OUP user menu

Intercalated disc abnormalities, reduced Na+ current density, and conduction slowing in desmoglein-2 mutant mice prior to cardiomyopathic changes

Stefania Rizzo, Elisabeth M. Lodder, Arie O. Verkerk, Rianne Wolswinkel, Leander Beekman, Kalliopi Pilichou, Cristina Basso, Carol Ann Remme, Gaetano Thiene, Connie R. Bezzina
DOI: http://dx.doi.org/10.1093/cvr/cvs219 409-418 First published online: 3 July 2012


Aims Mutations in genes encoding desmosomal proteins have been implicated in the pathogenesis of arrhythmogenic right ventricular cardiomyopathy (ARVC). However, the consequences of these mutations in early disease stages are unknown. We investigated whether mutation-induced intercalated disc remodelling impacts on electrophysiological properties before the onset of cell death and replacement fibrosis.

Methods and results Transgenic mice with cardiac overexpression of mutant Desmoglein2 (Dsg2) Dsg2-N271S (Tg-NS/L) were studied before and after the onset of cell death and replacement fibrosis. Mice with cardiac overexpression of wild-type Dsg2 and wild-type mice served as controls. Assessment by electron microscopy established that intercellular space widening at the desmosomes/adherens junctions occurred in Tg-NS/L mice before the onset of necrosis and fibrosis. At this stage, epicardial mapping in Langendorff-perfused hearts demonstrated prolonged ventricular activation time, reduced longitudinal and transversal conduction velocities, and increased arrhythmia inducibility. A reduced action potential (AP) upstroke velocity due to a lower Na+ current density was also observed at this stage of the disease. Furthermore, co-immunoprecipitation demonstrated an in vivo interaction between Dsg2 and the Na+ channel protein NaV1.5.

Conclusion Intercellular space widening at the level of the intercalated disc (desmosomes/adherens junctions) and a concomitant reduction in AP upstroke velocity as a consequence of lower Na+ current density lead to slowed conduction and increased arrhythmia susceptibility at disease stages preceding the onset of necrosis and replacement fibrosis. The demonstration of an in vivo interaction between Dsg2 and NaV1.5 provides a molecular pathway for the observed electrical disturbances during the early ARVC stages.

  • Arrhythmias
  • Sudden cardiac death
  • Electrophysiology
  • ARVC
  • Desmoglein 2

1. Introduction

Arrhythmogenic right ventricular cardiomyopathy (ARVC) is an important cause of ventricular arrhythmias and sudden cardiac death, especially in the young and in athletes.1 Mutations in one or more genes encoding desmosomal proteins are found in ∼50% of patients.26 Although uncovering the genetic basis of ARVC provided us with molecular leads to investigate the pathogenesis of the disease,7 the mechanisms of the early cardiac electrical instability in ARVC are still unclear.

Desmosomes are highly conserved structures that, together with adherens junctions and gap junctions, connect cardiac myocytes end to end at the level of the intercalated discs (ID) and thereby play a crucial role in maintaining proper myocardial function.8 Altered desmosomal organization, as the result of mutations, is thought to lead to myocardial damage and replacement fibrosis, the classical histopathologic pattern of ARVC.1 In advanced stages of the disease, focal scars cause electrical isolation of cardiomyocytes within non-conducting fibrous tissue, resulting in slow conduction and delayed activation, thus forming the substrate for re-entrant circuits and ventricular electrical instability.

However, arrhythmias have also been described early in the disease process,9 before overt structural changes of the myocardium.10 This phenomenon may be due to abnormalities in electrical coupling between cardiomyocytes that are already present before the onset of cardiomyopathic changes. Recent findings indicating the presence of mixed-type junctions (the area composita)11 and crosstalk between protein complexes pertaining to the different types of junctions1217 have markedly changed the perception of the ID. These data support the concept of cross-regulation between structural and electrical components at the ID.

