Fig 2: Validation of the anti‐BMP4 antibodies used in the study. Panel (a) Representative immunoblot demonstrating detection of BMP4 expression with anti‐BMP4 antibody (Abcam, Cat#ab39973) in protein extracts derived from: tibia, heart, bone cells and skeletal muscle (vastus lateralis muscle). Panel (b) Representative immunoblot demonstrating loss of the BMP4 expression detection with the antibody ab39973, after blocking the tibia, heart, bone cells, and vastus lateralis muscle‐derived protein extracts with the BMP4 peptide (Abcam, Cat#ab40140). Panel (c) Membrane loaded with protein extracts derived from heart and tibia of sedentary wild‐type (WT) mice and of sedentary mice with chronic heart failure (Tgαq*44) incubated only with anti‐BMP4 antibody and (panel d) Ponceau‐S staining of the same membrane. Panel (e) Membrane loaded with protein extracts derived from heart and tibia of sedentary wild‐type (WT) mice and of sedentary mice with chronic heart failure (Tgαq*44) incubated only with the secondary antibody conjugated with horseradish peroxidase (negative control) and (panel f) Ponceau S staining of the same membrane. Protein standard ladder presented in the panels (a and b) is Magic Mark XP Western Protein Standard (Invitrogen, Cat#LC5602). Protein standard ladder presented in panels (c–f) is a visible Precision Plus Protein Dual Color Standards (Biorad, Cat#1610374)

Fig 3: Downregulation of BMP4/SMAD4 signaling and osteogenesis-related markers in atrophic non-union fractures. a Histopathological analysis of atrophic nonunion (ANU) (n = 5) and standard healing fracture (SHF) specimens (n = 5) using H&E and Alcian Blue staining. IHC analysis of BMP4 and SMAD4 protein expression in ANU and SHF specimens. Scale bars = 100 μm. b qRT-PCR analysis of miR-1323 expression in ANU and SHF specimens. miR-1323 expression normalized to the SHF median value. c Representative immunoblots and d quantitation of BMP4, SMAD4, ALP, Col I, and RUNX2 protein expression in ANU and SHF specimens. Protein expression normalized to the SHF median values. e ALP activity in ANU and SHF specimens. ALP activity reported as the optical density (OD) 405 nm value per mg protein. f, g Pearson correlation analyses of (f) miR-1323 expression and BMP4 expression as well as (g) miR-1323 expression and SMAD4 expression in ANU specimens. *p < 0.05, **p < 0.01 [SHF vs. ANU]. b, d, e Data presented as medians ± interquartile ranges (boxes) and absolute ranges (whiskers)

Fig 4: Icariin treatment/miR-23a-3p knockdown induces the expression of BMP-2, BMP-4, Runx2, p-Smad5, Wnt1 and β-catenin in BMSCs. (A) RT-PCR was used to detect the expression of BMP-2, Runx2, Smad5, Wnt1 and β-catenin mRNA. (B and C) Western blotting was used to detect the protein expression of BMP-2, BMP-4, Runx2, p-Smad5, Wnt1 and β-catenin. (D) Dual-luciferase reporter assay was used to verify the relationship of miR-23a-3p and Rnux2. Data are represented as the means ± SD (n = 3). *P < 0.05 vs. its miR-23a-NC group. @P < 0.05 vs. miR-23a-NC + SI group. &P < 0.05 vs. miR-23a-inhibitor group.

Fig 5: Transcriptome comparison between the VELs and the primary VECs.a KEGG analysis of the shared genes between the VELs and the primary VECs, showing the enrichment of TGF-β, WNT, BMP, and NOTCH signaling pathways. b Heat map analysis showing that the majority of top 100 genes were highly expressed in the primary VECs and hPSC-derived VELs, but lowly in HAEC/HUVEC. c Validation of the indicated genes by qRT-PCR analysis. d WB results showing that BMP4 was expressed at much higher levels in day 7 hPSC-derived VELs than HUVEC. e IF staining results showing that the indicated markers were expressed at higher levels in day 7 hPSC-derived VELs than in HUVEC. Scale bar: 100 μm. f The t-SNE map showing the second-level clustering of VECs into two subpopulations, designated as 0 and 1. g Violin plots showing the expression of indicated genes for subcluster 0 and 1, suggesting that subcluster 0 may represent ventricular side specific VECs and subcluster 1 may represent aortic side specific VECs. h tSNE maps showing that the indicated marker genes were indeed highly expressed in the VEC subclusters. i qRT-PCR results showing that the indicated genes were expressed much higher in VECs than HAEC and HUVEC. VEC-A and VEC-V represent the aortic and the ventricular side of VECs, respectively. The paired t test in Graphpad software was used for the statistical analysis. Significant levels are: *p < 0.05; **P < 0.01; ***P < 0.001.