Nature 450, 819-824 (6 December 2007) | doi:10.1038/nature06321; Received 15 June 2007; Accepted 28 September 2007

Engraftment of connexin 43-expressing cells prevents post-infarct arrhythmia

Wilhelm Roell1,2,9, Thorsten Lewalter3,9, Philipp Sasse1,9, Yvonne N. Tallini6, Bum-Rak Choi7, Martin Breitbach1, Robert Doran6, Ulrich M. Becher1,3, Seong-Min Hwang8, Toktam Bostani1,2, Julia von Maltzahn4, Andreas Hofmann5, Shaun Reining6, Britta Eiberger4, Bethann Gabris8, Alexander Pfeifer5, Armin Welz2, Klaus Willecke4, Guy Salama8, Jan W. Schrickel1,3, Michael I. Kotlikoff6 & Bernd K. Fleischmann1

  1. Institute of Physiology I, Life and Brain Center,
  2. Department of Cardiac Surgery,
  3. Department of Internal Medicine II,
  4. Institute of Genetics,
  5. Institute of Pharmacology, University of Bonn, Bonn 53105, Germany
  6. Department of Biomedical Sciences, College of Veterinary Medicine, Cornell University, Ithaca, New York 14853-6401, USA
  7. Cardiovascular Research Center, Rhode Island Hospital and Brown Medical School, Providence, Rhode Island 02903, USA
  8. Department of Cell Biology and Physiology, University of Pittsburgh, School of Medicine, Pittsburgh, Pennsylvania 15261, USA
  9. These authors contributed equally to this work.

Correspondence to: Michael I. Kotlikoff6Bernd K. Fleischmann1 Correspondence and requests for materials should be addressed to B.K.F. (Email: or M.I.K. (Email:


Ventricular tachyarrhythmias are the main cause of sudden death in patients after myocardial infarction. Here we show that transplantation of embryonic cardiomyocytes (eCMs) in myocardial infarcts protects against the induction of ventricular tachycardia (VT) in mice. Engraftment of eCMs, but not skeletal myoblasts (SMs), bone marrow cells or cardiac myofibroblasts, markedly decreased the incidence of VT induced by in vivo pacing. eCM engraftment results in improved electrical coupling between the surrounding myocardium and the infarct region, and Ca2+ signals from engrafted eCMs expressing a genetically encoded Ca2+ indicator could be entrained during sinoatrial cardiac activation in vivo. eCM grafts also increased conduction velocity and decreased the incidence of conduction block within the infarct. VT protection is critically dependent on expression of the gap-junction protein connexin 43 (Cx43; also known as Gja1): SMs genetically engineered to express Cx43 conferred a similar protection to that of eCMs against induced VT. Thus, engraftment of Cx43-expressing myocytes has the potential to reduce life-threatening post-infarct arrhythmias through the augmentation of intercellular coupling, suggesting autologous strategies for cardiac cell-based therapy.

Cell transplantation has emerged as a potential treatment strategy for heart failure secondary to acute or chronic ischaemic heart disease (reviewed in ref. 1). Currently two autologous cell types, namely bone marrow (BM) cells and SMs, are used in clinical trials in patients after myocardial infarction (reviewed in ref. 2). Both of these cell types seem to provide only modest improvement of contractile heart function3, 4, 5 because neither BM cells6, 7 nor SMs8 adopt a cardiac cell fate or couple electrically with the host myocardium6, 9. Moreover, VT has been reported in several of the patients transplanted with SMs5, 10, raising concerns as to whether engraftment of skeletal-muscle-derived cells enhances the risk of VT, the most frequent cause of sudden death after myocardial infarction11, 12. This led to the recommendation that cardiac SM engraftment be limited to patients with an implantable cardioverter defibrillator13. Transplantation of eCMs in animal models results in effective engraftment with expression of gap-junction proteins and modest augmentation of heart performance14, 15. Here we show that eCMs confer marked protection against ventricular arrhythmias by enhancing intercellular electrical coupling within the engrafted infarct. We also show that genetic modification of SMs to express Cx43 not only eliminates the pro-arrhythmogenic effect of SM transplantation but also provides arrhythmia protection that is equivalent to that of engraftment with eCMs. Thus, augmentation of intercellular coupling by cell-based therapy may be an effective therapeutic strategy for the prevention of post-infarct VT.


Vulnerability testing in vivo

We assessed electrical vulnerability in vivo by burst (Fig. 1a and Supplementary Fig. 1b) and extrastimulus pacing (Supplementary Fig. 1c) protocols in mice with left ventricular infarcts (Fig. 1b, left panel) 11–14 days after injury and transplantation of SMs, BM cells, cardiac myofibroblasts or eCMs. In vivo pacing protocols induced VT in 38.9% (n = 18) of non-infarcted control mice (HIM:OF1 and CD1 background) (Figs 1a and 2h, and Supplementary Fig. 1b). In contrast, monomorphic or polymorphic, self-terminating VT (Fig. 1c and Supplementary Fig. 1c) could be evoked in 96.4% of infarcted and vehicle-injected (sham-injected) mice (n = 28; Fig. 2h), although only minor alterations to the electrocardiogram (ECG) were observed under baseline conditions (Fig. 1d, left panel, and Supplementary Fig. 1a). VT was proved by the presence of atrio-ventricular dissociation at the His bundle level (Fig. 1c, lower trace). Thus, localized myocardial infarction markedly enhances VT inducibility in mice, allowing a systematic evaluation of the consequences of the engraftment of different cell types on electrical vulnerability.

Figure 1: VT in control, sham-injected and SM-engrafted hearts.
Figure 1 : VT in control, sham-injected and SM-engrafted hearts. Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, or to obtain a text description, please contact

a, Surface ECG from a control mouse including magnified inset of the burst stimulation (asterisks) protocol. Arrows indicate ventricular response; VT is not evoked. b, Sirius red staining proves transmural fibrotic scar 2 weeks after sham injection (left) or engraftment of SMs into the scar (right; inset shows EGFP-positive SMs). c, Burst stimulation induces self-terminating VT with atrio-ventricular dissociation in sham-injected mouse. Top trace, surface ECG; middle trace, ventricular apex electrogram; bottom trace, His-level electrogram. A, atrium; V, ventricle; a′, atrial farfield; v′, ventricular farfield. d, ECG recording from sham-injected mouse and SM-engrafted mouse. e, Burst pacing induces sustained VT in SM-engrafted mouse; note constant atrio-ventricular dissociation. Upper trace, surface ECG; lower trace, His-level electrogram. f, Engrafted EGFP-positive (green) SMs; native myocardium is marked by yellow autofluorescence. g, Engrafted SM fibres are elongated and multinucleate with central nuclei (inset). h, Transplanted EGFP-positive SMs (green in inset); double cross-striation shown with nebulin staining (red). Scale bar, 1.3mm (b), 150μm (f), 13μm (g), 16μm (g, inset), 7μm (h), 18μm (h, inset).