In this study, we employed our ARVC mouse model with overexpression of mutant Desmoglein-2 (Dsg2-N271S).7 This mutation is the homologue of the DSG2-N266S mutation identified in an ARVC patient from Padua.5 A total of 45 mutations in DSG2, mostly affecting the adhesive extracellular domains of Dsg2, have been reported thus far,18 accounting for approximately 8% of the mutations associated with human ARVC. We used this model to investigate whether ID remodelling, as a consequence of a mutation in a component of the desmosome, impacts on cardiac electrophysiological properties at early disease stages that is before the onset of fibrosis and other cardiomyopathic changes. We show that mutant Dsg2 induces widening of the ID at the level of the area composita. This coincides with conduction slowing stemming from a reduced action potential (AP) upstroke velocity caused by a reduction in the cardiac Na+ current (INa) density. Furthermore, we demonstrate that Dsg2 and the cardiac Na+ channel (NaV1.5) interact in vivo, pointing to a molecular mechanism for the development of conduction slowing and arrhythmia in ARVC prior to gross and histological changes of the heart.

2. Methods

2.1 Animal husbandry and mouse lines used

We studied the effects of heart-specific overexpression of mutant Dsg2 in young mice before the development of gross and histological abnormalities. The transgenic mouse models with heart-specific overexpression of either mutant (Dsg2-N271S) or wild-type (Tg-WT) Dsg2 were generated previously.7 Of the two Dsg2-N271S lines we generated, which respectively had high (Tg-NS/H) and low (Tg-NS/L) overexpression of the transgene, the Tg-NS/L line was used in this study. Although both lines, i.e. Tg-NS/H and Tg-NS/L, develop an ARVC phenotype, the latter line was selected as it develops the phenotype at a slower pace which makes it more amenable for the dissection of the early, pre-cardiomyopathic, phenotype of the model. Tg-WT and wild-type (WT) mice were used as controls in all experiments throughout the study. The mice were studied at three different ages: <2 weeks, 3–4 weeks, and 6–9 weeks. At least four mice of each group were studied except where specifically indicated otherwise. Mice were sacrificed by cervical dislocation after sedation with O2/CO2 for 1 min. All experiments were approved by the local animal welfare committee and followed Dutch law concerning experimental animal welfare and conformed with the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH Publication No. 85-23, revised 1996) (PHS assurance number A5549-01).

2.2 Morphological analysis

For pathological studies, hearts were isolated and snap-frozen immediately after excision in liquid nitrogen and stored at −80°C. In parallel, tissue samples were fixed in 4% paraformaldehyde (in PBS) for light microscopy or in glutaraldehyde for transmission electron microscopy (TEM) (see below). Seven-µm-thick paraffin embedded sections were cut and routinely stained with haematoxylin and eosin (H&E) and Heidenhain's trichrome to examine the myocardium and to detect the presence and amount of necrosis, inflammation, and fibrosis.

2.3 Transmission electron microscopy

Transmission electron microscopy was used to characterize the desmosomes and the other intercellular junctions in situ as described before.11 Electron micrographs were taken by systematic random sampling and analysed by two independent expert pathologists (S.R. and C.B.) blinded to the genotype of the mouse. For evaluation, ∼100 IDs were analysed per mouse. Morphometric analysis of ID space was performed according to a previously described method.19 We also calculated the percentage of junctions with an intercellular space > 30 nm.20

2.4 Electrical analysis

The electrical properties of the heart were studied in vivo by surface electrocardiograms (ECG), ex vivo in Langendorff-perfused hearts by epicardial mapping and in isolated cardiomyocytes by patch-clamp analysis. Shortly: epicardial mapping: ventricular extracellular epicardial electrograms were recorded from the right ventricle (RV) and left ventricle (LV). Maximal conduction velocities in both longitudinal and transverse directions were measured from RV and LV activation maps. Cellular electrophysiology: mouse ventricular myocytes were isolated by enzymatic dissociation as described previously in detail.21 APs and INa were recorded with the amphotericin-B-perforated patch-clamp and ruptured patch-clamp technique, respectively. Current-clamp experiments: APs were measured at 36°C and data from 10 consecutive APs were averaged. Voltage-clamp experiments: INa was measured using a two-step protocol at room temperature (RT), with a holding potential of −120 mV and a cycle time of 5 s (Figure 6D). Details of all electrical experiments are given in Supplementary material online.

