Early ACM is characterized by ventricular arrhythmias out of proportion to the degree of myocardial remodelling [1]. In later stages, the myocardium undergoes progressive degeneration and CMs are replaced by fibrofatty tissue [1]. Classical ACM affects primarily the right ventricle. More recent evidence, however, suggests that sole left ventricular and biventricular disease forms exist [1].

Over 60% of ACM patients bear one or more mutations in genes encoding for the cardiac desmosomal proteins: DSP, PKP2, DSG2, DSC2 and JUP [1]. Non-desmosomal gene mutations are also associated with ACM, where studies indicate the presence of a common final signaling pathway underlying disease pathogenesis (Table 1) [1].

The cardiac ID involves 3 major protein complexes: AJs, GJs and desmosomes (Fig. 2). AJs anchor N-cadherin to the actin cytoskeleton via α- and β-catenin. Catenins have a dual role; both as adhesion molecules but also as transcriptional regulators [7]. GJs are comprised of connexins and are responsible for electrical conduction [7]. Desmosomes contain the desmosomal cadherins DSG2 and DSC2, which link adjacent CMs as well as the armadillo and plakin proteins DSP, PKP2 and JUP, which anchor the cadherins to the intermediate filaments [7]. IDs also contain ion channels, such as the voltage-gated sodium channels, responsible for action potential generation. The main protein subunit of the cardiac sodium channels is Nav1.5, coded by the SCN5A gene [7]. Multiple lines of evidence suggest that desmosomal gene mutations perturb the cWnt and Hippo signalling pathways, which in turn promote the fibrogenic and adipogenic phenotypes characterizing ACM [8, 9].

Diagrammatic representation of the cardiac ID. Among others, the multi-protein structure contains adherens junctions and desmosomes (involved in mechanical cell–cell adhesion) as well as gap junctions (involved in electrical propagation). Figure created with powerpoint

The highly conserved Wnt signaling pathway, originally recognized for its role in embryonic development and tissue homeostasis, has emerged as a crucial player in the pathogenesis of several human disorders and greatly contributes to disease progression with potential therapeutic implications [10]. CM differentiation from iPSCs is critically dependent upon Wnt regulation. Following initial Wnt activation, mesendoderm is generated. Thereafter, maintenance of Wnt signaling is critical to direct cell fate into cardiac mesoderm [11]. Wnt is comprised of canonical and non-canonical components. The canonical pathway is responsible for retaining the proliferative state of cardiac tissue during development and is an essential regulator of the expansion of mesenchymal cells populating the outflow tract cushions [12], whereas the non-canonical pathway primarily promotes precursor differentiation [13].

It is the canonical component that has been implicated in ACM pathogenesis [14, 15]. Efforts to target cWnt activation in experimental models have led to down-regulation of both Nav1.5 and the main ventricular gap junction protein Connexin43 (GJA1; Cx43) resulting in decreased cardiac conduction velocities [16]. Indeed, altered distribution and expression of both proteins is regarded as a phenotypic hallmark of ACM [17]. It is therefore, unsurprising that aberrant activation of cWnt may contribute to the pathogenesis of ACM.

Glycogen synthase kinase-3 (GSK3) is a highly conserved serine/threonine protein kinase that is ubiquitously expressed as two isoenzymes; GSK3α and GSK3β. It was originally recognized for its ability to phosphorylate and inhibit glycogen synthase and hence promote insulin resistance. The ability of lithium to reverse this action led to its classification as a GSK3 inhibitor [18, 19]. Later, a class of maleimides (including SB2 and SB4) were shown to be more potent GSK3 inhibitors that act by competitively binding to the ATP-binding site [20, 21]. Wnt signaling regulates GSK3 activity by displacing GSK3 from its binding partners: axin and adenomatous polyposis coli (APC) in the so-called destruction complex. In the absence of Wnt ligand binding, β-catenin is phosphorylated by GSK3β and targeted for ubiquitination and proteasomal degradation. Upon binding to Frizzled/LRP5-6 receptors, Wnt ligands displace GSK3β precluding the degradation of β-catenin, which is then free to enter the nucleus and bind to the T cell factor (TCF)/lymphoid enhancer factor (LEF) transcription factors, leading to the transcription of Wnt target genes (Fig. 3) [22]. Expression of all Wnt1, GSK3β and β-catenin is significantly increased in the hearts of hypertensive rats of various aetiologies [23] while the ATP-competitive GSK3β inhibitor CHIR reverses pathological electrical remodelling in aged rats via restoring Nav1.5 and Cx43 levels at the cardiac IDs [24]. Evidence from patient hearts as well as murine and cellular ACM experimental models suggests that the Hippo pathway is also aberrantly activated in this disease [25]. Specifically, ID disruption in the presence of ACM-causing mutations causes loss of submembrane localization of protein kinase C alpha (PKCα). This in turn aberrantly activates the Hippo kinase cascade. Specifically, macrophage-stimulating protein 1/ 2 (MST1/2) phosphorylates the Large tumour suppressor kinase 1/ 2 (LATS1/2) and its scaffold protein Mps one binder 1 (MOB1). Active LATS1/2 then phosphorylates and inactivates the Yes-associated protein/ transcriptional coactivator with PDZ binding motif (YAP/TAZ) preventing it from translocating into the nucleus and binding to the transcriptional enhanced associate domain (TEAD) transcription factors (Fig. 4). Phosphorylated YAP can be driven to the IDs through binding α-catenin. However, it may also interact with the destruction complex enhancing β-catenin degradation [25].

