Fibro-fatty remodelling in arrhythmogenic cardiomyopathy

Cardiomyocyte transdifferentiation into adipocytes during AC progression was first suggested based on the morphological examination of the ventricular myocardium of a female transplant patient [18]. Cardiomyocytes adjacent to fibro-fatty tissue contained sarcoplasmic vacuoles with a lipidic nature that highly resembled pre-adipocytes [18]. The authors therefore suggested a cardiomyocyte-to-adipocyte switch, as some of these cells also stained positive for vimentin, a marker expressed by adipocytes [18]. However, vimentin is a mesenchymal marker which is not exclusively expressed by adipocytes [19]. The presence of intracellular lipid droplets in cardiomyocytes has been also described in biopsied myocardial tissue of another AC patient [30]. These lipids were found in degenerating cardiomyocytes and were often discharged into the interstitial space upon cell membrane dissociation [30]. Despite the informative ultrastructural examination of explanted AC hearts, these studies were based on the histological observation of single cases.

In an in vivo setting, one report could show that some adipocytes arising in a Dsp knockout mouse model located at the sub-epicardium, but not at the midwall, originate from a cardiomyocyte lineage labelled by MLC2v [55]. Presence of a common “cardiomyocyte-adipocyte progenitor” cell population in normal hearts was further proposed in a study by Dorn et al. [24]. The authors described an Isl1⁺/Wt1⁺ progenitor cell population in normal hearts, which under different stimuli primes towards a myocytic or adipocytic fate, and hence could potentially contribute to adipocyte differentiation in AC [24].

Due to difficulties with recapitulating the fibro-fatty phenomenon in many AC mouse models, the use of hiPSCs presented an alternative tool to study the cardiomyocyte transdifferentiation hypothesis in vitro. In three subsequent reports, hiPSC-derived cardiomyocytes (hiPSC-CMs) generated from AC patients demonstrated several AC features such as reduced densities of desmosomal proteins and electrical instabilities [10, 42, 56]. Additionally, exposure of hiPSC-CMs to adipogenic stimuli induced lipid droplet accumulation, which suggested an underlying predisposition to adipocytic differentiation in AC [10, 42, 56]. However, the exposure of cells to an adipogenic environment is rather artificial and does not mimic human disease, hence the cardiomyocyte-to-adipocyte transdifferentiation theory requires further investigation.

The presence of unique progenitor cell populations in the heart with a multipotent differentiation potential has gained much attention in the recent years. Different reports have suggested that during AC disease progression, resident cardiac progenitors can differentiate into fibroblasts, adipocytes, or both. Below, we discuss the different populations proposed.

The epicardium is the outermost layer of the heart composed of mesothelial cells that mainly remain quiescent in adult hearts. During development and after disease, epicardial cells can undergo epithelial-to-mesenchymal transition (EMT) giving rise to different cardiac cell populations [34, 70]. Given their multipotent cell potential [34, 70], the sub-epicardial predominance of fibro-fatty infiltrates in AC patients [45, 76], and the high epicardial expression of desmosomal genes [43, 60], epicardial cells have been also suggested as candidates to originate fibro-fatty tissue.

In the normal murine heart, fat tissue is often confined at a specific region named the atrial–ventricular (AV) groove [82]. A study by Yamaguchi et al. demonstrated that the AV groove fat originates from the epicardium through the EMT and PPARγ pathway activation [82]. Later, using lineage tracing on ischemic injury models, adipocytes emerging at the peri-infarct region following MI were shown to also partially derive from the adult epicardium [51, 86]. Additionally, epicardial-to-fibroblast differentiation due to adult epicardial EMT has been also suggested in models of MI [26, 68, 80].

Interestingly, in studies on atrial fibrillation, which is characterized by the extensive atrial remodelling with fibro-fatty tissue, fibroblasts and adipocytes were shown to arise from the epicardium [74]. This appeared to be due to a pre-programmed state of subsets of adult epicardial-derived cells (EPDCs) towards either fibroblast or adipocyte cell fates, which when activated undergo fibro-fatty differentiation and infiltration into the diseased atria [74].

In the context of AC, Matthes et al. were the first to show that epicardial explants from neonatal rat hearts express PKP2, which when silenced promotes cellular proliferation, migration, lipogenesis and cellular differentiation into α-SMA⁺ cells [60]. These data suggested an important role of the desmosome in maintaining the mechanical integrity of epicardial cells, which when lost could potentially promote cellular differentiation.