2.5 Immunofluorescence microscopy

Cryosections of 5 µm were fixed in methanol (5 min) followed by acetone (20 s), at −20°C, air-dried, and rehydrated in PBS. Permeabilization was done in 0.2% Triton X-100 for 5 min. Primary antibodies were applied for 1 h at RT, followed by three washes in PBS (5 min each), incubation with the secondary antibodies (30 min, RT), and 3 × 5 min washes with PBS. Sections were mounted with 50% glycerol in PBS. Images were recorded with a confocal laser scanning microscopy (Leica CTR 5500). Details of the used antibodies are given in the Supplementary material online.

2.6 Protein isolation and western blot analysis

Proteins were isolated from LV tissue from a snap frozen mouse heart and western blotting was performed according to the standard procedures. In short: LV protein (60 µg) was run on denaturing SDS-page gels. The gels were blotted on a pre-equilibrated PVDF Immobilon-P membrane (Millipore) by means of a semidry system. Blots were cut at appropriate heights and probed with primary antibodies. HRP conjugated secondary antibodies were detected with ECL-Plus (Amersham). Chemiluminescent signals were visualized using a digital image analyzer (LAS-4000 Lite; Fujifilm) and quantified using the Aida software package (Aida Image Analyzer v.4.26). Details of the used antibodies and protein isolation are given in the Supplementary material online.

2.7 Co-immunoprecipitation

Aliquots of 100 µg LV whole cell lysate protein of 3–4-week-old mice were incubated rotating overnight at 4°C with washed agarose beads with either conjugated M2 Flag antibody or protein A (both Sigma) and 1 µL of normal mouse IgG (Santa Cruz) in 1 mL of PBS supplemented with protease inhibitors (Complete Mini; Roche) and 0.5 mM Sodium Orthovanadate (PBS++). Beads were washed three times with 1 mL of PBS++, and the bound proteins were eluted with Flag peptide (Sigma). Untreated protein and elutes from both precipitations were treated with Laemli buffer and analysed by western blot as described above.

2.8 Surface ECGs

Mice were anaesthetized using isoflurane inhalation (0.8–1.0% volume in oxygen), and efficacy of the anaesthesia was monitored by watching breathing speed and tail suspension. Four-lead surface ECGs were recorded from subcutaneous 23-gauge needle electrodes attached to each limb using the Powerlab acquisition system (ADInstruments). Lead II was analysed for heart rate (RR interval) and PR, QRS, and QT duration using Chart5 Pro analysis software (ADInstruments). QT intervals in mice were corrected for heart rate using the following formula: QTc = QT/(RR/100)1/2 (RR in ms).

Detailed descriptions of morphological analysis, TEM, antibodies, immunofluorescence microscopy, protein isolation, western blot analysis, co-immunoprecipitation, surface ECG, epicardial mapping experiments, cellular electrophysiology, and statistical analysis are presented in the Supplementary material online.

2.9 Statistical analysis

Data are expressed as mean ± SEM. Values are considered significantly different if P < 0.05 in an unpaired two-sided t-test or in two-way repeated measures of analysis of variance (two-way repeated measures ANOVA) followed by pairwise two-sided comparison using the Student–Newman–Keuls test, after testing for normal distribution of the data. Minimal sample sizes were calculated prior to the experiments based upon a power calculation with a (1−β) = 80%, α = 5%, mean control = 7.0, mean observed = 9.5, and SD = 1.0 (effect size of 2.5, which represents the expected effect sizes for the electrical mapping) which gives n = 4 animals per test group. The intraclass correlation coefficient was used to test the interobserver variability in the morphometric measurements. Statistical tests were performed using the PASW statistics software, version 18.0.2 (IBM, New York, NY, USA) and SigmaStat, version 3.1 (Aspire Software International, Ashburn, VA, USA).

