Novel pathogenic role for galectin-3 in early disease stages of arrhythmogenic cardiomyopathy

BACKGROUND Arrhythmogenic cardiomyopathy (AC) is a myocardial disease due to desmosomal mutations whose pathogenesis is incompletely understood. OBJECTIVE The purpose of this study was to identify molecular pathways underlying early AC by gene expression pro ﬁ ling in both humans and animal models. METHODS RNA sequencing for differentially expressed genes (DEGs) was performed on the myocardium of transgenic mice overexpressing the Desmoglein2 -N271S mutation before phenotype onset. Zebra ﬁ sh signaling reporters were used for in vivo validation. Whole exome sequencing was undertaken in 10 genotype-negative AC patients and subsequent direct sequencing in 140 AC index cases. RESULTS Among 29 DEGs identi ﬁ ed at early disease stages, Lgals3/ GAL3 (lectin, galactoside-binding, soluble, 3) showed reduced cardiac expression in transgenic mice and in 3 AC patients who suffered sudden cardiac death without overt structural remodeling. Four rare missense variants of LGALS3 were identi ﬁ ed in 5 human AC pro-bands. Pharmacologic inhibition of Lgals3 in zebra ﬁ sh reduced Wnt and transforming growth factor- b signaling, increased Hippo/YAP-TAZ signaling, and induced alterations in desmoplakin membrane localization, desmosome integrity and stability. Increased LGALS3 plasma expression in genotype-positive AC patients and CD98 activation supported the galectin-3 (GAL3) release by circulating macrophages pointing toward the stabilization of desmosomal assembly at the injured regions. CONCLUSION GAL3 plays a crucial role in early AC onset through regulation of Wnt/ b -catenin signaling and intercellular adhesion. open access article under the BY (http:// galectin-3(GAL3)reduction at the intercalated discs in comparison with control samples and end-stage AC myocardium. Massive in ﬂ ammatory response in acute-stage disease is characterized by CD68 1 and CD98 1 cells, whereas in the early-stage disease CD98 1 cells, designating alternatively activated macrophages, are completely missing, similar to the control samples (Ctrl). SD 5 sudden death; Tx 5 transplantation.


Introduction
Arrhythmogenic cardiomyopathy (AC) is an inherited heart muscle disorder that causes life-threatening arrhythmias and cardiac sudden death, particularly in the young and in athletes. [1][2][3][4] The inheritance pattern of the disease is mostly autosomal dominant with low penetrance. [5][6][7] Despite recent advances in AC research, the molecular and cellular mechanisms leading from an aberrant desmosomal protein to myocardial tissue replacement remain incompletely understood.
Animal-based studies previously demonstrated Wnt/bcatenin inhibition leading to translocation of junction plakoglobin (JUP) protein from intercalated discs to the nucleus, where it acts as a b-catenin competitor. 8 Wnt/b-catenin signaling suppression was also confirmed by the subsequent identification of a glycogen synthase kinase-3b (GSK3b) inhibitor, which rescued the AC phenotype in zebrafish models, and by the discovery of aberrant Hippo/YAP signaling pathway activation leading to b-catenin cytoplasmic sequestration and JUP nucleus translocation. [9][10][11][12] However, abnormal JUP nuclear translocation and decreased b-catenin activity are not sufficient to reproduce in vitro the AC phenotype. 13 A systematic study in a transgenic AC mouse model was adopted to identify differentially expressed genes (DEGs) during the early disease stage. 14 We found the galectin-3 (LGALS3/GAL3) gene as a regulator of Wnt/b-catenin signaling pathway in early stages and validated its effects in zebrafish reporter lines. Mechanical and signaling functions of this molecule likely link cell-to-cell stabilization to myocardial injury, inflammation, and fibrogenesis, thereby constituting a potential biomarker of AC. Accordingly, pharmacologic inhibition of Lgals3 in zebrafish reporter lines reduced Wnt signaling and induced alterations in desmoplakin (DSP) membrane localization. Thus, our findings demonstrate a potential role for GAL3 in early AC stages.

