Author + information
- Received May 20, 2005
- Revision received July 8, 2005
- Accepted July 11, 2005
- Published online November 15, 2005.
- Saskia L.M.A. Beeres, MD⁎,
- Douwe E. Atsma, MD, PhD⁎,⁎ (, )
- Arnoud van der Laarse, PhD⁎,
- Daniël A. Pijnappels, MSc⁎,
- John van Tuyn, MSc⁎,†,
- Willem E. Fibbe, MD, PhD‡,
- Antoine A.F. de Vries, PhD†,
- Dirk L. Ypey, PhD§,
- Ernst E. van der Wall, MD, PhD⁎ and
- Martin J. Schalij, MD, PhD⁎
- ↵⁎Reprint requests and correspondence:
Dr. Douwe E. Atsma, Department of Cardiology, Leiden University Medical Center, P.O. Box 9600, 2300RC Leiden, the Netherlands
Objectives We evaluated whether human adult bone marrow-derived mesenchymal stem cells (hMSCs) could repair an experimentally induced conduction block in cardiomyocyte cultures.
Background Autologous stem cell therapy is a novel treatment option for patients with heart disease. However, detailed electrophysiological characterization of hMSCs is still lacking.
Methods Neonatal rat cardiomyocytes were seeded on multi-electrode arrays. After 48 h, abrasion of a 200- to 450-μm–wide channel caused conduction block. Next, we applied adult hMSCs (hMSC group, n = 8), human skeletal myoblasts (myoblast group, n = 7), rat cardiac fibroblasts (fibroblast group, n = 7), or no cells (control group, n = 7) in a channel-crossing pattern. Cross-channel electrical conduction was analyzed after 24 and 48 h. Intracellular action potentials of hMSCs and cardiomyocytes were recorded. Immunostaining for connexins and intercellular dye transfer (calcein) assessed the presence of functional gap junctions.
Results After creation of conduction block, two asynchronously beating fields of cardiomyocytes were present. Application of hMSCs restored synchronization between the two fields in five of eight cultures after 24 h. Conduction velocity across hMSCs (0.9 ± 0.4 cm/s) was approximately 11-fold slower than across cardiomyocytes (10.4 ± 5.8 cm/s). No resynchronization occurred in the myoblast, fibroblast, or control group. Intracellular action potential recordings indicated that conduction across the channel presumably occurred by electrotonic impulse propagation. Connexin-43 was present along regions of hMSC-to-cardiomyocyte contact, but not along regions of cardiomyocyte-to-myoblast or cardiomyocyte-to-fibroblast contact. Calcein transfer from cardiomyocytes to hMSCs was observed within 24 h after co-culture initiation.
Conclusions Human mesenchymal stem cells are able to repair conduction block in cardiomyocyte cultures, probably through connexin-mediated coupling.
Autologous cell therapy has become a novel therapeutic option for patients with ischemic heart disease (1). Cardiac cell therapy aims to regenerate myocardial tissue and to improve neovascularization. To this purpose, bone marrow-derived cells, myoblasts, and peripheral blood-derived cells have been used in clinical trials (2–7). Initial enthusiasm for cardiac myoblast transplantation has somewhat faded because life-threatening arrhythmias have been observed (6). Grafted myoblasts were found to differentiate into hyperexcitable myotubes with a contractile activity electrically independent of neighboring cardiomyocytes, because the differentiated myoblasts did not express cardiac connexins (8,9). These findings highlight the importance of electrical coupling by any therapeutic cell type to be applied to damaged myocardium.
Several clinical studies have shown safety and feasibility of autologous bone marrow cell transfer in patients with myocardial infarction, refractory ischemia, or heart failure (2–5,7). In addition, it was shown that both symptoms and clinical parameters improved after cell therapy (2–5,7). In these studies, repetitive Holter monitoring showed no proarrhythmogenic effect, and electrophysiological studies showed no increased inducibility of ventricular arrhythmias (2). However, detailed electrophysiological characterization of adult human mesenchymal stem cells (hMSCs) injected in the myocardium is still lacking.
Recently, in vitro immunostaining showed the presence of cardiac connexins (Cxs) along regions of intimate cell-to-cell contact between hMSCs and between hMSCs and canine cardiomyocytes (10). The presence of Cxs resulted in electrical coupling between adjacent cells, but whether such electrical coupling permits activation of a mass of cardiac tissue over extended distances is not known.