In addition to the previously discussed studies from Marian’s group suggesting the contribution of different progenitors to fibro-fatty differentiation [52‐54], they could recently also demonstrate the role of the epicardium in AC pathogenesis [85]. Using a reporter mouse model carrying an inducible epicardial-specific Dsp deletion, the authors could show the epicardial origin of fibroblasts [85]. These cells were shown to express paracrine factors such as TGFβ1 and FGF, which mediate EMT as well as apoptosis, arrhythmias and cardiac dysfunction [85].

In a recently published report, our group further underscored the role of the epicardium in fibro-fatty remodelling. [43]. In this study, we made use of hiPSC-derived cardiac cultures generated from AC patients, their isogenic controls and healthy donors to study AC pathogenesis in vitro. We could show that hiPSC-derived epicardial cells undergo EMT and spontaneous fibro-fatty cellular differentiation upon desmosomal gene suppression, due to either intrinsic mutations or targeting siRNAs. Using single cell RNA sequencing, we identified transcription factor TFAP2A to mediate this process by enhancing EMT signalling in the diseased cells. Furthermore, we observed that human AC hearts display increased epicardial thickening and WT1 expression indicative of epicardial activation. Additionally, cells located at the sub-epicardial mesenchyme stained positive for WT1 and TFAP2A, which further suggested that an epicardial-derived subset of cells underlies the fibro-fatty phenotype.

Various studies have tried elucidating the cellular origin of fibroblasts contributing to scar tissue formation in different forms of cardiovascular disease. Although not particularly studied in AC, these identified fibroblast progenitor populations might also potentially contribute to fibrosis in AC patients and hence are also discussed below.

All resident fibroblasts in the adult heart derive from the epicardium during development and remain quiescent under homeostatic conditions [31]. Several reports could demonstrate the presence of cardiac fibroblast progenitor cell populations which upon injury can contribute to the pathological fibroblast differentiation. In one study, an adult stem cell population labelled by PW1 was found to contribute to fibrotic remodelling after MI [83]. PW1⁺ cells were mainly found near infarct areas of human and mouse ischemic hearts and could give rise to fibroblasts labelled by FSP1 and α-SMA [83]. In another study, peri-vascular precursors, labelled by GLI1 expression, were also shown to contribute to fibrosis after injury in different organs including the heart [44]. Lineage tracing experiments showed that 60% of infarct fibroblasts arise from GLI1 + vascular progenitors [44].

Bone marrow-derived cells have been also suggested to differentiate into cardiac fibroblasts, although the contribution seemed to vary between different cardiac disorders. In different studies, mice were transplanted with genetically labelled or sex-mismatched bone marrow cells which were shown to significantly contribute to scar tissue fibroblasts after MI [61, 79] and in dilated cardiomyopathy [15].

Another study described a circulating population of leukocytes termed fibrocytes as a source of injury fibroblasts in models of angiotensin-II-induced cardiac hypertrophy [38] and ischemic cardiomyopathy [39]. These cells normally express both hematopoietic markers (CD34, CD45) and fibroblast markers (procollagen I, vimentin) [48]. Upon stimulation, fibrocytes are recruited to the injury site where they adopt a myofibroblast phenotype expressing α-SMA and produce collagen [48].

In contrast to these reports, other lineage-tracing studies could show that scar tissue fibroblasts mainly originate from pre-existing resident epicardium-derived fibroblasts and not from bone marrow or stem cell populations. This was observed in models of pressure overload [1, 62] and MI [29, 41, 67, 84].

Despite the potential of the described reports to identify the origin of fibroblasts in different cardiac disorders, it is unknown whether these findings also apply to AC where fibroblasts infiltrate massive regions of adipose tissue.

Canonical WNT signalling is the most commonly accepted pathway involved in AC pathogenesis. β-catenin, together with PKG, also known as γ-catenin, acts to link cadherin proteins to the actin cytoskeleton [88]. In response to WNT ligands, intracellular β-catenin levels are stabilized leading to its translocation to the nucleus, where it activates target genes via binding to TCF/LEF1 transcription factors [88]. Since PKG functions as a constituent of the desmosome and has been also shown to inhibit the transcriptional activity of β-catenin [64, 88], WNT signalling has been suggested to be implicated in AC pathogenesis.