Tissue samples collection
Myocardial samples of 4 transgenic mice without cardiac structural changes overexpressing the desmoglein-2 (dsg2) N271S mutation (TgNS,2wks) were collected. Agematched, wild-type (WT) FVB/N strain littermates and overexpressing WT dsg2 (TgWt) mice were used as controls to exclude interference of transgene expression. 14 Paraffin-embedded heart tissue from genotype-positive AC patients was used for experimental validation, including patients who died in the concealed, acute, and overt phases of the disease. Control samples were derived from patients who died of trauma.
Human, zebrafish, and mouse samples were collected according to government and institutional guidelines. The protocols were approved by the University Hospital ethical committee (University of Padua) and the Animal Experimentation Commission of the Academic Medical Center (University of Amsterdam), respectively. All animal experiments conformed to the guidelines from Directive 2010/63/EU of the European Parliament on the protection of animals used for scientific purposes and adhered with the ARRIVE Guidelines for the care and use of laboratory animals. Human tissue studies were in compliance with the Declaration of Helsinki as revised in 2013.

Gene expression and ontology analysis
Transcriptome assembly and annotation, quantitative polymerase chain reaction, and western blot validation methods are described in the Supplemental Materials and Methods.
Transcript abundance estimation was performed using Cufflinks Version 2.2.1 (open source software). For differential expression analysis at the gene level, DESeq2 (Bioconductor Version 1.6.3; open source software) with false discovery rate cutoff of 0.05 and log2 fold-change (FC) cutoff of Log 2 FC 1 2, Log 2 FC 2 -2 was used.
Gene ontology (GO) and molecular pathway analysis were carried out with the "Panther classification system" to analyze the main function of DEGs and to identify enriched biological themes within a gene dataset. Predicted proteinprotein interaction networks based on DEGs were produced using STRING: Protein-Protein Interaction Networks (http://string-db.org/). Network statistics were obtained using the NetworkAnalyst tool (http://www.networkanalyst.ca/). Zebrafish lines, chemical treatment, and wholemount RNA in situ hybridization Zebrafish (Danio rerio) embryos and adults were raised, staged, and maintained at the Zebrafish Facility of the University of Padua, under standard conditions and according to national laws and EU regulations on animal welfare. In vivo analyses of Wnt, transforming growth factor-b (TGFb), and Hippo/YAP-TAZ signaling and DSP protein detection are described in the Supplemental Materials and Methods.

Genetic testing
Whole exome sequencing was performed in 10 selected AC patients negative for disease-related gene variants using the SureSelect Clinical Research Exome V2 kit (Agilent, Santa Clara, CA). Sanger sequencing of exon and exon-intron boundaries of LGALS3 (lectin, galactoside-binding, soluble, 3; NM_002306.3) was performed on 140 index cases.
These 29 DEGs are mainly involved in cellular (27.3%, GO: 0009987, P 5 .0434) and metabolic (24.2%, GO: 0008152, P 5 .000444) processes, and response to stimulus (9.1%, GO: 0050896, P 5 9.68E-8). This gene set was not of sufficient size to obtain significant functional enrichment by pathway analysis; therefore, DEGs were manually characterized based on literature search. The gene Lgals3, which was found to be significantly decreased in TgNS compared to Wt and TgWt hearts (Log 2 FC 5 -4.01, P 5 1.38E-04), is engaged in different biological events, tissues and plays a relevant role in cardiovascular diseases among other clinical conditions. 15

GAL3 quantification in TgNS mice and AC human specimens
We hypothesized an early GAL3 role in disease pathogenesis, as previous studies showed that GAL3 directly binds to DSG2 and enhances intercellular adhesion, 16

and that
Lgals3 regulates GSK3b activity leading to Wnt signaling suppression. 17 Validating the RNA-Seq results, quantitative polymerase chain reaction analysis of ventricular tissue demonstrated an approximately 4-fold reduction in Lgals3 mRNA expression in TgNS,2wks compared to Wt and TgWt of the same age. In contrast, no differences were observed in mice aged .3 weeks ( Figure 1A). Western blot analysis confirmed that the observed decrease in Lgals3 mRNA levels was accompanied by a significant reduction in GAL3 protein expression in ventricular tissue of TgNS,2 wks mice.
To extend these findings to human AC, we investigated GAL3 expression in paraffin-embedded heart tissue from control patients and genotype-positive patients with early and late disease stages. Immunohistochemical staining demonstrated GAL3 labeling in the intercalated discs of cardiomyocytes ( Figure 1B).
Compared to control tissue, GAL3 expression was reduced in tissue from 3 young DSP and JUP carriers (7, 14, and 15 years, respectively) who suffered sudden death at early disease stages (ie, in the absence of fibrofatty replacement). In contrast, no significant changes of GAL3 expression were detected in patients with advanced AC stages ( Figure 1B).
CD98 is a glycoprotein highly expressed in macrophages when alternatively activated by GAL3 binding. In the absence of GAL3 (early disease stages), no CD981 cells were observed, whereas during acute inflammatory response and in more advanced disease stages CD98 1 macrophages were found ( Figure 1B).