3. Results

3.1 Absence of cardiomyopathic changes in Tg-NS/L mice younger than 6 weeks of age

The early phase of the disease was studied by investigating mice at three different age groups: (i) <2 weeks of age, (ii) 3–4 weeks of age, and (iii) 6–9 weeks. On gross examination and histologically, hearts from Tg-NS/L mice at all three age groups appeared normal, with no evidence of replacement-type fibrosis (Supplementary material online,Figure S1). Other cardiomyopathic changes, consistent with those we previously reported for Tg-NS/H mice7 and which included necrosis, focal myofibrillar lysis, dilated sarcoplasmatic reticulum and T-tubules, and mitochondrial clustering, were observed exclusively in Tg-NS/L mice ≥6 weeks of age (Figure 1). These analyses established that Tg-NS/L mice <6 weeks of age were devoid of cardiomyopathic changes allowing for the examination of the early electrophysiological phenotype prior to and in the absence of cardiac remodelling.

Figure 1

At the ultrastructural level, widening of the intercellular space at the level of the desmosomes/adherens junctions compared with WT mice (A) and Tg-WT (B) is visible in Tg-NS/L mice both at 3–4 weeks (D and E) and at 6–9 weeks (F). This feature is observed also in a Tg-NS/L mouse aged <2 weeks (C). The structure of the gap-junctions is preserved in both Tg-NS/L and Tg-WT mice at all ages (B–F). Scale bars 500 nm.

3.2 Widening of intercellular space at the desmosome/adherens junctions

We next examined ID structures in detail by TEM. No consistent differences were observed in the general organization of the cell–cell junctions between Tg-NS/L and control mice. Gap junctions appeared structurally normal in all groups (Figure 1). In addition to desmosomes, gap junctions, and adherens junctions, intermediate structures were observed that displayed features of both desmosomal and adherens junctions concurring with recent reports describing this ‘area composita’.11,17

Separation of the opposed membranes, resulting in larger intercellular spaces, was observed at the level of desmosomes/adherens junctions. These changes were seen in otherwise morphologically normal cardiomyocytes in all Tg-NS/L mice ≥3 weeks of age and in 1 out of 4 Tg-NS/L mice <2 weeks old (Figure 1). None of the control mice (Tg-WT, WT) displayed any intercellular space widening. Morphometric analysis showed that the average intercellular space was significantly widened in Tg-NS/L mice at 3–4 weeks compared with controls (Supplementary material online, Figure S2A). The percentage of widened cellular junctions increased with age (Supplementary material online, Figure S2B). In some Tg-NS/L cardiomyocytes with intercellular space widening, myofibrils appeared to have undergone focal lysis at their points of attachment to desmosomes/adherens junctions (Figure 1E).

3.3 Conduction slowing in Tg-NS/L hearts from 3 to 4 weeks of age

To investigate the occurrence of electrophysiological abnormalities prior to the development of cardiomyopathic changes, we first performed surface ECG measurements in anaesthetized mice. In mice aged 3–4 weeks, no statistically significant differences in heart rate (RR interval), QRS duration, PR interval, and QTc interval were observed between Tg-NS/L and controls (Figure 2A and B and Supplementary material online, Table S1). However, fractionation of the QRS complex was observed in some Tg-NS/L mice aged 3–4 weeks, indicating the presence of discrete ventricular conduction slowing (Figure 2A). In the 6–9 weeks age group, Tg-NS/L mice developed significant QRS prolongation and abnormal QRS morphology (including marked fractionation), in addition to spontaneous ventricular rhythm disturbances consisting of single or multiple ventricular extra systoles and short runs of non-sustained ventricular tachycardia (Figure 2C). No spontaneous arrhythmias were observed in Tg-NS/L mice aged 3–4 weeks.

Figure 2

(A) Surface ECG examples of WT, Tg-WT, and Tg-NS/L mice (scale bar: 20 ms). In a subset of Tg-NS/L mice aged 3–4 weeks, discrete fractionation of the QRS complex was observed (top right example). In the 6–9 weeks age group, Tg-NS/L mice developed significant QRS prolongation and abnormal QRS morphology (including substantial fractionation of the QRS complex). (B) Average values (mean ± SEM) for QRS duration in WT, Tg-WT, and Tg-NS/L mice in age groups 3–4 weeks (n = 7, 7, and 12, respectively) and 6–9 weeks (n = 10, 9, 12, respectively) (# denotes P < 0.05 vs. WT; $ denotes P < 0.05 vs. Tg-WT). Mean values for all ECG parameters are presented in Supplementary material online, Table S1. (C) Examples of spontaneous ventricular rhythm disturbances observed in Tg-NS/L mice aged 6–9 weeks, including single or multiple ventricular extra systoles and short runs of non-sustained ventricular tachycardia (scale bar: 100 ms).