High resolution image and legend (199K)

Figure 2: VT protection in eCM-engrafted hearts.
Figure 2 : VT protection in eCM-engrafted hearts. Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, or to obtain a text description, please contact

a, Sirius red staining. b, EGFP-positive eCMs engraft in the infarct and extend to the border zone. c, ECG recording of eCM-engrafted mouse. d, Ventricular burst stimulation does not induce VT. Upper trace, surface ECG; lower trace, His-level electrogram. e, EGFP-positive eCMs integrate into the infarct. f, EGFP-positive eCMs are striated (α-actinin staining, red) and are in direct contact with EGFP-negative host cardiomyocytes. g, Cx43 staining (red) illustrates gap-junction formation between engrafted EGFP-positive eCMs (arrows) and with native cardiomyocytes (arrowheads). h, Summary of VT inducibility. Note strongly elevated susceptibility to VT induction in sham-injected group (Sh) compared with control group (Con, P<0.0001); eCM engraftment reduces VT inducibility compared with sham injection (P<0.0001). Conversely, VT remain frequent after transplantation of SMs, BM and cardiac myofibroblasts (cFib) (P = 0.0011 and 0.0005; 0.015 and 0.008; 0.019 and 0.016 versus control and eCMs, respectively). i, Improvement of left ventricular ejection fraction after eCM or SM transplantation. Asterisk, P<0.005. Numbers above bars indicate n; error bars show s.d. Scale bar, 1.6mm (a, b), 60μm (e), 5μm (f), 11μm (g).

High resolution image and legend (117K)


Engraftment of SMs enhances the severity of VT

We next tested the effect of SM transplantation on electrical stability of infarcted hearts by injecting embryonic enhanced green fluorescent protein (EGFP)-positive (ref. 16) skeletal-muscle-derived cells (Supplementary Fig. 1d). These cells formed a confined layer within the infarct two weeks after injury (Fig. 1b, right panel, and Fig. 1f, g). The engrafted SMs were elongated and multinucleate with central nuclei (Fig. 1g, inset) and expressed the skeletal muscle marker nebulin (Fig. 1h), indicating ongoing skeletal differentiation and, as reported earlier17, no evidence of cardiac ‘transdifferentiation’. SM-engrafted hearts did not express the gap-junction protein Cx43 in the graft region. Similarly to untreated infarcted animals, ST- and T-wave alterations were minor, but a split QRS complex without significant QRS prolongation was commonly observed (Fig. 1d, right panel, and Supplementary Table 1), suggesting inhomogeneity of ventricular activation. In all except one mouse containing EGFP-positive SMs (n = 16), VT could be evoked by pacing protocols (Figs 1e and 2h), and in 25% of mice these were found to degenerate into polymorphic VT and ventricular fibrillation (sustained arrhythmias; Fig. 1e). Although the induction of VT was almost 100% in both sham-injected and SM-engrafted mice the incidence of sustained arrhythmias was significantly increased in SM-transplanted mice (P = 0.0134). The equivalent incidence of arrhythmias in non-engrafted or SM-engrafted mice indicates that transplantation of non-electrically coupled cells does not decrease a major risk factor associated with myocardial infarction.


eCM engraftment decreases vulnerability to VT

Transplanted eCMs engrafted in a more diffuse pattern (Fig. 2b, e) than SMs, resulting in an apparently less thickened myocardial wall (compare right panel of Fig. 1b with Fig. 2a, b). The engrafted EGFP-positive SMs and eCMs could be clearly distinguished from the native myocardium, as confirmed by anti-EGFP staining (Supplementary Fig. 1e–h). Quantitative morphometry indicated engraftment in the range 3,000–20,000 eCMs (n = 11, median 4,480), similar to results reported earlier15. Moreover, immunostaining revealed Cx43 expression between the engrafted, differentiated eCMs (Fig. 2g). Occasionally, contact between graft and host myocardium and Cx43 staining between transplanted eCMs and native cardiomyocytes in the border zone of the infarct was observed (Fig. 2f, g). Despite the relatively modest degree of engraftment, transplantation of eCMs, which increases the survival rate of infarcted mice15, markedly improved electrical stability. In contrast to the sham-injected group, eCM engraftment caused a marked decrease in the incidence of VT that was similar to values of the non-infarcted control group (35.7%, n = 22, and 38.9%, n = 18, for eCMs and control, respectively; Fig. 2d, h). Quantitative morphometry did not reveal an association between contact of the grafted eCMs with the host myocardium in the border zone of the infarct and VT protection, because, in a series of combined electrophysiological investigation and morphometry, in two-thirds of protected mice (n = 6) no contact was found; VT protection was also observed in hearts with a relatively low degree of engraftment (n = 3). Surface ECGs were similar to those of controls; in contrast to the SM group, no deformation of the QRS complex was seen (Fig. 2c and Supplementary Table 1). The observed difference in VT incidence is unlikely to have been due to engrafted non-muscle cells, because both SM (Supplementary Fig. 1d) and eCM15 preparations contain similar percentages of fibroblasts. Despite the markedly different consequences of SM and eCM transplantation with respect to arrhythmogenicity, both procedures augmented left ventricular function to a similar extent (Fig. 2i), excluding heart failure as the mechanism underlying differences in electrical stability. We also tested the effect of transplanting BM cells, a second cell type currently used in clinical trials, on electrical stability. Engraftment of BM cells (n = 17) did not decrease the incidence or severity of induced VT (Fig. 2h). Similar results were also obtained after transplantation of cardiac myofibroblasts (n = 9; Fig. 2h and Supplementary Fig. 3a–c).


