Beyond mechanical adhesion, desmosomes have been implicated in signaling processes such as apoptosis, cell migration, and proliferation.1,2
In particular, a functional link between desmosomes and gap junctions has become evident through cardiac disease. In ARVC, mutations in genes coding for cardiac desmosomal proteins affect gap junction morphology10
and this pathologic situation is thought to contribute to the ventricular arrhythmias that characterize the disease. In Naxos disease, gap junction remodeling has been shown to possibly precede fibrofatty replacement of cardiomyocytes.11
In this study, a patient who presented with delayed RV endocardial conduction and local fractionation of electrograms in the triangle of dysplasia was found to be a carrier of two desmosomal mutations. Despite the patient's benign phenotype, pathologic changes were evident at the molecular level. Immunoreactive signal for PG was remarkably reduced at the intercalated disks (). This feature was found to characterize remodeling processes occurring in the myocardium of ARVC patients independent of the underlying gene mutation.23
The observation that total PG levels were unchanged (A) supports the hypothesis that PG is redistributed to other cellular pools.25
Moreover, changes in the main cardiac gap junction protein Cx43 were evident in the patient. Reduced Cx43 protein levels were detected (A), and, more importantly, a shift in electrophoretic mobility was observed (B). In the patient's sample, faster migrating Cx43 was more prominent, suggesting a lower proportion of the highly phosphorylated protein.26
The biologic relevance of this posttranslational modification at multiple sites in Cx43 is complex, as phosphorylation events are thought to regulate gap junction assembly, disassembly, and channel activity.24
The slight reduction in gap junction protein ZO-1 indicates that the entire structure might be affected.
A series of in vitro
experiments performed provides insight into the molecular mechanisms that could underlie the observed changes. Functional studies on the DSG2
A517V change failed to identify any pathogenic potential (Figure S4
). However, we cannot exclude completely that this conservative change contributes to the phenotype. The truncated DSC2a Q851fsX855 protein was also found to be normally incorporated into cardiac desmosomes (); however, it completely abolished binding to PG and DSP, whereas binding to PKP2, which occurred in both DSC2 isoforms, was not affected ( and S5
). It is likely that the presence of the mutant DSC2a protein in the cardiac desmosomes reduces the capacity of these structures to retain PG, which could contribute to the observed redistribution of PG from junctional pools (), as shown for other ARVC causing DSC2 mutations.27
In particular, nuclear translocation of nonjunctional PG may aberrantly modulate Wnt signaling pathways, driving adipogenesis and fibrogenesis.25,28
Due to the limited availability of patient material, we could not stain for other desmosomal proteins in our myocardial samples. Therefore, the assumption that PKP2 still is present at the intercalated disks is based solely on in vitro
binding data () and is somewhat speculative.
The truncation mutation in DSC2
only affects the DSC2a isoform, whereas the DSC2b splice variant is not affected. A recent report suggested low expression levels of the DSC2a isoform in the heart and indicated that truncation of the last five amino acids of DSC2a by the A897fsX900 variant is not sufficient to cause ARVC29
despite reduced binding to PG.27
In contrast, in our patient with slow conduction and borderline diagnosis of ARVC, truncation of a substantial portion of the DSC2a ICS domain by the Q851fsX855 mutation causes complete loss of binding to PG and DSP. Moreover, our functional data shed new light on the unique cellular functions of the DSC2a splice isoform in cardiac tissue. Only this isoform (in its WT form), and not its DSC2b counterpart, provides a direct physical link to Cx43. Whereas the interaction is not affected by the A897fsX900 variant, the Q851fsX855 mutation abolishes this cellular function of DSC2a (B). This may, among many other factors,30,31
contribute to the conduction abnormalities observed in the patient.
Oxford et al32
reported an interaction between desmosomal PKP2 and Cx43. However, the binding profiles of WT DSC2a and DSC2b do not suggest that the DSC2a Cx43 interaction is mediated by PKP2 (). Moreover, PKP2 was not detectable in our immunocomplexes using a Cx43 antibody (D). This discrepancy may be attributable to differences in experimental setups. Additionally, the multimolecular complex analyzed by Oxford et al might also have contained DSC2a or other bridging molecules. Nevertheless, the observation that knock-down of PKP2 affects Cx43 expression and localization as well as cell coupling properties supports our hypothesis that desmosomal proteins are essential for proper gap junction function. Down-regulation of PKP2 affected localization of DSP32
and, most likely, other desmosomal proteins, such as DSC2, as well. Hence, the lack of other desmosomal components at the intercalated disks, rather than the loss of PKP2 itself, could have triggered the redistribution and down-regulation of Cx43.
Despite the observed molecular changes and evidence of significant reductions in both RV endocardial conduction time and activation gradient (a surrogate of conduction velocity), no ventricular arrhythmia was evident in the patient. A similar phenomenon of significantly high “conduction reserve” has been reported in animal models. Mice with a heterozygous deletion of Cx43 in the heart have no overt phenotype, and a reduction of Cx43 protein down to 10% of normal levels is required to induce spontaneous arrhythmias.33
These data indicate that reduced Cx43 expression contributes to an optimal milieu that facilitates arrhythmogenesis in ARVC. However, a specific physiologic trigger (e.g., increased adrenergic tone) promoting ventricular ectopy or further modulation of the substrate by inflammatory processes may be required to increase the probability of a clinically important arrhythmic event. These modulations in the substrate are expected to increase the heterogeneities in conduction and repolarization kinetics in the ventricle promoting conduction block, reentry, and the generation of ventricular tachycardia and fibrillation.
Analysis of this patient with a subclinical ARVC phenotype provides valuable insight into disease mechanisms. Changes in gap junction protein Cx43 occur before any clinical manifestation significant enough to fulfill Task Force criteria for diagnosis but is, nevertheless, detectable utilizing high-density electrophysiologic mapping of the substrate. Moreover, the newly identified interaction between DSC2a and Cx43 may contribute to the interdependence of desmosomal integrity and gap junction function and thereby opens avenues for new therapeutic approaches in ARVC.