Interplay of cell-cell contacts and RhoA/MRTF-A signaling regulates cardiomyocyte identity

Abstract

Cell-cell and cell-matrix interactions guide organ development and homeostasis by controlling lineage specification and maintenance, but the underlying molecular principles are largely unknown. Here, we show that in human developing cardiomyocytes cell-cell contacts at the intercalated disk connect to remodeling of the actin cytoskeleton by regulating the RhoA-ROCK signaling to maintain an active MRTF/SRF transcriptional program essential for cardiomyocyte identity. Genetic perturbation of this mechanosensory pathway activates an ectopic fat gene program during cardiomyocyte differentiation, which ultimately primes the cells to switch to the brown/beige adipocyte lineage in response to adipogenesis-inducing signals. We also demonstrate by in vivo fate mapping and clonal analysis of cardiac progenitors that cardiac fat and a subset of cardiac muscle arise from a common precursor expressing Isl1 and Wt1 during heart development, suggesting related mechanisms of determination between the two lineages.

Keywords: MRTF/SRF; RhoA/ROCK signaling; cardiac fat; cardiac progenitors; lineage conversion.

Figure EV1. Expression of desmosomal, pro‐apoptotic, and adipocytic genes in PKP2^mut CMs

1. qRT–PCR analysis of PKP2 reveals similar expression levels in wt and PKP2^mut CMs (n = 3).

2. Left, Western blotting using an antibody directed against the N‐terminus of PKP2 demonstrates reduced expression of wild‐type PKP2 protein (PKP2 wt, 97 kDa) and absence of truncated A587fsX655‐PKP2 product (PKP2 mut, 72 kDa) in total lysates of PKP2^mut CMs. Pan‐Cadherin is shown as a loading control. Right, densitometric readings for PKP2‐wt bands reveal an almost 50% reduction in PKP2^mut myocytic lysates compared to wt ones. Values are expressed as the integrals (area × mean density) of each band normalized to Pan‐Cadherin and relative to wt; n = 3; *P < 0.01 vs. wt; t‐test.

3. Immunofluorescence analysis indicates an interrupted desmoplakin expression (DSP, red) at the plasma membrane of PKP2^mut CMs compared to wt cells. cTNT (green) marks cardiomyocytes. Nuclei are stained with Hoechst 33258 (blue). Scale bars, 12.5 μm.

4. qRT–PCR analysis of pro‐apoptotic genes and white adipocytic markers shows similar expression levels in wt and PKP2^mut CMs over time in culture in adipogenic medium. Brown/beige adipocytic markers were upregulated in PKP2^mut compared to wt cells; n = 3; *P < 0.05, **P < 0.01 vs. wt CMs; t‐test.

5. Brown/beige adipocyte‐specific genes were similarly upregulated in PKP2^mut CMs at 28 days in culture in adipogenic medium and wt iPSC‐derived brown/beige adipocytes at day 20 of differentiation compared to wt CMs as determined by qRT–PCR; n = 3; *P < 0.05, **P < 0.01, ***P < 0.001 vs. wt; t‐test.

6. qRT–PCR analysis of indicated genes shows similar activation in PKP2^mut CMs and day‐20 iPSC‐derived brown/beige adipocytes after treatment with 1 mM 8‐Br‐cAMP for 48 h. Expression values are relative to basal conditions before treatment in each group; n = 3; *P < 0.05, **P < 0.01, ***P < 0.001 vs. corresponding PKP2^mut CMs or brown/beige adipocytes at basal conditions; t‐test.

Data information: All data are shown as means ± SEM.Source data are available online for this figure.

Figure 1. CMs with pathological PKP2 mutation convert into adipocytes in vitro

1. A

   Immunostaining of cTNT (green) and PKP2 (red) in wt and PKP2^mut CMs. Scale bars, 25 μm. Insets show a magnified image of PKP2 expression; scale bars, 10 μm.

2. B

   Immunodetection of cTNT (blue) and lipid stain ORO (red) in wt and PKP2^mut CMs over time in culture. Scale bars, 50 μm.

3. C–E

   Bar graphs show the percentage of cTNT⁺/ORO⁺ cells (C), cTNT⁺/ORO⁺ cells with organized (org), disorganized (disorg) and dissolving sarcomeres (loss) (D), and adipocyte‐like cells (E) in wt and PKP2^mut CMs over time (n = 3; N = 45–300 cells per time point in each group; *P < 0.05, **P < 0.001 vs. wt at the same time point; t‐test).