We investigated whether adult hMSCs could provide resynchronization of two previously asynchronously beating fields of cardiomyocytes. To this purpose, an in vitro model was developed in which a monolayer of beating neonatal rat cardiomyocytes was separated in two halves by a channel, thereby creating a line of anatomical conduction block. Next, hMSCs were seeded in a channel-crossing pattern and the presence of electrical coupling between the cardiomyocyte fields across the hMSCs was assessed using extracellular and intracellular electrophysiological techniques. Immunostaining and intercellular dye transfer were used to investigate presence and function of cardiac Cxs.
Cardiomyocytes and cardiac fibroblasts
Animal experiments were approved by the institutional animal experiments committee and complied with the European Convention of Animal Care. Cultures of cardiomyocytes were prepared as described earlier (11). Cardiac ventricles of two-day-old Wistar rats were minced and dissociated using collagenase and DNase. The cells were suspended in Ham’s F10 medium (ICN Biomedicals, Irvine, California) with 10% horse serum (Invitrogen, Eugene, Oregon) and 10% fetal bovine serum (FBS) (Invitrogen), and pre-plated to allow preferential attachment of non-cardiomyocytes. After 1 h, the non-adherent cells were collected and plated in micro-electrode arrays (MEA) (Multi Channel Systems, Reutlingen, Germany) or in six-well plates containing uncoated glass coverslips. Culture medium consisted of Dulbecco’s modified eagle medium (DMEM) (Invitrogen) with 10% FBS, 100 U/ml penicillin and 100 μg/ml streptomycin (Invitrogen). Cells were grown at 37°C in a humidified CO2incubator. Spontaneously beating cardiomyocytes were observed within 48 h after culture initiation.
Cardiac fibroblasts remained attached to the culture dish after removal of the cardiomyocytes and were grown in DMEM containing 10% FBS and the aforementioned antibiotics. After 24 h, the cardiac fibroblasts were trypsinized, resuspended, and seeded into culture dishes with uncoated glass coverslips.
Adult hMSCs were obtained from the posterior iliac crest of three patients with ischemic heart disease scheduled for cell therapy after approval of the institutional medical ethical committee and after written informed consent. Bone marrow cells were separated over a Ficoll density gradient, and the mononuclear cells were harvested and washed with phosphate-buffered saline (PBS) containing 0.5% human serum albumin. Cells were pelleted by centrifugation and resuspended in DMEM with 10% FBS and antibiotics. Twenty-four hours after seeding the cells at a density of 106cells/cm2in culture flasks, the non-adherent cells were discarded and hMSCs were expanded by serial passage, and used from passages three to six. The hMSC surface antigen profile characterized by a FACSort flow cytometer and CellQuest Software (Becton-Dickinson, Palo Alto, California) matched previously published data (12). Their ability to differentiate into adipocytes and osteoblasts after appropriate stimulation confirmed that the cultured cells were hMSCs. For identification, hMSCs were infected with an adenoviral vector-encoding enhanced green fluorescent protein (eGFP;hAd5/F50.CMV.eGFP). For immunostaining and dye transfer experiments, hMSCs were labeled with red fluorescent protein (DsRed;hAd5/F50.hEF1α.DsRed) (13). The generation, propagation, purification, and titration of these Ad vectors were carried out as described elsewhere (14).
Human skeletal myoblasts
Human skeletal myoblasts (a gift of Dr. D. Trono, University of Geneva, Geneva, Switzerland) were prepared as previously described (15). Myoblasts were cultured in Ham’s F10 medium supplemented with 15% fetal calf serum, bovine serum albumin (0.5 mg/ml), fetuin (0.5 mg/ml), epidermal growth factor (10 ng/ml), dexamethasone (0.39 μg/ml), insulin (0.18 mg/ml), creatine (1 mM), pyruvate (100 μ g/ml), and uridine (50 μg/ml).
In vitro model for conduction block
Cardiomyocytes were plated on standard planar MEAs containing 60 electrodes (inter-electrode distance, 200 μm; electrode diameter, 30 μm). The MEAs were pre-coated with Ham’s F10 medium containing 10% horse serum and 10% FBS. The cultures were maintained at 37°C during the measurements. The MEA data acquisition system allowed simultaneous recording of 60 extracellular electrograms (sample rate 5 kHz/channel), which were analyzed using MC-Rack software (version 188.8.131.52, Multi Channel Systems).
