An endothelial SOX18–mevalonate pathway axis enables repurposing of statins for infantile hemangioma

Figures

• 

  Figure 1. R(+) propranolol reduces MVP transcripts in IH-derived HemSCs undergoing endothelial differentiation. HemSCs in A–C were treated with or without R(+) propranolol for 2 hours before RNA isolation. (A) KEGG pathway analysis and Gene Ontology Biological Processes (GO-BP) analysis of bulk RNA-Seq data from HemSCs treated on day 6 of endothelial differentiation (n = 6 biological replicates) with or without R(+) propranolol, showing the number of differentially expressed genes (DEGs), defined as log2 fold change greater than 1 and adjusted P less than 0.05. (B) Heatmap of differentially expressed MVP genes at day 4 and day 6. (C) Volcano plots of DEGs on day 4 and day 6 of HemSC to HemEC differentiation (–log10 for adjusted P value). (D) SOX18 mRNA levels on day 4 and day 6 (n = 3 biological replicates; 1 of the biological replicates was analyzed in 2 independent experiments, yielding n = 4 data points). (E) HemSCs induced to undergo endothelial differentiation were treated with or without R(+) propranolol for 2 hours on day 6 or treated continuously from day 2 to day 6. qPCR analyses for HMGCS1, HMGCR, and MVK (n = 3 biological replicates; 1 of the biological replicates was analyzed…

• 

  Figure 2. SOX18 fine-tunes endothelial MVP gene expression. (A) ChIP-Seq dataset in HUVECs identifies SOX18-binding sites within the HMGCS1 and HMGCR gene loci. (B) Lentiviral SOX18 overexpression in undifferentiated HemSCs (HemSCSOX18OE) versus control HemSCs (HemSCCtr). RNA was analyzed for HMGCS1, HMGCR, MVK, and NOTCH1 by qPCR (n = 3 independent experiments). (C) Treatment of HemSCSOX18OE with or without R(+) propranolol for 24 hours (n = 5 independent experiments). (D) Treatment of IH-derived control HemECs with R(+) propranolol for 24 hours. RNA was analyzed for HMGCS1, HMGCR, MVK, and NOTCH1 by qPCR. (E) HemECs with lentiviral knockdown of SOX18 (HemECshSOX18) were treated with R(+) propranolol for 24 hours followed by qPCR analyses (n = 3 biological replicates, performed in 2 independent experiments, yielding n = 6 data points; lentiviral knockdown of SOX18 was performed twice in each of the 3 HemEC lines with an efficiency cutoff of >70%). P values were calculated using a 2-tailed, unpaired t test (B–E). Data are shown as the mean ± SD.

• 

  Figure 3. Biosynthetic output of the MVP is mediated via a SOX18-/SREBP2-dependent mechanism. (A and B) HUVECs were incubated with MBCD for 16 hours, followed by treatment with or without R(+) propranolol or with or without Sm4 for 16 hours. Endogenous cholesterol levels were measured by mass spectrometry (n = 3 biological replicates). (C and D) Overexpression of ragged-opossum (RaOp) in HUVECs followed by immunofluorescence staining of HMGCS1 and HMGCR (n = 3 biological replicates). P values were calculated using a 2-tailed, unpaired t test (A and B) and by a 1-way ANOVA multiple-comparison test with Tukey’s correction (C and D). Data are shown as the mean ± SD. (E) Schematic of SREBP2 maturation by SCAP, site 1 protease (S1P), and S2P to produce the bHLH domain that translocates to the nucleus. (F and G) HemSCs undergoing endothelial differentiation for 6 days were treated for 2 hours with or without R(+) propranolol, lysed, and analyzed by WB with anti-SREBP2, anti-SOX18, and anti-GAPDH (n = 3 biological replicates; 1 of the biological replicates was analyzed in 2 independent experiments, yielding n = 4 data points). (H and J) Relative gene expression of SREBP2, SCAP, S1P, and S2P in…

• 

  Figure 4. Nuclear colocalization of SOX18 and SREBP2 in IH and congenital vascular tumors. (A–D) Human age-matched skin, proliferating IH, involuting IH, and regrowing IH stained with anti-SREBP2 (magenta), anti-SOX18 (cyan), and the human endothelial cell–specific lectin UEA1 (yellow). Cell nuclei stained with DAPI (blue). (E and F) Rapidly involuting and non-involuting congenital hemangiomas (RICH and NICH) stained in the same manner. Boxed areas are shown enlarged in the bottom panels. (G) SOX18+SREBP2+ double-positive cell nuclei (arrowheads) were quantified and expressed relative to total endothelial cells. P values were calculated using 1-way ANOVA with Šidák correction. Data are shown as the mean ± SD. Data represent 15 representative images each for patient sample sizes n = 4 for age-matched skin controls, n = 10 each for proliferating- and involuting-phase IH, n = 4 for regrowing IH, and n = 3 each for RICH and NICH. See Table 1 for detailed patient information. Scale bars: 50 μm (top panels) and 25 μm (bottom magnified panels).

• 

  Figure 5. Statins inhibit HemSC endothelial differentiation and blood vessel formation. (A) HemSCs were induced to undergo endothelial differentiation for 6 days in the presence of simvastatin (0.5 μM), atorvastatin (0.1 μM), or R(+) propranolol (10 μM) (n = 3). VE-cadherin (circles) and CD31 (triangles) were measured by qPCR. R(+) propranolol served as positive control. (B) Schematic of IH xenograft model. (C) Top: Matrigel implants harvested after 7 days. Middle panels: H&E staining highlights blood vessel lumens with red blood cells. Bottom panels: Anti–human CD31 staining (red) indicates human blood vessels. Cell nuclei were stained with DAPI (blue). Arrows point at perfused, human CD31 positive vessels, arrowhead in the treatment group points at an immature vessel. Scale bars: 1 cm (Matrigel implants). Scale bars: 100 um (top); 50 um (bottom). (D) Left: Vessels/mm2 in H&E-stained sections. Right: Human CD31+ vessels/mm2. (E) Dose response to simvastatin. (F and G) Body weight (F) and blood glucose levels (G) of mice over 8 days. (H and I) Immunofluorescence staining for Ki67 and human CD31 in sections from control and simvastatin-treated mice (n = 8 biological replicates). Data were…

• 

  Figure 6. Drug modulation along the SOX18-MVP axis in HemSCs results in upregulation of LDL receptor mRNA levels. (A) LDL receptor mRNA was measured in HemSCs following 24-hour treatment with 10 μM R(+) propranolol; 0.5 μM simvastatin; 0.1 μM atorvastatin; 20 nM rapamycin; 28 nM OX03050, a squalene synthase 1 inhibitor; or 0.1 μM of the farnesyltransferase inhibitor tipifarnib. PBS [for R(+) propranolol] and DMSO (for all other drugs) served as vehicle controls. P values were calculated using 1-way ANOVA with Šidák correction. Data are shown as the mean ± SD (n = 2 biological replicates; 1 of the replicates was used in 2 independent experiments, resulting in 3 data points). (B) Schematic illustrates points of inhibition of the different drugs along the endothelial SOX18-MVP axis.