4. F

   qRT–PCR analysis of myocytic and adipocytic genes in wt and PKP2^mut CMs over time in culture (n = 3; *P < 0.05, **P < 0.01, ***P < 0.001 vs. wt at the same time point, t‐test).

5. G

   Immunodetection of cTNT (magenta) and lipid stain BODIPY (green) in wt and PKP2^mut CMs after 8 weeks of culture. Scale bars, 50 μm. Lower panels show magnified images of areas indicated by the white box for BODIPY (green) and PPARγ (red). Scale bars, 25 μm.

6. H

   Glycerol release of wt and PKP2^mut CMs after 8 weeks in culture in the absence or presence of 10 μM forskolin (n = 3; *P < 0.01 vs. wt; t‐test).

Data information: All data are shown as means ± SEM.Source data are available online for this figure.

Figure 2. Cardiac muscle and fat derive from Isl1/Wt1‐expressing progenitors

1. Immunofluorescence analysis of Isl1 derivatives in Isl1 ^Cre/+ ;R26 ^mTmG/+ mouse hearts at postnatal d28 (P28): Plin1 (magenta), membrane GFP (mG, green), and membrane Tomato (mT, red). The three panels on the right show a magnified view of the area indicated by the white box in the left panel. Arrows indicate Plin1⁺/mG⁺ adipocytes. la, left atrium; lv, left ventricle; ra, right atrium; rv, right ventricle. Scale bars, 500 μm (left panel), 25 μm (right panels); n = 3.

2. Immunostaining of cTNT, Isl1 or Wt1 (all magenta), mG (green), and mT (red) in sagittal, sequential sections of Isl1 ^Cre/+ ;R26 ^mTmG/+ embryos at E9.5. The boxed region in panel a is shown at higher magnification in the consecutive section in b^I. c^I and c^II are magnifications of the boxed area in b^I (in consecutive sections). Isl1‐Cre‐mediated mG labeling was observed in CMs of the forming heart (a, box b), cardiac progenitors (b^I, marked by the asterisk), and a proportion of Wt1⁺ proepicardial cells in the PEO (c^I, indicated by arrows, and c^II). Scale bars, 250 μm (a), 150 μm (b^I), 25 μm (c^I and c^II); n = 3.

3. Immunostaining for Isl1 and Wt1 in transverse sections of E8 embryos. A representative section corresponding to the position indicated by the dashed line drawn through the adjacent embryo view (i) is shown in panel ii after Hoechst 33258 staining of nuclei (blue). Panels iii show the boxed area in panel ii after Isl1 (red) and Wt1 (green) co‐immunostaining. Isl1⁺/Wt1⁺ progenitors and marked by the arrows in (iii). Scale bars, 2 mm (i), 200 μm (ii), 10 μm (iii). ca, caudal; cr, cranial; d, dorsal; fg, foregut; ht, heart; l, left; nt, neural tube; r, right; v, ventral; n = 3.

4. Temporal restriction of Cre‐mediated labeling of Isl1⁺ and Wt1⁺ progenitors and their derivatives using tamoxifen‐inducible Isl1 ^MerCreMer/+ ;R26 ^mTmG/+ and Wt1 ^CreERT2/+ ;R26 ^mTmG/+ mouse lines. Labeling was induced by tamoxifen treatment at E7.5, and heart sections were analyzed at postnatal day 28 (P28). Shown are representative immunostainings of cTNT or PLIN1 (magenta), mG (green), and mT (red) in sections of Isl1 ^Cre/+ ;R26 ^mTmG/+ (left panels) and Wt1 ^CreERT2/+ ;R26 ^mTmG/+ (right panels). Arrows indicate cTNT⁺/mG⁺ CMs. Scale bars, 25 μm; n = 3 per genotype.

