Fig 2: TAK1 mediates ADAM12 induction in vitro and in vivo. (A) Results of the signaling screen: only TAK1 expression resulted in the activation of ADAM12. Yellow and red dots represent ADAM12 normalized Nanostring counts in IMR90 infected with the kinase, each dot representing one kinase. Black dots represent ADAM12 expression level in control cells (infected with empty vector). (B) RT-qPCR validation of ADAM12 reactivation by TAK1 in IMR90 cells. ADAM12 was activated by the wild-type form of TAK1 but not by the catalytic dead form of TAK1 (TAK1 CD). (C) Western blot analysis of ADAM12 protein levels following the overexpression of wild-type and catalytic dead TAK1 (TAK1CD). Ctl: cells infected with the empty vector. The white arrows represent the exogenous TAK1; TAK1CD is lower because it does not self-phosphorylate. Tubulin (TUB) is the loading control. For ADAM12, concanavalin A enrichment was performed. (D) Schematic representation of canonical and non-canonical TGF-ß pathways. TAK1 is a component of the non-canonical TGF-ß pathway. (E) RT-qPCR on the indicated genes in the presence or absence of the TAK1 kinase inhibitor (5Z)-7-Oxozeaenol (5Z). IMR90 were pre-treated with 0.3µM 5Z or DMSO for two hours, followed by stimulation with 5 ng/ml of TGF-ß for 6 h. ADAM12 induction by TGF-ß is abolished by 5Z. (F) Control western blot showing the phosphorylation of SMAD3, indicating the activation of canonical TGF-ß pathway even though TAK1 was inhibited by 5Z. (G) Testing the dependence of ADAM12 on TAK1 in vivo: reporter mice with the ADAM12 promoter (PADAM12) driving GFP expression were subjected to an injury (cardiotoxin injection into the tibialis muscle). Some of the mice were treated by intraperitoneal injection of TAK1 inhibitor (5Z) at 5 mg/kg before and after injury, while the controls were injected only with solvent (DMSO). Three days after the injury, mice were sacrificed and muscle tissue was dissected. (H) Immunofluorescence on injured muscle shows that 5Z treatment reduces GFP induction. The arrows indicate GFP-positive cells. Scale bar: 150 µm. (I) Quantification of panel H (10 fields counted in each condition). All the experiments were performed at least three time except for ADAM12 western blot and mice experiments. The statistical analysis was performed with one way ANOVA followed by Dunnett's test, except for panel I where a Mann–Whitney test was performed. In all figures, we used the following conventions: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Fig 3: Targeting BMP signaling suppresses onset of AD in mice. (A) Monocle analysis showing the ordering of FB2 in pseudotime. Each dot is a cell. Each color indicates a cell state. (B) Heatmap to display different blocks of top 1,000 differentially expressed genes (DEGs) along the pseudotime trajectory (A). (C) Schematic of in vivo intervention experiment showing the administration time point of inhibitors or activators of pathways selected in Figure S8. (D) Representative images showing the effects of different compounds on aortas in mouse AD model. (E) Maximal ascending aortic diameter in indicated groups. *p < 0.05 compared to Saline group (One-way ANOVA). (F) Systolic blood pressure at different time points following treatment. **p < 0.01, ***p < 0.001 compared to Saline group (One-way ANOVA). For D-F, saline, n = 9; PDTC, n = 9; LDN-193189, n = 9; GW7647, n = 9; Oroxin B, n = 9. (G) Incidence of aortic dissection following treatment. (H) Co-staining of dFB2 marker PDPN, dFB2-specific ligand ADAM12 and corresponding SMC receptor ITGB1in mice aorta sections. White arrows indicate PDPNhigh cells, red arrows indicate interaction between ADAM12 and ITGB1. Scale bars, 50 µm (left image); 25 µm (magnification). FL, false lumen. Images are representative of 3 independent experiments (3 samples each time point).

Fig 4: Adam12 and Gap43 expression identify immune-responsive muscle-resident mesenchymal cell populations(A and B) Single-cell (A) feature and (B) violin plots for cluster 5 (Adam12) and cluster 6 (Gap43) genes and type 2 immune-receptor genes (Il4ra, Il13ra1).(C) LipidTOX/DAPI-stained PDGFRa+ sorted (C1) ADAM12+ cluster 5, (C2) GAP43+ cluster 6, (C3) CLU+ cluster 7, (C4) HSD11B1 clusters 8/9 cells cultured in adipogenic medium for 5 days.(D) LipidTOX/DAPI-stained PDGFRa+ sorted (D1) ADAM12+ cluster 5, (D2) GAP43+ cluster 6, (D3) CLU+ cluster 7, (D4) HSD11B1 clusters 8/9 cells cultured in adipogenic medium supplemented with recombinant IL-4 for 5 days.(E) LipidTOX/DAPI-stained PDGFRa+ sorted (E1) ADAM12+ cluster 5, (E2) GAP43+ cluster 6, (E3) CLU+ cluster 7, (E4) HSD11B1 clusters 8/9 cells cultured in adipogenic medium supplemented with recombinant IL-13 for 5 days.(F) Quantification of adipogenic differentiation as visualized by %LipidTOX/DAPI. Data are represented as mean ± SEM. Intergroup analysis performed with two-way ANOVA and post hoc Tukey’s test. Intragroup analysis performed with paired Student’s t test. n = 9 per cluster and condition. Significance denoted by an asterisk.(G) Flow-cytometry contour plots for PDGFRa and either IL-4Ra (cluster 5/6), CLU (cluster 7), or HSD11B1 (cluster 8/9) gated on tdT+ live cells obtained from BaCl2-injured muscle and NaCl-injected contralateral controls at 4 days post injury as well as glycerol-injured muscle and NaCl-injected contralateral controls 3 days post injury.(H) Percent quantification of tdT+/PDGFRa+/IL4Ra (cluster 5/6), tdT+/PDGFRa+/CLU+ (cluster 7), and tdT+/PDGFRa+/HSD11B1 (cluster 8/9) cells in BaCl2-injured muscle or NaCl-injected contralateral controls. n = 6 per cluster and condition. Data are represented as mean ± SEM. Significance determined using two-way ANOVA and post hoc Tukey’s test. Significance denoted by an asterisk. NS, not significant.(I) Percent quantification of tdT+/PDGFRa+/IL4-Ra (cluster 5/6), tdT+/PDGFRa+/CLU+ (cluster 7), and tdT+/PDGFRa+/HSD11B1 (cluster 8/9) cells in glycerol-injured muscle or NaCl-injected contralateral controls. Data are presented as mean ± SEM. n = 5 per cluster and condition. Significance determined using two-way ANOVA and post hoc Tukey’s test. Significance denoted by an asterisk. NS, not significant.All scale bars represent 200 µm.

Fig 5: Epistasis and interaction between KAT2A and TAK1. (A) RT-qPCR on the indicated genes in MRC5 cells transfected with siCtl (non-targeting siRNA), siKAT2A and/or siTAK1. ADAM12 is induced by the knockdown of KAT2A but this is abolished by the simultaneous knockdown of TAK1. Statistical analysis was performed with a two-way ANOVA followed by a Dunnett's test. (B) Western blot showing expression level of TAK1 and KAT2A after siRNA transfection. (C) Co-immunoprecipitation of endogenous TAK1 from MRC5 cells with or without stimulation by 5 ng/ml TGF-ß revealed interaction with KAT2A. TAB1, a known interactor of TAK1, served as a positive control.