Local activation time (LAT) was determined by the timing of the maximal negative intrinsic deflection (−dV/dtmax) of the electrogram recorded at each electrode. The LAT values of all electrodes were used for the generation of color-coded activation maps using two-dimensional plotting software (S-Plus, version 6.0, Insightful, Seattle, Washington).
Two days after seeding cardiomyocytes on the MEAs, activation maps were constructed to assess impulse propagation. Next, a 200- to 450-μm-wide channel was abraded in the cell monolayer perpendicular to the activation direction using a 200-μm-wide pipette tip. After ensuring the presence of conduction block, we applied either 50,000 eGFP-labeled hMSCs (hMSC group, n = 8), or 50,000 myoblasts (myoblast group, n = 7), or 50,000 cardiac fibroblasts (fibroblast group, n = 7) in a channel-crossing pattern. In seven cultures the channel was not filled with extraneous cells (control group, n = 7).
Impulse propagation was assessed at 24 and 48 h after seeding the cells. In this model, the depolarization wave front has to traverse two to five cells in the channel. The two previously asynchronously beating cardiomyocyte fields were considered electrically coupled if the activation map showed conduction of electrical activity through the channel. To rule out the possibility of coincident spontaneous activation in the lower field of cardiomyocytes, graphs were created plotting consecutive LATs of the upper field versus the LATs of the lower fields over a period of 30 s. The first LAT on an electrode in the upper field was correlated with the first LAT on an electrode in the lower field, the second LAT on the same electrode in the upper field was correlated with the second LAT on the same electrode below in the lower field, and so on. Therefore, each point on each graph represents one single excitation wave measured in the upper and lower field simultaneously. A culture was considered synchronized only if there was a constant time interval between LATs on either side of the channel leading to a 1:1 correlation between the LATs of both fields.
Cultures of cardiomyocytes with hMSCs seeded in a 200- to 450-μm-wide channel were studied 24 to 48 h after hMSC seeding and after ensuring resynchronization. Action potentials of beating cardiomyocytes and hMSCs in the channel were recorded by the use of standard patch electrode techniques (glass capillaries filled with (in mM) 10 Na2ATP, 115 KCl, 1 MgCl2, 5 EGTA, 10 HEPES/KOH (pH 7.4), and a tip resistance of 2.0 to 2.5 MΩ) (16).
Mixed cultures of cardiomyocytes and DsRed-labeled hMSCs or skeletal myoblasts or cardiac fibroblasts on glass coverslips were subjected to immunostaining with Cx40-, Cx43-, or Cx45-specific antibodies. After 48 h of co-incubation, the mixed cultures were fixated for 30 min in PBS-1% formalin (Merck, Darmstadt, Germany), permeabilized using 0.1% Triton X-100 (BDH Laboratories, Poole, England) and incubated with goat anti-Cx40 (Santa Cruz Biotechnology, Santa Cruz, California), rabbit anti-Cx43 (Sigma, St. Louis, Missouri) or goat anti-Cx45 (Santa Cruz Biotechnology) antibodies at a dilution of 1:100 in PBS and 1% FBS for 24 h at 4°C. The cells were then washed with PBS and incubated with secondary fluorescein isothiocyanate (FITC)-conjugated anti-goat antibodies (dilution 1:100 in PBS; Sigma) or FITC-conjugated anti-rabbit antibodies (dilution 1:100 in PBS; Sigma) for 1 h at 4°C. Then, the coverslips were incubated in a solution containing Hoechst 33342 (dilution, 1:1000 in PBS; Molecular Probes, Eugene, Oregon) for 8 min at 4°C to stain cell nuclei. Finally, the coverslips were mounted onto glass slides using Vectashield mounting medium (Vector Laboratories, Burlingame, California) and sealed with transparent fingernail polish. Stained cells were examined using a fluorescence microscope (Nikon Eclipse, Nikon Europe, Badhoevedorp, the Netherlands) equipped with a digital camera (Nikon DXM1200).