Figure EV2. Localization and quantification of mG‐labeled cells in Isl1 ^MerCreMer/+; R26 ^mTmG/+ and Wt1 ^{Cre ERT 2/+};R26 ^mTmG/+ mouse embryos at E9.5

1. Temporal restriction of Cre‐mediated labeling of Isl1⁺ and Wt1⁺ progenitors and their derivatives using tamoxifen‐inducible Isl1 ^MerCreMer/+;R26 ^mTmG/+ and Wt1 ^CreERT2/+;R26 ^mTmG/+ mouse lines. Labeling was induced by tamoxifen treatment at E7.5, and embryos were analyzed at E9.5. Immunostaining of cTNT (red), Wt1 (magenta), Isl1 (magenta), and mG (green) in Isl1 ^MerCreMer/+;R26 ^mTmG/+ (upper panels) and Wt1 ^CreERT2/+;R26 ^mTmG/+ (lower panels). The boxed regions in the left panels are shown in higher magnification (in consecutive sections) in the four right panels a–d and a'–d' for Isl1 ^MerCreMer/+;R26 ^mTmG/+ and Wt1 ^CreERT2/+;R26 ^mTmG/+, respectively. Isl1 ^MerCreMer/+‐mediated mG labeling was absent in the PEO (a), but it was observed in epicardial cells (b), Isl1⁺ SHF progenitors (c), and CMs (d), as indicated by arrows. Wt1 ^CreERT2/+‐mediated mG labeling was detected in Wt1⁺ cells of the PEO (a') and in epicardial cells (b'), as indicated by arrows; mG expression was absent in Isl1⁺ SHF progenitors (c') and CMs (d'). Scale bars, 100 μm (left panels), 25 μm (right panels).

2. Table summarizing the regional distribution and number of mG‐labeled cells in Isl1 ^MerCreMer/+;R26 ^mTmG/+ and Wt1 ^CreERT2/+;R26 ^mTmG/+ mouse embryos at E9.5 as determined in (A). Epi, epicardium; LV, left ventricle; RV, right ventricle.

3. Bar graph depicting the percentage of the regional distribution (PEO, epicardium, SHF and myocardium) of mG⁺ cells in Isl1 ^MerCreMer/+;R26 ^mTmG/+ and Wt1 ^CreERT2/+;R26 ^mTmG/+ mouse embryos at E9.5. Major contribution of Isl1‐derivatives was detected in the SHF and myocardium, while most Wt1‐derivatives were found in the PEO. Both Isl1‐ and Wt1‐expressing progenitors contributed to the epicardium.

Figure EV3. Localization and quantification of mG‐labeled cells in Isl1 ^MerCreMer/+ ;R26 ^mTmG/+ and Wt1 ^{Cre ERT 2/+} ;R26 ^mTmG/+ adult mouse hearts at P28

1. Temporal restriction of Cre‐mediated labeling of Isl1⁺ and Wt1⁺ progenitors and their derivatives using tamoxifen‐inducible Isl1 ^MerCreMer/+;R26 ^mTmG/+ and Wt1 ^CreERT2/+ ;R26 ^mTmG/+ mouse lines. Labeling was induced by tamoxifen treatment at E7.5, and adult mouse hearts were analyzed at postnatal day 28 (P28). Immunostaining of mG (green) and mT (red) of Isl1 ^MerCreMer/+ ;R26 ^mTmG/+ (left panel) and Wt1 ^CreERT2/+ ;R26 ^mTmG/+ (right panel). Nuclei are stained with Hoechst 33258. Scale bars, 100 μm. la, left atrium; lv, left ventricle; ra, right atrium; rv, right ventricle.

2. Table depicting the regional distribution and the amount of the mG‐labeled CMs (as number of clusters and CM number/cluster) and adipocytes (as percentage of total AV groove adipocytes) in Isl1 ^MerCreMer/+ ;R26 ^mTmG/+ and Wt1 ^CreERT2/+ ;R26 ^mTmG/+ adult mouse hearts at P28. Shown are schematic representations of the distribution of mG‐expressing cells per each analyzed heart and as a summarized overview for each line. Size of CM clusters is illustrated. LA, left atrium; LAV, left atrioventricular; LV, left ventricle; RA, right atrium; RAV, right atrioventricular; RV, right ventricle.

Figure 3. A single Isl1/Wt1‐expressing progenitor gives rise to both CMs and adipocyte lineages in vitro

1. A

   Scheme for isolation of Isl1⁺/YFP⁺ cardiac precursors from the blastocyst‐derived mouse Isl1 ^Cre/+ ;R26 ^YFP/+ ESCs and their clonal analysis.

