Desmoglein-2 mutations in propeptide cleavage-site causes arrhythmogenic right ventricular cardiomyopathy/dysplasia by impairing extracellular 1-dependent desmosomal interactions upon cellular stress

Abstract

Aims

Desmoglein-2 (DSG2) mutations, which encode a heart-specific cadherin crucial for desmosomal adhesion, are frequent in arrhythmogenic right ventricular cardiomyopathy/dysplasia (ARVC/D). DSG2 mutations have been associated with higher risk of biventricular involvement. Among DSG2 mutations, mutations of the inhibitory propeptide consensus cleavage-site (Arg-X-Arg/Lys-Arg), are particularly frequent. In the present work, we explored the functional consequences of DSG2 propeptide cleavage site mutations p.Arg49His, p.Arg46Trp, and p.Arg46Gln on localization, adhesive properties, and desmosome incorporation of DSG2.

Methods and results

We studied the expression of mutant-DSG2 in human heart and in epithelial and cardiac cellular models expressing wild-type or mutant (p.Arg49His, p.Arg46Trp, and p.Arg46Gln) proDSG2-GFP fusion proteins. The consequences of the p.Arg46Trp mutation on DSG2 adhesiveness were studied by surface plasmon resonance. Incorporation of mutant p.Arg46Trp DSG2 into desmosomes was studied under low-calcium culture conditions and cyclic mechanical stress. We demonstrated in human heart and cellular models that all three mutations prevented N-terminal propeptide cleavage, but did not modify intercellular junction targeting. Surface plasmon resonance experiments showed a propeptide-dependent loss of interaction between the cadherin N-terminal extracellular 1 (EC1) domains. Additionally, proDSG2 mutant proteins were abnormally incorporated into desmosomes under low-calcium culture conditions or following mechanical stress. This was accompanied by an epidermal growth factor receptor-dependent internalization of proDSG2, suggesting increased turnover of unprocessed proDSG2.

Conclusion

Our results strongly suggest weakened desmosomal adhesiveness due to abnormal incorporation of uncleaved mutant proDSG2 in cellular stress conditions. These results provide new insights into desmosomal cadherin regulation and ARVC/D pathophysiology, in particular, the potential role of mechanical stress on desmosomal dysfunction.

Introduction

Arrhythmogenic right ventricular cardiomyopathy/dysplasia (ARVC/D) is an inherited cardiomyopathy characterized by the fibro-fatty replacement of cardiomyocytes, predominantly observed in the right ventricle. The main complication of ARVC/D is the occurrence of life-threatening ventricular arrhythmias but heart failure due to progressive dilatation of right or both ventricles is also an important cause of death.¹ Arrhythmogenic right ventricular cardiomyopathy/dysplasia is associated with mutations in genes encoding the main components of the cardiac desmosome, mostly plakophilin-2 (PKP2) but also desmoglein-2 (DSG2), desmocollin-2 (DSC2), desmoplakin, and plakoglobin (PG).¹ Although less frequent than PKP2 mutations, DSG2 mutations are observed in 10–20% of patients.^1,DSG2 mutations have been associated with more frequent biventricular involvement and were associated with a higher risk of evolution to end-stage heart failure, suggesting that DSG2 mutations could be associated with a specific molecular mechanism leading to an increased risk of myocardial dysfunction and heart failure.^2,3

Within the desmosomal complex, DSC2 and DSG2 cadherins are responsible for intercellular trans-interactions and are critical for cell-to-cell adhesion. Desmosomal cadherin interaction relies on Ca²⁺-dependent trans-interactions mediated through their extracellular domains as well as cis-interactions through both intra- and extracellular domains.⁴ Besides their function in maintaining desmosome cohesion, they also regulate several signalling-pathways involved in cell differentiation, proliferation, and invasion. Although the precise role of DSC2 and DSG2 in the structural organization of desmosomes remains unclear, a recent study suggests that each desmosomal cadherin has a specific function in desmosome assembly and adhesion.⁵ Classical and desmosomal cadherins are synthesized with an N-terminal propeptide that is cleaved during the maturation process, resulting in cadherins with adhesive properties.⁶ A hotspot of DSG2 heterozygous missense mutations p.Arg46Trp (R46W), p.Arg46Gln (R46Q), p.Arg49His (R49H), and p.Lys48Asn (K48N) is located in the fully conserved DSG2 propeptide consensus cleavage-site (Arg-X-Arg/Lys-Arg), targeted by Kex2-like proprotein convertases (www.ARVDdatabase.info). These mutations, that account for ∼40% of DSG2 mutations,² are predicted to abolish DSG2 propeptide cleavage.^7,8 This process has been shown in classical cadherins to unmask the N-terminal EC1 domain and its tryptophan residue at position 2 (Trp-2), elements essential for Ca²⁺-dependent trans-interactions.⁶ In the present study, we explored the regulation of desmosomal cadherins in ARVC/D by investigating the molecular consequences of propeptide cleavage site (PCS) mutations on DSG2 function. For this purpose, we performed in vitro and ex vivo experiments using both cardiac and epithelial cellular models. We hypothesized that PCS mutations prevent effective maturation, thereby altering DSG2 adhesiveness. Thus we explored mutation consequences on DSG2 protein stability and integration into desmosomes in different culture conditions.