Aberrant WNT signalling in AC was first reported by Garcia-Gras et al. in 2006 [32]. The authors showed that loss of DSP expression in cultured atrial myocytes or in mice leads to the nuclear translocation of PKG, where it competes with β-catenin for binding to TCF/LEF1 and therefore inhibits β-catenin-mediated WNT signalling [32]. Since WNT signalling inhibition is known to promote adipogenesis [65], the authors suggested PKG nuclear translocation as an underlying mechanism to adipocyte differentiation in AC [32]. Reduced WNT signalling activity has been further demonstrated in other AC mouse models overexpressing mutant forms of Jup [53] and Dsg2 [8] and in Dsp-deficient zebrafish [35]. However, conflicting to these data, cardio-restricted Jup knockout was shown to increase β-catenin stabilization and TCF/LEF transcriptional activity, which was suggested to underlie a cardiac hypertrophic phenotype in these mice with no signs of cardiac adipogenesis [47]. In another cardiac-specific Jup knockout mouse model, WNT signalling was shown to be unaltered [46]. In addition to its involvement in adipogenic remodelling, WNT signalling has been also linked to cardiomyocyte apoptosis and electrical instabilities [2, 37], suggesting a broader role for WNT signalling in AC pathogenesis. However, further investigation is needed to understand the causal mechanisms underlying WNT pathway dysregulation in AC patients.

The Hippo pathway is well known for its roles in regulating cellular proliferation and apoptosis, thereby controlling growth of tissues [81]. Particularly in AC, a study showed that in human AC hearts, as well as in cell culture and mouse models of AC, Hippo signalling was activated [13]. This led to the phosphorylation of downstream Hippo molecules including the Hippo-effector molecule Yes-associated protein (YAP), which was found to interact with β-catenin and PKG, ultimately enhancing the adipogenic phenotype [13].

TGFβ signalling is key in regulating cardiac fibrosis and hence has been linked with fibrotic remodelling in AC [23, 49]. The association of TGFβ signalling with AC was first demonstrated by the identification of mutations in the UTR regions of TGFβ3 in two familial cases of AC, which were shown to increase the activity of TGFβ by twofold [7]. Furthermore, in a Jup knockout mouse model, which showed massive fibrosis and no alterations in WNT signalling, TGFβ signalling was significantly induced [46]. The authors suggested that the elevated TGFβ activity could be contributed to an increase in myocardial wall stress as a result of desmosomal loss [46]. Interestingly, the increase in TGFβ activity has recently also been validated in plasma collected from AC patients, which coincided with an increase in fibrotic markers in endomyocardial biopsies [57]. In this study, the authors found that MSCs isolated from AC patients were more prone to fibrotic differentiation in response to TGFβ1 treatment compared to healthy donors [57]. In another study, knockdown of Pkp2 in cardiomyocytes induced a TGFβ-mediated induction of pro-fibrotic genes [27]. This effect was corroborated in tissues isolated from Pkp2 and Dsp-deficient mouse and zebrafish models [27, 35]. In addition to its role in fibrotic differentiation, TGFβ is known for its crucial roles in promoting EMT signalling in the heart which can precede fibrotic remodelling [6, 25]. Recently, we could show that hiPSC-epicardial cells and primary human atrial EPDCs undergo excessive EMT in response to TGFβ1 treatment or desmosomal suppression [43]. This has been further corroborated in an epicardial-specific Dsp knockout mouse model which demonstrated an epicardial-derived origin of fibroblasts due to enhanced EMT and TGFβ signalling [85]. However, the exact link between AC-related mutations and EMT remains to be elucidated.

PPARγ is a nuclear receptor that functions as a master regulator of lipid uptake and adipogenesis [11]. Induction of PPARγ signalling together with increased lipogenesis have been described in different models of AC. These include Dsp- and Pkp2-deficient cardiomyocytes [24, 32, 42] and epicardial cells [43] as well as human AC hearts [22]. The exact mechanisms by which desmosomal dysregulation alters PPARγ signalling are not fully understood. However, one possible link could be the negative regulation of PPARγ by β-catenin [50], which is widely suggested to be inhibited in AC [8, 32, 35, 53].