2. B, C

   Assessment of YFP expression by bright field/epifluorescence microscopy (B) and qRT–PCR analysis of cardiac progenitor markers (C) during EB differentiation of Isl1 ^Cre/+ ;R26 ^YFP/+ ESCs. Scale bars, 200 μm; n = 4; *P < 0.05, **P < 0.01 vs. d0; t‐test. Data are shown as means ± SEM.

3. D

   Flow cytometry analysis of YFP expression and qRT–PCR analysis of cardiac progenitor markers in YFP⁺ and YFP⁻ cells sorted from 4‐day‐old Isl1 ^Cre/+ ;R26 ^YFP/+ EBs. FITC, fluorescein isothiocyanate; PE, phycoerythrin. n = 3; *P < 0.05 vs. YFP⁻; t‐test. Data are shown as means ± SEM.

4. E

   Immunostaining of cTNT (red) and PPARγ (red) in conjunction with lipid stain BODIPY (green) of sorted cells from 4‐day‐old Isl1 ^Cre/+ ;R26 ^YFP/+ EBs after differentiation in pro‐myogenic and pro‐adipogenic culture conditions. Scale bars, 50 μm. The insets show higher magnifications, scale bars, 25 μm.

5. F

   Immunostaining of Isl1 (cyan), Wt1 (red), and YFP (green) in a representative clone derived from a single Isl1⁺ progenitor sorted from 4‐day‐old Isl1 ^Cre/+ ;R26 ^YFP/+ EBs and expanded on CMCs for 6 days. Scale bar, 100 μm.

6. G

   Immunostaining of cTNT (red) and PPARγ (red) in conjunction with BODIPY (green) in cell derivatives from a single Isl1⁺/Wt1⁺ clone after differentiation in pro‐myogenic (left) or pro‐adipogenic (right) culture conditions for 6 and 28 days, respectively. Scale bars, 50 μm. The inlays show higher magnifications; scale bars, 25 μm.

7. H

   RT–PCR profile of 10 representative clones after expansion on CMCs for 6 days. Isl1⁺/Wt1⁺ clones differentiated into both adipocytic and myocytic lineages, and Isl1⁺/Nkx2‐5⁺/Wt1⁻ clones differentiated only into the myocytic lineage. CMC cDNA was used as a negative control. Adipo, adipocytes; CMC, cardiac feeder cells; L, DNA ladder; Myo, myocytes.

Source data are available online for this figure.

Figure 4. Cooperative role of WT1 and PPARγ in cardiomyocyte‐to‐adipocyte conversion

1. A

   Immunostaining of endogenous WT1 (magenta), F‐actin (Phalloidin, green), and PPARγ (red) in wt and PKP2^mut CMs at 0 and 28 days in culture. Arrows indicate PKP2^mut cells with nuclear translocation of WT1, sarcomere disarray, and PPARγ activation. Scale bars, 25 μm.

2. B, C

   Immunocytochemistry of wt CMs after 28‐day infection with lentiviral constructs encoding the nuclear‐tagged form of WT1 (NLS‐WT1) plus GFP or GFP alone showing GFP (green), cTNT (red) and PPARγ (magenta) (B), and GFP (green, left) and cTNT (blue) in conjunction with lipid stain ORO (red, right) (C). Scale bars, 25 μm (B) and 50 μm (C).

3. D–F

   Immunocytochemistry of wt CMs after lentiviral overexpression of PPARγ alone (PPARγOE, D), PPARγ and NLS‐WT1 (NLS‐WT1/PPARγOE, E) or PPARγ, and shRNA targeting WT1 (shWT1/PPARγOE, F) at 14 (top) and 21 (bottom) days after infection. All constructs co‐express GFP. Representative images show GFP (green, left) and cTNT (blue) in conjunction with ORO stain (red, right) in the same cells. Scale bars, 50 μm.

4. G, H

   qRT–PCR analysis of myocytic and adipocytic genes in PPARγOE (G), NLS‐WT1/PPARγOE (H), and shWT1/PPARγOE (H) conditions at the indicated time points; n = 3 (PPARγOE, NLS‐WT1/PPARγOE) and n = 4 (shWT1/PPARγOE). *P < 0.05, **P < 0.01, ***P < 0.001 vs. GFP control (G) or PPARγOE (H) at the same time point (t‐test). Data are shown as means ± SEM.