Methods

Details are provided in the Supplementary material online, Material and Methods section.

Human heart tissue analysis

We performed immunofluorescence (IF) analysis of desmosomal proteins DSG2, PKP2, PG, Connexin-43 (Cx43), and β-catenin in explanted heart tissues samples collected from PCS mutations-carrier and a control subject with dilated cardiomyopathy, as described previously.³ DSG2 cleavage was studied using a custom propeptide antibody specifically directed against the DSG2 propeptide. All patients gave their informed consent for use of tissue samples for research purpose. The use of human tissue was performed conform to the declaration of Helsinki and the study was approved by a local ethics review board.

Plasmid and adenovirus constructions

We constructed p.EGFPN3 vectors expressing the wild-type (WT)-, R46Q-, R46W-, and R49H-DSG2-GFP fusion proteins with a C-terminal GFP. The three mutations R46Q, R46W, and R49H were introduced with the QuikChange II XL Site-Directed Mutagenesis kit (Agilent Technologies^®). Adenoviruses expressing DSG2-WT or DSG2-R46W were constructed subsequently after sub-cloning the DSG2 cDNA into the p.Shuttle vector using the Adeasy™ Adenoviral Vector System (Agilent technologies^®) according to manufacturer’s instructions. cDNA encoding the EC1 and proEC1 domains were cloned into the pET6xHN-C vector, resulting in expression of EC1 and pro-EC1 domains fused with 6 HN tags at the C-terminus (pET Express & Purify kit –His60, Clonetech^®).

Cellular immunolabelling

WT- and mutant-DSG2-GFP were transiently expressed in neonatal rat cardiomyocytes (NRC) using lipofectamine 2000 reagent (Invitrogen^®). NRC isolation and culture is described in Supplementary material online. Twenty-four hours after isolation, NRC were transiently transfected using lipofectamine 2000 reagent (Invitrogen^®) in OPTIMEM medium. After 4 days of culture, cells were fixed in ice-cold methanol (−20°) and co-immunolabelled with either anti-DSG2, anti-GFP, anti-PG, anti-PKP2, anti-Cx43 (Cx43) or anti-N-cadherin, and anti-α-actinin2 or anti-cMyBP-C antibodies.

Co-immunoprecipitation assays

Monolayers of HaCat cells were infected with adenovirus either expressing WT or mutant R46W-DSG2-GFP in low-calcium medium (LCM, 0.09 mM Ca²⁺). One day post-infection, LCM was replaced by standard DMEM medium (1.8 mM Ca²⁺) and incubated for 24 h. Proteins were extracted in cytoskeleton extraction buffer and immuno-precipitated using the rabbit anti-GFP antibody or anti-β-catenin antibody as a control (β-catenin was previously shown to not co-immunoprecipitate with DSG2⁹) and M-280 anti-rabbit dynabeads (Invitrogen^®). Immuno-precipitated proteins were analysed by western blot with the anti-DSG2 6D8, anti-GFP, and anti-α-tubulin antibodies and imaged on an Odyssey^® imager (Sciencetec^®).

Extracellular cross-linking study

HaCat cells were infected with adenovirus expressing WT or R46W-DSG2-GFP in LCM. Twenty-four hours after infection, LCM was replaced with standard DMEM medium for 24 h, then incubated for 30 min at room temperature (RT) with SEGS extracellular cross-linker (1 mM ethylene glycolbis[sulfosuccinimidyl-succinate]; Thermoscientific^®). Cells were subsequently incubated in Hank's Balanced Salt Solution (HBSS) with 2 mM EGTA for 30 min at RT and immunolabelled with the anti-GFP antibody as previously described.