Wnt signaling inhibition following GAL3 inhibition in zebrafish
Previous studies have demonstrated a link between Lgals3 and GSK3b activity/Wnt signaling regulation, 17 with aberrant GSK3b distribution and Wnt suppression apparently key to explaining disease progression. 10 To explore the potential functional consequences of reduced Lgals3/ GAL3 expression, we inhibited GAL3 pharmacologically in zebrafish using the compound TD139, an inhibitor targeting the glycan binding pocket of GAL3 with high binding affinity for the arginine residues (Arg48 and Arg73) of the unique carbohydrate-recognition domain (CRD) of GAL3. 18 The in vivo impact of TD139 on Wnt signaling was investigated in a transgenic Wnt reporter zebrafish line. TD139 administration into 2-and 3 day-old embryos led to reduced Wnt signaling activity as indicated by a significant decrease in Wnt-responsive fluorescent signal in the cardiac region comparable with that obtained by administration of Xav, a known Wnt signaling antagonist ( Figure 2).
Reduced Wnt signaling also has been linked to TGFb and Hippo/YAP-TAZ signaling alterations. 11 In this setting, transgenic reporter zebrafish lines were treated by TD139 showing inhibition of GAL3 activity, reduced TGFb, and increased Hippo/YAP-TAZ signaling, confirming the interconnection among these pathways ( Figure 3).

Alterations in desmosome structure following GAL3 inhibition in zebrafish
We next evaluated the distribution of Dsp after TD139 treatment. Two-day-old TD139-treated zebrafish embryos displayed abnormal Dsp localization in epidermal cells compared to controls. In particular, in treated embryos, Dsp was not equally distributed in the whole cell membrane but showed a focally reduced or absent signal ( Figure 4B, arrows).
Desmosome integrity and localization were evaluated through ultrastructural investigation (Figure 4). In the epidermis of TD139-treated embryos, desmosomes appeared disconnected from the cytoskeletal complex and not regularly distributed ( Figure 4E, arrows). Moreover, internal and external desmosomes plaques displayed a "pale" phenotype ( Figure 4E, arrowhead) compared to controls. Taken together, these observations indicate a functional role for Gal3 in mediating proper desmosomal structure and integrity, as well as colocalization with desmosomal proteins.
LGALS3 expression in blood samples of AC patients To determine the potential of LGALS3 as a biomarker, we assessed a series of 40 AC patients who fulfilled the Task Force Criteria (TFC) 5 and carried desmosomal pathogenic/likely pathogenic variants. Significantly increased levels of LGALS3 were found in the blood of patients compared to controls (Log 2 FC 5 5.02, P 5 1.50E-03), confirming GAL3 recruitment during healing processes as reported in a previous study. 19 LGALS3 rare genetic variants in AC index cases To determine whether genetic variants in LGALS3 may contribute to AC, we performed genetic sequencing in a cohort of 150 AC index cases identifying 5 LGALS3 missense variant carriers (4%; 4 males; mean age 39 6 11 years) ( Table 2). All 4 LGALS3 rare missense variants were highly conserved among species (with maximum and positive PhastCons and PhyloP scores, respectively) and predicted by in silico tools to affect the protein structure. Of note, we found the rare missense variant c.485G.A, which occurred in the unique CRD of GAL3 conferring the loss of its function ( Figure 5).
Ad hoc assessment of the genetic burden in LGALS3 demonstrated a large excess of rare nonsynonymous variants among our AC cohort compared to the general population ( Table 3). The analysis highlighted a 5.8-fold enrichment of LGALS3 rare variants in our AC patients (Fisher exact test, P 5 2.10E-03).