We next performed epicardial mapping in isolated Langendorff-perfused hearts from WT, Tg-WT, and Tg-NS/L mice of the three age groups to further assess cardiac conduction in detail. In mice aged <2 weeks, no significant differences were found between the groups for LV or RV activation time or effective refractory periods, but a tendency towards lower longitudinal and transversal conduction velocities was observed in Tg-NS/L mice compared with controls (Figure 3A–D, Supplementary material online, Table S2). In contrast, a significantly prolonged epicardial LV activation time in addition to decreased LV conduction velocity was observed in Tg-NS/L mice from the age of 3–4 weeks onwards compared with controls. Both longitudinal and transversal conduction velocities were equally affected, as indicated by the unaltered L/T ratio (Supplementary material online, Table S2). In the RV, activation time and conduction velocity were only significantly prolonged at the age of >6 weeks. Arrhythmia inducibility was tested using up to three extra stimuli and burst pacing. No arrhythmias could be induced in WT, Tg-WT, or Tg-NS/L mice younger than 2 weeks. In contrast, ventricular arrhythmias were inducible in almost half of all Tg-NS/L mice aged 3–4 weeks and >6 weeks, but only sporadically in control mice of the same age groups (Supplementary material online, Table S2).

Figure 3

(A) Typical examples of LV activation maps of isolated Langendorff-perfused hearts from WT, Tg-WT, and Tg-NS/L mice aged 3–4 weeks. Arrows indicate directions and distances used for measurements of longitudinal (CV-L) and transversal (CV-T) conduction velocities. Crowding of isochrones (1 ms) in the Tg-NS/L heart indicates areas of conduction slowing. (B) Average values (mean ± SEM) for LV and RV total activation time. (C and D) Average values (mean ± SEM) for LV and RV longitudinal and transversal conduction velocities (B–D) in hearts from WT, Tg-WT, and Tg-NS/L mice aged <2 weeks (n = 6, 4, 5, respectively), 3–4 weeks (n = 5, 6, 6, respectively), and >6 weeks (n = 6, 5, 6, respectively) (* denotes P < 0.05 vs. WT; $ denotes P < 0.05 vs. Tg-WT). Average values for all parameters and ANOVA P-values are presented in Supplementary material online, Table S2. (E) Examples of ventricular arrhythmias induced in isolated Langendorff-perfused hearts of Tg-NS/L mice aged 3–4 weeks (scale bar: 100 ms). Top panels show induction of non-sustained polymorphic ventricular tachycardia induced by either three extrasimuli (top left panel) or burst pacing (top right panel). The lower panel depicts an example of a sustained monomorphic ventricular tachycardia induced through burst pacing.

3.4 Localization and levels of the intercalated disc proteins

We hypothesized that the conduction slowing observed in hearts at 3–4 weeks of age, i.e. prior to the onset of cardiomyopathic changes, could be related to altered localization or reduced levels of components of the ID. We used immunofluorescence to characterize the distribution of ID proteins in Tg-NS/L mice and control mice. A normal immunoreactive signal was detected for Flag-tagged Desmoglein-2, Desmocollin-2, Plakophilin-2 (Pkp2), Plakoglobin (PG), Desmoplakin, and NaV1.5 as well as for the classical adherens junction proteins N- and pan-Cadherin and α- and β-Catenin and the gap junctional Connexin43 (Cx43) (Figure 4A, Supplementary material online, Figure S3, and data not shown). Note the normal co-localization of Cx43 with pan-Cadherin in all three genotype groups; no clear lateralization of Cx43 was observed (Figure 4A).