Surface plasmon resonance experiments

DSG2-EC1-6xHN and DSG2-proEC1-6xHN fusion proteins were expressed in BL21 Escherichia coli and concentrated in an HBS-EP buffer according to manufacturer’s instructions (pET Express & Purify Kit—His60, In-Fusion^® Ready; Clontech^® 631428). Recombinant DSG2-EC1 and DSG2-proEC1 were then immobilized in sodium-acetate buffer (10 mM NaAc, pH 5.0) on a ‘CM5 sensor chip’ for 20 min. BSA acted as a negative control. Binding and kinetics studies were performed in HBS-P buffer in the absence or presence of 5 mM calcium. Equilibrium dissociation constants (Kd) were calculated using the BIAevaluation software v4.1 (Biacore AP^®). All binding and kinetic studies were performed in duplicate.

Calcium sensitivity study

HeLa (human epithelial cell line) and HaCat cells were infected by adenovirus (MOI of 20) expressing WT- or R46W-DSG2-GFP in LCM. Twenty-four hours after infection, cells were cultured in LCM supplemented with Ca²⁺ ranging from 0.09 to 1.8 mM for 24 h. Cells were then fixed in ice-cold methanol and immunolabelled with the anti-GFP, anti-PKP2, or anti-PG antibodies as previously described. To study the role of the epidermal growth factor receptor (EGFR) in DSG2 internalization, cells were treated with or without the EGFR inhibitor PKI166 (7 µM, Santa Cruz^®) for 5 h at 37° before labelling in 0.5 mM Ca²⁺ medium. Quantification of fluorescence was performed using the image J software (https://fiji.sc).

Mechanical stress experiments

HaCat cells were cultured in latex chambers (ST-CH-04, B-Bridge international, Inc.^®) covered with gelatin and fibronectin in standard DMEM medium and infected with the WT- or mutant R46W-DSG2-GFP adenovirus. Two days after infection, cells were submitted to mechanical longitudinal stretching for 1 h at 37° using the ST-140-04 STREX apparatus (B-Bridge International, Inc^®). Total proteins were extracted with TritonX-100 buffer. The supernatant was considered as the Triton-soluble fraction and the pelleted fraction, considered insoluble, was further extracted in a SDS-urea buffer. Samples were analysed by western blot with the anti-GFP and alpha-tubulin antibodies and imaged on the Odyssey imager. Signal intensity was quantified using Image J software (v1.41, National Institutes of Health, Bethesda, MD, USA). Experiments were performed in duplicate (n = 4).

Statistical analysis

Statistical analysis was performed with the GraphPad Prism 5 software (GraphPad Software^®, Inc., CA, USA) using the Mann–Whitney non-parametric test. A P < 0.05 was considered as statistically significant.

Results

Cleavage site mutations prevent cleavage of DSG2 propeptide

We first aimed to demonstrate that ARVC/D PCS mutations led to inhibition of DSG2 maturation. Because of the small size of the propeptide compared to full-length DSG2-GFP, no difference in size could be observed by western blot. We, therefore, used an in-house antibody specifically directed against the propeptide (anti-proDSG2) in neonate rat cardiomyocytes (NRC) expressing WT-DSG2-GFP. Specificity of the antibody was first established through western blot of recombinant DSG2-EC1-HN fragments (Supplementary material online, Figure S1A) and immuno-fluorescence studies in MDCK cells (Supplementary material online, Figure S1B). As expected, no propeptide signal was observed in NRC expressing WT-DSG2-GFP, except for occasional faint peri-nuclear staining in few cells, suggesting that propeptide cleavage occurs close to the nucleus before DSG2 moves to the membrane (Figure 1A and Supplementary material online, Figure S2). In contrast, clear propeptide staining was observed at intercellular junctions in patient cells expressing various PCS pro-DSG2 mutants (R49H, R46W, and R46Q; Figure 1A). This was strongly corroborated by specific proDSG2 staining at the intercalated disks in ex vivo heart samples from the R49H mutation carrier (Figure 1B). These results clearly indicate striking consequences for all PCS mutations, which inhibit DSG2 propeptide cleavage. However, they also demonstrate that PCS mutations do not affect DSG2 localization at intercalated disks and intercellular junctions under standard culture conditions or in patient heart samples. Moreover, other junction proteins (PG, PKP2, N-cadherin, and Cx43) were normally localized in NRC suggesting that overexpressed mutant proDSG2 had no direct dominant-negative effect on the localization of these proteins under standard culture conditions (Supplementary material online, Figure S3). The results of desmosomal protein immunofluorescence staining and electronic microscopy in PCS mutation carriers heart samples were described previously.³ Desmosomal proteins PKP2, PG, DSG2, and β-catenin showed normal localization at intercalated disks without intracellular accumulation (Supplementary material online, Figure S4).³ The signal of DSG2 appeared decreased in the R46W heart sample. Delocalization of Cx43 was observed in ARVC/D samples. The electronic microscopy study of heart tissue from the R46W mutation-carrier showed the presence of small desmosomes associated with abnormal inter-membrane vacuoles in both ventricles.³