eCM grafts couple to native myocardium in vivo

We proposed that augmented electrical stability after eCM transplantation could result from the electrical integration of transplanted cardiomyocytes, because eCMs have previously been shown to couple to native myocytes after transplantation into the non-injured myocardium ex vivo18. We therefore examined whether eCMs integrate functionally within infarcted tissues in vivo by transplanting cells isolated from mouse hearts expressing the genetically encoded fluorescent Ca2+ indicator GCaMP2 (ref. 19) (Fig. 3f) or transduced with lentivirus expressing GCaMP2 (Fig. 3a–e and Supplementary Fig. 1i, j). Both approaches allowed us to track the function of implanted cells in vivo, including the degree of activation by the surrounding myocardium during normal sinoatrial rhythm, as well as the amplitude and time course of Ca2+ transients in transplanted cells. In vivo optical recordings of Ca2+-dependent fluorescence in open-chested mice 9–16 days after transplantation of eCMs indicated that Ca2+ signals from engrafted cells were entrained to the heart rhythm in infarct areas. Ca2+ signals were often coupled below the sinus rate (1:2 or 1:4) and reflected a conduction delay as indicated by simultaneous ECG and Ca2+-dependent fluorescence recordings (Fig. 3c and Supplementary Video 1). In three such experiments, the delay between the QRS minimum and activation of the GCaMP2+ cells was 70.1±9.5ms (mean±s.e.m.); this value is an overestimate because of the delay between electrical excitation and Ca2+ and the on-rate of GCaMP2 (ref. 19). Engrafted eCMs could be stimulated by pacing from an electrode placed outside the infarct area, resulting in synchronized Ca2+ transients at the stimulus rate within the infarct zone (Fig. 3f). In addition, transplanted cells showed uncoupled activity that conducted slowly throughout the engrafted myocytes, although ectopic beats or arrhythmias associated with this activity were never observed. In Langendorff-perfused hearts, in which heart motion could be constrained, high-resolution imaging of engrafted GCaMP2+ cells in the infarct revealed their sequential activation and hence electrical coupling (Fig. 3d, e). Thus, activation signals were transferred across the border zone to cells within the infarct region, over a distance of as much as 1mm, and transplantation of eCMs resulted in two types of functional coupling in vivo: heterologous coupling between transplanted cells and host cardiomyocytes, and autologous or regional coupling within islands of transplanted cells. Moreover, in Langendorff-perfused hearts loaded with Rhod-2, Ca2+ signals crossed the infarct border and were recorded within the infarct zone (left panels of Supplementary Fig. 2c, d, and lower panels of Supplementary Fig. 2e) in contrast to sham-injected hearts in which the Ca2+ wave largely bypassed the infarct, with minimal Ca2+ transients within the infarct zone (right panels of Supplementary Fig. 2c, d, and upper panels of Supplementary Fig. 2e). We reasoned that entrained Ca2+ signals in transplanted eCMs and a decreased incidence of conduction block of Rhod-2 Ca2+ transients reflect enhanced intercellular electrical coupling that probably underlies the marked decrease in arrhythmia vulnerability after eCM transplantation.

Figure 3: Entrainment of engrafted eCMs in vivo.
Figure 3 : Entrainment of engrafted eCMs in vivo. Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, or to obtain a text description, please contact

a, Image of a heart (taken ex vivo) with engrafted GCaMP2+ cells in the infarct (dark area). b, Magnified view of the yellow box in a. c, Combined in vivo recordings (left) of ECG (black) and Ca2+ transients (red, area marked in b) reveal 2:1 coupling of native myocardium with engrafted cells. Averaging of coupled QRS (green lines) and Ca2+ signals yields their temporal correlation (middle). Pseudocolour images of GCaMP2 fluorescence at time points indicated by numbers (right). d, e, High-resolution Ca2+ fluorescence from engrafted GCaMP2+ cells (d, left) in a Langendorff-perfused heart. Image series (d, middle) demonstrates coupling of engrafted cells, with sequential activation (e) of individual regions (d, right). f, Entrainment of Ca2+ fluorescence in vivo to 2-Hz stimulation from electrode in normal myocardium (intrinsic heart rate slowed with the use of deep anaesthesia). Scale bar, 1.5mm (a), 360μm (b, d), 3.2mm (c).

High resolution image and legend (289K)


Enhanced conduction in eCM-engrafted infarcts

Although the mechanisms underlying post-infarction re-entry arrhythmias are not completely understood, wave breaks at the boundary of an anatomic block20, 21 and focal activity including early and delayed afterdepolarizations arising from the infarct border zone22 are the characteristic electrophysiological features of arrhythmogenesis after infarction. We performed optical mapping by using Langendorff perfusion23 in sham-injected and eCM-engrafted hearts two weeks after infarct or engraftment and observed both wave breaks and focal activity (Fig. 4b, upper panels). Under sinus rhythm and unipolar stimulation from outside the infarct zone, action potentials propagated transmurally, emerged on the epicardium, and travelled around the infarct zone in sham-injected hearts (n = 5; Fig. 4a, upper panels, and Supplementary Fig. 2a, upper panels); pacing revealed prominent conduction blocks at the boundaries of infarcted regions (Fig. 4b, upper left panel). In contrast, action potentials propagated into the infarct zone in eCM-engrafted hearts (Fig. 4b, lower panel) and were recorded within areas of the infarct (n = 5, Fig. 4a, lower panels, and Supplementary Fig. 2a, upper panels). Additionally, non-engrafted hearts showed a higher incidence of focal activity emanating from the infarct border zone (Fig. 4b, upper right panel); ectopic beats were observed in 38% of sham-injected (n = 8) and 14% of eCM-engrafted (n = 7) hearts. In addition, premature ventricular stimuli designed to evoke re-entry phenomena induced prominent spiral electrical waves in sham-injected hearts (n = 3; Fig. 4d and Supplementary Video 2), whereas the same protocols did not induce re-entry waves in eCM-engrafted hearts (n = 5, Supplementary Video 3), indicating that the conditions required to elicit stable re-entrant circuit movement24 were eliminated in these hearts. Thus, eCM engraftment decreases the incidence of conduction block and wave breaks, and suppresses border-zone focal activity, the two hallmarks of post-infarct ventricular arrhythmia.

Figure 4: eCM engraftment improves conduction and prevents re-entry in Langendorff-perfused hearts.
Figure 4 : eCM engraftment improves conduction and prevents re-entry in Langendorff-perfused hearts. Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, or to obtain a text description, please contact

a, Sequential images from a series showing the first derivative (2-ms interval) of action-potential propagation during pacing. The infarct region is outlined by a black circle. b, Representative isochronal maps depicting the activation wavefront and conduction delays near the infarct border zone. The activation bypasses the sham-injected infarct (upper left), or initiates from an ectopic focus at the border zone (right), whereas it proceeds through the eCM-injected lesion (lower left). Scale bar indicates local activation times. c, Local conduction velocities. There is a more than fourfold increase within eCM-engrafted infarcts (P<0.001; numbers above bars indicate n; error bars show s.e.m.). d, Sequential di-4-ANNEPS fluorescence images show induction of spiral waves in a sham-injected heart (images cropped around infarct zone). The infarct region is ringed in the leftmost panel.

