Constitutive Activation of mTORC1 in Endothelial Cells Leads to the Development and Progression of Lymphangiosarcoma through VEGF Autocrine Signaling

Abstract

Angiosarcoma/lymphangiosarcoma is a rare malignancy with poor prognosis. We generated a mouse model with inducible endothelial-cell-specific deletion of Tsc1 to examine mTORC1 signaling in lymphangiosarcoma. Tsc1 loss increased retinal angiogenesis in neonates and led to endothelial proliferative lesions from vascular malformations to vascular tumors in adult mice. Sustained mTORC1 signaling was required for lymphangiosarcoma development and maintenance. Increased VEGF expression in tumor cells was seen, and blocking autocrine VEGF signaling abolished vascular tumor development and growth. We also found significant correlations between mTORC1 activation and VEGF, HIF1α, and c-Myc expression in human angiosarcoma samples. These studies demonstrated critical mechanisms of aberrant mTORC1 activation in lymphangiosarcoma and validate the mice as a valuable model for further study.

Figure 1. Increased mTORC1 signaling in human angiosarcomas and mTORC1 activation promotes neonatal retinal angiogenesis in Tsc1^iΔEC mice

(A) IHC of CD31 (marker for ECs) and pS6 for representative human quiescent blood vessels in normal tissue and human angiosarcoma samples. Arrows mark the normal ECs. Scale bar, 50 µm. (B) Quantitative analysis of pS6 staining within normal ECs and angiosarcoma samples. Fisher exact test. p <0.01. (C) Schematic for the inducible gene deletion in postnatal retinal ECs. (D) X-gal staining for p5 retinas of Tsc1^f/f;Rosa 26;Scl-Cre-ER^T mice. Scale bar, 250 µm. (E, F) Lysates from primary lung ECs of control and Tsc1^iΔEC mice at P5 were analyzed by immunoblotting using various antibodies as indicated (E) and the levels of TSC1, pS6, and pAKT are quantified (F). Mean ±SEM, n=3, *, p <0.05. (G, H) Representative images of whole-mount staining for isolectin B4 of retinas from control and Tsc1^iΔEC mice at P5 (G) and quantification results of branch points and vessel area per field (H). Scale bar, 100 µm. Mean ±SEM. (I, J) Representative images of whole-mount retinal vasculatures of control and Tsc1^iΔEC at P5 stained by isolectin B4 (green) and pH3 (red) (I) and quantification of mitotic cells (pH3 positive, marked by arrows) per field (J). Scale bar in the top panels, 100 µm. Scale bar in the bottom panels, 50 µm. Mean ±SEM. *, p <0.05. See also Figure S1.

Figure 2. Tsc1^iΔEC mice develop vascular tumors

(A) Representative images of tail and paw tumors of Tsc1^iΔEC mice. (B) MRI images of control and Tsc1^iΔEC mouse livers. Arrows mark the tumors. (C) Macroscopic appearance of representative livers of control and Tsc1^iΔEC mice. Arrows mark the tumor masses. (D, E) Percentage of control and Tsc1^iΔEC mice with liver and cutaneous tumors at 3–4 months (D) and 6–8 months (E). N.D., not detected. (F) Kaplan-Meier survival analysis of control and Tsc1^iΔEC mice. (G) H&E staining of liver and cutaneous sections of control and Tsc1^iΔEC mice. The boxed regions are enlarged to show the cell morphology. Scale bar, 50 µm. (H) Representative microCT images of vascular casting of control and Tsc1^iΔEC liver vasculatures. The boxed regions in the right panel are enlarged to show the multiple irregular vascular channels in the neoplasm area. (I) IHC for CD31 and PROX1 of liver and cutaneous sections of Tsc1^iΔEC mice. Scale bar, 50 µm. See also Figure S2.