Figure 4

(A) Immunohistochemistry of Cx43 (green) and pan-Cadherin (Cdh, red) on LV tissue of WT, Tg-WT, and Tg-NS/L mice at 3–4 weeks. (B) Western blot analysis of whole cell lysate of hearts of WT, Tg-WT, and Tg-NS/L mice aged 3–4 weeks for Calnexin, NaV1.5, and Cx43. (C) Quantification of the NaV1.5 and Cx43 western blot signals (n = 3) relative to Calnexin normalized for WT, error bars denote standard errors, and * denotes P < 0.05 vs. WT.

On western blot analysis in 3–4-week-old hearts, no differences were observed in the levels of PG, Pkp2, NaV1.5, Pan-Cadherin, and Cx45 between the three genotype groups (Figure 4B and Supplementary material online, Figure S4). However, we observed a significantly reduced level of Cx43 in Tg-WT compared with WT hearts (Figure 4B and C). Additionally, we observed a shift in the height of the Cx43 bands from predominantly phosphorylation state P2 as previously described22 in WT to P1 and P0 in both Tg-WT and Tg-NS/L lines (Figure 4B).

3.5 Reduced AP upstroke velocity in isolated cardiomyocytes

In parallel, we investigated the possible cellular electrophysiological changes underlying the ventricular conduction slowing in the 3–4-week-old Tg-NS/L. AP and INa characteristics of LV cardiomyocytes isolated from 3 to 4-week-old mice were measured using patch clamp methodology. Figure 5A shows typical APs of Tg-WT and Tg-NS/L myocytes; Figure 5B summarizes the average AP characteristics of WT, Tg-WT, and Tg-NS/L myocytes. No differences were noted between Tg-NS/L, Tg-WT, and WT in resting membrane potential (RMP), AP amplitude (APA), and AP duration (APD) at 20, 50, and 90% repolarization (APD20, APD50, and APD90, respectively). However, cardiomyocytes from Tg-NS/L mice showed a significantly lower AP upstroke velocity (Vmax) compared with age-matched WT and Tg-WT mice (Figure 5B). On average, Vmax in Tg-NS/L myocyte was reduced with ≈17 ± 4%. The AP upstroke is predominantly determined by Na+ influx through voltage-gated Na+ channels. The observed lower Vmax thus indicates that cardiomyocytes of Tg-NS/L mice have a reduced functional Na+ channel availability.21

Figure 5

(A) Representative APs of LV cardiomyocytes isolated from a Tg-NS/L (light grey) and Tg-WT (black) mouse heart of 3–4 weeks. Inset: First derivatives of the AP upstrokes. (B) Average AP characteristics of Tg-NS/L (n = 3, n = 11), Tg-WT (n = 3, n = 13), and WT (n = 3, n = 11) cardiomyocytes aged 3–4 weeks. RMP, resting membrane potential; APA, maximal AP amplitude; Vmax, maximal upstroke velocity; APD20, APD50 and, APD90 = AP duration at 20, 50, and 90% repolarization, respectively.

The reduced functional Na+ channel availability during the AP upstroke may be due to a decrease of INa density and/or changes in voltage-dependency of (in)activation. Voltage-clamp experiments performed at RT using a double pulse protocol (Figure 6D) demonstrated no differences in the voltage dependencies of the activation (Figure 6A) and inactivation (Figure 6B) of INa. Figure 6C shows the current–voltage relationships of INa in WT, Tg-WT, and Tg-NS/L myocytes. Maximal peak currents were smaller in Tg-NS/L myocytes compared with WT and Tg-WT myocytes (Figure 6D). On average, the maximal peak current was ≈19 ± 6% lower; thus, the reduction in INa density was in the same order as the Vmax reduction.

Figure 6

Na+ current (INa) characteristics in WT, Tg-WT, and Tg-NS/L myocytes. (A) Voltage dependency of activation. (B) Voltage dependency of inactivation. (C) Current-voltage relationships of INa. (D) Voltage clamp protocol. (E) Average maximal peak currents, * denotes P < 0.05 vs. WT.