Functional gap junction coupling between cardiomyocytes and hMSCs was assessed using a fluorescent dye transfer assay (17). After two days of culture on glass coverslips, cardiomyocytes were loaded with 10 μmol/l calcein-AM (Molecular Probes) for 45 min. Calcein-AM, a non-fluorescent cell-permeable compound, is converted by intracellular esterases to calcein, a 622-Da green fluorochrome retained in the cytoplasm, which can diffuse through functional gap junctions. After incubation of cardiomyocytes with calcein-AM, culture medium was replaced by HEPES (ICN Biomedicals)-buffered salt solution containing 2.5 mmol/l probenecid (Sigma) to prevent cellular release of calcein. Thereafter, DsRed-labeled hMSCs were added to the calcein-loaded cardiomyocytes. If functional gap junctions between cardiomyocytes and hMSCs had been established, calcein transfer from cardiomyocytes to hMSCs should occur through these channels, and hMSCs should show both green and red fluorescence. Calcein transfer was assessed 24 h after co-culture initiation using a computer-controlled fluorescence microscope (Zeiss Axiovert 200M, Göttingen, Germany) equipped with imaging software (OpenLab, Improvision, Coventry, England).
Statistical analysis was performed using SPSS 11.0 for Windows (SPSS Inc., Chicago, Illinois). Data were compared using the Student ttest, analysis of variance, or the two-sided chi-square test when appropriate. Synchronization between two beating cardiomyocyte fields was assessed using linear correlation analysis (Pearson correlation coefficient). A p value < 0.05 was considered statistically significant.
Synchronization of electrical activity
A spontaneously beating monolayer of cardiomyocytes had formed in all cultures two days after seeding the cells into the MEA (Fig. 1A).Synchronized beating was present in all cultures at baseline (Fig. 2A).Application of an anatomical central line of block created two asynchronously beating cardiomyocyte fields in all cultures (Fig. 1B). Average channel width was similar in the four groups (Table 1).Conduction block was evident by lack of correlation between LATs recorded on electrodes at either side of the channel, indicating asynchronous beating of the two cardiomyocyte fields (Fig. 2B). The cross-channel application of 50,000 eGFP-labeled hMSCs resulted in restoration of conduction and re-establishment of synchronization between the two cardiomyocyte fields in five of eight cultures after 24 h (Fig. 1C). Resynchronization was reflected by the re-establishment of a 1:1 correlation between LATs at either side of the channel, as shown by a consistent time interval between LATs on either side of the channel (Fig. 2C). In resynchronized cultures, channel width was 300 ± 71 μm, whereas the two cultures that were not resynchronized after 24 h had a channel width of 450 μm. The remaining eighth cardiomyocyte culture was electrically inactive at 24 h, precluding any measurement. After 48 h, synchronization was present in five of eight cultures, including one of the two cultures with a channel width of 450 μm. In the remaining three hMSCs cultures, the cell monolayers had already detached from the MEA culture dish, precluding further analysis. In none of the experiments with added myoblasts or fibroblasts, nor in control cultures, was synchronization restored after 24 or 48 h.
Conduction velocity across hMSC-seeded channels
Activation maps of cardiomyocyte cultures were generated before induction of a conduction block (Fig. 3A),in the presence of a conduction block (Fig. 3B), and 24 h after hMSC seeding (Fig, 3C). Figure 3C indicated the presence of slow conduction through hMSCs in the channel. This is confirmed by the rightward shift of the trend line in Figure 2C compared with Figure 2A: local activation of the electrode in the upper part is 80 ms delayed compared with local activation of the electrode in the lower part.
After 24 h, conduction velocity across hMSCs (0.9 ± 0.4 cm/s) was 11-fold slower than across cardiomyocytes (10.4 ± 5.8cm/s; p = 0.005). After 48 h, hMSC conduction velocity was 0.8 ± 0.4 cm/s (vs. 0.9 ± 0.4 cm/s at 24 h; not significant). Cardiomyocyte conduction velocities were similar both after 24 and 48 h in the four different groups (Table 1).
Before channel abrasion, the whole culture showed synchronized activity (Figs. 3A and 4A).Immediately after channel abrasion, no electrical activity was recorded by the electrodes located at the side of the channel (Figs. 3B and 4B), indicating the presence of anatomical conduction block. Furthermore, no electrotonic interaction between the two fields of cardiomyocytes was recorded (Fig. 4E). In the hMSC-seeded channel, −dV/dtmaxwas 5 ± 5 μV/ms, in contrast to a −dV/dtmaxof 373 ± 243 μV/ms in the cardiomyocyte fields (p < 0.001). The amplitude of electrical activity of hMSCs (64 ± 19 μV) was significantly lower compared with that of cardiomyocytes (478 ± 156 μV; p < 0.001), whereas duration of electrical activity did not differ (cardiomyocytes, 35 ± 44 ms vs. hMSCs, 34 ± 22 ms; p = NS). Typical examples of extracellular recordings of cardiomyocytes and hMSCs are shown in Figures 4D and 4F. The configuration of the electrograms of cardiomyocytes was similar in all electrodes, whereas slopes and amplitudes of the hMSC electrograms within the channel were decremental along the path of conduction.