Representation of canonical Wnt signaling. LEFT: Upon binding of Wnt ligands to Fzd/LRP5/6 receptors, Dvl recruits GSK3β, APC and axin into the destruction complex. Active GSK3β phosphorylates serine/threonine residues on the N-terminal domain of β-catenin targeting it for ubiquitination and proteasomal degradation. RIGHT: In the absence of Wnt signals, Dvl activity is inhibited by Fzd. Accordingly, the destruction complex is not formed and β-catenin is free to enter the nucleus where it interacts with the TCF/LEF transcription factors to drive expression of target genes including c-Myc and Cyclin D1. Figure created with draw.io

Representation of canonical Wnt/Hippo pathway crosstalk. Upon activation, MST1/2 phosphorylates LATS1/2 and its scaffold protein MOB1. Active LATS1/2 then phosphorylates and inactivates YAP/TAZ preventing it from translocating into the nucleus and binding to the TEAD transcription factors. There is bidirectional modulation and regulation between the pathways, where YAP/TAZ is inhibited by the Axin/APC/GSK-3β destruction complex. Figure created with draw.io

The importance of GSK3β in ACM pathogenesis was initially recognized through animal models. Notwithstanding the heterogeneity of studies in review, SB2 seems to show several key trends in limiting ACM-driven defects.

The first evidence stemmed from the high-throughput screening of a zebrafish model with cardiac-specific expression of JUP2157del2 variant, known to underlie a syndromic form of ACM in patients (Naxos disease) [3, 26]. Zebrafish ventricular CMs expressing JUP2157del2 showed a 70–80% reduction of the inward sodium current INa and the inward rectifying potassium current IK1, responsible for maintaining the resting-membrane potential. SB2 completely prevented and reversed these ionic current abnormalities [3]. These results were replicated in investigations of NRVMs transfected to express ACM-causing variants. SB2-treated ACM-NRVMs also showed restoration of localization of key proteins including plakoglobin, Cx43 and GSK3β [3, 27, 28]. Additionally, SB2 was shown to prevent and reverse redistribution of SAP97, a molecule implicated in the trafficking of Nav1.5, the potassium channel protein Kir2.1 (driving IK1) and plakoglobin to the membrane [3].

Later literature showed additional promising results with in vivo models. Chelko et al. investigated the role of GSK3β in two murine models: a DSG2 knock-in model (Dsg2mut/mut) and a transgenic mutant JUP model (JUP2157del2). SB2-treated mouse strains (onset at 3 weeks for Dsg2mut/mut mice and 6 months for JUP2157del2 mice, prior to disease manifestation) showed improved EF, reduced arrhythmic load, myocardial inflammation and fibrosis and restoration of key ID and signalling proteins (JUP, Cx43, GSK-3β, Nav1.5, SAP97) compared to vehicle-treated litter mates [4]. Of note, SB2 also significantly improved all functional parameters and reversed key protein remodelling in mice treated after disease onset [4].

Heterozygous Dsg2mut/+ mice do not show ACM hallmarks at rest. However, upon endurance exercise, deleterious re-distribution of JUP/Cx43 occurs in addition to arrhythmia development. SB2 administration prior to exercise prevented these defects [4]. Studies have shown efficacy of SB2 in preventing ACM-related abnormalities also in iPSC-CMs models. Specifically, iPSC-CMs derived from patients bearing PKP2 mutations show significantly reduced INa current densities as well as subcellular redistribution of Nav1.5, restored both by SB2 and CHIR [13]. In a related study, SB2 restored Cx43 localization, electrical coupling and calcium (Ca2+) waveforms in mutant iPSC-CM pairs derived from ACM patients bearing PKP2 variants [29]. This supports previous work showing that Ca2+ overload may contribute to the high levels of apoptosis and myocardial remodelling characterizing ACM [30].

