo. Comparable expression of nuclear palladin was documented in all groups (Figure 5A), excluding the likelihood of pathological effects of increased palladin levels in the nucleus. Interestingly, the 65-kDa MypnQ526X peptide was detected in the nuclear fraction of heterozygote hearts, but very little was detected in homozygote hearts, suggesting a potential cause of the RCM phenotype. A concordant decrease in total Carp levels was detected in MypnWT/Q526X hearts.
For further clarification of MYPN-CARP interactions, different combinations of human MYPNWT and MYPNQ529X cDNAs fused with GFP or mixtures of both MYPNWT and MYPNQ529X (MYPNWT/Q529X) were cotransfected with a human CARP-V5 construct in HEK293 cells and pulled down. As shown in Figure 5B, the GFP antibody pulled down the 65-kDa MYPNQ529X as did the 145-kDa MYPNWT in all transfected cells, except controls and CARP-transfected ones. In contrast, levels of V5-pulled 40-kDa CARP were reduced in MYPNQ529X cells compared with those in MYPNWT. Further, the 40-kDa CARP-V5 was undetectable in MYPNWT/Q529X mixed lysates, presumably due to greater reduction of CARP expression compared with mutant MYPNQ529X cells. These effects support the in vivo data of reduced murine Carp levels in heterozygous hearts and the idea that nuclear 65-kDa MypnQ526X may suppress Carp expression.
Subsequently, levels of Carp, Mlp, and Des mRNA were evaluated in the myocardial tissues of heterozygote and homozygote mice to further elucidate whether differential phenotypes were associated with nuclear MypnQ526X-mediated alteration in transcriptional activities of mechanosensitive molecules. As shown in Figure 5C, divergent changes in CARP mRNA levels were detected in MypnWT/Q529X and MypnQ529X hearts compared with WT controls. Carp transcription was reduced in MypnWT/Q529X, whereas MypnQ529X mice presented Carp up-regulation, and this variation between mutants was statistically significant. In contrast, no significant difference in transcription of Mlp and Des was revealed in any groups.
The characteristic features of autosomal dominant FRCM result in a high-risk disorder with poor clinical outcomes, including HF, arrhythmias, and SCD (2). These features include diastolic dysfunction and restrictive ventricular filling, with limited ability to augment cardiac stroke volume due to increased myocardial stiffness and interstitial fibrosis; however, a mechanistic etiology remains unclear (4). MYPN is a sarcomeric protein localized at the Z-disk, I-band, and nucleus. Expression of the N-terminal region of MYPN in cardiomyocytes resulted in severe disruption of the sarcomere (5), whereas N-terminal mutation MYPN-Y20C reduced CARP-MYPN binding and caused hypertrophic cardiomyopathy in vivo (Central Illustration) via perturbed MYPN nuclear shuttling and abnormal assembly of terminal Z-disks within intercalated disks (9). Mutations in MYPN, therefore, cause diverse cardiomyopathic phenotypes within a critical “final common pathway” (9,15).
The objectives of this study were to uncover the molecular pathogenetic mechanism(s) and discover the associated key signaling molecules in MYPN-Q529X mutation–induced restrictive remodeling using a knock-in mouse model. We demonstrated that homozygous MypnQ526X mice displayed no phenotype due to ablation of the mutant protein. Heterozygous MypnWT/Q526X mutants, corresponding to a human heterozygous MYPN-Q529X mutation, recapitulated the clinical and pathological features of human FRCM, including diastolic dysfunction with preserved systolic function by 12 weeks of age. Interstitial and perivascular fibrosis was seen before overt cardiac contractile dysfunction, whereas myocyte hypertrophy, apoptosis, and necrosis were absent.
From these data, we hypothesize that the RCM phenotype results from persistence of dysfunctional truncated MypnQ526X protein and consequent multiple pathological “hits.” First, MypnQ526X translocates to the nucleus, probably perturbing levels of Carp expression (Central Illustration). Second, MypnQ526X causes alterations in Mlp/Csrp3 and Des and blunts the phosphorylation of Mek1/2–Erk1/2, likely via negative nuclear feedback. Further, activation of Smad2 and Akt are reduced, likely to compensate for the mutation-induced “final common pathway” of RCM (15).
Balanced levels of CARP are reported to be essential for proper adaptive responses to different stresses in striated muscle (Central Illustration), including exercise (16), stretch (17), cytokines (18), hypoxia (19), adrenergic stress (20), and anthracycline (doxorubicin) toxicity (21). Multiple mechanisms, including 26S proteasome degradation, TGFβ-SMADs, and caspase 3 and calpain 3 proteases, are reported to be CARP regulators (22). Ablation of all forms of muscle ankyrin repeat proteins (MARPs), including CARP, enhances myogenic differentiation protein and MLP expression in skeletal muscle (23). Although ablation of CARP or MARPs has been shown to not affect cardiac function similarly to our MypnQ526X homozygotes (24), cardiac-restricted CARP overexpression results in an inhibition of hypertrophic and fibrotic remodeling via reduced levels of TGFβ and ERK1/2 (18). Our previous MypnY20C transgenic mouse model (Central Illustration) suggested a crucial role of nuclear WT Mypn by demonstrating the development of hypertrophic cardiomyopathy in vivo as a result of perturbed nuclear shuttling of Mypn (9). In the present MypnWT/Q526X model, the mutant MypnQ526X translocated to the nucleus, reducing the expression of Carp. Transcription of a key fibrotic molecule, Tgfβ1, was up-regulated, but Tgfβ1 protein levels were not changed, and Smad2 was down-regulated, suggesting an alternative TGFβ-SMAD–independent fibrotic pathway. Increased expression of 90-kDa palladin was shown to induce α-smooth muscle actin through Tgfβ1 and Erk1/2 activation (25), whereas decreased Ankrd2 and Mlp/Csrp3 levels were reported in forced skeletal muscle inactivity in mice (26). Therefore, decreased Carp and elevated Mlp/Csrp3 and Des expression likely led to fibrosis via periostin and osteopontin in our model (17). Moreover, up-regulation of intercellular adhesion protein, vascular cell adhesion protein, and interleukin-1β in mutants may contribute to perivascular fibrosis (27).