Propeptide cleavage-site mutations modify DSG2 adhesive properties and desmosome incorporation upon cellular stress

We hypothesized that abolishing the propeptide cleavage might alter the adhesive properties of DSG2. For this purpose, we focused our studies on DSG2-DSG2 homophilic interactions.¹⁰ To study the consequences of the mutation on DSG2 adhesive properties and desmosome incorporation, we used epithelial cell models such as HaCat and Hela, which endogenously express mature desmosomes, contrary to cardiac cellular model, which only express immature desmosomes¹¹(cf. limits).

Analysis of homophilic interactions

Co-immunoprecipitation experiments between the exogenously transfected WT- or mutant-DSG2-GFP fusion proteins (195 kDa), and endogenous DSG2 (165 kDa) in HaCat cells revealed the expected DSG2-GFP fusion protein and a second band of lower mass corresponding to endogenous DSG2. This demonstrated that mutant R46W-DSG2-GFP, although uncleaved, was still capable of homophilic interactions with native DSG2 (Figure 2A). To determine if this preserved homophilic interaction was at least in part due to trans-interaction, we measured mutant DSG2 intercellular trans-interaction in the presence of a specific cross linker (SEGS) in HaCat cells. The chelation of available Ca²⁺ with 2 mM EGTA induced internalization of both WT and mutant DSG2 and complete dissociation of cells in the absence of SEGS (Figure 2B). Pre-treatment with SEGS prevented both EGTA-induced cell dissociation and intracellular relocation of both WT- and R46W-DSG2, suggesting persistent intercellular direct or indirect trans-interactions of mutant proDSG2. Thus, these experiments suggest that despite the presence of the propeptide, mutant proDSG2 was still involved in direct or indirect extracellular trans-interaction.

Surface plasmon resonance reveals decreased in vitro pro-DSG2/DSG2 interaction

According to classical cadherin interaction models, several extracellular and intracellular domains are likely to be involved in DSG2 homophilic interactions,^4,10 In line with our previous results, we hypothesized that propeptide presence could specifically alter DSG2 trans-interactions via the first N-terminal EC1 domain, without significantly decreasing interactions for other domains. We purified DSG2-EC1 (EC1 alone) and proDSG2-EC1 (propeptide + EC1) recombinant domains and studied their binding properties in vitro using Surface Plasmon Resonance (Biacore™) technology (Figure 3). As anticipated, EC1/EC1 domain interactions were fully Ca²⁺-dependent. Compared to EC1/EC1, we observed a drastic 90% decrease of pro-EC1/EC1 binding signal in the presence of Ca²⁺, clearly showing propeptide inhibitory effects on EC1 domain adhesiveness. No binding was observed between pro-EC1 domains only. These results suggest that although mutant proDSG2 may still be able to interact with other DSG2 partners, retention of the propeptide drastically reduces EC1-mediated DSG2 binding properties with unmasked EC1 domains, which are critical for trans-interactions.