3.6 Physical interaction between Dsg2 and NaV1.5

Since our electrophysiological studies showed a reduced INa density in Tg-NS/L mice, and, considering the fact that recent studies in rat neonatal cardiomyocytes have demonstrated that desmosomal proteins interact with the Na+ channel complex,12,23,24 we next sought to investigate the possible in vivo interaction between Dsg2 and NaV1.5 by co-immunoprecipitation. We made use of the Flag-tag epitope present on the Dsg2 protein overexpressed in Tg-WT and Tg-NS/L mice to efficiently and specifically pull down Dsg2 from whole cell protein extracts of the LV of transgenic mice. This demonstrated an interaction between Flag-tagged Dsg2 and NaV1.5 in both Tg-WT and Tg-NS/L hearts (Figure 7).

Figure 7

Co-immunoprecipitation of NaV1.5 with Flag-Dsg2 in Tg-WT and Tg-NS/L mice of 3–4 weeks: western blot for NaV1.5 and Flag tagged Dsg2 on whole cell lysate (In), whole cell lysate precipitated with normal mouse IgG (IgG, negative control), and whole cell lysate precipitated with M2 α-Flag (M2).

4. Discussion

We here demonstrate for the first time that a mutation in a structural component of cardiac desmosomes impacts on ventricular conduction and arrhythmia susceptibility even before the onset of necrosis and replacement fibrosis. These effects occur through a reduction in AP upstroke velocity due to a reduced INa density. Furthermore, we show for the first time that the cardiac Na+ channel in an in vivo murine model forms part of a macromolecular complex that includes Dsg2. This structural link between the desmosomal protein complex and the NaV1.5 channel may explain the conduction disturbances and arrhythmias seen early in the ARVC disease process.

Our findings provide support to the recent proposition that at the ID, cross-talk exists between structures previously perceived as being independent.14 In a series of recent studies, the Delmar group demonstrated interactions between Pkp2, Cx43, NaV1.5, and AnkG, thus connecting the desmosomes to the gap junctions and the Na+ channel complex.12,23 In these previous studies, disruption of these protein complexes by downregulation of Pkp2 and/or AnkG in cultured neonatal rat cardiomyocytes led to reduced Na+ channel availability. The data presented in the current study now shows that Dsg2 interacts with NaV1.5 in the mouse heart in vivo. This observation coupled to the reduced AP upstroke velocity due to reduced INa density point to this functional link as the basis for the observed reduction in conduction velocity. It will be interesting to study the temporal and spatial changes in these interactions, as well as the interaction with the other molecules in this protein complex; these studies are unfortunately impossible with co-immunoprecipitation, but will need real-time in vivo imaging of the components of ID.

Unexpectedly, Cx43 protein levels were found to be significantly reduced in Tg-WT; furthermore, a shift towards less-phosphorylated (P0, P1) forms of Cx43 was observed. Nevertheless, Cx43 sub-cellular localization and distribution within the myocardium did not appear to be altered compared with WT mice, in particular no clear lateralization, which has been reported to correlate to the phoshporylation state of Cx4322, was observed. Previously, homogeneous reduction in Cx43 has been shown to be well tolerated. For instance, no differences in cardiac conduction were observed in mice carrying a heterozygous deletion of Cx43.25,26 The observed reduction of ∼25% of Cx43 in both Tg-NS/L and Tg-WT mice is therefore not expected to influence conduction parameters. Corroborating this, no differences on surface ECG and in epicardial mapping were observed between Tg-WT and WT animals. However, we cannot exclude the possibility that the observed reduction in Cx43 levels sensitizes the hearts of the Tg-NS/L mice to the consequences of the Dsg2 mutation. The fact that no other changes in level and localization of ID proteins were seen in Tg-NS/L mice at 3–4 weeks indicates that the observed changes in conduction are due to subtle changes of protein–protein interactions of the ID proteins rather than gross changes in their level and localization.