We hypothesized that the net result of decreased Erk1/2, Smads, and Akt signaling may blunt the hypertrophic response in heterozygotes (Figure 5C). Because Carp has been shown to regulate the Erk2-dependent titin-PEVK spring force at the I-band, we assume that reduction in Mek1/2–Erk1/2 phosphorylation might have a negative feedback origin through the nuclear MypnQ526X-induced perturbation of mechanosensitive proteins, Carp, Mlp, and Des (28).
An assessment of diastolic dysfunction ideally includes invasive measurements. This study adopted an integrated approach using echocardiography and CMR imaging in the same cohort of mice and provides compatible data comparable to invasive measurements.
We demonstrated that the 65-kDa MypnQ526X peptide caused diastolic dysfunction without contractile impairment via net changes in the mechanosensory proteins. To date, no pharmacological therapies have been shown to clearly improve outcomes in patients with RCM, with heart transplant being the definitive treatment option for many, especially in childhood (29). If our murine findings mirror the pathological hallmarks of RCM in humans, it could explain why the TGFβ1 suppressor angiotensin-converting enzyme inhibitors, beta-blockers, and angiotensin receptor blockers are ineffective in the treatment of patients with RCM (3). Further studies on time-dependent expression changes in CARP, MLP, DES, and ERK1/2 in patients with RCM may provide useful information for discovering diagnostic and therapeutic targets.
COMPETENCY IN MEDICAL KNOWLEDGE: Restrictive myocardial remodeling is responsible for the diastolic ventricular dysfunction that characterizes FRCM, which carries a poor prognosis, particularly in affected children. No pharmacological therapies have been found that clearly improve clinical outcomes in patients with this disorder, leaving few alternatives to cardiac transplant.
TRANSLATIONAL OUTLOOK: This study discovered early molecular markers of restrictive remodeling in a murine model of FRCM that caused diastolic dysfunction without contractile impairment. This points the way toward future studies in which the expression of genes regulating mechanosensory proteins could be modified as potential therapeutic targets.
The authors thank Dr. Siegfried Labeit (Klinikum Mannheim, Mannheim, Germany) for providing antibodies against myopalladin and nebulette.
For a supplemental Methods section as well as figures and tables, please see the online version of this article.
This work was supported in part by research grants from the American Heart Association, Children’s Cardiomyopathy Foundation (to Drs. Towbin and Purevjav), John Patrick Albright Foundation, and the National Institutes of Health (R01 HL53392 and R01 HL087000; to Dr. Towbin). All authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- cardiac ankyrin repeat protein
- extracellular signal–regulated kinase
- familial restrictive cardiomyopathy
- heart failure
- left ventricular
- muscle LIM protein
- restrictive cardiomyopathy
- restrictive physiology
- transforming growth factor
- Received June 17, 2014.
- Revision received August 18, 2014.
- Accepted September 4, 2014.
- American College of Cardiology Foundation
- Maron B.J.,
- Towbin J.A.,
- Thiene G.,
- et al.
- Webber S.A.,
- Lipshultz S.E.,
- Sleeper L.A.,
- et al.
- Bang M.L.,
- Mudry R.E.,
- McElhinny A.S.,
- et al.
- Duboscq-Bidot L.,
- Xu P.,
- Charron P.,
- et al.
- Purevjav E.,
- Arimura T.,
- Augustin S.,
- et al.
- Liu P.,
- Jenkins N.A.,
- Copeland N.G.
- Maiellaro-Rafferty K.,
- Wansapura J.P.,
- Mendsaikhan U.,
- et al.
- Walsh M.A.,
- Grenier M.A.,
- Jefferies J.L.,
- Towbin J.A.,
- Lorts A.,
- Czosek R.J.
- Gautel M.,
- Goulding D.,
- Bullard B.,
- Weber K.,
- Furst D.O.
- Mohamed J.S.,
- Boriek A.M.
- Lee M.J.,
- Kwak Y.K.,
- You K.R.,
- Lee B.H.,
- Kim D.G.
- Zolk O.,
- Marx M.,
- Jackel E.,
- El-Armouche A.,
- Eschenhagen T.
- Kanai H.,
- Tanaka T.,
- Aihara Y.,
- et al.
- Barash I.A.,
- Bang M.L.,
- Mathew L.,
- Greaser M.L.,
- Chen J.,
- Lieber R.L.
- Roberts M.D.,
- Childs T.E.,
- Brown J.D.,
- Davis J.W.,
- Booth F.W.
- Wilhelmi M.H.,
- Leyh R.G.,
- Wilhelmi M.,
- Haverich A.