Figure 3. Tumor progression and metastasis in Tsc1^iΔEC mice

(A) Representative images of edema of Tsc1^iΔEC paw and tail at 1 month after tamoxifen injection. (B) H&E staining of cutaneous sections of control and Tsc1^iΔEC mice 1 month after tamoxifen injection. Arrows mark the increased small vessels. Scale bar, 50 µm. (C) H&E staining and IHC for PROX1, CD31, and Ki67 of cutaneous vascular malformation and lymphangiosarcoma. Scale bar, 50 µm. Scale bar in the insets, 50 µm. (D) Mean ±SEM of percentage of Ki67⁺ ECs in liver and cutaneous sections. *, p <0.05. (E) A representative image of cutaneous tumor section shows the transition from malformation to lymphangiosarcoma. Boxed regions in the left panel are enlarged in the right. Note the ECs transit from normal flat shape (blue arrow) to atypical and neoplastic (red arrow). Scale bar, 50 µm. (F) H&E staining and IHC for Ki67 of liver vascular malformation and tumor. Yellow arrows mark proliferating ECs. Red arrows indicate blood cells that are also Ki67 positive. Scale bar, 50 µm. (G) H&E staining of a bone section of Tsc1^iΔEC mice showing infiltration by cutaneous lymphangiosarcoma. The boxed area is shown in more detail on the right. Scale bar, 250 µm. (H) IHC for CD31, VEGFR3 and VEGFR2 of cutaneous tumor sections of Tsc1^iΔEC mice. Yellow arrows mark a metastatic nodule and red arrows mark the lymphatic vessels. Scale bar, 50 µm. (I) An example of macrometastasis of cutaneous lymphangiosarcoma to the inguinal (a) and cervical (b) lymph nodes (red, marked by arrows) of Tsc1^iΔEC mice. H&E staining of lymph node metastasis (c) and an enlarged area (d, arrow marks tumor cells). Scale bar, 500 µm. (J) A representative macrometastasis on the sternal musculature of Tsc1^iΔEC mice (a) and H&E staining (b) and IHC for CD31 (c) and PROX1 (d) of this macrometastasis. Scale bar, 500 µm in (b) and 100 µm in (c, d). See also Figure S3.

Figure 4. Tsc1^iΔEC tumor cells show increased proliferation, migration and tubulogenesis in vitro and tumorigenicity in vivo

(A) Representative image and quantification of analysis for proliferation by BrdU incorporation assays of primary ECs isolated from cutaneous tumors from Tsc1^iΔEC mice or lungs of control mice. Scale bar, 100 µm. (B) Representative image and quantification of migration by wound closure assay of cells as in (A). Scale bar, 500 µm. (C) Representative image and quantification of tube formation on Matrigel of cells as in (A). Scale bar, 1mm. All quantifications shown as Mean ±SEM. *, p <0.05. (D–F) Isolated tumor ECs were subcutaneously injected in recipient nude mice. Tumor formation at injection site (D), H&E staining (E), and IHC for CD31, VEGFR3 and pS6 of tumor sections (F) are shown for recipients of Tsc1^iΔEC tumor cells. Scale bar, 50 µm. See also Figure S4.

Figure 5. Constitutive mTORC1 activation is required for vascular tumor initiation and maintenance

(A, B) Immunofluorescence for CD31 and pS6 in liver (A) and cutaneous (B) sections of control and Tsc1^iΔEC mice. Arrows mark individual ECs. Scale bar, 50 µm. (C) Schematic of rapamycin treatment (5 mg/kg IP every other day) to examine effects on vascular tumor initiation of Tsc1^iΔEC mice. (D) Percentage of Tsc1^iΔEC mice with tumors in rapamycin treated and untreated group. (E) H&E staining of liver and cutaneous sections of rapamycin treated and untreated Tsc1^iΔEC mice. Scale bar, 100 µm. (F) Schematic of rapamycin treatment (5 mg/kg IP every other day) to examine effects on established vascular tumors of Tsc1^iΔEC mice. (G) MRI images of the livers of Tsc1^iΔEC mice at various times after treatment with or without rapamycin. Arrows mark liver tumors. (H) Quantification (Mean ±SEM) of fold changes in tail tumor size at 3.5 months after rapamycin or vehicle treatment relative to before treatment (0 month). n=6, *, p <0.05. (I) Kaplan-Meier survival analysis of Tsc1^iΔEC mice treated with or without rapamycin. (J–L) IHC for Ki67 and pS6 (J), double immunofluorescence for CD31 and cleaved caspase 3 (K), and H&E staining (L) of vascular tumor sections from Tsc1^iΔEC mice treated with or without rapamycin. Scale bar, 20 µm (J, L) or 50 µm (K). Arrows in (K) mark apoptotic ECs. See also Figure S5.