5. I

   Wt CMs were transduced with HA‐tagged PPARγ and NLS‐WT1 and cultured in adipogenic medium for 14 days. Nuclear cell lysates were precipitated using anti‐HA or anti‐WT1 followed by immunoblot (IB) analysis with an anti‐WT1 or anti‐PPARγ, and IgG was used as control; n = 3. Arrows indicate WT1 (left panel) and PPARγ (right panel) bands.

Source data are available online for this figure.

Figure EV4. Upregulated expression and aberrant subcellular localization of WT1 in iPSC‐derived PKP2^mut CMs and native cardiac muscle cells from ARVC patient heart samples

1. A

   qRT–PCR analysis of WT1 reveals elevated expression levels in iPSC‐derived PKP2^mut compared to wt CMs; n = 3; *P < 0.05 vs. wt; t‐test. Data are shown as means ± SEM.

2. B

   Immunostaining of cTNT (green) and WT1 (magenta) in wt and PKP2^mut iPSC‐derived CMs. Note the filamentous‐like pattern of WT1 cytosolic expression. Nuclei are stained with Hoechst 33258 (blue). Scale bars, 25 μm.

3. C, D

   Analysis of adult human myocardium from patients affected and non‐affected by ARVC after immunostaining for cTNT (green) and WT1 (magenta). Nuclei are stained with Hoechst 33258 (blue). Low magnification of phase‐contrast images merged with immunoflourescence signals are shown in (C). Note intramyocardial fat infiltrations in ARVC conditions. Scale bars, 50 μm. High magnification images demonstrate nuclear localization of WT1 in CMs of ARVC patients (arrows), whereas in non‐ARVC individuals only cytosolic WT1 could be detected (D). Scale bars, 25 μm.

Source data are available online for this figure.

Figure EV5. Forced expression of PPARγ and WT1 drives adipocytic conversion of mouse adult CMs in vivo

1. A

   Scheme of experimental setup for injection of adeno‐associated virus serotype 9 (AAV9) encoding NLS‐WT1 (AAV9‐NLS‐WT1) and PPARγ (AAV9‐NLS‐ PPARγ) in Myh6 ^Cre/+ ;R26 ^mTmG/+ mice and their analysis after 1 or 5 weeks. 2.5 × 10¹² virus particles were injected intravenously via tail vein. Representative immunostainings of heart sections after 1‐week injection of each virus alone or in combination show CMs expressing the lineage marker mG (green) and the Wt1 (cyan) and PPARγ (red) transgenes. Nuclei are stained with Hoechst 33258. Arrowheads and arrows indicate mG⁺ CMs infected with one or both viruses, respectively. Scale bars, 50 μm. Quantification of transgene expressing mG⁺ CMs reveals high rate of co‐transduction. Three random heart slices per mouse and three mice per virus condition were analyzed.

2. B, C

   Analysis of heart sections 5 weeks after infection with AAV9‐NLS‐WT1 and AAV9‐PPARγ. (B) shows a representative phase‐contrast (PH) image merged with BODIPY fluorescence signal (green) visualizing a lipid‐filled cell within the myocardium of a mouse that received both viruses. Higher magnification is shown in the inset. Scale bar, 50 μm. Note that BODIPY⁺ cells with enlarged multilocular lipid droplets were detected exclusively in the heart of mice infected with both AAV9‐NLS‐WT1 and AAV9‐PPARγ at a frequency of 3–5 cells per heart section. Arrowhead indicates a BODIPY⁺ cell. In (C), subsequent immunofluorescence detection of mG (green), PPARγ (magenta), and the adipocyte marker PLIN1 (red) ultimately identifies infected CMs that underwent adipocytic conversion solely in mice treated with both viruses. Scale bar, 10 μm. The boxed region is shown in higher magnification (panels a^I and a^II). Scale bars, 25 μm.

Source data are available online for this figure.

Figure 5. Transcriptome and histone mark analyses reveal aberrant regulation of adipogenesis and impaired RhoA‐ and MRTF/SRF‐dependent gene programs in PKP2^mut CMs

1. A

   Functional annotation of genome‐wide differentially expressed genes in PKP2^mut CMs with gene set enrichment analysis (GSEA).