High resolution image and legend (440K)

To determine whether these effects result from augmented intercellular electrical coupling, we measured local conduction velocities from the activation time points of optical action potentials25 and found that conduction velocity within the infarct was increased more than fourfold in eCM-engrafted hearts (Fig. 4c), which is consistent with a marked increase in intercellular coupling. Further evidence of the critical nature of gap-junctional conductance within the infarct zone was obtained by perfusion of hearts with the connexin blocker carbenoxolone (0.5mM). After perfusion of the drug, the conduction velocity decreased more rapidly and to a greater extent in the infarct zone than in the intact myocardium of sham-injected hearts (29±9% versus 53±4%, n = 4), indicating a contribution by gap junctions to conduction, as well as a lower conduction reserve, within the infarct zone. Anti-Cx43 immunostaining and western blotting revealed significantly lower Cx43 expression in cardiac myofibroblasts than in eCMs and adult heart (Supplementary Fig. 3d–f). Although cardiac myofibroblasts do not themselves provide sufficient VT protection (Fig. 2h), electrical coupling of engrafted eCMs that do not extend to the border zone probably reflects the contribution of cardiac myofibroblasts to the electrotonic transfer of action potentials across the border zone26 and improved electrical coupling within the infarct zone. It is possible, however, that arrhythmia protection results from tonic effects on border-zone myocytes (for example, changes in resting membrane potential) associated with the engraftment of electrically coupled eCMs. To examine this possibility, we determined conduction velocities and action-potential rise times, parameters that would probably be altered in the event of dominant effects of border-zone cardiomyocytes, in the border zone of infarcts from optical mapping data. Conduction velocities were very similar in sham-injected (0.178±0.011mmms-1, n = 3) and in eCM-engrafted (0.164±0.037mmms-1, n = 4) hearts. In addition, action-potential rise times were almost identical in border-zone myocytes of sham-injected (5.6±1.5ms, n = 3) and eCM-engrafted (5.7±1.7ms, n = 4) hearts. However, these measurements do not exclude alterations in subregions of the border zone that could give rise to the decreased focal activity that we observed. Taken together, these data indicate that eCM engraftment markedly enhances intercellular electrical conduction within the infarct zone, resulting in the elimination or reduction of electrical events that initiate re-entry arrhythmias (focal activity and conduction blocks) and provide a mechanistic basis for the markedly decreased electrical vulnerability produced by eCM engraftment.


SM-Cx43 engraftment confers arrhythmia protection

We directly assessed the role of enhanced electrical coupling through Cx43 in arrhythmia protection, by engrafting SMs from transgenic mice expressing Cx43 (SM-Cx43+) under the control of a skeletal muscle promoter (see Supplementary Methods and Supplementary Fig. 4a–e). This was also intended to exclude the possibility that engraftment of eCMs indirectly enhances conduction by paracrine mechanisms27 and to explore the feasibility of generating an accessible and autologous cell source for potential clinical applications. Functional gap junctions are able to form between SM-Cx43+ myotubes, as shown by dye-transfer studies in vitro (Supplementary Fig. 4f). SM-Cx43+ stably engrafted into the infarcted myocardium (Fig. 5a, b) differentiated, forming multinucleate cells with a distinct cross-striation (Fig. 5f), and expressed Cx43 (Fig. 5f, g, and Supplementary Fig. 4g). Surface ECGs recorded from these mice showed only minor alterations compared with ECGs of control and sham-injected animals (Fig. 5c and Supplementary Table 1), and QRS splitting was observed after transplantation of wild-type SMs (Fig. 1d) but not SM-Cx43+ (Fig. 5c). Stable engraftment of SM-Cx43+ markedly improved electrical stability (Fig. 5d, h), because VT was induced in only 37.5% (n = 16) of SM-Cx43+-engrafted mice, an improvement in electrical stability almost identical to that achieved by the transplantation of eCMs. In contrast, 100% of mice transplanted with SMs not expressing Cx43 from control littermates (SM-Cx43-) developed VT (n = 13; Fig. 5e, h). These experiments clearly demonstrate that Cx43 expression is necessary for protection from arrhythmia and exclude paracrine mechanisms.

Figure 5: VT protection in Cx43-expressing SM-engrafted hearts.
Figure 5 : VT protection in Cx43-expressing SM-engrafted hearts. Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, or to obtain a text description, please contact

ac, Sirius red staining (a), native EGFP fluorescence (green, b) and ECG recording (c) in SM-Cx43+-engrafted hearts. d, e, Ventricular burst stimulation induces no VT in SM-Cx43+-engrafted mouse (d), but transient VT in SM-Cx43--engrafted mouse (e). Upper traces, surface ECG; lower traces, His-level electrogram. f, g, EGFP-positive (green) SM-Cx43+ engraft stably into the infarcted heart. Myotubes are multinucleate and striated (f, α-actinin staining in white). Cx43 immunostaining (red) shows punctate expression at the cell surface in longitudinal section (f) and cross-section (g). h, VT inducibility of SM-Cx43+-engrafted and SM-Cx43--engrafted hearts (P<0.0004). Numbers above bars indicate n. Scale bar, 1.2mm (a, b), 15μm (f, g).

High resolution image and legend (350K)



Taken together, our results show that cardiomyocyte transplantation has the potential to impart electrical stability to the injured heart, thereby markedly reducing the major factor leading to sudden death. This protective effect is independent of the documented modest augmentation of left ventricular function and is associated with improved electrical coupling within the infarct by the engraftment of Cx43-expressing eCMs. This enhanced coupling reduces vulnerability to VT by decreasing the incidence of conduction block within the infarct and/or by a modulatory effect on border-zone cardiomyocytes. Although conduction velocity and action-potential rise time are not altered in the border zone, subtle effects cannot be excluded and are supported by the observation of decreased ectopic activity after eCM engraftment. Expression of the cardiac gap-junction protein Cx43 is the critical factor underlying augmented intercellular electrical conduction and protection from arrhythmia, because engraftment of Cx43-expressing SMs protects against VT induction. Because autologous SM muscle can be isolated from humans and protection is achieved with relatively small numbers of engrafting cells, transplantation of SMs in combination with Cx43 gene transfer represents a promising therapeutic strategy to decrease the risk of potentially fatal arrhythmias.