Figure 6. Increased VEGF transcription through HIF1α and c-Myc in Tsc1^iΔEC tumor cells

(A) Double IHC for VEGFA and CD31 of the normal blood vessels in control mice and cutaneous tumor sections of Tsc1^iΔEC mice. Nuclei were counterstained with methyl green. Scale bar, 20 µm. (B) Lysates from primary lung ECs of control mice and cutaneous tumor ECs of Tsc1^iΔEC mice were analyzed by immunoblotting using various antibodies as indicated and the levels of TSC1, VEGFA, pS6 and pAKT were quantified. Mean ±SEM, n=3 *, p <0.05. (C) Mean ±SEM of mRNA (normalized to control ECs as 1) of VEGFA in cutaneous tumor cells of Tsc1^iΔEC mice. (D) Isolated control and tumor ECs were treated with rapamycin for indicated times. Lysates were analyzed by immunoblotting using various antibodies as indicated. (E) IHC for VEGFA of liver tumors from Tsc1^iΔEC mice treated with or without rapamycin. Scale bar, 20 µm. (F) IHC for CD31, HIF1α and c-Myc of cutaneous sections of control and Tsc1^iΔEC mice. Arrows mark normal ECs (left panels) and asterisks indicate elevated HIF1α and c-Myc in the nuclei of cutaneous tumor cells (right panels). Scale bar, 50 µm. (G) mRNA (normalized to control ECs as 1) of HIF1α and c-Myc in control and tumor cells. (H) HIF1α mRNA in tumor ECs treated with rapamycin (100 ng/ml) for the indicated times. (I) VEGFA mRNA in tumor ECs transfected with HIF1α and c-Myc siRNA for 48 hr. Mean ±SEM, n=3. *, p <0.05. N.S., Not Significant. See also Figure S6.

Figure 7. Autocrine VEGF signaling is required for increased proliferation of Tsc1^iΔEC tumor cells

(A) Lysates from primary lungs of control mice and cutaneous tumors of Tsc1^iΔEC mice were immunoprecipitated with anti-phosphotyrosine antibody 4G10 and analyzed by immunoblotting using various antibodies as indicated. (B) IHC for pMEK1/2 S221 of liver and cutaneous sections of control and Tsc1^iΔEC mice. Scale bar, 25 µm. (C) IHC for pERK1/2 T202/Y204 of liver and cutaneous sections of control and Tsc1^iΔEC mice. Scale bar, 25 µm. (D) Isolated cutaneous tumor cells of Tsc1^iΔEC mice were analyzed for proliferation, migration, and tube formation in the presence or absence of axitinib (50 nM) in vitro. Mean ±SEM from three independent experiments. *, p <0.05. (E) Schematic of axitinib treatment (25 mg/kg daily PO) to examine effects on established vascular tumors of Tsc1^iΔEC mice. (F, G) IHC (F) and quantification (G) for pERK, Ki67, and CD31 of cutaneous tumor sections of Tsc1^iΔEC mice that had been treated with or without axitinib. Scale bar, 50 µm. Mean ±SEM, *, p <0.05. (H, I) Double immunofluorescence for CD31 and cleaved caspase 3 of liver tumor sections of Tsc1^iΔEC mice treated with or without axitinib (H) and quantification for percentage of apoptotic cells (I). In (H), the boxed regions in the top panels are enlarged in the bottom panels. Scale bar in the top panels, 50 µm. Scale bar in the bottom panels, 10 µm. Mean ±SEM, *, p <0.05. (J) Tumor growth curves and tumor volumes of mice treated with placebo, axitinib (25 mg/kg daily), rapamycin (5 mg/kg every other day) and both axitinib and rapamycin. Tumor volumes were measured on day 35. Mean ±SEM, n=8. (K) Tumor growth curves and tumor volumes of tumor ECs with VEGFA, VEGFC or VEGFD knockdown. Tumor volumes were measured on day 38. Mean ±SEM, n=8. See also Figure S7.

Figure 8. Clinical relevance of mTORC1/VEGF signaling axis in human angiosarcomas

(A) Representative images of IHC for VEGFA, HIF1α and c-Myc of samples with high and low pS6. Scale bar, 20 µm. (B, C) Correlation analyses between pS6 and VEGFA, HIF1α, or c-Myc expression (B) and between VEGFA and HIF1α or c-Myc expression (C). (D) IHC for pS6 and Ki67. Scale bar, 20 µm. (E) Relationship between pS6 and Ki67 expression, High, >=15% Ki67 positive cells; Low < 15% Ki67 positive cells. The correlative relationship between two quantitative measurements was investigated using Fisher's exact test. A p value <0.05 was considered to be statistically significant.