Mutant pro-DSG2 is internalized under mechanical and calcium stress through an epidermal growth factor receptor-dependent mechanism

Despite decreased EC1-mediated interaction strength as displayed in vitro (Figure 3), cardiac and epithelial cultured cells (NRC, HaCat, HeLa) expressing PCS pro-DSG2 mutants under standard culture conditions (1.8 mM Ca²⁺) maintained significant levels of trans-interactions (Figures 1, 4, and Supplementary material online, Figure S5). We hypothesized that under Ca²⁺ depletion, PCS mutant DSG2 expression might induce significant decreases in cellular adhesiveness. Although cardiac cells in culture express desmosomal proteins, it has been showed that these cells often lose the intercalated disk structure in culture with immature desmosomes and such cells were unsuitable for these experiments.¹¹ We therefore chose Hela cells, an epithelial model largely used to study desmosomal function in vitro. HeLa cells were transfected with WT or mutant DSG2 at various Ca²⁺ concentrations. Under control conditions (1.8 mM), WT and mutant DSG2 were located at membranes and intercellular junctions (Figure 4A). Remarkably, when cells were cultured in low calcium (0.5 mM), although most of the WT-DSG2 remained localized at the intercellular junctions, mutant DSG2 was mostly located within intra-cytoplasmic vesicles, suggesting either mis-targeting or internalization (Figure 4A and C). Accordingly, the mean membrane/cytoplasm fluorescence ratio of R46W DSG2-GFP was significantly decreased compared to WT at 0.5 mM Ca2+ (P < 0.001, Figure 4D). Similar results were obtained in HaCat cells (Supplementary material online, Figure S5). Klessner et al.¹² demonstrated that DSG2 endocytosis is regulated by EGFR. In the presence of PKI 166 (PKI+), a specific EGFR inhibitor, mutant proDSG2 was relocated to the intercellular junctions of HeLa cells cultured at 0.5 mM Ca²⁺, indicating EGFR-dependent internalization of mutant proDSG2 under low-calcium conditions, rather than reduced membrane targeting (Figure 4B). The mean membrane/cytoplasm fluorescence ratio of R46W DSG2-GFP was significantly increased after PKI 166 treatment (P < 0.001) and comparable to WT at 0.5 mM Ca2+ (Figure 4D).

Another hypothetical mechanism of PCS mutant DSG2 expression might be an abnormal integration into desmosomes, resulting in decreased adhesiveness and increased mutant proDSG2 internalization. As cardiomyocytes experience permanent mechanical stress, we simulated this environment by submitting HaCat cells to longitudinal cyclic mechanical stress and investigated the consequences of proDSG2 incorporation into desmosomes. We observed a significant increase of the non-desmosomal, triton-soluble fraction, of stretched cells expressing mutant proDSG2 in comparison to cells maintained at rest, or to stretched cells expressing WT-DSG2 (Figure 5). The DSG2 signal from the insoluble desmosomal fraction of stretched cells was not significantly different between the protein extracts from WT- and mutant expressing cells. This suggests that upon mechanical stretch, a proportion of mutant proDSG2 is not incorporated into desmosomes and could either remain unbound at the membrane or in internalized pools.

Discussion

Several human diseases are caused by desmosome dysfunction. Among them, ARVC/D, which is directly caused by desmosomal gene mutations, provides an interesting model to study the effects of disease-causing mutations on desmosome function and regulation. Few studies have focused on DSG2 mutation pathophysiology in ARVC/D. This gene, although less frequently implicated in the disease than PKP2, is the second most frequent gene in most ARVC/D populations,¹ and could lead to higher risk of heart failure.² Within DSG2, PCS-mutations are particularly frequent (≈40%) (www.arvcdatabase.info). In the present study, we investigated the molecular and cellular consequences of these particular mutations.

Consequences of DSG2 PCS mutations on DSG2 maturation and adhesiveness

We demonstrated that PCS mutations prevented effective DSG2 maturation in NRC and in mutation-carrier heart samples, as suggested by previous data obtained in heterologous cellular systems or in vitro.⁷ Mutated DSG2 was correctly addressed to the intercalated disk, as previously observed in skin tissue from a R46Q mutation carrier.⁸ We also showed that DSG2 maturation occurs before the proteins are targeted to the plasma membrane, most likely in the late Golgi apparatus, similar to classical cadherins.

