The hippo pathway is activated and is a causal mechanism for adipogenesis in arrhythmogenic cardiomyopathy

Abstract

Rationale: Mutations in the intercalated disc proteins, such as plakophilin 2 (PKP2), cause arrhythmogenic cardiomyopathy (AC). AC is characterized by the replacement of cardiac myocytes by fibro-adipocytes, cardiac dysfunction, arrhythmias, and sudden death.

Objective: To delineate the molecular pathogenesis of AC.

Methods and results: Localization and levels of selected intercalated disc proteins, including signaling molecules, were markedly reduced in human hearts with AC. Altered protein constituents of intercalated discs were associated with activation of the upstream Hippo molecules in the human hearts, in Nkx2.5-Cre:Dsp(W/F) and Myh6:Jup mouse models of AC, and in the PKP2 knockdown HL-1 myocytes (HL-1(PKP2:shRNA)). Level of active protein kinase C-α isoform, which requires PKP2 for activity, was reduced. In contrast, neurofibromin 2 (or Merlin), a molecule upstream of the Hippo pathway and that is inactivated by protein kinase C-α isoform, was activated. Consequently, the downstream Hippo molecules mammalian STE20-like protein kinases 1/2 (MST1/2), large tumor suppressor kinases 1/2 (LATS1/2), and Yes-associated protein (YAP) (the latter is the effector of the pathway) were phosphorylated. Coimmunoprecipitation detected binding of phosphorylated YAP, phosphorylated β-catenin, and junction protein plakoglobin (the latter translocated from the junction). RNA sequencing, transcript quantitative polymerase chain reaction, and reporter assays showed suppressed activity of SV40 transcriptional enhancer factor domain (TEAD) and transcription factor 7-like 2 (TCF7L2), which are transcription factors of the Hippo and the canonical Wnt signaling, respectively. In contrast, adipogenesis was enhanced. Simultaneous knockdown of Lats1/2, molecules upstream to YAP, rescued inactivation of YAP and β-catenin and adipogenesis in the HL-1(PKP2:shRNA) myocytes.

Conclusions: Molecular remodeling of the intercalated discs leads to pathogenic activation of the Hippo pathway, suppression of the canonical Wnt signaling, and enhanced adipogenesis in AC. The findings offer novel mechanisms for the pathogenesis of AC.

Keywords: Wnt signaling pathway; adipogenesis; cardiomyopathies; genetics.

Figure 1. Molecular remodeling of intercalated discs (IDs) in human AC

A. Immunofluorescence (IF) panels showing reduced levels and localization of selected ID proteins; B. Immunoblots (IB) showing reduced levels of selected ID proteins in the human AC; C. Bar graphs showing relative levels of the ID proteins. (N=2 for normal hearts and N=4 for AC hearts, * p<0.05; ** p=0.01, *** p<0.001.

Figure 2. Activation of the Hippo pathway in the human hearts with AC

A. IB panels and quantitative data (B), showing increased levels of NF2 (total and phosphorylated form) and the effector of the Hippo pathway pYAP in the human hearts with AC (N=4 for each group, *p<0.05). Level of pTAZ (a co-activator of YAP) was unchanged. C. IF panels showing localization of pNF2 to the ID in the normal hearts but not in the hearts with AC. The Hippo pathway components pMST1 and LATS2 had a diffuse distribution in the human heart. Likewise, pYAP was predominantly localized to cell junctions in the hearts with AC. D. IF panels showing impaired localization of pNF2 to the ID and increased pYAP at the epicardium in the human hearts with AC.