Functional analysis of the LGALS3 c.485G.A variant in zebrafish
To assess the in vivo impact of the rare missense variant c.485G.A located in the unique CRD of GAL3, we injected human WT LGALS3 mRNA in zebrafish embryos, which caused growth delay, dysmorphology, and increased mortality compared to control-injected sibs. This suggests

Discussion
Transcriptome profiling in AC mutant mice preceding structural abnormalities 14 displayed decreased Lgals3 cardiac expression encoding for GAL3, a member of the galectin family, which is thought to have a dual positive or negative regulatory function during the assembly and stabilization of intercellular junctions. 20 Indeed, we observed a different expression of Lgals3/GAL3 in the early vs the late stage of the disease. Subsequent genetic analysis of a larger AC cohort highlighted LGALS3 rare variants in 4% of cases, corresponding to a significant 5.8-fold increase of LGALS3 rare variants in affected patients compared to the control population.

Intercellular adhesion and Gal3
Few studies to date have investigated disease pathways in the early stages of AC. Previous ultrastructural studies in AC patients and mechanotransduction in in vitro assays  indicated that abnormal cell-cell mechanical coupling, exacerbated by strenuous exercise, plays an important pathogenic role. 14,21,22 In our study, transcriptional data obtained from TgNS,2wks mice exhibited decreased expression of Lgals3, which is involved in a wide array of cellular activities, such as regulation of cellular trafficking of glycoproteins, signaling, and cell adhesion. 15 GAL3 is composed of a 21-amino-acid-long N-terminal stretch with 2 sites for serine phosphorylation, followed by 9 non-triple-helical collagen-like Pro/Gly-rich repeats, which harbor cleavage sites for diverse proteases, and the C-terminal CRD with a conserved sequence motif that confers affinity for b-galactoside-containing glycans. GAL3, as do other galectins that mediate signal transduction events on the cell surface, 23 enhances intercellular adhesion through CRD direct binding to the extracellular domain of cadherins and specifically with DSG2, as previously demonstrated in experiments in intestinal, corneal, and conjunctival epithelia. 16,[24][25][26] Also, GAL3-reactive sites colocalize with the DSP and DSG proteins in primary squamous carcinomas, pointing to the potential role of GAL3 in mediating intercellular contacts. The role of galectins as agonists or antagonists of adhesion depends on the expression level and location of these lectins, the glycosylation pattern of the receptors, the binding valence  of the interaction, and the repertoire of receptors clustered by these lectins. 20 GAL3-inhibited zebrafish embryos displayed abnormal and irregularly distributed desmosomes with "pale" internal and external plaques, as observed in previous studies in humans. 21 Thus, our findings underscore a role for GAL3 in modifying desmosomal ultrastructure, with reduced cellcell adhesion as a consequence of diminished GAL3 levels possibly leading to cardiomyocyte injury and death.