2. B

   GSEA of gene sets for adipogenesis (left). For each GSEA, the normalized enrichment score (NES) and P‐value are specified. Relative expression of the “leading edge” genes of the three adipogenic GSEAs is shown in heatmaps on the right. Expression level is represented as a gradient from low (blue) to high (red).

3. C

   Genome browser representation for H3K4me3, H3K27me3, and H3K27ac tracks at the PPARγ locus in wt CMs identified by ChIP‐Seq (top). qPCR analyses of H3K4me3‐, H3K27me3‐, and H3K27ac‐ChIP on p1 and p2 peaks (shaded in gray) at the PPARγ locus in wt and PKP2^mut CMs and human adipocytes (bottom); n = 2; *P < 0.05 vs. wt; t‐test.

4. D, E

   Plots of the BERENJENO_TRANSFORMED_BY_RHOA_FOREVER_UP (D) and SRF_C and SRF_01 (E) GSEAs. NES and P‐value are shown. Relative expression of the “leading edge” genes of the GSEA is shown in the concomitant heatmaps. Direct MRTF‐SRF downstream target genes are highlighted in blue. Expression level is represented as a gradient from low (blue) to high (red).

Source data are available online for this figure.

Figure 6. Decreased RhoA signaling downstream of defective cell–cell contacts leads to reduced nuclear MRTF‐A localization and derepression of PPARγ

1. A

   Immunostaining of cTNT (cyan), PKP2 (red), and RhoA (green) in wt and PKP2^mut CMs. Scale bars, 12.5 μm.

2. B

   RhoA activity in wt and PKP2^mut CMs at 0 and 1 day in culture; n = 3; *P < 0.05, **P < 0.001 vs. wt at the same time point; t‐test.

3. C

   Immunostaining of F‐actin (Phalloidin, green) and actin (red) in wt and PKP2^mut CMs at 0 and 7 days in culture. Scale bars, 12.5 μm. Bar graph shows quantification of G‐actin/F‐actin ratio in wt and PKP2^mut CMs; N = 10 (wt) and N = 9 (PKP2^mut) cells; *P < 0.01 vs. wt at the same time point; Mann–Whitney U‐test.

4. D, E

   Live imaging of MRTF‐A subcellular localization in wt and PKP2^mut CMs by transient expression of doxycycline‐inducible MRTF‐A‐GFP. Images in (D) show representative CMs at d0 in 20% serum (left) and after subsequent treatment with adipogenic medium for 1 h (middle). Images in (E) show representative CMs in adipogenic medium at 7 days, before (left) and after treatment with 20 nM LMB for 1 h (middle). Small panels on the right show magnified views of the boxed areas. Arrows indicate dynamic changes of nuclear MRTF‐A localization before and after corresponding treatment. Scale bars, 50 μm (left and middle panels) and 25 μm (right panels). Bar graphs show average values of nuclear/cytoplasmic ratio of GFP signal intensity in wt and PKP2^mut CMs at indicated conditions; (D): N = 60 (wt) and N = 37 (PKP2^mut) cells; (E): N = 105 (wt) and N = 64 (PKP2^mut) cells; *P < 0.001 vs. wt at the same condition; Mann–Whitney U‐test.

5. F

   Immunostaining of cTNT (magenta), GFP (green), and PPARγ (red) in wt and PKP2^mut CMs infected with MRTF‐A‐GFP and cultured in adipogenic medium for 7 days. Cells were treated as in (E) before fixation. Arrows indicate differential PPARγ expression dependent on levels of nuclear MRTF localization. Scale bars, 25 μm.

6. G

   Prevention of myocyte‐to‐adipocyte conversion in PKP2^mut CMs by continuous overexpression of MRTF‐A‐GFP. Immunostaining of GFP (green) and correspondent colorimetric immunodetection of cTNT (blue) in conjunction with lipid stain ORO (top) and cTNT (magenta), GFP (green), and PPARγ (red) (bottom) in PKP2^mut CMs overexpressing MRTF‐A‐GFP cultured in adipogenic medium for 28 days. Phase‐contrast image is merged with PPARγ and GFP. Scale bars, 50 μm (top), 12.5 μm (bottom). Dotted lines indicate the border separating GFP⁺ and GFP⁻ (left) as well as ORO⁺ and ORO⁻ (right) cells. Adjacent bar graphs indicate the percentage of adipocyte‐like cells (top) or PPARγ⁺ cells (bottom) that were positive and negative for GFP; n = 3; N = 840 cells (top graph) and N = 306 cells (bottom graph); *P < 0.001 vs. GFP⁻; t‐test.