Figure 3. Activation of the Hippo pathway in the hearts of mouse models of AC

A. IB of cardiac protein extracts from Nkx2.5-Cre:Dsp^W/F and Myh6:Jup mouse models of AC showing increased NF2 and reduced pNF2 levels and the corresponding increase in the pYAP level in the heart, as compared to non-transgenic (NTG) mice. B. Quantitative data for total NF2, pNF2 (inactive) and pYAP levels in the mouse models of AC as compared to controls (N=2, *p<0.05, **p<0.002, *** p <0.0001). C. IF panels of mid myocardial sections from Nkx2.5-Cre:DspW/F and Myh6:Jup mice showing localization of pNF2, pMST1, and pYAP in the control and mouse models of AC. As observed in the human hearts with AC, pYAP and pNF2 are predominantly localized to cell membrane. D. IF panel from the epicardial region of the heart in mouse models of AC showing distinct localization of pNF2, pMST1 and pYAP to the cell junction.

Figure 4. Activation of the Hippo pathway in the HL-1 myocytes upon knockdown (KD) of plakophilin 2 (PKP2)

A–C. Quantitative PCR (qPCR), IF and IB data showing KD of PKP2 mRNA and protein in the HL-1 myocytes (HL-1^PKP2:shRNA). PKP2 mRNA and protein were reduced by 60 to 80% (N= 6 for mRNA and N=4 for protein, *p<0.0001). D. Volcano plot of RNA-Seq data, demarcated at q<0.05. E. Heat plot of YAP-TEAD targets constructed from the RNA-Seq data and analyzed by the IPA (q<0.05). F–G. Quantitative data for selected low abundant (F) and high abundant (G) YAP-TEAD targets identified by RNA-Seq and represented as transcript per million reads (N=2, q<0.05). H. qPCR data showing reduced level of selected Hippo pathway downstream targets (N=19 for Ccnd1, N=12 for Ctgf, N=3–7 for other genes, *p<0.05, **p<0.005, ***p<0.0001). I. IB analysis of Hippo pathway components. Top panel shows Phos-tag blot for NF2 showing increased active (non-phosphorylated) NF2 level. Increased pMST1, pLATS1 and pYAP and reduced RAC1 and pERK44/42 levels, the latter two negatively regulated by the Hippo pathway. J. Phos-tag blot showing increased pYAP levels in HL-1^PKP2:shRNA cells. K. IF panels largely corroborating the IB findings. L. TEAD-luciferase reporter assay showing decreased transcriptional activity of TEAD transcription factor (N=3, *p<0.005).

Figure 5. Inactivation of PKC-α and activation of NF2 in AC

A. IF panels showing co-localization of PKP2 and pPKC-α in the normal human heart sections to the IDs, whereas localization of PKP2 and pPKC-α to IDs was reduced or absent in the human hearts with AC. B. IB analysis of cardiac proteins showing reduced pPKC-α in the human hearts with AC (61 ± 16%, N=2 for normal hearts and 4 for hearts with AC, p=0.007). C. IF panels showing localization of PKP2 and pPKC-α to the IDs in the NTG mouse hearts but substantially reduced localization of pPKC-α to the IDs in the mouse models of AC (Nkx2.5-Cre:DspW/F: 57± 5%, N=2, p=0.003; and Myh6:Jup: 70 ± 5%, N=2, p=0.002). D. IB of cardiac proteins showing reduced pPKC-α levels in the mouse models of AC as compared to NTG. E. IB showing reduced PKP2 and pPKC-α levels in the HL-1^PKP2:shRNA myocytes. F. IF panels showing reduced pPKCα levels in HL-1^PKP2:shRNA myocytes at the baseline, as compared to control HL-1 cells. G and H. Differential (reduced) activation of pPKC-α in the HL-1^PKP2:shRNA myocytes upon treatment of cells with two independent PKC activators Phorbol-Myristate-Acetate (PMA) and Oleoyl-2-Acetyl-sn-Glycerol (OAG). I. IB showing a differential increase in the pPKC-α level in the HL-1 and HL-1^PKP2:shRNA myocyte upon activation with PMA and OAG at two different time points. DMSO treated samples were used as control for the solvent. Corresponding changes in the level of pNF2 upon activation of pPKC-α is also shown for each set of experiments.