Intracellular recordings were performed in four cultures (Fig. 5).When comparing resting membrane potential, amplitude, and duration of intracellular hMSC and cardiomyocyte recordings, significant differences were found (Table 2).
Immunostaining revealed the presence of Cx43 along regions of intimate cardiomyocyte-to-hMSC and hMSC-to-hMSC contact (Figs. 6Aand 6B). Furthermore, Cx40, Cx43, and Cx45 were observed at the interface of juxtaposed cardiomyocytes (data for Cx43) (Fig. 6A). In addition to the presence of Cx43 at the margins of the hMSC, Cx43 was present in the cytoplasm of the hMSC (Figs. 6A and 6B). The hMSCs showed a cytoplasmatic distribution pattern of Cx40 and Cx45, without clustering at contact areas between either hMSCs or cardiomyocytes (data for Cx40 and Cx45 are not shown in Fig. 6).
Myoblasts showed a low intensity of punctate cytoplasmatic distribution of Cx43 and Cx45, but no Cx40. These connexins were undetectable at myoblasts-to-myoblasts or myoblasts-to-cardiomyocyte contact areas. No Cx40-, Cx43-, or Cx45-positive cardiac fibroblasts were detected in co-cultures with cardiomyocytes.
Calcein transfer from cardiomyocytes to hMSC
Calcein transfer experiments were performed in 10 cultures. Twenty-four hours after start of co-culture of DsRed-labeled hMSCs and calcein-loaded cardiomyocytes, transfer of calcein to adjacent hMSCs was observed (Fig. 7).Solitary hMSCs did not take up the dye, indicating that functional gap junctions are necessary for calcein transfer from cardiomyocytes to hMSCs.
The key finding of this study is that two previously asynchronously beating fields of cardiomyocytes can be resynchronized by the administration of hMSCs in the dividing channel. In other words, by conducting the electrical impulse and subsequent activation of a distal mass of cardiomyocytes, hMSCs could repair an experimentally induced conduction block. Impulse transmission across hMSCs within 48 h of co-incubation is characterized by slow conduction, reduced depolarization rates, and low-amplitude electrical activity decaying with distance.
Conduction across hMSCs
Successful impulse conduction across an hMSC-seeded channels and resynchronization of previously asynchronously beating fields of cardiomyocytes occurred within 24 h after administration of hMSCs. Gap junctional proteins were found both at the interfaces between hMSCs and between cardiomyocytes and hMSCs. Calcein dye transfer confirmed the presence of functional gap junctions between cardiomyocytes and hMSCs. The slow deflections of extracellular hMSC electrograms occurring synchronously with the rapid deflections in cardiomyocyte electrograms support the presence of intact electrotonic interaction. Intracellular recordings indicated that conduction across hMSCs occurred mainly by passive electrotonic current flow as hMSC action potentials showed a reduced depolarization rate and a reduced amplitude decaying with distance. Outgrowth of cardiomyocytes over seeded hMSCs is unlikely to be responsible for the results because cardiomyocytes are found not to migrate within 48 h and resynchronization did not occur in the control group.
Although impulse conduction across hMSCs is relatively slow, it is in line with the conduction velocity across non-excitable Cx43-positive cells reported by Gaudesius et al. (18). Furthermore, cardiomyocyte conduction velocity in our study is in accordance with previously reported data (19). Although slow conduction could be considered as proarrhythmogenic, until now, clinical studies with hMSCs showed no increased inducibility of arrhythmias (2–5,7).