Methods Summary

Harvesting of cells and transplantation

Single cells were dissociated from embryonic hearts (embryonic development day (E)14.5–E16.5) and skeletal muscles (E18.5/E19.5) of human cardiac α-actin–EGFP transgenic mice16 (CD1 and HIM:OF1 background). Cx43+ SMs and their controls were obtained from MCK-tTA/Cx43TetOEGFP embryos and their negative littermates (C57/Bl6 background). Cardiac myofibroblasts were obtained by pre-plating of embryonic heart cells (E14.5/E15.5). BM cells were aspirated from hindlimbs of adult wild-type CD1 mice28. Dissociated cells (105 SMs, eCMs or cardiac myofibroblasts, or 3×106 BM cells) or vehicle (sham-injected) were injected into the cryolesioned anterior-lateral left ventricular wall of recipient mice29. Non-operated mice were used as a control group.

In vivo electrophysiology

ECG parameters and inducibility of VT were evaluated by a blinded investigator 11–14 days postoperatively by transvenous electrophysiological investigation by using burst and premature-beat stimulation30. VT was defined as at least four ectopic ventricular beats characterized by atrio-ventricular dissociation.

Histology and immunohistochemistry

After electrophysiological investigation, hearts were imaged by fluorescence macroscopy, fixed in paraformaldehyde and cryosectioned. The extent of myocardial lesions was determined by staining with Sirius red. Engrafted cells were identified by their native EGFP fluorescence and characterized by immunohistochemistry.

In vivo imaging and ex vivo optical mapping

GCaMP2+ eCMs were derived from transgenic embryos19 or obtained by lentiviral transduction of plated eCMs (CMV-GCaMP2)31. In vivo experiments were performed in ventilated and thoracotomized mice. GCaMP2 signals from transplanted cells were recorded with a fluorescence macroscope ( OV100 or MVX10; Olympus) and an electron-multiplied, charge-coupled-device camera at frame rates of 67–128Hz.

For ex vivo recordings, hearts were Langendorff perfused, suspended in an optical chamber and labelled with the membrane-potential-sensitive dye di-4-ANEPPS or the fluorescent calcium indicator Rhod-2. Voltage or Ca2+ signals were recorded with a complementary-metal-oxide semiconductor (CMOS) camera at 1,000Hz. Conduction velocity within the infarct or normal regions was measured under point stimulation at a cycle length of 200ms for 20 beats.

Full methods accompany this paper.