Intercellular adhesion and Wnt signaling suppression
Aberrant GSK3b distribution has been proposed to explain morphofunctional and arrhythmic features of the disease progression. 10,27 Lgals3 is an upstream regulator gene of GSK3b. Recent studies demonstrated that lower expression of Lgals3 leads to reduced levels of phospho-GSK3b at serine 9, thus increasing GSK3b activity, and phospho-GSK3b reduction leads to phosphorylation of b-catenin and its subsequent degradation, suppressing Wnt signaling. 17 To prove this concept, Gal3 expression was pharmacologically inhibited in a Wnt reporter zebrafish line showing clearly suppression of the Wnt signaling and reduced expression levels of downstream targets. Taken together, our findings indicate that suppression of Wnt/b-catenin signaling mediated by a desmosomal gene defect downregulates Lgals3 expression at early disease stages, specifically in the intercalated disc. Through a feedback loop, this reduced GAL3 expression triggers additional desmosomal instability and inflammation, leading to a compensatory increase in systemic GAL3 mediated by alternative macrophage activation CD98 1 during the acute inflammatory phase in an effort to counterbalance the destabilization of the desmosomal complex. GAL3: Inflammation, wound repair and heart failure GAL3 was previously reported as the link between macrophages and fibrosis. 15 Myocardial injury generates inflammatory processes that stimulate GAL3 secretion, which leads to interleukin-4-induced alternative macrophage activation, upregulation of profibrotic genes stimulating matrix production, enhanced wound healing through scar production, and ultimately cardiac remodeling. 28,29 Overexpression and secretion of GAL3 are associated with several diseases and have been extensively studied in the context of fibrosis and heart failure. 19,30,31 However, little is known, and often with contradictory findings, on the mechanism responsible for GAL3 inhibition or downregulation. 32 In this context, our experiments have demonstrated that overexpression of Lgals3 in an in vivo model causes increased mortality rates typical of heart failure Wnt dysregulation effects, 30,31 whereas inhibition of GAL3 expression causes aberrant desmosome assembly. 16,20 Overall, the effects of GAL3 at the myocardial level may depend on its dual positive or negative regulatory function during the assembly and stabilization of intercellular junctions.
In our series of 40 AC patients with definite TFC and genotype-positive for desmosomal mutations, we found expression levels of LGALS3 significantly increased in blood, thus confirming recruitment of GAL3 through inflammatory cells (monocytes, macrophages) in healing and fibrogenic processes. 26,28,32 We further showed the myocardial recruitment of CD98 1 macrophages at the acute inflammatory disease phase, supporting our hypothesis that upon cardiac reduction of GAL3, circulating macrophages release GAL3 at the injured region and pointing toward the stabilization of desmosomal assembly. As a whole, our data suggest that GAL3 cardiac reduction triggers compensatory GAL3 production, which activates CD98 and alternative macrophage activation by a GAL3 feedback loop mechanism. GAL3/LGALS3 compensatory serum and tissue levels have been also related to dilated cardiomyopathy and myocarditis and very recently its release from the heart into the circulation has been proven. [33][34][35][36] Circulating LGALS3 levels may not be disease-specific but probably reflect in general healing and fibrotic conditions. 37 The transitional absence and subsequent presence of GAL3 have also demonstrated in other conditions (eg, epidermis lesions). 38 In order to explore complement system activation and the role of myocardial GAL3 expression during the early stage of AC, a time-course analysis of cardiac and serum GAL3 in animal models should be undertaken in future studies.
LGALS3 rare genetic variants in AC index cases Whole exome sequencing and direct sequencing in 150 AC index cases with definite TFC highlighted 4% of LGALS3 rare genetic variants. Genomic comparison among species showed that LGALS3 rare missense variants were highly conserved and predicted to significantly affect protein structure conferring the loss of GAL3 function. In this regard, our data demonstrated a possible association between low expression levels of LGALS3 and loss of function rare missense variants on key functional domains. Specifically, we demonstrated no effects on disease status when zebrafish embryos overexpressed c.485G.A LGALS3 compared to their WT counterparts; instead, overexpression of the WT LGALS3 in zebrafish leads to increased mortality rates, reduced development, and dysmorphology. These findings support further that LGALS3 genetic variants confer the complete loss of protein function, which ultimately impacts Wnt signaling, decrease intrinsic cell adhesion reserve, and eventually increase susceptibility to AC. Nevertheless, the causal role of LGALS3 rare variants in AC pathogenesis should be supported by cascade family screening.

Study limitations
Several molecules have been developed as GAL3 inhibitors, but TD139 currently is being examined in several clinical trials, and its specific interaction with the CRD binding site of GAL3 has been thoroughly demonstrated. Although no direct readout to measure the degree of GAL3 inhibition by TD139 was developed, our data and that of other investigators demonstrated that Lgals3 downregulation and Wnt suppression mutually occur. 17,37 As such, we can confidently state that TD139 suppresses Wnt signaling. The link with recently reported inflammatory biomarkers such as anti-DSG2 antibodies should be undertaken. 39

Conclusion
Our data support a potential pathogenic role of LGALS3 in the myocardial injury of AC mediated by destabilization of desmosomes. Reduced Lgals3 myocardial expression levels in prephenotypic disease stages suppress Wnt signaling pathway and activate an excess of proinflammatory processes by recruitment of circulating GAL3. Understanding the role and function of this molecule in disease pathogenesis might enable identification of high-affinity and selective galectin inhibitors to regulate inflammatory processes and disease progression.