Conduction through fibroblasts or myoblasts
Recently, Gaudesius et al. (18) showed that rat cardiac fibroblasts can couple electrically to cardiomyocytes and activate cardiac tissue over distances. They reported a gradual decline of conduction through fibroblasts with increasing channel width. Electrical propagation failed at a width >302 μm, whereas in the present study average channel width was 336 ± 76 μm. Fibroblasts used by Gaudesius et al. (18) stained positive for both Cx43 and Cx45. However, data on electrical coupling between fibroblasts and between cardiomyocytes and fibroblasts are conflicting. Rook et al. (20) and Laird and Revel (21) also observed electrotonic interaction between cardiomyocytes and fibroblasts. However, no Cx43-positive gap junctions between fibroblasts and only a few punctate Cx43-positive spots at cardiomyocyte-fibroblast interfaces were found (20,21). Other in vitro studies showed that rabbit cardiac fibroblasts stained positive for Cx40 and Cx45, but not for Cx43 (22). Feld et al. (23) reported that cardiac fibroblasts did not cause any significant electrophysiological changes when added to cardiomyocyte cultures. Furthermore, de Maziere et al. (24) showed absence of robust gap junction coupling between cardiomyocytes and cardiac fibroblasts in intact and healthy hearts. Oyamada et al. (25), performing a dye transfer assay in cultures of neonatal rat cardiomyocytes, showed that despite the presence of a few Cx43-positive spots between cardiomyocytes and fibroblasts, no dye transfer to fibroblasts occurred. The results of our study are in line with these observations. No electrical coupling and consequently no resynchronization was observed in any of the fibroblast experiments, which is possibly caused by the absence of Cx43 in these cells. Similarly cardiomyocytes devoid of Cx43 caused conduction block as reported by Fast et al., whereas hepatocytes transfected with a Cx43 expression vector were able to propagate the electrical impulse in contrast to wild-type hepatocytes without Cx43 (18,19).
Concerning the electrical properties of skeletal myoblasts, several studies showed that these cells lack Cx43 and cannot couple electrically to neighboring cardiomyocytes (26). Only in myoblasts in very early stages of differentiation was the presence of Cx43 reported. During differentiation, Cx43 was downregulated (27). Similarly, in our study skeletal myoblasts were unable to resynchronize two cardiomyocyte fields, which can be explained by the absence of Cx43 at the contact areas.
To perform long-term recordings, the attachment of cardiomyocytes to the surface of the MEA should be improved. Although several kinds of coatings of the MEA have been used, the majority of the monolayers detached from the MEA 48 h after channel abrasion. In addition, electrophysiological properties of hMSCs should ideally be tested in co-culture with adult human cardiomyocytes. Unfortunately, cardiomyocytes from healthy people are hard to obtain and dedifferentiate rapidly during culture (28). Therefore, we performed our experiments with neonatal rat cardiomyocytes, which are easily to obtain and show spontaneous electrical activity.
This study showed that two previously asynchronously beating fields of cardiomyocytes can be resynchronized by the administration of hMSCs. Impulse transmission across hMSCs is characterized by slow conduction, reduced depolarization rates, and low-amplitude electrical activity decaying with distance.
- Abbreviations and Acronyms
- Dulbecco’s modified eagle medium
- fetal bovine serum
- human mesenchymal stem cell
- local activation time
- micro-electrode array
- phosphate-buffered saline
- Received May 20, 2005.
- Revision received July 8, 2005.
- Accepted July 11, 2005.
- American College of Cardiology Foundation
- Fuchs S.,
- Satler L.F.,
- Kornowski R.,
- et al.
- Perin E.C.,
- Dohmann H.F.R.,
- Borojevic R.,
- et al.
- Menasché P.,
- Hagége A.A.,
- Vilquin J.-T.,
- et al.
- Britten M.B.,
- Abolmaali N.D.,
- Assmus B.,
- et al.
- Léobon B.,
- Garcin I.,
- Menasché P.,
- Vilquin J.-T.,
- Audinat E.,
- Charpak S.
- Pittenger M.F.,
- Marin B.J.
- Goldberg G.S.,
- Bechberger J.F.,
- Naus C.C.
- Gaudesius G.,
- Miragoli M.,
- Thomas S.P.,
- Rohr S.
- Fast V.G.,
- Darrow B.J.,
- Saffitz J.E.,
- Kleber A.G.
- Rook M.B.,
- van Ginneken A.C.,
- de Jonge B.,
- El Aoumari A.,
- Gros D.,
- Jongsma H.J.
- Laird D.W.,
- Revel J.P.
- Camelliti P.,
- Green C.R.,
- Le Grice I.,
- Kohl P.
- Feld Y.,
- Melamed-Frank M.,
- Kehat I.,
- Tal D.,
- Marom S.,
- Gepstein L.
- Oyamada M.,
- Kimura H.,
- Oyamada Y.,
- Miyamoto A.,
- Ohshika H.,
- Mori M.
- Reinecke H.,
- MacDonald G.H.,
- Hauschka S.D.,
- Murry C.E.
- Jacobson S.L.,
- Banfalvi M.,
- Schwarzfeld T.A.