  1. Laflamme, M. A. & Murry, C. E. Regenerating the heart. Nature Biotechnol. 23, 845|[ndash]|856 (2005) | Article |
  2. Murry, C. E., Field, L. J. & Menasche, P. Cell-based cardiac repair: reflections at the 10-year point. Circulation 112, 3174|[ndash]|3183 (2005) | Article | PubMed | ISI |
  3. Wollert, K. C. et al. Intracoronary autologous bone-marrow cell transfer after myocardial infarction: the BOOST randomised controlled clinical trial. Lancet 364, 141|[ndash]|148 (2004) | Article | PubMed | ISI |
  4. Janssens, S. et al. Autologous bone marrow-derived stem-cell transfer in patients with ST-segment elevation myocardial infarction: double-blind, randomised controlled trial. Lancet 367, 113|[ndash]|121 (2006) | Article | PubMed | ISI |
  5. Smits, P. C. et al. Catheter-based intramyocardial injection of autologous skeletal myoblasts as a primary treatment of ischemic heart failure: clinical experience with six-month follow-up. J. Am. Coll. Cardiol. 42, 2063|[ndash]|2069 (2003) | Article | PubMed | ISI |
  6. Murry, C. E. et al. Haematopoietic stem cells do not transdifferentiate into cardiac myocytes in myocardial infarcts. Nature 428, 664|[ndash]|668 (2004) | Article | PubMed | ISI | ChemPort |
  7. Nygren, J. M. et al. Bone marrow-derived hematopoietic cells generate cardiomyocytes at a low frequency through cell fusion, but not transdifferentiation. Nature Med. 10, 494|[ndash]|501 (2004) | Article |
  8. Reinecke, H., Poppa, V. & Murry, C. E. Skeletal muscle stem cells do not transdifferentiate into cardiomyocytes after cardiac grafting. J. Mol. Cell. Cardiol. 34, 241|[ndash]|249 (2002) | Article | PubMed | ISI | ChemPort |
  9. Leobon, B. et al. Myoblasts transplanted into rat infarcted myocardium are functionally isolated from their host. Proc. Natl Acad. Sci. USA 100, 7808|[ndash]|7811 (2003) | Article | PubMed | ChemPort |
  10. Hagege, A. A. et al. Skeletal myoblast transplantation in ischemic heart failure: long-term follow-up of the first phase I cohort of patients. Circulation 114, I-108|[ndash]|I-113 (2006) | Article |
  11. Solomon, S. D. et al. Sudden death in patients with myocardial infarction and left ventricular dysfunction, heart failure, or both. N. Engl. J. Med. 352, 2581|[ndash]|2588 (2005) | Article | PubMed |
  12. Henkel, D. M. et al. Ventricular arrhythmias after acute myocardial infarction: a 20-year community study. Am. Heart J. 151, 806|[ndash]|812 (2006) | Article | PubMed |
  13. Makkar, R. R., Lill, M. & Chen, P. S. Stem cell therapy for myocardial repair: is it arrhythmogenic? J. Am. Coll. Cardiol. 42, 2070|[ndash]|2072 (2003) | Article | PubMed | ISI |
  14. Soonpaa, M. H., Koh, G. Y., Klug, M. G. & Field, L. J. Formation of nascent intercalated disks between grafted fetal cardiomyocytes and host myocardium. Science 264, 98|[ndash]|101 (1994) | Article | PubMed | ISI | ChemPort |
  15. Roell, W. et al. Cellular cardiomyoplasty improves survival after myocardial injury. Circulation 105, 2435|[ndash]|2441 (2002) | Article | PubMed | ISI |
  16. Fleischmann, M. et al. Cardiac specific expression of the green fluorescent protein during early murine embryonic development. FEBS Lett. 440, 370|[ndash]|376 (1998) | Article | PubMed | ISI | ChemPort |
  17. Pouzet, B. et al. Is skeletal myoblast transplantation clinically relevant in the era of angiotensin-converting enzyme inhibitors? Circulation 104, I-223|[ndash]|I-228 (2001) | Article |
  18. Dowell, J. D., Rubart, M., Pasumarthi, K. B., Soonpaa, M. H. & Field, L. J. Myocyte and myogenic stem cell transplantation in the heart. Cardiovasc. Res. 58, 336|[ndash]|350 (2003) | Article | PubMed | ISI | ChemPort |
  19. Tallini, Y. N. et al. Imaging cellular signals in the heart in vivo: Cardiac expression of the high-signal Ca2+ indicator GCaMP2. Proc. Natl Acad. Sci. USA 103, 4753|[ndash]|4758 (2006) | Article | PubMed | ChemPort |
  20. Pertsov, A. M., Davidenko, J. M., Salomonsz, R., Baxter, W. T. & Jalife, J. Spiral waves of excitation underlie reentrant activity in isolated cardiac muscle. Circ. Res. 72, 631|[ndash]|650 (1993) | PubMed | ISI | ChemPort |
  21. Takahashi, T. et al. Optical mapping of the functional reentrant circuit of ventricular tachycardia in acute myocardial infarction. Heart Rhythm 1, 451|[ndash]|459 (2004) | Article | PubMed |
  22. Grant, A. O., Hondeghem, L. M. & Katzung, B. G. Effects of droperidol on depolarization-induced automaticity, maximum upstroke velocity (vmax) and the kinetics of recovery of vmax in guinea-pig ventricular myocardium. J. Pharmacol. Exp. Ther. 205, 193|[ndash]|203 (1978) | PubMed |
  23. Baker, L. C., London, B., Choi, B. R., Koren, G. & Salama, G. Enhanced dispersion of repolarization and refractoriness in transgenic mouse hearts promotes reentrant ventricular tachycardia. Circ. Res. 86, 396|[ndash]|407 (2000) | PubMed |
  24. Allessie, M. A. et al. Experimental electrophysiology and arrhythmogenicity. Anisotropy and ventricular tachycardia. Eur. Heart J. 10 (Suppl. E). E2|[ndash]|E8 (1989)
  25. Salama, G., Kanai, A. & Efimov, I. R. Subthreshold stimulation of Purkinje fibers interrupts ventricular tachycardia in intact hearts. Experimental study with voltage-sensitive dyes and imaging techniques. Circ. Res. 74, 604|[ndash]|619 (1994) | PubMed |
  26. Gaudesius, G., Miragoli, M., Thomas, S. P. & Rohr, S. Coupling of cardiac electrical activity over extended distances by fibroblasts of cardiac origin. Circ. Res. 93, 421|[ndash]|428 (2003) | Article | PubMed | ISI | ChemPort |
  27. Gnecchi, M. et al. Paracrine action accounts for marked protection of ischemic heart by Akt-modified mesenchymal stem cells. Nature Med. 11, 367|[ndash]|368 (2005) | Article |
  28. Kolossov, E. et al. Engraftment of engineered ES cell-derived cardiomyocytes but not BM cells restores contractile function to the infarcted myocardium. J. Exp. Med. 203, 2315|[ndash]|2327 (2006) | Article | PubMed | ISI | ChemPort |
  29. Roell, W. et al. Cellular cardiomyoplasty in a transgenic mouse model. Transplantation 73, 462|[ndash]|465 (2002) | Article | PubMed | ISI |
  30. Kreuzberg, M. M. et al. Connexin30.2 containing gap junction channels decelerate impulse propagation through the atrioventricular node. Proc. Natl Acad. Sci. USA 103, 5959|[ndash]|5964 (2006) | Article | PubMed |
  31. Pfeifer, A., Ikawa, M., Dayn, Y. & Verma, I. M. Transgenesis by lentiviral vectors: lack of gene silencing in mammalian embryonic stem cells and preimplantation embryos. Proc. Natl Acad. Sci. USA 99, 2140|[ndash]|2145 (2002) | Article | PubMed | ChemPort |
  32. Rubart, M. et al. Physiological coupling of donor and host cardiomyocytes after cellular transplantation. Circ. Res. 92, 1217|[ndash]|1224 (2003) | Article | PubMed | ISI | ChemPort |
  33. Miragoli, M., Gaudesius, G. & Rohr, S. Electrotonic modulation of cardiac impulse conduction by myofibroblasts. Circ. Res. 98, 801|[ndash]|810 (2006) | Article | PubMed |
  34. Berul, C. I., Aronovitz, M. J., Wang, P. J. & Mendelsohn, M. E. In vivo cardiac electrophysiology studies in the mouse. Circulation 94, 2641|[ndash]|2648 (1996) | PubMed |
  35. Mitchell, G. F., Jeron, A. & Koren, G. Measurement of heart rate and Q-T interval in the conscious mouse. Basic Res. Cardiol. 274, H747|[ndash]|H751 (1998)
  36. Schrickel, J. W. et al. Induction of atrial fibrillation in mice by rapid transesophageal atrial pacing. Basic Res. Cardiol. 97, 452|[ndash]|460 (2002) | Article | PubMed |
  37. Thevenaz, P., Ruttimann, U. E. & Unser, M. A pyramid approach to subpixel registration based on intensity. IEEE Trans. Image Process. 7, 27|[ndash]|41 (1998) | Article |
  38. Choi, B. R. & Salama, G. Simultaneous maps of optical action potentials and calcium transients in guinea-pig hearts: mechanisms underlying concordant alternans. J. Physiol. (Lond.) 529, 171|[ndash]|188 (2000) | Article | PubMed |
  39. Gottlieb, G. J., Kubo, S. H. & Alonso, D. R. Ultrastructural characterization of the border zone surrounding early experimental myocardial infarcts in dogs. Am. J. Pathol. 103, 292|[ndash]|303 (1981) | PubMed |
  40. Traub, O. et al. Immunochemical and electrophysiological characterization of murine connexin40 and -43 in mouse tissues and transfected human cells. Eur. J. Cell Biol. 64, 101|[ndash]|112 (1994) | PubMed |

Supplementary Information

Supplementary information accompanies this paper.