One of the earliest signs of disease observed in the Tg-NS/L mice was a widening of the ID space at the level of the desmosome/adherens junction with focal lysis of the myofilaments, as previously described in humans.19 In our previous studies on Tg-NS/H mice,7 which have a faster disease development, the widening of the intercellular space was detected only in the setting of concomitant necrosis and inflammation and as such was interpreted as a secondary phenomenon. In the current study, the use of the Tg-NS/L line, which is characterized by a later disease onset, allowed us to show that ID space widening is an early feature of the disease and precedes cell injury and inflammation. Such ID widening could be explained by the fact that the N271 residue, located between the second and third extracellular cadherin domains of Dsg2, has been shown to be critical for co-ordination of Ca2+ binding, a phenomenon essential to the adhesive intercellular interactions of junctional cadherins.27 Similar observations were also made by Kant et al.28 who studied mice carrying a deletion of the adhesive extracellular domain of Dsg2. They suggested that mutant Dsg2 results in compromised adhesion at ID and mechanical cell stress during postnatal heart development, eventually inducing cardiomyocyte death, inflammation, and fibrotic replacement. These features are similar to those observed in our Tg-NS/L mouse model, showing an age-related development of structural lesions, although in the current investigation we focused on the early stages when the heart is still grossly and histologically normal.

In 3–4-week-old mice, intercellular space widening coincided with the onset of conduction slowing. Here, two scenarios are possible. In one, desmosomal interactions27 are weakened by the N271S mutation leading to intercellular space widening, consequently disrupting the multiprotein Na+ channel complex. In a second scenario, the Dsg2 mutation leads to a conformational change affecting the functional interaction between the desmosomal complex and the Na+ channel complex independent of the intercellular space widening.

The conduction slowing and increased arrhythmia inducibility in the 3–4-week-old Tg-NS/L mice point to electrical instability prior to overt structural changes. However, no spontaneous arrhythmias were observed at this age on surface ECG, although we cannot exclude their occurrence since long-term telemetric ECG recordings are not feasible at this young age. Clearly, the observed conduction slowing likely sensitizes the Tg-NS/L mice for development of spontaneous arrhythmias at more advanced disease stages. Our experimental findings of altered ventricular conduction and increased arrhythmia susceptibility even before the onset of necrosis and replacement fibrosis support the hypothesis that conduction disturbances and electrical instability could develop in human carriers of ARVC gene mutations without structural changes (pre-clinical phase of ARVC). This finding underlines the need of a diagnostic tool targeting conduction changes, especially in the cardiological screening of first-degree relatives of ARVC probands carrying gene mutations.

Epicardial mapping in Tg-NS/L mice indicates that the LV is more affected than the RV. This is in line with the increasing recognition of biventricular involvement in ARVC.1,5,9,29,30 The relative rarity of the left-dominant involvement in published ARVC populations is likely a consequence of restrictive inclusion criteria and low sensitivity of diagnostic tools.31 In the setting of family history of ARVC, even signs of isolated LV involvement should be carefully investigated, as they could be the only marker of the underlying genetically determined cardiomyopathy.9,30 Noteworthy, all the reported experimental animal models irrespective of the affected gene show both LV and RV involvement.32

In summary, we here dissect the early stages of disease development in mice overexpressing a Dsg2 mutation associated with ARVC in humans. Intercellular space widening at the level of the ID (desmosomes/adherens junctions) and a concomitant reduction in AP upstroke velocity as a consequence of reduced functional Na+ channel availability leads to slowed conduction and increased arrhythmia susceptibility at disease stages preceding the onset of replacement fibrosis. The demonstration of an in vivo interaction between Dsg2 and NaV1.5 provides a molecular pathway for the observed electrical disturbances during the early ARVC disease process.


This work was supported by the Netherlands Heart Foundation (2009B051 and 2005T024), the Netherlands Heart Institute (ICIN, 061.02), and the Division for Earth and Life Sciences (ALW) of the Netherlands Organization for Scientific Research (NWO) (836.09.003); the Registry for Cardio-Cerebro-Vascular-Pathology, Veneto Region, Venice; Pricard Conacuore, Modena; and the CARIPARO Foundation, Padua, Italy. During this investigation, Dr S. Rizzo was a visiting researcher from the University of Padua at the Academic Medical Center, University of Amsterdam.


We thank Jacques M. de Bakker for his careful reading of the manuscript.

Conflict of interest: none declared.


  • Both authors contributed equally to this work.


View Abstract