Fig 2: In Vivo Assessment of Migration and Engraftment of DLL4 and PDGF-BB-Treated SCs(A) In vivo analysis of limb muscles from Sgca-null/scid/beige mice intra-arterially transplanted with 106 GFP+/β-GAL+ control or DLL4 and PDGF-BB-treated SCs after 3 weeks. Upper panel: exposure to X-gal revealed engraftment of β-GAL+ donor cells (blue; arrowhead) in the knee region and lower limb muscles of mice transplanted with DLL4 and PDGF-BB-treated SCs. Lower panel: assessment of X-gal in the gastrocnemius muscle of mice transplanted with treated SCs, revealing nuclear localization of β-GAL in a GFP+ myofiber (arrowheads). Graph quantifies the area of X-gal+ signal in muscles intra-arterially transplanted with treated or untreated SCs (lateral view of knee region) normalized to the volume of cell suspension delivered (N = 4).(B) Quantification of transplanted cells that engrafted intramuscularly injected TA muscles (N = 5).(C) Distribution across muscle length of X-gal+ nuclei per section of TA muscles transplanted with treated and untreated SCs (N = 5).(D) Quantification of the localization (i.e., inside or outside myofibers) of X-gal+, treated/untreated, donor-derived nuclei in transplanted muscles (N = 5).(E) Immunofluorescence panel showing a representative Scga-null/scid/beige muscle transplanted with 106 GFP+/nLacZ+ DLL4 and PDGF-BB-treated SCs intramuscularly 3 weeks before explant. SGCA (red) and X-gal (violet): donor-derived fibers and nuclei, respectively. LAMININ (green) and Hoechst (blue): extracellular matrix and nuclei, respectively.(F) Quantification of the experiment in (E) showing the average number of donor-derived fibers from three mice.Data: means ± SEM. Statistical significance based on unpaired Welch's t test. ∗p < 0.05; n.s., not significant. Scale bars, 2 mm (A, left panel), 250 μm (A, middle panel), 50 μm (A, right panel), and 50 μm (E).

Fig 3: ZEB1 depletion epigenetically reduces Dll4 and Notch1 expressions in bone ECs.a Schematic representation of reporter constructs of wild-type (WT) and mutated (MUT) murine Dll4 and Notch1 promoters. b Luciferase reporter analysis of cultured control and ZEB1-deleted bone ECs that were electroporated with WT or MUT Dll4 and Notch1 promoter reporter constructs (n = 3 independent experiments). c Luciferase reporter analysis of cultured bone ECs that were co-electroporated with WT or MUT Dll4 and Notch1 promoter reporter constructs in combination with pcDNA3.1 or pcDNA3.1-HA-ZEB1 (n = 3 independent experiments). d, e ChIP-PCR analysis of interactions of ZEB1-Dll4 promoter and ZEB1-Notch1 promoter within the proximal and distal regions in cultured bone ECs. Agarose gel images as shown are from three independent experiments (d) and promoter occupancy relative to corresponding input was quantified (e; n = 3 independent experiments). prox. proximal, dist. distal. f ChIP-qPCR for analyzing the enrichments of H3K4Ac, H3K14Ac, H3K18Ac, and H3K27me3 on Dll4 and Notch1 promoters in control and ZEB1-deleted bone ECs (n = 3 independent experiments). g Immunoprecipitation (IP) analysis of the physical interaction between ZEB1 and CBP or p300 in nuclear extracts of bone ECs. Images as shown are from three independent experiments. h Sequential ChIP-PCR analysis confirming the co-occupancy of ZEB1 and CBP or p300 on Dll4 and Notch1 promoters in bone ECs. Images as shown are from three independent experiments. i, j Luciferase reporter assays for analyzing Dll4 and Notch1 promoter activity in HEK293T cells that were co-transfected with the indicated constructs (i). Immunoblot analysis confirming ectopic expression of HA tagged ZEB1, p300, and CBP in nuclear extracts of HEK293T cells (j). Images as shown are from three independent experiments. Arrow marks specific bands with the expected molecular weights, respectively. All data are represented as mean ± SD. **P < 0.01, *P < 0.05; NS not significant. Differences are tested using one-way ANOVA with Tukey’s post hoc test (b, c, i) and unpaired two-tailed Student’s t-test (e, f). The source data are provided as a Source Data file. Unprocessed original scans of blots are shown in Source Data file.