We thank D. Fuerst for providing the anti-nebulin antibody; L. Field for providing the transgenic α-MHC–EGFP mouse line; W. Bloch for advice on immunostaining; C. Schaffer and N. Nishimura for technical advice; R. Gilmour Jr and N. Otani for comments on the manuscript; C. Fuegemann for the preparation of cardiac myofibroblasts; and H. Begerau, M. Czechowski, B. Eixmann, F. Holst, H. Doerr, K. Granitza and C. Russell for technical help. This study was supported by grants from the Deutsche Forschungsgemeinschaft (to W.R. and B.K.F.), the Federal Ministry of Education and Research, Germany (to T.L.), the European Commission (to B.K.F.), BONFOR (to J.W.S., A.H.) and the National Institutes of Health (to Y.N.T., M.I.K. and G.S.).

Author Contributions W.R., T.L. and P.S. contributed equally to this work. W.R. performed microsurgery, left-ventricular catheterization and analysis of data. T.L. supervised and analysed in vivo electrophysiology. P.S. performed in vivo imaging experiments, analysis and immunohistochemistry. Y.N.T. was involved in the in vivo imaging experiments and immunohistochemistry. B.-R.C. was involved in optical imaging and analysis of data. M.B. was involved in morphometry, immunohistochemistry and establishment of cardiac fibroblasts. R.D. was involved in the analysis of in vivo imaging experiments. U.B. was involved in the preparation of skeletal myoblasts, and in morphometry and immunohistochemistry. S.-M.H. was involved in the optical-imaging experiments and analysis of data. T.B. was involved in microsurgery, morphometry, left ventricular catheterization and immunohistochemistry. J.V.M. was involved in mouse breeding, immunohistochemistry and western blotting of tissues from the Cx43-expressing mouse. A.H. was involved in the generation of the lentivirus constructs. S.R. was involved in mouse breeding and in vivo imaging experiments. B.D. was involved in generation of the Cx43-expressing mouse model. B.G. was involved in Langendorff perfusion and optical-imaging experiments. A.P. supervised lentiviral work. A.W. supervised the microsurgery. G.S. was involved in the optical-imaging experiments, their analysis and writing of the manuscript. J.W.S. performed electrophysiological experiments in vivo and analysed data. M.I.K. designed experiments, analysed in vivo imaging experiments and wrote the manuscript. B.K.F. initiated the study, designed experiments and wrote the manuscript.


Online Methods

All animal experiments were performed in accordance with National Institutes of Health animal protection guidelines and were approved by the local authorities.

Preparation and characterization of donor cells

Cardiac and skeletal muscle cells expressing EGFP under control of the human α-actin promoter were harvested from transgenic mice (HIM:OF1 and CD1 background)16 and transplanted into syngeneic wild-type mice. Skeletal muscle was harvested from hindlimb muscles of E18.5/E19.5 embryos and dissected; single cells were obtained by dissociation with collagenase and trypsin. Embryonic hearts (E14.5–E16.5) were enzymatically dissociated into single cells as reported previously29. To assess the percentage of eCMs after dissociation, flow cytometry was performed. Cardiomyocytes in transgenic α-MHC-EGFP+ (ref. 32) embryonic hearts (E15.5) were found to be 46.8±5.6% (n = 3) using flow cytometry.

SMs from the hindlimb and diaphragm of embryonic (E18.5/E19.5) and postnatal (day 1/day 2) Cx43tetOeGFP/MCKtTA (SM-Cx43+) transgenic mice were harvested and isolated as described above. For negative controls, skeletal muscle was harvested and processed in an identical fashion from monoallelic littermates (Cx43tetOeGFP/- and -/MCKtTA: SM-Cx43-). These cells were transplanted either directly after dissociation or after 24h of culture. Cardiac myofibroblasts (cFib, E14.5–E16.5) were isolated, cultured and characterized in a similar manner to that reported previously33. BM cells were harvested by BM aspiration by flushing the femur and tibia of 2–3-month-old wild-type CD1 mice with a 27-gauge needle (see also ref. 28).


Reproducible transmural cryolesions at the anterior-lateral left ventricular wall were generated in recipient mice (CD1, HIM:OF1, C57/Bl6, 10–12 weeks) with liquid-N2-cooled copper probes 3 or 4mm in diameter as reported previously28, 29 and cells suspended in 5–6μl were injected intramyocardially immediately thereafter. SM-Cx43+ and SM-Cx43- cells (C57/Bl6 background) were transplanted into CD1 mice. For non-syngeneic transplantations, the recipients were immunosuppressed by using daily intraperitoneal injections of cyclosporinA (20mgkg-1; Novartis).

Left ventricular catheterization

Left ventricular catheterization was performed 2 weeks after the operation by a blinded investigator with a 1.4 french Millar Aria1 catheter (Millar Instruments Inc.)28.

Surface ECG and in vivo transvenous electrophysiological investigation

Mice (HIM:OF1 or CD1 background) were put under inhalative anaesthesia and a surface 6-lead ECG (lead III is shown in figures) was obtained for at least 3min. R–R intervals, P-wave durations, PQ intervals, QRS durations and QT and JT intervals were measured by successive evaluation as described previously30, 34, 35. A schematic illustration of the measured surface ECG parameters is given in Supplementary Fig. 1a. JT was defined as the interval from the S-wave meeting the isoelectrical line to the end of the T-wave. QT and JT intervals were also rate-corrected as reported previously35. After amplification, all data were sampled at 4kHz ( Bard stamp amplifier; C.R. Bard Inc.). For transvenous electrophysiological investigation a 2 french octapolar mouse-electrophysiological catheter ( CIBer Mouse; NuMed Inc.) was inserted into the right jugular vein and positioned in the right atrium and ventricle. Bipolar electrograms were obtained from adjacent electrode pairs. Twice pacing threshold rectangular stimulus pulses were administered by a multi-programmable stimulator ( Model 5328; Medtronic). Evaluation of electrophysiological investigation parameters included sinus node recovery period (SNRP), Wenckebach periodicity (WBP) as well as atrial refractory periods (ARP), atrio-ventricular-nodal refractory periods (AVNRP) and ventricular refractory periods (VRP). Vulnerability to VT, characterized by atrio-ventricular dissociation, was tested by ventricular burst stimulations performed at a stimulation cycle length (S1S1-CL) starting at 50ms with 10-ms stepwise reduction down to 10ms (refs 30, 36). In addition, extrastimulus pacing was used with eight fixed-rate stimuli at S1S1-CL of 120, 110 and 80ms, followed by up to three short-coupled extra beats with successive reduction of the coupling interval until VRP was reached. Stimulations were performed at twice pacing threshold. VT was defined as at least four ventricular beats; the mean VT duration was 3.4±5s (n = 9). For all mice, complete vulnerability testing was performed; ECG analysis was not performed in some sham-injected and eCM-engrafted mice.