The consequences of propeptide retention in desmosomal cadherins such as DSG2 have been poorly studied, in contrast to classical cadherins. Abnormal propeptide cleavage for both E- and N-cadherins has been clearly shown to prevent EC1-mediated calcium-dependent trans-interactions, most likely by masking the Trp2 residue of mature forms.⁶ Our results also suggest a role for mutant pro-cadherin through modified interactions at the desmosomal level. However, the model for desmosomal cadherin interaction appears to be much more complex and involves homophilic, heterophilic, as well as cis/trans interactions.^4,10,13 Different interaction models involving one or more of the four extracellular domains have been proposed.¹⁴ Here, using in vitro Biacore™ technology, we demonstrated that the propeptide prevented Ca²⁺-dependent homophilic interactions between recombinant DSG2-EC1 domains. EC1-mediated interactions have been shown to be critical for cadherin trans-interactions and their abolition may have major consequences on desmosome adhesiveness.¹⁰ However, immature mutant proDSG2 was still correctly targeted to intercalated disks upon standard culture conditions. Co-immunoprecipitation and extracellular cross-linking experiments suggest the ability of mutant DSG2 to homodimerize and participate in direct or indirect trans-interactions, that could be facilitated by WT DSG2 or mediated by DSC2 or other non-desmosomal cadherin. Such a mechanism was also suggested in a recent study of mutant DSG2 expressing the p.Trp2Ala mutation, which is also predicted to abolish EC1-mediated interactions.⁵ The observed persistent homophilic interactions might involve other domains, in particular, the EC3/4 and DUR domains, which have previously been shown to participate in DSG2 cis-interactions.⁴ We proposed that PCS mutations probably selectively prevent EC1-mediated DSG2 trans-interactions, but interactions with other extracellular domains may be marginally modified, maintaining molecular interaction under normal unstressed conditions (Figure 6).

Effect of mutations under stress conditions

We anticipated that weakened trans-interactions of mutant proDSG2-containing desmosomes would result in modified cellular contact compared to WT-DSG2. This result differs from those from Gaertner et al.,⁷ who found that the mutated R46Q DSG2 displayed increased binding properties using flow-cytometry based assay. Indeed, we unmasked decreased cellular adhesion and increased intracellular labelling of pro-DSG2 with extracellular calcium depletion. Similarly, mechanical stretch also associated with increased triton-soluble pools of mutant proDSG2. Whether this is due to intracellular sequestration of de novo synthesized proDSG2, or to internalization of the membranous pools, remains unclear. Moreover, increased mutant proDSG2 vesicular staining and cellular disjunction with low calcium concentrations were inhibited when EGFR-dependent internalization was chemically blocked with PKI. Both results suggest that PCS mutations weaken calcium-dependent proDSG2 trans-adhesion and trigger internalization by abolishing EC1-mediated interactions.

In addition, cyclical mechanical stress mimicking in vivo heart conditions increased the soluble, non-desmosomal, mutant-DSG2 fraction in comparison to WT. Such observations may result from changes in DSG2 dimerization states, as described for E-cadherin dimerization, which involves several extracellular domains at rest conditions, but only EC1 domains during stretch (Figure 6).¹⁵

Pathophysiological consequences of arrhythmogenic right ventricular cardiomyopathy/dysplasia PCS mutations

DSG2 is a critical component of the cardiac desmosome. Defective DSG2 trans-interactions leading to DSG2 internalization and desmosome disruption appear to be a major consequence of PCS mutations.

The fate of internalized proDSG2 is unclear, but degradation of the internalized mutant proDSG2 leading to a haplo-insufficiency-like mechanism is an interesting hypothesis, similar to what was observed for DSG1 and DSG3 when either of their maturation processes or interactions were perturbed.¹⁶ In accordance with this hypothesis, we have shown specific decreases in both DSC2 and DSG2 in ARVC/D patient heart samples, including two patients carrying PCS mutations, suggesting that the loss of desmosomal cadherins is strongly associated with ARVC/D and may play a central role in the pathophysiology of the disease.³ Recently, auto-antibodies directed against DSG2 have been found in the serum of ARVC/D patients.^17,In vitro data have shown potential pathogenous effect of these antibodies similarly from what has been described in pemphigus.¹⁶ These data altogether suggest that desmosomal cadherin play a critical role in ARVC/D pathophysiology. These effects correlate with desmosomal loss and the inter-cellular cell widening observed in both ARVC/D patient and animal models heart micrographs.^3,18 Interestingly, the cellular phenotype associated with PCS mutations is only revealed upon stress conditions and could explain how exercise can influence ARVC/D penetrance and worsen phenotypic expression.¹⁹ Although calcium depletion allowed unmasking weaker mutant DSG2 adhesion properties, the clinical relevance of this finding remains unknown. The adhesion defect probably does not solely explain all of the phenotypic traits observed in ARVC/D. Widening of intercellular spaces observed in an animal model over-expressing mutant N271S-DSG2 was associated with reduced Na-current density and conduction velocities. Therefore, loss of intercellular interactions could also disrupt electrical complexes located at the intercalated disks leading to conduction slowing, that could play a role in the genesis of ventricular arrhythmias.²⁰