Fig 4: Effect of DLL4 and PDGF-BB Treatment on Human SC-Derived Myoblasts(A) Representative images of CD56+ FACS-purified human SC-derived myoblasts from two individuals (HuSC 3 and 5) expanded in control conditions (upper panel) or treated with DLL4 and PDGF-BB (lower panel) for 1 week, then pulsed for 2 h with EdU (green) and co-immunostained with Ki67 (red).(B) Quantification of the percentage of EdU+ cells (N = 5).(C) Quantification of Ki67+ cells (N = 3).(D) Real-time qPCR analysis of myogenic (PAX7, MYOD, and MYOGENIN), perivascular (PDGFRB, TNAP, and CD146) and Notch signaling (HES1 and HEY1) genes in response to DLL4 and PDGF-BB treatment in HuSC 3 and 5 (n = 3 per line). Graphs show fold change to control conditions, statistical significance based on ΔCt.(E) Phase contrast images of AP enzymatic activity (violet) and Hoechst (blue) in untreated and treated HuSCs.(F) Differentiation of HuSCs in response to DLL4 and PDGF-BB treatment and Notch inhibition. Cells expanded in control or treatment conditions for 1 week and induced to differentiate for 4 days in the presence or absence of L-685,458. Pictures representatives from HuSC1, and graphs quantify the percentage of MyHC+ cells (N = 4).(G) Representative images of the lower chambers of transwell assays. Untreated/treated HuSCs and human muscle pericyte-derived mesoangioblasts (MABs; technical control) were seeded for 8 h on HUVECs. Right graph: number of 6-CFDA+ migrated cell/mm2 in each condition. N = 3; means ± SEM.Statistical significance based on paired Student's t test (B, C and cell-line-specific data in G), unpaired Student's t test (D and pooled data in G) or one-way ANOVA with Tukey's multiple comparison (F). ∗p < 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001, ∗∗∗∗p ≤ 0.0001; n.s., not significant. Scale bars, 50 μm (A), 75 μm (F), and 100 μm (E and G).

Fig 5: Administration of Zeb1?EC mice with r.Dll4 protein efficiently restores the impaired bone angiogenesis and osteogenesis.a, b Control and ZEB1-deleted bone ECs seeded on vehicle- or r.Dll4-precoated chamber wells (or culture dishes) were subjected to NICD1 immunostaining (a) and immunoblotting (b) analyses, respectively. Representative confocal or immunoblot images as shown are from three independent experiments. Scale bar, 20 µm. c Representative confocal images of CD31/EMCN and VEGFA immunostaining in the tibia of Zeb1WT and Zeb1?EC mice that were i.p. injected with 1 µg/g r.Dll4 protein at P4 for 2 consecutive weeks before analysis at P21 (n = 5, each). Scale bar, 100 µm. d Quantification of type H vessel density (top panel) and VEGFA staining levels (bottom) in the tibia as shown in c (n = 5 independent experiments). e Micro-CT analysis of the tibia showing improved bone formation in trabecular bone (top panels) and cortical bone (bottom panels) of Zeb1?EC mice treated with r.Dll4 protein as described in c (n = 5, each). Scale bar, 0.2 mm. f Quantification of bone architectures in the tibia as shown in e (n = 5 independent experiments). g Runx2, Osterix, and ALP immunostaining of the tibia as described in c (n = 5, each). Scale bar, 100 µm. h Quantification of Runx2+, Osterix+, and ALP+ cells in the tibia as shown in g (n = 5 independent experiments). All data are represented as mean ± SD. **P < 0.01. Differences are tested using one-way ANOVA with Tukey’s post hoc test (d, f, h). The source data are provided as a Source Data file. Unprocessed original scans of blots are shown in Source Data file.