In vivo imaging

GCaMP2+ eCMs were obtained from transgenic embryos19 and transplanted into non-transgenic littermates or wild-type C57/Bl6 mice. Alternatively, eCMs were harvested from CD1 wild-type embryonic hearts (E14.5–E16.5) and transduced overnight in vitro with lentiviral vectors expressing GCaMP2 under the control of the cytomegalovirus (CMV) promoter (LVGCaMP2)31. In brief, LVGCaMP2 was cloned by replacing the EGFP in LV-GFP31 with the GCaMP2 complementary DNA. Lentiviral particles were prepared as described previously31. The cells were trypsinized and thoroughly washed after lentivirus transduction and injected (105 to 4×105 cells) into recipient CD1 mice. Imaging experiments were performed 7–16 days after operation in ventilated and thoracotomized mice in which the heart was exposed and partly immobilized. Engrafted cells were revealed with a fluorescence macroscope ( OV100 or MVX10; Olympus). GCaMP2 was excited at 470±20nm, emitted light was collected at 525±25nm with a electron-multiplied, charge-coupled device camera ( iXon 860–BI or iXion 885KS; Andor Technology). Data were acquired at frame rates of 67–128Hz with the Andor Solis software. ECG was recorded by an EXT-02F amplifier (npi electronic) connected to a PowerLab 26T AD converter (AD Instruments). Data were processed with ImageJ software (NIH) and custom-written software ( Labview 7.1; National Instruments). In some experiments, the concentration of isoflurane was increased to decrease the heart rate to less than 300 beats per minute. Some of the hearts were explanted and imaged while being perfused with a Langendorff system. Registration using the ‘turbo reg’ plugin of ImageJ37 was used to eliminate translating and rotating movement. Pseudo-coloured images were generated after individual normalization of each pixel.

Optical mapping

For optical-mapping experiments, hearts (CD1 or C57/Bl6 background) were excised, perfused in a constant-flow Langendorff chamber and stained with di-4-ANEPPS or Rhod-2 as described previously23. Fluorescence (di-4-ANEPPS: λex = 540±30nm, λem>640nm; Rhod-2: λex = 540± 30nm, λem = 585±20nm) was collected with a camera lens ( 50mm focal length, f 0.95; Navitar), focused on a CMOS camera ( 100 pixels×100 pixels; Ultima-L SciMedia). Excitation light was delivered by epi-illumination with a custom-designed 300-W tungsten–halogen lamp (University of Pittsburgh machine and electronic shops). The CMOS camera (1×1cm2) viewed the anterior surface of the heart, with each pixel recording signals from four to ten adjacent cardiomyocytes (70×70μm2 per pixel with a depth of field of about 130–140μm); the integration time was set to 1ms (1,000 framess-1) to reduce motion blurring caused by rapid wave propagation. To measure the conduction velocity from the infarct region and its healthy surroundings, point stimulation (Ag+ electrode 200μm in diameter; 2ms duration; 20% above threshold) was performed at a cycle length of 200ms for at least 20 beats. Activation time-points were determined by dF/dtmax after applying a Savitzky–Golay smoothing filter (11 point width, third order), and isochronal maps of activation were generated as described previously38. The local conduction velocity vector was calculated from the activation time-point at each pixel compared with that of its eight nearest neighbours19, 23, 25. The rise time of action potentials was measured from fluorescence signals that showed a signal-to-noise ratio of more than about 20:1, at a sampling rate of 5,000 framess-1 and a spatial resolution of 100×100μm2 per pixel. The rise time was measured from 10% to 90% of action-potential amplitude from at least five consecutive action potentials and low-pass filtered with an averaging kernel of 1ms. Border zones39 were identified from activation maps where slow propagation occurred and from autofluorescence images.

Histology and Immunohistochemistry

After electrophysiological investigation, hearts were harvested, imaged with a fluorescence stereomicroscope ( Leica MZ 16F; Leica Microsystems) and a ProgRes C10+ camera (Jenoptik), arrested in diastole by coronary perfusion with cardioplegic solution at stable haemodynamic pressure, and fixed with 4% paraformaldehyde by perfusion. The hearts were cryopreserved and cut into tissue slices 8 or 20μm thick (for morphometry and immunohistochemistry) and 20μm thick (for histology)15. The extent of myocardial lesions in the different groups of animals was determined by staining with Sirius red (Aldrich Chemical Company). Immunostaining with anti-EGFP (1:50 dilution from Santa Cruz Biotechnology, or 1:20 dilution from Chemicon) was used to corroborate EGFP or GCaMP2 expression. To evaluate cross-striation, primary antibodies against sarcomeric α-actinin were used. The differentiation of engrafted SMs was confirmed by using a mouse monoclonal antibody directed against nebulin (1:10 dilution). Cx43 immunostaining was performed with a commercially available antibody (1:400 dilution; Bio Trend) or an antibody provided by K.W.40. All primary antibodies were revealed by appropriate secondary Cy3-conjugated and Cy5-conjugated donkey antibodies (1:400–1:1,000 dilution; Dianova). Nuclei were stained blue with Hoechst dye. Samples were imaged with a Zeiss Axiovert 200 microscope equipped with an ApoTome and AxioCam MRm; images were acquired with the Zeiss software AxioVision.

Statistical analysis

Statistical analysis of the electrophysiological investigation data was performed with a multivariant one-way analysis of variance with post-hoc subgroup testing when appropriate (Tukey–Kramer multiple comparisons test). Non-parametric variables were evaluated by using Kruskal–Wallis testing with Dunn post-hoc testing when appropriate. Discrete variables were analysed by two-sided Fisher’s exact test. Statistical analysis of the other data was performed by Student´s t-test. P0.05 was regarded as statistically significant; unless stated otherwise, errors are given as s.d.

Readers' Comments

If you find something abusive or inappropriate or which does not otherwise comply with our Terms and Conditions or Community Guidelines, please select the relevant 'Report this comment' link.

There are currently no comments.

Add your own comment

This is a public forum. Please keep to our Community Guidelines. You can be controversial, but please don't get personal or offensive and do keep it brief. Remember our threads are for feedback and discussion - not for publishing papers, press releases or advertisements.

You need to be registered with Nature and agree to our Community Guidelines to leave a comment. Please log in or register as a new user. You will be re-directed back to this page.