Figure 6. Suppression of the canonical Wnt signaling in the human hearts with AC and the mouse models of AC

A. IB panels showing expression of phosphorylated and total β-catenin in the control and human hearts with AC. B. Quantitative data (* p≤0.01, N=4 for each group). C. IF panels showing mislocalization of the β–catenin in the human hearts with AC. Nuclear localization of CTNBB1 was not specifically analyzed. D–E. IB and the quantitative data showing increased pβ-catenin level in the hearts of Nkx2.5-Cre:Dsp^W/F and Myh6:Jup mice as compared to control NTG mice (N=2, *p<0.05, **p<0.001) F. IF staining of myocardial sections showing membrane localization of the β-catenin and pβ-catenin in the NTG, Nkx2.5-Cre:Dsp^W/F and Myh6:Jup mice.

Figure 7. Suppression of the canonical Wnt signaling in HL-1^shRNA-Pkp2 myocytes

A-B. Heat maps of canonical Wnt target genes (A) and Wnt inhibitors (B), constructed from the RNA-Seq data (q<0.05). C. Down-regulation of selected canonical Wnt targets, as detected by qPCR (N=3–7, ***p<0.0001). D. IF panel showing reduced nuclear localization of the canonical Wnt target CCND1 in the HL-1^PKP2:shRNA myocytes. E. IB analysis of the components of canonical Wnt signaling pathway showing increased pβ-catenin and reduced CCND1 levels. Phos-tag blot (bottom panel) showing increased β-catenin levels. F–G. IF staining for β-catenin (F) and pβ–catenin (G) showing reduced membrane localization of β–catenin in the HL-1^PKP2:shRNA myocytes. H. TCF luciferase reporter assay showing reduced TCF7L2 activity by approximately 55% in the HL-1^PKP2:shRNA myocytes (N= 8,*p<0.05) I. Representative Oil Red O (ORO) and PPARG stained panels showing increased number of cells containing fat droplets and cells expressing PPARG, respectively, in the HL-1^PKP2:shRNA, as compared to control cells. L,K: Corresponding quantitative data for adipogenesis (1000 to 1,250 cells per group, **p<0.005).

Figure 8. Binding of pYAP, pβ-catenin and JUP and the effects of inhibition of CK1, GSK3-β

A. Co-immunoprecipitation (Co-IP) of human cardiac proteins with an anti YAP antibody and detection of β-catenin (CTNNB1) or JUP by IB showing binding of CTNNB1 with YAP as well as binding of YAP with JUP. B. Corresponding Co-IPs on cell lysates obtained from the HL-1 and HL-1^PKP2:shRNA myocytes showing binding of YAP with β-catenin and with JUP. C–E. Effects of treatment with CK1 inhibitor IC261(C) or GSK3β inhibitor BIO (D) and ubiquitin proteasome inhibitor MG132 (E) on levels of pβ-catenin and pYAP, showing differentially increased pβ-catenin and pYAP levels in the HL-1^PKP2:shRNA myocytes.

Figure 9. Rescue of adipogenesis by KD of LATS1 and LATS2

A–B. Suppression of Lats1 (A) and Lats2 (B) mRNAs in wild type HL-1 and HL-1^PKP2:shRNA myocytes upon simultaneous targeting of LATS1 and LATS2 by shRNA pairs; C. KD of LATS1 protein after LATS1 and LATS2 KD by 4 shRNA pairs. The antibody against LATS2 protein did not detect a specific band; D. Increased PPARG levels after one week of adipogenic induction in the HL-1^PKP2:shRNA myocytes and normalization upon KD of LATS1/2. E. Rescue of adipogenesis upon KD of LATS1/2, as detected by Oil Red O and IF staining for PPARG, corroborating the findings of increased PPRAG protein detected by IB; F–G: Quantitative data showing increased number of PPARG positive and Oil Red O positive cells in the HL-1^PKP2:shRNA myocytes and their normalization upon KD of LATS1/2 (N~500 cells per group, one-way ANOVA p value<0.0001, Tukey post-hoc test for pair wise comparison; *p<0.05, **p<0.01, ***p<0.001,###p<0.0005).