Hamstra et al. corroborated this proposition, investigating the effects of SB2 in cytosolic Ca2+ handling [31]. The SERCA pump isoform, SERCA2a, maintains homeostasis by actively transporting cytosolic Ca2+ ions into the SR [32]. GSK3β inhibition starting at 3 weeks of age in DSGmut/mut mice demonstrated an increase in SERCA2a density/activity, in contrast to vehicle-treated mice [31]. This elucidates another potential mechanism of SB2 in preventing ACM tachyarrhythmias. Likely, this may explain overall survival rates of exercising mice reported by Chelko et al.[4]. Notably, a mouse model with postnatal CM-ANK2 deletion shows structural abnormalities reminiscent of ACM consistent with the identification of rare ANK2 variants in ACM probands (Table 1) [33]. SB2 administration at 4 weeks of age (prior to disease manifestation) led to improved EF and reduced fibrosis in the ANK2 mutant mice coupled with reduced levels of phosphorylated β-catenin [33]. Another mouse model expressing the TMEM43-S358L mutation recapitulates the human disease exhibiting CM death and severe fibrofatty replacement, preventable both by SB2 and CHIR. iPSC-CMs bearing the same mutation show marked contractile dysfunction prevented by the GSK3β inhibitor [34]. Of note, myocardial injury was independent of GSK3β pharmacological inhibitors and GSK3β levels in a mouse model of myocardial infarction highlighting differences in pathology between different heart diseases [35].

In another preclinical study, Asimaki et al. cultured buccal mucosa cells from ACM patients bearing desmosomal gene mutations. SB2 exposure of cultured cells led to restoration of Cx43/JUP signal distribution [36]. Additionally, HeLa cells expressing JUP2157del2 show a dramatic decrease of ID-localized Cx43 as well as marked microtubule disassembly, restored by SB2 [37].

Giuliodori et al. performed an in vivo cell signalling screen using pathway-specific reporter transgenes in a DSP-deficient zebrafish model. Three pathways (Wnt, TGF3β and Hippo/YAP-TAZ) were significantly altered, with Wnt being the most dramatically affected. Interestingly, under persistent DSP deficiency, the phenotype was rescuable by SB2 [38]. Furthering this work, Celeghin et al. created a DSP knock-out zebrafish line characterized by cardiac alterations, oedema and bradycardia at larval stages. Adult hearts showed reduced contractile structures, abnormally-shaped ventricles, myocardial layer thinning, adipocyte infiltration and disorganized desmosomes. Intensive physical training caused a global worsening of the cardiac phenotype accelerating disease progression. The mutant fish showed a dramatic decrease of Wnt signalling activation as well as Hippo/YAP-TAZ and TGFβ pathway dysregulation. SB2 administered at 1–3 days post fertilization rescued all pathway expression and cardiac abnormalities restoring the heart rhythm [39]. Although several of the studies cited above only examined the efficacy of SB2 in preventing ACM-related abnormalities, certain studies also showed that the GSK3β inhibitor can reverse disease phenotypes in varying experimental models. This is crucial given how potential clinical trials would primarily enrol already symptomatic patients with existing disease.

Caution should be exerted when evaluating trials using non-CM cell-types [36, 37]. There are also caveats when evaluating results from iPSC-CMs, as these cells demonstrate an immature phenotype. Structural variation may account for reduced ion densities, as immature CMs are not as polarized as adult variants, exhibiting different sodium channel distribution across the membrane [40]. Furthermore, results from ex vivo models may also pose differences too due to minimal inflammatory and hormonal influences compared to in vivo conditions.

Caution must also be exerted in light of a study published by Li et al. showing that SB2 can potentiate arrhythmic events in human cardiac slices [41]. Combined computational modelling and experimental approaches showed that the GSK3β inhibitor can decrease sodium-channel conductance and tissue conductivity underlying the observed arrhythmic phenotypes [42]. Whether or not this is due to dosing regimens, nuance at the molecular level, or modelling variance requires further discourse beyond the scope of this review. A summary of the experimental results reviewed is shown in Table 2 below.

Trials utilizing CHIR in ACM models are limited [13, 34] perhaps due to its propensity to binding other kinases at high micromolar concentrations causing collateral alterations [42]. Current literature reports no use of the GSK3β inhibitors lithium and SB4 in ACM experimental models. Caution should be raised with lithium, a pervasive mood stabilizer, as it may have abnormal electrophysiologic effects by blocking sodium channels [43]. However, the GSK3β inhibitor tideglusib, has been used in phase II clinical trials for Alzheimer’s disease [44] and myotonic dystrophy [45] while the TaRGET trial, aiming to assess its efficacy in ACM patients, launched in February 2024 [5]. Of note, tideglusib is a non-ATP competitive GSK3β inhibitor. Most kinase inhibitors are designed to bind to highly homologous ATP-binding sites, which leads to promiscuity and possible off-target effects. Allosteric inhibitors, however, exhibit high specificity and selectivity minimizing potential adverse effects [46]. Consequently, the mode of action of tideglusib alone may classify it as a superior molecule of choice as a mechanistic inhibitor of ACM.