In our study, pro-DSG2 internalization appeared to be dependent on EGFR pathway. The EGFR pathway induces DSG2 intracellular cleavage via cysteine-proteases, leading to apoptosis, a mechanism frequently observed in ARVC/D patients or animal models and potentially involved in cardiomyocyte loss.¹² Our results suggest that EGFR pathways could be involved in the pathophysiological mechanisms underlying ARVC/D.

Limitations

The main limit of this work is the use of epithelial cells model expressing endogenous DSG2 for low calcium and for stretch experiments. However, these cells have mature desmosomes and are thus more suitable than cardiac cultured cells to study consequences of mutants of desmosome adhesion whereas cardiac cultured cells often lose the intercalated disk and desmosome structure in culture.¹¹ Our preliminary experiments on various cardiac cells (NRC, HL1, IPS-derived cardiomyocytes) showed weak only intercellular adhesion under standard culture conditions. Thus, we cannot exclude different pathophysiological consequences of these mutations in cardiac myocytes. Although the anti-propeptide antibody showed a good specificity for the pro-EC1 fragment in western blot and in immunofluorescent experiments, it lacked specificity for full-length pro-DSG2 in western blot experiments and couldn’t be used for this purpose. Although it was demonstrated that the three PCS mutations shared the same consequences on propeptide cleavage and membrane targeting at rest, the study of its consequences on DSG2 interactions upon stress were only investigated with the R46W mutation. We were not successful studying the DSG2 fluorescent signal redistribution upon mechanical stress because of strong auto-fluorescence of the culture chambers. In addition, only homophilic interactions of DSG2 were investigated and we cannot rule-out that these mutations could also affect heterophilic DSG2/DSC2 interactions. Although the mechanisms described in this study are probably shared between all PCS DSG2 mutations, this may not account for all genetic causes of ARVC/D.

Perspectives

DSG2 mutations associated with ARVC/D provide useful tools to study their consequences on desmosomal cadherin regulation in the setting of human disease. Our results show how these mutations impair DSG2 adhesiveness due to abnormal incorporation of uncleaved mutant proDSG2 into desmosomes upon cellular stress conditions and strengthen the relation between cadherin-mediated desmosomal adhesiveness and ARVC/D pathophysiology. In addition, the deleterious effects of DSG2 mutations were only observed upon stress conditions, which could explain how exercise can influence ARVC/D penetrance and worsen phenotypic expression. However, although the results clearly show a central defect associated with pro-cadherin maturation, mechanisms have to be more thoroughly studied (precise involvement of EGFR signalling, signalling consequences of cadherin loss) in cardiac context. This could also help to understand the higher risk of biventricular involvement and heart failure associated with DSG2 mutations.

Conclusion

Our results strongly suggest weakened desmosomal adhesiveness due to abnormal incorporation of uncleaved mutant proDSG2 in cellular stress conditions. These results provide new insights into desmosomal cadherin regulation and ARVC/D pathophysiology, in particular the potential role of mechanical stress on desmosomal dysfunction.

Acknowledgements

We thank Dr Bouceba for the Biacore experiments. We thank the Myobank—AFM (Institut de Myologie Hôpital Pitié-Salpêtrière, Paris, France) for the ARVC/D patient heart samples, and we thank Dr Rachel Peat for critical reading of the manuscript. HeLa cells were a kind gift from Dr A. Lombes.

Funding

This work was supported by the Assistance Publique-Hôpitaux de Paris (PHRC programme hospitalier de recherche AOM n°05073), Fédération Française de Cardiologie/Société Française de Cardiologie, Promex Stiftung für die Forschung, the Ligue contre la cardiomyopathie, and the European Union FP7 (Best Ageing network grant n^o306031).

Conflict of interest: none declared.

References

Author notes