Data information: All data are shown as means ± SEM. Source data are available online for this figure.

Figure 7. Pharmacological and genetic perturbation of RhoA‐mediated cytoskeleton remodeling induces myocyte–adipocyte conversion

1. A, B

   Immunostaining of F‐actin (Phalloidin, green) and PKP2 (red) in wt CMs treated and non‐treated with 30 μM ROCK inhibitor Y‐27632 for 14 days (A). Immunofluorescence analysis of WT1 (magenta), F‐actin (Phalloidin, green), and PPARγ (red) in wt CMs treated and non‐treated with 30 μM Y‐27632 for 28 days. Arrow indicates a cell with severe remodeling of the actin cytoskeleton, nuclear shuttling of WT1, and PPARγ expression (B). Scale bars, 25 μm.

2. C

   Immunodetection of cTNT (blue) and lipid stain ORO (red) in wt CMs untreated and treated with 30 μM Y‐27632 for 28 days. Scale bars, 50 μm. Bar graph shows the percentage of adipocyte‐like cells in treated and untreated conditions; n = 3; N = 600 (untreated) and N = 592 (Y‐27632) cells; *P < 0.05 vs. untreated cells; t‐test.

3. D, E

   Immunostaining of cTNT (green) and NMIIB (red) (D) or cTNT (cyan), PKP2 (red), and RhoA (green) (E) in wt and MYH10^mut CMs. Scale bars, 12.5 μm.

4. F

   Immunostaining of F‐actin (Phalloidin, green) and actin (red) and quantification of G‐actin/F‐actin ratio in wt and MYH10^mut CMs. Scale bars, 12.5 μm. N = 11 (wt) and N = 13 (MYH10^mut) cells; *P < 0.05 vs. wt; Mann–Whitney U‐test.

5. G

   Live imaging of MRTF‐A subcellular localization in wt and MYH10^mut CMs by transient expression of doxycycline‐inducible MRTF‐A‐GFP. Shown are representative myocytes at d0 in 20% serum (left) and after subsequent treatment with adipogenic medium for 1 h (middle). Small right panels show magnified views of the boxed areas. Arrows indicate dynamical changes of nuclear MRTF‐A localization. Scale bars, 50 μm (left and middle) and 25 μm (right). Bar graph presents quantification of nuclear/cytoplasmic ratio of GFP signal intensity in wt and MYH10^mut CMs; N = 55 (wt) and N = 53 (MYH10^mut) cells; *P < 0.001 vs. wt; Mann–Whitney U‐test.

6. H

   qRT–PCR analysis of myocytic and adipocytic genes in wt and MYH10^mut CMs over time in culture; n = 3; *P < 0.05, **P < 0.01 vs. wt at the same time point; t‐test.

7. I

   Immunostaining of F‐actin (Phalloidin, green), WT1 (magenta), and PPARγ (red) in wt and MYH10^mut CMs at 28 days of culture. Arrows indicate cells with severe remodeling of the actin cytoskeleton, nuclear WT1, and PPARγ expression. Scale bars, 25 μm.

8. J

   Colorimetric immunodetection of cTNT (blue) and lipid stain ORO (red) in wt and MYH10^mut CMs at 28 days of culture. Scale bars, 50 μm. Bar graph shows the percentage of adipocyte‐like cells in wt and MYH10^mut conditions; n = 3; N = 567 (wt) and N = 520 (MYH10^mut) cells; *P < 0.05 vs. wt (t‐test).

Data information: All data are shown as means ± SEM. Source data are available online for this figure.

Figure 8. Model describing cell–cell adhesion‐dependent regulation of cardiomyocyte‐to‐adipocyte lineage conversion

Impaired intercellular contacts at the intercalated disk of developing cardiac myocytes lead to a reduction in RhoA‐GTPase activity and remodeling of actin cytoskeleton resulting in increased level of cytosolic G‐actin. G‐actin binding to MRTF prevents its nuclear translocation, resulting in repression of the MRTF/SRF‐regulated myogenic lineage commitment and activation of the alternative adipocytic gene program triggered by PPARγ.